Introduction

On 30 October South Korea’s KOSPI fell 3.48% in a single session. The night before, Alphabet lost more than 12% on Wall Street after its third-quarter revenue disappointed and revived doubts about whether the enormous investment in artificial intelligence will pay off. SK hynix, one of the largest makers of memory for AI servers, lost 7.08% on the day. Over the whole of October the KOSPI is down 16.19% and the Nasdaq 100 6.65%. A month earlier the Nasdaq 100 was the strongest index in our table. Now September’s winners are among the biggest losers, while the Russell 2000 and the broad European STOXX Europe 600 finish the month higher.

In the same three days the three large central banks met: the Fed on 28 October, the ECB on 29 October and the Bank of Japan on 30 October. None raised rates. Yet at all three a hike was on the table. At the Fed two members voted for one. At the ECB the Governing Council debated whether one was appropriate. At the Bank of Japan one member proposed 1.50%. This is not a pause in the usual sense. It is a held breath: the decision has been postponed, not cancelled.

Beneath the pause, quantities are almost standing still. Net Fed Liquidity is $17.6 billion smaller, euro area excess liquidity €33.3 billion smaller, the Bank of Japan’s balance sheet is practically motionless, and China’s central bank added liquidity in September. But even with a held breath the body keeps burning oxygen. Since 23 October SOFR, the rate on secured overnight money, has been above the rate the Fed pays on bank reserves for five sessions in a row. The question this month is what happens inside the system while the central banks wait.

Central Banks

Three of the four banks held rates, but holding does not mean doing nothing with the quantity of money. The Fed is neutral on both price and quantity. The Eurosystem keeps withdrawing liquidity, and the market tightens on its behalf through long-term rates. The Bank of Japan pauses its quantitative shrinkage for the first 20 days of October. China’s central bank is the only one that added liquidity in the latest measured month.

Federal Reserve

The Fed balance sheet stands at $6.746 trillion as of 28 October, against $6.743 trillion on 30 September. The difference is $2.6 billion, so the balance sheet is standing still. Maturing Treasuries are rolled over, and principal from mortgage-backed securities is reinvested in Treasury bills. The instruction to the New York Fed also keeps one permission in place: to buy Treasury bills when needed to maintain ample reserves. A permission is not a purchase, and no such purchases show up in the October balance sheet.

On 28 October the Fed kept the target range at 3.75-4.00% by a vote of 10 to 2. Lisa Cook and Beth Hammack dissented because they wanted a 25 basis point hike. A month earlier the hike was unanimous, 12 to 0. Now the decision is a hold and the dissent points towards tightening. Interest on reserve balances, the rate the Fed pays banks on their reserves, stays at 3.90%. Net Fed Liquidity falls by $17.6 billion over the month, as the Net Fed Liquidity section shows in detail.

Regime: neutral. Price and quantity are standing still, but two members already want more expensive money.

European Central Bank

The Eurosystem balance sheet grows from €5.897 trillion on 25 September to €5.935 trillion on 23 October, an increase of €37.7 billion. This is not an injection. The quarterly revaluation of gold and foreign currency at the end of September, booked in the statement as of 2 October, adds €56.5 billion to assets. Without it the balance sheet is €18.8 billion smaller. This is exactly why the headline size of the balance sheet should not be read in a revaluation month. A larger balance sheet after a revaluation is not more money for banks.

Excess liquidity, the measure revaluations do not touch, confirms the direction. It falls from €2.109 trillion on 30 September to €2.076 trillion on 29 October, a decline of €33.3 billion, after a fall of €33.0 billion the month before. The APP and PEPP portfolios keep shrinking because maturing principal is not reinvested. On 29 October the ECB kept the deposit rate at 2.50%, the main refinancing rate at 2.65% and the marginal lending rate at 2.90%. Christine Lagarde said the Governing Council had debated whether a hike was appropriate but decided unanimously to wait. Part of the work has already been done by the market: the average 10-year government bond yield in the euro area is about 35 basis points higher than when the September projections were made.

Regime: tightening. Rates are on hold, the quantity keeps draining, and long-term yields are tightening on the ECB’s behalf.

Bank of Japan

In the previous issue the Bank of Japan’s balance sheet was ¥641.281 trillion as of 20 September. Ten days later, on 30 September, it was ¥625.025 trillion. Almost the entire fall of ¥16.3 trillion comes from the government bonds on the balance sheet, which drop from ¥520.303 trillion to ¥504.008 trillion. This is the continuation of the plan for smaller purchases, under which maturing bonds are not fully replaced. In the first 20 days of October the balance sheet is practically motionless.

Current deposits, which are the banks’ reserves, rise by ¥1.8 trillion after falling sharply in September. Part of this comes from the government spending out of its account at the central bank. At the same time the Bank of Japan absorbs ¥3.2 trillion through repo transactions, in which it hands out bonds against cash. The monetary base for September is 15.2% lower than a year earlier. The October figure is published on 4 November and will enter the December issue.

On 30 October the Bank of Japan kept its policy rate at around 1.25% by a vote of 8 to 1, with Hajime Takata proposing 1.50%. The board decided unanimously that from January 2027 sales of exchange-traded funds from its portfolio will accelerate from about ¥330 billion to about ¥860 billion a year at book value. This is a future pace, not an October sale.

Regime: tightening, with a pause in quantities. The monetary base is shrinking 15.2% year on year, but in the first 20 days of October the balance sheet does not move.

People’s Bank of China

The latest published balance sheet of the People’s Bank of China is for September, and it reverses August’s direction. Assets grow by CNY 643.6 billion to CNY 50.500 trillion. Reserve money grows by CNY 1.118 trillion, more than the balance sheet itself, because government deposits at the central bank fall by CNY 380.8 billion. When the state spends from its account, the money goes to the banks. September’s operations confirm the direction, with a net injection of CNY 556.4 billion.

The money comes through the medium-term instruments, not the targeted ones. The targeted programmes are net negative, and the new quotas of CNY 700 billion announced on 29 September do not yet show up in September’s operations. The one-year Loan Prime Rate, the benchmark for bank lending, stays at 3.00% and the five-year at 3.50%. The latest official figure for the average reserve requirement ratio, the share of deposits banks must hold at the central bank, is 6.3%, given by Deputy Governor Zou Lan on 15 January 2026. The PBoC’s monthly operations tables from January to September 2026 show no change.

In October the PBoC also announced new safety nets. On 19 October Governor Pan Gongsheng said the bank would consider moving its main operations from seven days to overnight. On 26 October it created BNSF, a facility that lets large fund managers obtain financing during mass redemptions from bond funds. No size or rate has been announced. A safety net is not liquidity that has been provided.

Regime: easing in September. The balance sheet, reserve money and net operations all grow. October has not yet been measured.

Central Bank Synthesis

Money and Credit Transmission

In the previous issue bank credit was the detour along which the system kept moving without help from the central banks. But the data then were for August, before the three hikes. The September data are the first real test. The detour held. In the US the traffic on it is even accelerating. In the euro area volumes are still growing, but banks are already closing some lanes.

United States

US M2 stands at $23.414 trillion in September, up 0.49% on the month and 5.71% on the year. The August level has been revised from $23.343 trillion to $23.301 trillion, so the monthly change should not be calculated from the figure published last month. Commercial and industrial loans, bank lending to businesses, grow 0.84% in September to $2.959 trillion, and the annual pace accelerates to 9.96%.

August’s increase has been revised from 1.62% to 1.60%. The weekly data run to 21 October. In the three weeks after 30 September bank loans and leases grow by $76.3 billion and deposits by $76.7 billion. In the previous issue loans added $51.1 billion over four weeks, so the weekly pace is almost twice as high. And unlike August and September, banks are this time also buying securities, by $44.3 billion, so total bank credit grows by $120.6 billion. One technical note: in the week to 30 September banks sold $5.2 billion of auto loans to non-bank lenders. That is a transfer of credit, not credit that has disappeared. The offset has not only survived the hike. It is accelerating.

Euro Area

M3 stands at €17.717 trillion in September, up 3.5% on the year and unchanged from August. Adjusted loans to non-financial corporations grow 4.2% and loans to households 3.1%, also unchanged.

Here we owe a correction. Last month we wrote that the corporate bond channel was closing, because firms’ net debt issuance in August was -€0.2 billion. The revised data do not confirm it.

August has been revised to plus €0.3 billion, and September is the strongest of the five months. The bond channel narrowed over the summer, but it did not close.

The volumes are the good news. The surveys are the bad news. They show what banks and firms think about the coming months, and credit standards, the conditions banks set for lending, usually tighten before credit slows.

These are net shares from surveys, not percentage falls in the volume of credit. Banks tightened conditions for firms more than they themselves expected three months ago. Demand for housing loans is falling sharply and is expected to fall further. According to the bank lending survey, banks’ access to funding from household deposits is deteriorating at the sharpest pace since the series began in 2011. Volumes are still growing, but banks are already tightening the terms.

China

China’s M2 stands at CNY 359.63 trillion in September, up 0.79% on the month. The annual pace, however, keeps slowing, from 7.5% in August to 7.2%, and the data are preliminary. Total social financing, the broadest measure of credit to the real economy, adds CNY 3.548 trillion in September, almost the same as a year earlier (CNY 3.530 trillion).

The share of government bonds falls from 60.9% to 43.3%, and this time bank loans carry more than a third of the flow. Firms, however, are paying down bonds on a net basis. Yuan bank loans grow 4.8% year on year, against 4.9% in August. From the cumulative data, enterprises and organisations add about CNY 1.29 trillion in September, while households add practically nothing.

The PBoC’s quarterly report on where credit goes shows two different worlds beneath these figures. Corporate loans grow 8.0% year on year, and loans to small and medium-sized technology firms 20.1%. Household loans fall 2.0%, and property loans 5.2%. Fixed asset investment excluding rural households is 7.2% lower over the first nine months, and private investment 10.4% lower. The enterprises in these data are not only private firms, so we cannot say how much of the corporate credit reaches the private sector. Credit in China goes where the state directs it, not where households seek it. The central bank can add liquidity, but it cannot make households borrow. That is pushing on a string.

Japan

Japan’s M2 stands at ¥1,295.6 trillion in September, up 2.0% on the year, and M3 at ¥1,638.3 trillion, up 1.2%. Seasonally adjusted M2 rises 0.20% on the month. Broad money remains practically motionless. Bank lending by major, regional and shinkin banks reaches ¥682.0 trillion, up 5.3% on the year, with August’s pace revised from 5.4% to 5.3%. At major and regional banks alone the growth is 5.7%. The Bank of Japan’s survey of 50 large banks shows slightly stronger demand from firms, with the index at 3 against -1 in the previous survey. Standards barely change, but banks are widening their lending margins. The central bank is draining the water, and commercial banks are still putting it back: the monetary base is shrinking 15.2% year on year while bank credit grows 5.3%.

Money and Credit Synthesis

Net Fed Liquidity and Its Relationship to M2

Net Fed Liquidity is the Fed balance sheet minus the Treasury’s account at the Fed and minus overnight reverse repo. It shows how much of dollar liquidity is not locked in the government’s account or parked back at the central bank.

From this issue all three components are taken on the same Wednesday. Until now the Treasury account entered as a weekly average, while the balance sheet and reverse repo were single-day values, so we were mixing an average with a point. The method is already used in Credit Pulse from October. The change affects the comparison with the previous issue. We then published Net Fed Liquidity of $5.794 trillion for 30 September. Under the new method the same date is $5.747 trillion. The difference of $46.6 billion comes from the method, not from the economy.

Net Fed Liquidity

As of 28 October Net Fed Liquidity is $5.730 trillion, against $5.747 trillion on 30 September. The decline is $17.6 billion, or 0.31%.

The Fed balance sheet barely plays a part. The whole movement is between the Treasury account and reverse repo. Reverse repo on 30 September is an extreme case: at quarter end banks shrink their balance sheets for the reporting date, and money market funds park cash at the Fed for a day. Afterwards it is practically zero again.

The month has two halves. Until 14 October the Treasury spends from its account and Net Fed Liquidity rises to $5.870 trillion, the highest since 22 July. The account then refills by $129.8 billion in two weeks and reaches $1,015.6 billion. That is the highest Wednesday level since the start of 2025, the beginning of our extracted data. Reverse repo has nothing left to give. Every refill of the Treasury account runs through bank reserves. When the buffer is empty, every dollar the government collects leaves the banks.

Net Fed Liquidity is not the accounting level of reserves. Other items on the Fed’s balance sheet move them too, and in October Wednesday reserves even rise slightly, from $2,881.7 billion to $2,895.2 billion. Two Fed notes from October explain why the total is not the whole story. The first, from 16 October, shows that total reserves and a single average rate can hide an uneven distribution across individual banks. The second, from 30 October, estimates that for every dollar of new longer-term Treasuries the balance sheets of large dealers grow by about 21 cents, of which 13 cents is repo financing. More issuance means more demand for overnight money. That explains why the price of overnight money is reacting even though the quantity barely changes, as the Credit Stress section shows.

M2 vs Net Fed Liquidity

The comparison covers the same one-year window, September 2025 to September 2026, using the last Wednesday of each month for Net Fed Liquidity.

The annual decomposition of Net Fed Liquidity: the Fed balance sheet adds $134.6 billion, the drain of reverse repo adds another $17.6 billion, and the higher Treasury account removes $226.1 billion. The three parts sum exactly to -$73.8 billion. Over one year the Treasury has drawn about 1.7 times as much out of the system as the Fed balance sheet has added. Last month the annual decline was -$201.7 billion. The two figures are not comparable, because the window then ended in August and the method was the old one, with the weekly average of the Treasury account. The direction, however, is the same. This is not a measure of privately created money. M2 is a monthly average, Net Fed Liquidity a weekly snapshot. The divergence says something narrower: broad money is growing, while the federal liquidity plumbing is smaller than a year ago.

Read more

How far the leak travelled, which European country led it instead of the usual suspect, and which region is less calm than it looks: the full analysis is for subscribers.

Read more

This is the monthly Macro Pulse for Japan for August 2026. All the main August releases are now out, the last being the final wage data on 23 October. For the first time in a Japan issue we also compare the main figures with market expectations, to show what surprised and in which direction. Data released after August, including the national CPI for September and the flash PMI for October, serve as a check on the direction rather than as August observations.

Where the economy stands

The phase in August remains growth with slowing momentum. Surveys of managers describe the strongest manufacturing upswing since 2014, while the measured data show a plateau. The composite PMI, the combined survey of managers in manufacturing and services in which a reading above 50 signals improvement, rose to 53.5, and new orders at factories grew at the fastest pace since January 2018. In the same month industrial production fell 2.4%, against a market expectation of a 1.7% rise. That is the biggest surprise of the month. In the other direction, prices further up the supply chain and wages came in above forecasts.

The engine is still global demand for chips and equipment for artificial intelligence. Machinery orders are up 18.6% from a year earlier and companies are raising their investment plans, but exports are growing more and more in value and less and less in volume. The brake comes from households. Their real spending is 3.1% below last year, and the broader consumption index compiled by the Bank of Japan (BoJ) fell on the month. Prices send the BoJ the opposite signal. Without the government’s energy measures, headline inflation would be around 2.5%, and wages are growing by close to 4%. Our reading is therefore on hold with a tightening bias. Weak production argues for a pause at the meeting on 29 and 30 October, while prices and wages argue for one more hike before the end of the year.

Core indicators

The trajectory

One month cannot decide the phase, so the six-month window compares August with February and May and shows whether it is a continuation or a turn. The most important thing in it is a contradiction. The composite PMI is turning up: it is higher than in May, though still below February. The manufacturing PMI is above both. Industrial production, the measured volume, is below both. The survey and measured output diverge in exactly the month when the survey is at its most optimistic, which is why we devote the section on the underestimated detail to this gap.

Consumption comes next. The BoJ’s consumption index fell 0.8% on the month and sits below its level in both February and May. Its average for July and August is 1.0% below the second-quarter average. That is not the whole third quarter, but it is two thirds of it. The household spending survey points the same way, with a real fall of 3.1% from a year earlier, and nominal retail sales are turning down after their May peak. So the two indicators that are turning move in opposite directions: the survey of the whole private sector is turning up, while sales in shops are turning down. If we had to pick the more reliable turn, we would pick the measured one, because sales are yen actually recorded, while the survey is an opinion about direction. Both, however, are a single month, so in the section on what to watch we set concrete September thresholds that will settle the question.

On wages we need to correct our July issue. We wrote that total cash earnings, meaning wages together with overtime and bonuses, grew 4.7% in July and real wages 2.4%. The final data show 4.3% and 2.0%. For August the final figures are 3.9% and 1.6%, 0.1 point above the preliminary ones. Under our direction rule, nominal earnings are still accelerating against February and May, while real wages are at their May level and below February. Base pay for full-time employees is rising 4.0%. The income base remains solid, but it is not broadening, and inflation is eating a growing share of it. Prices themselves, however, are accelerating more quietly than it seems. Core CPI, the overall index excluding fresh food, is above both comparison points, by 0.2 and 0.3 points. That is a small move, and the government’s energy measures are holding it down. The measure excluding fresh food and energy, known as core-core, stayed at 1.9% for a second month.

The engine: chips grow, the rest of industry lags

Japan’s engine in August has the same name as in July: demand linked to artificial intelligence. What has changed is how much of it reaches volume. Exports rose 19.3% from a year earlier in yen, but according to the Ministry of Finance the volume of goods exported rose only 2.4%, after 5.2% in July. Export prices in yen are 16.5% above last year, so about 85% of export growth is price, not quantity. That is the effect of the weak yen: every good sold in dollars brings in more yen without the factory shipping more containers. September continued in the same vein, with exports up 16.7% in yen and 1.9% in volume. The BoJ’s measure of real exports, however, is 1.0% above the second quarter, so net exports may give a small lift to third-quarter GDP.

At the core of the engine, volume is still moving. Exports of chipmaking equipment rose 40.1% in value and 43.1% in quantity, so quantity is growing even faster than value. Exports of semiconductors and electronic components are 52.3% above last year in value. In industrial production, however, only two industries grew. The first is production machinery, the machines other factories use to make things, chipmaking machines among them: up 5.9% on the month and 24.7% on the year. The second is transport equipment excluding cars, up 2.5%. Everything else fell, led by cars, down 6.8% on the month. The engine is therefore smaller than it sounds. Chips and the equipment to make them are a narrow segment of industry, and industry is a smaller part of Japan’s economy than consumption. The segment can grow 40% and still fail to lift total output when cars and general machinery are falling. That is exactly what happened in August.

Orders confirm that the core is alive. Core machinery orders rose 5.2% on the month, against 3.0% expected by the market, and are 18.6% above August 2025. The answer to the question we left open in July is also positive. In the September Tankan, the BoJ’s quarterly survey of companies, investment plans for fixed assets in the fiscal year that began in April point to growth of 7.6%. Companies did not cut them. They raised them by 0.8% from the June survey. These are intentions, not money spent, and the check will come with business investment in the first GDP estimate on 16 November. There is, however, a new warning. In the September survey the manufacturing PMI slipped to 54.1, its lowest in six months. Some firms explain slower orders by customers running down inventories after earlier overstocking. That admission matters, because if part of the summer upswing in the surveys was stockbuilding, it borrowed from the months ahead rather than adding new growth.

What could stop the engine? First, a reversal in global investment demand for AI infrastructure. Japanese equipment suppliers sit at the start of the chain and are the first to feel any delayed project. Second, a stronger yen. If the exchange rate snaps back, the price part of export growth will shrink and the modest 2.4% volume growth will come to the fore. Third, energy. The surveys keep naming supply disruptions around the Strait of Hormuz as a reason for high costs.

The brake: households are not coming back

The brake is where it was: in household spending. Real spending by households of two or more people is 3.1% below last year. That is less bad than the 3.6% fall the market expected, but the decline has exceeded 3% for a third month in a row, after 3.3% in June and 3.6% in July, whereas in April and May it was below 1%. The monthly improvement is only 0.1%, against 0.5% expected. The broader measure matters more. The BoJ’s consumption index gathers data on sales, services and spending from many sources, which brings it closer to consumption in GDP than the household survey. Excluding purchases by foreign tourists, it fell 0.8% in August, and its average for July and August is 1.0% below the second quarter. Unless September brings a strong rebound, consumption will subtract from third-quarter growth. Sentiment is improving, but slowly. The Cabinet Office consumer confidence index stands at 35.5, against 34.9 in July and far below February’s 39.7. The Economy Watchers survey of people in direct contact with customers remains below 50, meaning more of them see deterioration than improvement.

Here we need to correct our own explanation from July. We wrote then that strong wages and weak spending describe different people: those in work and pensioners. That remains a plausible hypothesis, but it is not proven. The wage survey covers employees at firms with five or more staff, while the spending survey covers households of two or more people. The two have different coverage, and without a breakdown of spending by age and household type we cannot say how much of the gap comes from ageing. The simpler explanation is also closer to the data: real wages are growing 1.6%, and that is not enough to offset the loss from dearer food and energy since the start of the war in the Middle East.

The labour market shows where caution sits. Unemployment rose from 2.4% to 2.5%, above the 2.4% expected, and the ratio of job openings to applicants held at 1.18. In services, hiring is the slowest in a year according to the firms themselves in the PMI survey, and some are not replacing staff who leave. Factories are hiring at the fastest pace since 2018, while the services that live on household spending are saving on people.

Is the brake cyclical or structural? Mostly cyclical. Energy, the weak yen and dearer food can recede. The government’s electricity and gas measures are already reducing bills, and the yen was stronger in September than in July. If real wages stay positive, households have room to catch up. Beneath the cyclical layer, however, lies a structural base: an ageing population and a large share of households on fixed incomes. It makes Japanese consumption slow to respond to good news, but on its own it does not explain the decline of the past two months.

Prices: the measures hide the pressure

On the surface, consumer prices are pausing. Headline CPI held at 1.9%. Japan’s standard core CPI, excluding fresh food, eased from 1.8% to 1.7%, below the 1.8% expected, and the measure excluding fresh food and energy stayed at 1.9%. Beneath the surface the picture is different. Energy swung from a 0.6% rise in July to a 0.7% fall in August. The Statistics Bureau of Japan attributes the turn to the start of the government’s programme to ease electricity and gas bills, and it estimates the combined effect of the measures: they cut the annual rate of the headline index by 0.62 point. The largest part, 0.38 point, comes from the removal of the temporary surcharge on petrol tax, and electricity contributes another 0.17. Without the measures, headline inflation in August would have been around 2.5%, not 1.9%. This matters because the measures are temporary and the BoJ looks through them.

Goods prices rose 2.6% and services 1.2%. Services are the part that depends most on wages, and pressure there is still moderate. Surveys of firms, however, say this is changing. Service companies are raising their selling prices at the second-fastest pace in the history of the PMI survey, and for the private sector as a whole selling-price inflation is the fastest since the combined series began almost twenty years ago. Households expect the same. In the BoJ survey conducted in August and early September, 88.6% expect prices to keep rising, although the share expecting a “significant” rise fell from 40.5% to 34.5%.

Further up the chain, pressure remains strong. The CGPI, the BoJ’s index of prices at which firms sell goods to one another, rose 7.6% from a year earlier, above the 7.4% expected. On the month the index fell 0.3%, mainly because of cheaper electricity, gas and petroleum products. Import prices in yen are 25.1% above last year, and in September the CGPI slowed to 6.8%. The gap between the CGPI and core CPI is 5.9 points. It does not measure corporate margins and does not forecast future CPI, because the two indices cover different things with different weights. It shows only that costs along the chain are rising much faster than prices in the shops. Whether firms will pass them on to customers is what the surveys tell us, and they say they will.

The national CPI for September, published on 23 October, shows that the acceleration is not just an effect of administrative charges in Tokyo. Core CPI accelerated from 1.7% to 2.1%, as the market expected, and core-core from 1.9% to 2.4%. The energy measures now cut the headline rate by 0.82 point, so without them it would be around 2.9%. Energy plays almost no part in the acceleration. The lead comes from accommodation and other leisure services, water, processed food and household durables. Water charges are again administrative, but the rest are not. This is September and does not change the phase for August, but it strengthens the tightening bias.

The underestimated detail: managers versus factories

The PMI does not measure how much is produced. It asks several hundred managers whether their output is higher, lower or the same as in the previous month, and a reading above 50 means more of them answer “higher”. That makes the index fast and broad, but it does not measure quantity. Physical volume is measured by the industrial production index of METI, Japan’s Ministry of Economy, Trade and Industry. The manufacturing PMI also has a technical quirk. Longer supplier delivery times push it up, because the survey reads them as a sign of strain. Deliveries around the Strait of Hormuz were slow over the summer, but S&P Global reports that delays eased slightly over the past two months, so the August rise in the index came from orders, not from late lorries.

In August the two measures say opposite things. S&P Global reports the fastest growth in new orders since January 2018 and the strongest quarterly production trend since early 2014. METI reports a 2.4% monthly fall in output, with factory shipments down 2.9%. Average output in July and August is practically equal to the second-quarter average. The gap is not only between the PMI and production. Firms in METI’s own survey expected 6.4% growth in August. METI, which corrects for the systematic optimism of company plans, expected 3.1%, and the market 1.7%. The result was -2.4%. Everyone who asks firms saw growth, and measured output fell.

Why is this underestimated? Because the PMI makes the headlines. It comes out first, has a long history and sounds decisive. And the September survey already offers one explanation: customers who had been stockpiling are starting to run inventories down. If that is right, part of the summer optimism was orders for someone else’s warehouse. The flash PMI for October, published on 23 October, adds a warning. The manufacturing index fell from 54.1 to 53.3, below the 54.5 expected, and its output component grew at the slowest pace since March. Two components, however, are holding the index up for technical reasons: delivery times are lengthening markedly, and firms are building stocks of inputs at the fastest pace in almost four years. Both lift the manufacturing PMI without meaning more demand.

The strongest counter-argument is METI’s forecast. Firms expect growth of 3.2% in September and 3.1% in October, while METI’s adjusted estimate for September is a more modest 1.6%. A rough calculation shows how much rides on September. It assumes the whole index moves like the forecast, although the forecast covers only part of manufacturing. If September comes in at 1.6%, the third quarter will be practically level with the second. If it comes in at 3.2%, it will be about 0.6% above it, and production will catch up with the survey. The preliminary September data are due on 30 October, and the first GDP estimate for the third quarter on 16 November. That will show who was right.

What this means

The August data give the BoJ arguments in both directions. Growth is weaker than the surveys say: output is falling, consumption is shrinking and unemployment is rising. Prices are stronger than the headline index shows: without the temporary measures inflation is around 2.5%, core-core is 1.9%, and wages are growing by close to 4%. The last confirmed quarterly position, for the second quarter, was the most uncomfortable one: growth weakening while inflation pressure builds. August does not change that. What is confirmed is that the core of the engine, chips and the equipment to make them, is growing in volume too, that investment plans were raised, and that households are not spending more. What remains an assumption is whether the summer upswing in the surveys was real growth or stockbuilding, and the answer depends on September production.

Our reading is therefore on hold with a tightening bias. On 18 September the BoJ raised its rate from 1.0% to 1.25% by a vote of 7 to 2. The next meeting is on 29 and 30 October, with a new Outlook Report, the bank’s quarterly forecast report. On 15 October Policy Board member Koeda said underlying inflation is anchoring around 2% and that the bank should not be overly cautious about raising rates. That is the view of one member, not a decision, but it shows where part of the Board is looking. Weak August production is the best argument for a pause in October. If September refutes it, the argument falls away. The real interest rate, the policy rate minus inflation, remains negative: at 1.25% and core CPI of 1.7%, it is around -0.45 point. Monetary conditions still support the economy, as Koeda himself says.

What does this mean for liquidity? Beyond the policy rate, tightening is coming through two more channels. The first is the BoJ’s balance sheet. The monetary base, the money the BoJ creates directly, was 15.7% below last year in August and 15.2% below in September. In the “Outflow” issue of the Global Liquidity Snapshot we described the BoJ’s regime as tightening. Banks, however, are compensating: their lending is growing 5.3% from a year earlier. That is slightly below the 5.4% we cited in “Outflow”, because the August data were revised. The BoJ’s survey of bank loan officers, published on 22 October, shows the same from the banks’ side. Corporate demand for credit is back in positive territory, with an index of 3 against -1 three months earlier, and lending standards are barely changing. Banks are, however, widening the spreads on corporate loans, so credit remains available but is getting dearer. There are also two quieter signals. M1, cash and overnight deposits, was already 0.2% below last year in September. And from 27 November the BoJ will value the longest bonds it accepts as collateral one point lower, which makes liquidity against them slightly more expensive.

The second channel is the bond market. The two-year yield on Japanese government bonds (JGBs), which tracks expectations for the BoJ’s rate, was 1.74% at the end of August and 1.95% at the end of September, and on 22 October it was 1.91%. The ten-year yield was 3.06% at the end of September and 2.91% on 22 October. In the September Credit Pulse, the Japanese two-year yield crossed our warning threshold. The market already prices another hike, and since September some of that bet has been unwound.

The yen remains the main unknown. The average rate fell from 162.45 yen per dollar in July to 158.81 in August and 156.23 in September, and on 23 October it stood at 156.96. The Ministry of Finance reports intervention of 15.4 trillion yen between 30 July and 26 August, which we covered in the July issue. In the following reporting period, to 28 September, there was none. Around the intervention, the yen strengthened by close to 4 yen per dollar. The monthly report, however, does not say in which direction the trades went, so we do not know how much of the appreciation came from the intervention and how much from expectations for the BoJ’s rate. In October the rate has moved between 156 and 158 yen per dollar without a clear direction. The intervention coincided with a stronger yen, but whether it caused it will be shown by the quarterly report with the dates and direction of the trades, expected in early November.

For the ordinary person, the picture is familiar from July. Someone working at a large company is seeing the biggest pay rises in years, and jobs are easy to find. Electricity and petrol bills are temporarily cheaper thanks to the government, but food is not. Someone with a floating-rate mortgage is paying more since the September hike. According to the BoJ, rising wages are absorbing the higher payment for now, but more and more people are stretching mortgage terms beyond 35 years to keep the payment bearable.

Risks in both directions

The strongest argument against our phase is that we give too much weight to one bad month. The manufacturing PMI has signalled improvement for an eighth month in a row. Machinery orders are 18.6% above last year, investment plans were raised, and firms in METI’s survey expect growth of more than 3% in September and October. If that comes true, the third quarter will be stronger than the second, and the August fall in production will look like a pause rather than a plateau. The BoJ’s regional report of 8 October is also more optimistic than we are: all nine regions report a moderate recovery, two raised their assessment and none lowered it, and according to the bank’s branches consumption is holding up despite higher prices.

The downside risk has three sources. The first is inventories. If the destocking that firms describe in the September survey continues, orders will slow just when consumption cannot take up the slack. The second is households. Two more months of real declines in spending would turn consumption into a negative contribution to third-quarter GDP. The third is energy. Government measures hold down bills only temporarily, and if the Middle East pushes oil prices up again, the measures will cost more and purchasing power will fall again. The BoJ’s Financial System Report of 23 October adds a second channel to the AI risk. Japanese banks are lending more and more to foreign funds that finance AI data centres. The share of these loans is still small, but it is growing. If expectations for AI profits fall, the blow will come not only through equipment orders but also through the banks.

There is also an upside risk that is easy to miss. If the yen keeps strengthening, as it did in September, imports will become cheaper, real wages will grow faster and households will get relief without the BoJ doing anything. The cost is slower export growth in value. For the BoJ itself, the risk remains asymmetric. If it waits too long, inflation excluding energy could anchor above target, supported by wages. If it moves too early, just as households are cutting back, it could stall a recovery that has not yet begun.

What to watch

The national CPI for September has already shown core-core at 2.4%, and the final wage data confirmed real growth of 1.6%. On 30 October come preliminary production for September and Tokyo CPI for October. If September production is above 1.6%, METI’s adjusted estimate, output is starting to catch up with the survey. If it is below zero, the plateau becomes a decline. On 29 and 30 October the BoJ decides on rates and publishes its new Outlook Report. On 10 November household spending for September will show whether the decline is continuing.

The thresholds for a change of phase remain the same as in the July issue. Towards acceleration: a positive contribution to GDP from both private consumption and business investment in the third quarter, with the composite PMI above 52. The first GDP estimate is due on 16 November. Towards stagnation: the composite PMI below 50 for two consecutive months, together with a new fall in production. Here we correct one sentence from July. We wrote then that if real household spending stayed negative from a year earlier, the threshold for acceleration would not be met. That does not follow logically, because the household survey is not consumption in GDP, and an annual decline does not rule out quarterly growth. The test is GDP itself.

Closing

Japan’s surveys describe the strongest manufacturing upswing since 2014, METI’s measured output shows a plateau, and between them stand households spending less than last year. The engine is real but narrow, and a growing share of its power comes from price rather than volume. Without the temporary measures, prices are higher than they look, which is why the BoJ is on hold with a tightening bias. We do not yet know whether the summer optimism was growth or inventories. September production on 30 October will give the first answer.

Sources

• S&P Global, Japan Manufacturing PMI, August 2026

• S&P Global, Japan Services and Composite PMI, August 2026

• S&P Global, Japan Manufacturing PMI, September 2026

• S&P Global, Japan Services and Composite PMI, September 2026

• S&P Global, Japan Flash PMI, October 2026

• METI, Industrial Production, final data for August 2026

• METI, Industrial Production, preliminary data for August 2026, with the September and October forecast

• METI, Commercial Dynamics (retail sales), August 2026, e-Stat

• Cabinet Office, Machinery Orders, August 2026

• Cabinet Office, GDP for Q2 2026, second preliminary estimate

• Cabinet Office, Consumer Confidence Survey

• Cabinet Office, Economy Watchers Survey, August 2026

• Statistics Bureau of Japan, CPI, 2025 base, e-Stat

• Statistics Bureau of Japan, national CPI for August 2026, summary of results, e-Stat

• Statistics Bureau of Japan, national CPI for September 2026

• Statistics Bureau of Japan, Tokyo CPI for September 2026, preliminary, e-Stat

• Statistics Bureau of Japan, Family Income and Expenditure Survey, August 2026

• Statistics Bureau of Japan, Labour Force Survey, August 2026

• MHLW, Monthly Labour Survey, August 2026, final data

• Ministry of Finance, Trade Statistics, September 2026, preliminary

• Ministry of Finance, Trade Statistics, August 2026

• Ministry of Finance, seasonally adjusted trade statistics

• Ministry of Finance, Balance of Payments, August 2026, with export volume and prices

• Ministry of Finance, Foreign Exchange Intervention Operations, 30 July - 26 August 2026

• Ministry of Finance, Foreign Exchange Intervention Operations, 27 August - 28 September 2026

• Ministry of Finance, JGB yields

• Bank of Japan, CGPI, September 2026, with revised August figures

• Bank of Japan, Real Exports and Imports, September 2026

• Bank of Japan, Consumption Activity Index

• Bank of Japan, Tankan, September 2026

• Bank of Japan, Statement on Monetary Policy, 18 September 2026

• Bank of Japan, speech by Policy Board member Koeda, 15 October 2026

• Bank of Japan, Monetary Base, September 2026

• Bank of Japan, Time-Series Data Search: monetary base, M2, lending, USD/JPY

• Bank of Japan, Senior Loan Officer Opinion Survey, October 2026

• Bank of Japan, Financial System Report, October 2026

• Trading Economics, Japan economic calendar, market consensus

For information and analysis only. Not financial advice.

In four weeks the White House made seven moves against the price of diesel. It signed a law imposing the toughest sanctions yet on Russian energy, considered a ban on US diesel exports, agreed a release of G7 emergency reserves, allowed dyed off-road diesel on the highways, struck a deal with Putin for Russian diesel, announced an “energy ceasefire”, and finally suspended its own law two days before it was due to take effect. Each move answered the one before, like a game of chess in which the opponent is not Russia but the price of diesel. Policy changes with a signature. Diesel arrives by tanker. One week after the deal with Moscow, the average pump price of diesel tracked by AAA is $6.3276 a gallon, 0.72% higher than a week earlier and 73.72% higher than a year ago. Distillate inventories are 12.2% below their seasonal norm, and 29% below it on the East Coast. The Russian volumes announced through December equal about three days of US distillate consumption. The midterm elections are two weeks away. The deal can work through the global market, but not in time for the data published before the vote.

Sanctions can be lifted in a day. Inventories take months to refill.

Energy: seven moves in four weeks

The September edition ended on the conclusion that neither the Fed nor the White House could produce a spare gallon of diesel within a week. Over the following month Washington tried to do exactly that. Not with refineries, but with decisions.

The three October moves deserve the closest reading, because they pull in opposite directions on the same law. On September 18 the President signed the Lindsey O. Graham Sanctioning Russia and Iran Act of 2026. The Senate had passed it 86 to 11. It provided for a 500% duty on Russian goods and a 100% duty on goods from the five largest importers of Russian oil and gas if they made new purchases from October 18. On October 9 OFAC, the Treasury office that administers sanctions, issued General License 135. A general license is an exemption from sanctions that applies automatically to everyone, with no need to apply for a separate permit. Until April 7, 2027, it authorises the sale, delivery, offloading and importation of Russian diesel, including into the United States. The sanctions on Russia remain in place, and the license can be withdrawn at any time. On October 16, two days before the deadline, the President used his authority under section 115 of the law to waive all of the new duties and most of the new sanctions as being “in the national interests of the United States”.

The White House logic is about energy, not foreign policy. Had the 100% duty taken effect, China and India would have had to choose between Russian oil and the US market. Every barrel left unsold because of that choice drops out of the global balance at a time when Hormuz is still not operating normally. A sanction on the buyer of oil works like a cut in supply, only slower. The waiver is therefore consistent with the goal of cheaper fuel, even where it contradicts the purpose of the law. Reuters cites analysts who said the new measures could have pushed oil prices higher weeks before the election. China and India, two of the largest buyers of Russian oil, had protested against the law.

The market believed it at first. On October 8 ULSD in New York Harbor, the physical spot price of ultra-low sulfur diesel, was $5.305 a gallon. On October 9, the day of the deal, it fell to $4.910, a 7.45% drop in a single session. By October 14 it had recovered 39% of the fall and stood at $5.066. The Kremlin did not confirm the “energy ceasefire”, and it did not last a week. According to Zelenskiy, Russia kept striking Ukraine’s power system, and on October 15 Ukraine hit Rosneft’s refinery in the Ryazan region. On October 16 November WTI settled at $101.41 a barrel, up 3.51% on the day, and the NYMEX HO futures contract at $4.8991 a gallon. That session closed before the Graham Act waiver was signed late in the evening.

The pump moves more slowly than the exchange, but here the direction matters. According to the EIA, the average highway diesel price was $6.179 a gallon in the week of October 12. That is 5.36% below the nominal record of $6.529 set in the week of September 21. AAA’s daily data, however, show a new rise after the deal: $6.2824 on October 10 and $6.3276 on October 17. Gasoline is holding up better. Regular gasoline tracked by AAA is $4.3085, 2.93% below its level a month ago, but still above the $4 political threshold.

The physical buffer explains why the relief is only partial. As of October 9, US distillate inventories stood at 102.6 million barrels. By our calculation that is 12.2% below the average for the same week in 2021-2025. The East Coast deficit is 29%, and New England holds 40.7% less than a year earlier just before the heating season. Refineries are running at 93.5% of capacity, down from 96.8% in mid-September but far above the 85.7% of a year ago. High throughput is not enough, because exports remain strong. Over the last four weeks the US exported an average of 1.499 million barrels of distillates a day, against 1.336 million a year earlier.

Crude oil tells a different story. Brent for physical delivery was $126.55 on October 14 according to the EIA, while WTI at Cushing was $101.91. The $24.64 gap is twice the $12-13 threshold we set in our analysis of the diesel export ban. The seaborne barrel carries the premium of war and freight, while the American barrel stays on land. The American side of that gap is not a shortage of crude. It is a shortage of finished fuel. Natural gas is the counterpoint to the idea of a general energy shortage. Working gas in storage was 3,567 billion cubic feet on October 9, 45 billion above the five-year average. Henry Hub spot is $3.22 per MMBtu. American gas has its own balance, and for now it is sufficient.

The US commodity dashboard

The dashboard shows a market that is splitting apart. US crude is almost unchanged over the month, wholesale gasoline is down 12.51%, and diesel is standing still. Agricultural prices jumped in September, but October futures quotes are already lower. The CRB is down 1.0% for the month and the BDI has lost almost a quarter. The shock is no longer broad. It has narrowed to diesel.

Urea is quoted on the international benchmark, FOB Middle East, not US Gulf. The USDA regional report for Oklahoma puts it at $638.08 a ton at the dealer as of October 16. The two prices belong to different markets and are not compared as a percentage.

The two Brent prices are not an error. $126.55 is the EIA spot price for a physical cargo. $109.86 is the NYMEX futures contract for December delivery. The difference shows how much more a barrel costs today than a barrel two months out. That is a tight market, but the curve is not a clean forecast of future prices. A Fed analysis of October 9 warns that under the current shock futures also include a risk premium and reflect low liquidity.

The dashboard is built from spot prices and monthly averages. Futures contracts are shown separately, because their official history is not publicly available, so for them we give only the latest settlement, on October 16.

The Russian deal in barrels

The deal is announced in tonnes, while America measures fuel in barrels and gallons. So we convert first. At 7.45 barrels per tonne, the firm volumes of 300,000, 500,000 and 1 million tonnes add up to 1.8 million tonnes, or about 13.4 million barrels. With the conditional tranche of another 3 million tonnes “depending on the condition of Russia’s refineries”, the total becomes 4.8 million tonnes, or about 35.8 million barrels.

The numbers say two things at once. The firm tranches almost match the US deficit against the seasonal norm, at 94% of it. That sounds substantial. Yet the same 13.4 million barrels equal about three and a half days of US distillate consumption and a little more than one week of US exports. The October tranche covers half a day. And the entire firm volume through December is about half of one normal month of Russian exports before the crisis.

The deal is not for the American market. It is for the world market. Diesel is fungible: every tonne of Russian fuel that reaches Morocco, Turkey or Brazil reduces demand for American diesel there. That is the strongest argument in the White House’s favour. US refiners can export less without anyone banning them, and inventories at home can refill. The channel is real, but it is indirect and slow. It runs through lower prices in foreign markets, then weaker demand for US exports, and only then higher inventories in New York and Boston.

The law is part of the delivery too. The OFAC license authorises transactions under the sanctions regulations, but does not mention the 2022 law ending imports of Russian oil. That law bans imports of Russian petroleum products into the United States, and the President can lift it only with a certification to Congress. Senators from both parties argue that the license violates the law. European and British sanctions remain in force. The EU price cap for Russian diesel is $100 a barrel. European tankers, insurers and banks can take part only if the fuel was bought at or below that price. Diesel in New York costs about $213 a barrel. Russian diesel can become legal in America and still not find a legal ship. Nor does the license cover the blocked tankers of the “shadow fleet”.

Then there is the refinery in Russia. Ukrainian strikes have cut Russian refining, and Russia itself restricted exports to keep its own market supplied. Moscow partially lifted the restrictions from October 10. On October 15 Ukraine hit Rosneft’s refinery in the Ryazan region, about 5% of Russian refining. The more Russian capacity goes offline, the less is left for the promised tonnes.

The deal also has a price that does not show up in fuel. Criticism came from both parties in Congress, and Zelenskiy called the easing of sanctions a “gift” to Putin. The license explicitly does not authorise debits to accounts of the Russian central bank, the National Wealth Fund or the Ministry of Finance. It cannot, however, avoid generating revenue for Russian oil companies. On October 9 Rosneft shares rose 9%. We are not assessing foreign policy here. We note that the license can be revoked at any time. The market will price it with a risk discount until it sees a steady flow.

The transmission chain: from decision to receipt

Channel 1: crude oil to the gasoline pump

Gasoline is the only channel where the relief is clear. The wholesale price in New York Harbor fell 12.51% over the month to $2.987 a gallon. The EIA average for all grades of gasoline was $4.471 in the week of October 12, against $4.61 in the week of September 21. For drivers that is a little relief, but the annual increase remains around 40%. The $4 political threshold has not been broken to the downside.

Channel 2: ULSD to freight and producer prices

Here the September forecast was fully confirmed. We said in the September edition that the September PPI would carry a stronger diesel imprint. The BLS producer price of diesel jumped 21.0% in a single month and is 115.2% above a year earlier. Truck transportation of freight rose 3.8% over the month, and long-distance truckload trucking 4.6%. Energy within final demand goods is up 10.8%. Overall PPI is 1.3% higher on the month and 6.3% on the year. These are preliminary data and subject to revision. Freight is already in business costs. The question is how fast it reaches the shelf, and the September CPI gives an unexpected answer. Food prices were unchanged over the month, at 2.5% annually. Expensive diesel does not automatically raise food prices in the same month. Part of the cost is absorbed by carriers and retailers. Part will arrive later. Part may never arrive if diesel gets cheaper before contracts are renewed.

Channel 3: gas and urea to the next season

Urea rose 4.49% in September on the international benchmark, but remains 11.62% below its level a year ago. The World Bank fertilizer index is almost unchanged. US gas is cheap, and storage is above normal. This is the weakest of the four channels. The risk to fertilizer use next spring depends more on gas in Europe, which is 128.6% more expensive than a year ago, than on Henry Hub. Farm diesel in Oklahoma is $5.46 a gallon, 7 cents cheaper than in the previous report. The relief for farmers is smaller than the relief on the exchange.

Channel 4: biofuels hold up demand

Biofuels do not depend on a fixed WTI price threshold. They depend on EPA mandates and on the price of diesel. The USDA expects 5.650 billion bushels of corn for ethanol and co-products in 2026/27, 50 million more than in the September forecast. Soybean oil for biofuels is projected at 17.8 billion pounds, against 14.8 billion last year. Expensive diesel makes biodiesel more competitive and lifts demand for soybean oil. That is one reason soybeans remain above $12 a bushel with planted area 7% larger than last year.

Corn: the harvest is bigger than the fear

The September edition presented corn as the strongest fundamental case for expensive grain. The forecast of a weaker crop, the costly harvest and a price of $5.28 a bushel reinforced one another. A month later, two of the main premises have weakened. The USDA raised its forecast. The average yield is 181.2 bushels per acre, 2.7 bushels above the September estimate. Production rises to 16.034 billion bushels, 234 million more. Beginning stocks were raised by 173 million bushels after the quarterly stocks report of September 30. As a result, ending stocks for 2026/27 jump 18% to 1.849 billion bushels. The stocks-to-use ratio, the main measure of how much grain remains as a reserve, rises from 9.68% to 11.34%.

The exchange took it in quickly. The December contract is at 474.25 cents a bushel, about 10% below the 527.50 cents in our September dashboard. The two prices come from different sources, so the comparison is a guide rather than an exact return. The normalisation signal we set in September was corn below $4.75. It has now fired. The opposite signal, corn above $5.25 until harvest passes 25%, failed. Harvest is 33% complete as of October 11, and the price is half a dollar below the threshold. That does not mean America has a surplus. Production is still 5.48% below last year’s record, and ending stocks remain below the 2.095 billion bushels of 2025/26. The difference is that the shortage is no longer deepening. Harvest is running behind average, at 33% against 39%, and 56% of the crop is rated good or excellent. A slow harvest with expensive diesel means a costlier harvest, but not less grain.

The futures curve supports this picture. The March contract is 486.5 cents, 2.58% above December. That premium mainly covers storage and financing of the grain until spring. It looks more like a market with enough grain than a market that pays any price for immediate delivery. Soybeans and wheat do not follow corn. Soybeans are at $12.61, supported by biodiesel and exports, with ending stocks almost unchanged at 315 million bushels. Wheat is at $6.61. It lost its September break above $7, even though the US crop is 22.9% smaller than last year’s. The reason is weak exports: the USDA cut them by another 25 million bushels. A small harvest does not make grain expensive if the world does not want to buy it. The conclusion for food is calmer than in September. Diesel makes harvesting and freight more expensive. The grain itself, however, is getting cheaper, and grain is the larger part of the cost of feed, flour and ethanol.

CRB at 542: the impulse stalls without reversing

The CRB Index stood at 542.15 points on October 16. That is 1.0% below its level a month ago and 48.59% above a year ago. The index remains above the 500-point threshold we have watched since July, but it is no longer making new highs. We need to correct a description from earlier editions. From the close of May 8, 2026, FTSE Russell changed the index composition. It now contains 22 commodities, with weights of 39% energy, 34% agriculture and livestock, 9% precious metals and 18% industrial metals. The old split of 39% energy and 41% agriculture, which we also used in September, no longer applies. The change is methodological, not a market move. It means metals weigh more and grain less than we wrote.

World Bank monthly data show how narrow the September shock was. The energy index rose 25.65% over the month, while the non-energy index rose only 1.32%. Fertilizers were almost unchanged, and precious metals fell 1.91%. Copper is 44.97% above a year ago, but up only 1.03% for the month. This does not look like a general flight into commodities. It looks like one expensive product surrounded by very calm markets. The Baltic Dry Index lost 24.73% over the month. The BDI tracks dry bulk cargoes such as grain, ore and coal, not oil tankers. Its decline says nothing about the cost of shipping diesel. It does say that physical demand for bulk commodities is not following the diesel shock. A full cycle reversal would require the CRB below 400, WTI below $70 and a broad simultaneous decline in every group. None of the three is in place. The more accurate description is a correction within the cycle. Energy keeps the index high, while agriculture and metals have stopped pushing it higher.

The Fed: the Russian diesel decision is a monetary move

The Federal Reserve has held its target range at 3.75-4.00% since the September 16 increase. The next meeting is on October 27-28, six days before the election. The September projections put the median rate at the end of the year at 4.1%, which leaves room for one more small step up. The September CPI does not make the decision easier. Headline inflation is 3.7% annually and 0.6% for the month, up from 3.4% in August. Core inflation, excluding food and energy, accelerated to 2.6% annually from 2.4%. Energy is 20.4% above a year ago, gasoline 34.2% and fuel oil 74.2%. PCE, the Fed’s preferred index, was 3.4% in August and core PCE 3.0%. The September data are not due until November 4, after the Fed meeting and the day after the election. The Fed will therefore decide without seeing September’s energy surge in its preferred index. PPI at 6.3% annually shows that pressure in business costs is stronger than pressure in the consumer basket.

This is where the two institutions meet. The minutes of the September meeting, published on October 7, warn that the longer energy stays expensive, the greater the risk that sector-specific pressure spreads across the whole economy. The Fed cannot produce diesel. The White House is trying to secure it through sanctions policy. When the interest rate cannot increase supply, the government starts conducting monetary policy with other instruments. If Russian tonnes and G7 reserves bring diesel down, the Fed gains room to pause. If they fall short, the Fed will have to respond to 0.6% monthly headline inflation with a tool that hits demand rather than the shortage. The supply shock trap remains the same, but who pays for it is changing. In September it was paid by the carrier and the farmer. In October part of the bill was shifted onto the sanctions regime against Russia. That is a political choice, not a market outcome, and therefore easier to reverse.

What we are watching over the next 30 days

The coming month has a clear calendar. The FOMC meets on October 27-28. The election is on November 3. The October CPI will be published in mid-November. The Russian November tranche of 500,000 tonnes must be loaded over the same weeks. To judge the deal, it is not enough for diesel to get cheaper for one session. We need higher inventories and a lower pump price at the same time.

Signals that the promise is being delivered

• EIA diesel falls below $6 a gallon for two consecutive weeks by mid-November. It is now $6.179. This would be the first relief at the pump that carriers feel in their fuel surcharges.

• The distillate inventory deficit against the seasonal average narrows below 8% by mid-November, and the East Coast below 20%. It is now 12.2% and 29%. This is the only proof that global tonnes are reaching the American buffer.

• NYMEX HO / ULSD falls below $4.25 a gallon and stays there for at least two weeks. It is now $4.8991. Without this, the retail price has no room to come down.

• Publicly tracked Russian cargoes consistent with the November tranche of 500,000 tonnes. An announced volume is not a loaded volume.

Signals that the promise stays on paper

• OFAC amends or withdraws General License 135, or the license is challenged in Congress or in court. The legal knot with the 2022 law is the easiest point of failure.

• New strikes reduce Russian capacity and Moscow tightens export restrictions again. The Ryazan strike shows that the “energy ceasefire” is not working.

• AAA diesel returns above the $6.5276 record. It is now $6.3276. That would mean the deal has been swallowed by a new shock.

• The October PPI shows another monthly increase in truck transportation of more than 1%. That would confirm that cost pressure continues regardless of the policy measures.

Scenario framework

Our base case for the next 30 days is Scenario B: the promise stays on paper. That does not mean the Russian deal is pointless or that it will never work. It means it is too slow for this window, and the November 3 election is two weeks away. Scale is the first reason. The 300,000 tonnes of Russian diesel promised for October come to about 2.2 million barrels. Americans burn that much in distillates in roughly half a day. On public data, only one Russian cargo has been tracked so far: about 40,000 tonnes, loaded on October 13 and heading to Morocco. That is just over one eighth of the amount promised for the month. The reserves are the second reason. On October 2 the G7 promised 100 million barrels from its emergency stocks, but did not say how much each country would provide. The US Strategic Petroleum Reserve holds crude oil, not diesel, and as of October 9, a week after the pledge, it had not released a single barrel. A week is a short time for a sale of this kind, but so far the G7 pledge does not show up in the US data. The pump is the third reason. If the market believed Russian diesel would soon ease the shortage, the retail price would have started to fall. Instead, AAA diesel rose from $6.2824 on October 10 to $6.3276 on October 17.

Scenario A, in which the promise is delivered, is more likely for the winter. The condition is that Russia really does return its diesel exports to 3-3.5 million tonnes a month, its pre-crisis level, and that its refineries are not hit again. The world market would then receive more diesel every month than the entire US inventory shortfall, and prices would have room to fall. That would come after the election, not before it.

The strongest counterargument deserves to be set out in full. Russian Deputy Prime Minister Alexander Novak, who oversees energy, said diesel exports could return to 3-3.5 million tonnes a month, around 22-26 million barrels. According to Reuters this could happen by the end of the year, while Interfax quoted him as saying “in the medium term”. If he is right, every month would bring more than the entire US inventory deficit. Diesel is fungible, so America benefits without a single Russian tanker in New York. G7 reserves add another 100 million barrels of oil and fuel over four months. Gas is above normal, corn is more plentiful than expected, and food prices did not rise in September. The wholesale market has already shown it can fall 7% in a day. If flows start, prices could fall faster than inventories rebuild. The weak point of this argument is that it depends on Russia, Ukraine and the American courts at the same time.

The White House changed the instrument, but not the timing. Inventories refill over weeks, and the election is in two.

Washington has made seven moves. Diesel has not moved yet.

Sources

• U.S. Treasury, OFAC, Russia-related General License 135 and the notice of October 9, 2026: scope, expiry on April 7, 2027, and the financial exclusions.

• The White House, Implementing the Lindsey O. Graham Sanctioning Russia and Iran Act of 2026: the duties, the sanctions and their waiver on October 16, 2026.

• Interfax, Novak on diesel exports, October 12, 2026: the announced tranches and the quote from the President’s post.

• CNBC, Trump says Russia will supply diesel, October 9, 2026 and CNBC, critics of the diesel agreement: the conditional 3-million-tonne tranche and reactions in Congress and Ukraine.

• CNBC, Trump waives Graham Act sanctions, October 17, 2026 and NBC News: the October 18 deadline, the Senate vote and reactions.

• Reuters, Zelenskiy criticizes Trump’s waiver of sanctions on Russia, October 17, 2026: analysts on the oil-price risk, the protests by China and India, and Russian strikes on Ukrainian energy.

• Clyde & Co, OFAC authorizes US imports of Russian diesel and Lloyd’s List Intelligence: the 2022 law, European restrictions and the $100 price cap.

• Ukrainska Pravda, citing Reuters, October 13, 2026: the target of 3-3.5 million tonnes a month; Le360, citing LSEG: the first tracked cargo, bound for Morocco.

• Reuters, Ukraine strikes Ryazan refinery, October 15, 2026 and NBC News, energy ceasefire: the “energy ceasefire” and its end.

• G7 Leaders’ Statement on Global Energy Security, October 2, 2026: the release of 100 million barrels.

• The White House, Executive Order 14435 and IRS IR-2026-122: dyed diesel relief.

• U.S. Energy Information Administration, spot prices and NY Harbor ULSD: WTI, Brent, gasoline and ULSD through October 14, 2026.

• EIA, Weekly Retail Diesel Prices and gasoline, all grades: weekly retail prices.

• EIA, Weekly Petroleum Status Report, Tables 1, 6 and 9: inventories, the SPR, refinery throughput, exports and product supplied for the week through October 9, 2026.

• EIA, Weekly Natural Gas Storage Report: gas storage as of October 9, 2026.

• AAA Gas Prices: retail prices as of October 17, 2026.

• CME Group, Daily Bulletin, Energy and Agriculture; USDA AMS 3148: futures settlements as of October 16, 2026.

• World Bank, Pink Sheet, October 2026: monthly prices and indices for September 2026.

• Trading Economics, CRB and Baltic Dry: levels as of October 16, 2026.

• FTSE Russell, CoreCommodity CRB Index Methodology, May 2026: the new weights.

• USDA, WASDE-676 and USDA NASS, Crop Progress, October 13, 2026: supply and demand balances and harvest progress.

• USDA AMS, Oklahoma Production Cost Report: fertilizer and farm diesel as of October 16, 2026.

• U.S. Bureau of Labor Statistics, CPI September 2026 and PPI September 2026: consumer and producer prices.

• U.S. Bureau of Economic Analysis, Personal Income and Outlays, August 2026: PCE.

• Federal Reserve, FOMC statement, September 16, 2026, Summary of Economic Projections and FOMC calendar.

• Federal Reserve, FEDS Note, October 9, 2026: how to read oil futures during a geopolitical shock.

• EPA, Final Renewable Fuel Standards for 2026 and 2027: biofuel mandates.

This publication is for informational and educational purposes only. It does not constitute investment advice, a recommendation, or a solicitation to buy or sell any financial instrument. Past performance is not indicative of future results. Every investor should conduct independent research and assess their personal tolerance for risk.

Liquidity Desk | liquiditydesk.org

This issue examines the eurozone in August 2026, using information available through 17 October. The core monthly releases are in, although official August services output is still missing. September inflation, the ECB’s interest rate decision and its October analysis show how the story developed after the reporting month. They are subsequent evidence, rather than August results.

An economy growing unevenly

Our working assessment is weak, uneven growth, without confirmation of a broad acceleration. Services and parts of investment are supporting activity, while industrial production and household purchases remain subdued. Rising energy bills and more expensive finance constrain the recovery. For the European Central Bank, the question is whether price pressure is spreading far enough to justify adding more interest rate pressure.
In July’s analysis, we called the picture stagnation. August failed our pre-announced test for acceleration because industrial output fell again. Yet industry is only part of the economy, and the positive services signals deserve more weight than we gave them then. Failing our acceleration test does not establish that aggregate growth was zero.

The monthly figures expose the difference. The composite PMI, a survey of changes in current business activity, stayed at 52.0. A reading above 50 signals expansion. Industrial production nevertheless fell 0.5%, while retail sales rose just 0.1% after July’s decline.
There are identifiable sources of resilience. Digital services, investment associated with artificial intelligence and public spending support activity. Some additional borrowing, however, pays larger bills rather than expands productive capacity. Credit growth alone cannot settle the argument.
Prices also send a mixed message. Headline inflation rose to 3.2%, while core inflation eased to 2.4%. That supports a conditional case for waiting after September’s rate increase. The central question is how long the economy’s growth engines can keep paying the rising bill.

The main indicators

The direction becomes clearer over time

Business surveys recovered from the spring shock, but measured output and purchases did not follow convincingly. We compare August with February, six months earlier, and May, three months earlier, so that a single better month cannot erase the weakness that preceded it.

Three divergences matter. The first is between services and industry. Services returned to survey expansion after the spring decline, but their activity reading was still below February’s. Manufacturing surveys improved more clearly, even as recorded production fell again. Differences in coverage, the mix of firms and measurement dates can help explain the gap. They do not justify dismissing either source.
The second divergence is between August’s small retail rebound and the weaker medium-term trend. Purchases remained below June’s level after July’s fall and August’s gain. A positive monthly sign does not mean households have recovered the purchasing power they lost.
The third concerns prices. Producer costs are accelerating sharply, yet core consumer inflation is not following them. The question for the coming months is whether firms can pass on more of the bill or whether weak sales force them to absorb it.

The quarterly backdrop also argues against a hasty verdict of economy-wide contraction. The current estimate puts real GDP growth in the second quarter at 0.6% from the first. The ECB’s measure adjusting for distortions in Irish data is 0.3%. It does not replace the official figure, but explains why the headline needs care.
Employment grew 0.1% in the second quarter. Negotiated wages rose 2.44% from a year earlier, after a revised 2.59% in the first quarter. These are quarterly supports, rather than measurements of August. Our previously corrected claim that inventories drove second-quarter growth does not return in this account.

What keeps the economy moving

Services are the most important support. The latest official output figure is for July and shows monthly growth of 0.4%. August’s survey remained in expansion. These are different sources for different months, but together they make a stronger case for continuing activity than the industrial aggregate alone.
In an economy with a large services sector, weakness among goods producers can coexist with overall growth. The issue is whether orders and revenue elsewhere can compensate. August’s surveys give us reason to keep that possibility central while we await official volumes.
There is a more specific story within services. The ECB’s October analysis highlights digital activity, investment in artificial intelligence and demand for European components and equipment serving that investment. Europe supplies and adopts these technologies, although its investment wave is smaller than America’s.
Capital goods production rose 0.4% in August. That is consistent with resilient investment segments, but it cannot all be attributed to AI: the statistical category covers much more than technology equipment. We have no measured August contribution from AI to aggregate growth, and do not need to invent one to explain the mechanism.

Fiscal policy is the second support. German defence and infrastructure programmes and spending under Next Generation EU sustain demand. This is support across the year, rather than a quantified August effect. It is also temporary: the ECB expects fiscal support to weaken after 2026.
The two engines have different staying power. A public contract can fill a production schedule, but depends on budgets and execution. Technology investment can sustain demand for longer, but depends on expected returns, available funding and confidence that new equipment will be used. Announced spending is not output already delivered.

Funding sources matter too. In the ECB’s survey of planned AI investment, most firms intending to invest expected to use internal funds. A strong balance sheet can allow one business to invest even as bank finance becomes more expensive. Another, dependent on new borrowing, will react more quickly to interest rates.
This helps explain why particular activities can look strong while the broader economy still struggles to find support. The engine is real, but unevenly distributed. The best confirmation would be growth spreading into more services and actual purchases, rather than remaining concentrated in a few investment chains.
Durability also depends on what remains after the initial order. Building a data centre creates work now; its continuing contribution depends on subsequent use and the services surrounding it. Buying software can improve a firm’s operations, but the purchase itself does not establish higher productivity. Time and risk separate investment spending from sustainable income.
The benefits must spread too. An equipment supplier can place more orders with its partners and support employment. If that impulse remains narrow, it will not be felt equally in retail, home purchases and other services. We are looking for a stronger connection between investment, labour income and demand, as well as more investment itself.

The bill is rising faster than purchases

August’s European Commodity Snapshot focused on expensive energy even without an immediate physical shortage. The new macroeconomic releases show how that problem reaches the external balance and household budgets. Available supplies and affordable supplies are different conditions for recovery.
The seasonally adjusted goods trade balance deteriorated to a deficit of EUR 6.8 billion in August, after July’s EUR 1.4 billion surplus. Our older July article reported a larger surplus; this comparison uses the revised figures. Exports fell 1.2% on the month and imports rose 2.0%.
Energy is an important part of the story. August’s energy deficit reached EUR 26.7 billion, compared with EUR 21.3 billion a year earlier. But the deterioration extended beyond fuel: machinery and vehicles also moved from a surplus to a deficit compared with the previous year.
These flows are measured in current euros. Higher imports can reflect prices, quantities or a different product mix. The nominal balance cannot be translated directly into a real GDP contribution. Despite the weak August result, exports over June-August exceeded the preceding three months. External demand is not undergoing a one-way collapse.

The household constraint appears in purchases. Non-food sales grew in August, but total retail volume barely moved. Retail excludes many consumer services and therefore does not measure all household consumption. It still shows how fragile the recovery in goods purchases remains.
The labour market is a buffer for now. Unemployment stayed at 6.4%, but the number of unemployed people increased by 26,000 on the month behind that unchanged rounded rate. This does not establish a sharp labour market reversal. It does mean that one decimal place offers no guarantee of calm.
There is also evidence against an excessively bleak household reading. Nominal disposable income continued to grow in the second quarter, while the saving rate fell from 14.6% to 14.4%. Lower saving can support some spending. That option is very different for a household with accumulated reserves and one already financing daily expenses with debt.

The brake has both cyclical and structural elements. The energy price surge and the effect of interest rates can ease. Dependence on imported energy, supply contracts and countries’ different energy mixes change more slowly. A lower interest rate can reduce a loan instalment, but cannot provide cheaper deliveries through the Strait of Hormuz.
Releasing the brake therefore requires more than an ECB decision. Energy must become more affordable, incomes must catch up and sales must justify investment. Until these conditions improve together, firms and households will experience growth very differently.
There is an external dimension too. When the region pays more for necessary imports, income that could have supported domestic goods, services and investment goes to overseas suppliers. That can depress growth even if the quantity of energy imported holds steady and production continues uninterrupted.
Fiscal compensation can redistribute this loss, but cannot remove it for the economy as a whole. Supporting the most affected households can preserve essential purchases. Broadly offsetting every price increase, however, also sustains the demand the central bank is trying to cool. The difficulty is helping vulnerable households without making the cost pressure more persistent.

Higher inflation, different sources

Headline inflation accelerated in August, but its composition matters more than the aggregate. Energy prices rose 14.3% from a year earlier, after 10.3% in July. Services inflation slowed from 3.3% to 3.0%. This was not a simultaneous acceleration throughout the consumer basket.
Core inflation, which excludes energy, food, alcohol and tobacco, also eased. For the central bank, it is evidence of how far the shock has travelled beyond directly affected prices. One softer reading does not mean the delayed transmission is over.

Producers face much stronger pressure. Industrial producer prices rose 8.2% annually, with their energy component up 21.0%. Excluding energy, the increase was 3.4%. The contrast identifies where the shock is concentrated instead of suggesting that every firm faces the same cost increase.
The gap between headline producer and consumer inflation is 5.0 percentage points. It is neither a ready-made forecast of future inflation nor a measure of business profits. The indices have different baskets and weights. Industrial producer prices exclude consumer services and do not measure import prices.
What matters is the response between a firm’s costs and its sales. A business with strong orders can raise prices. A business facing weak sales may accept lower profits, cut production or postpone investment. Those responses have different consequences for inflation, employment and growth.

The ECB’s newer analysis of gas and electricity clarifies the timing. National central banks expect wholesale gas price changes to reach consumer gas inflation within 1-3 months in more than half the euro area, measured using the consumer index’s weights. Electricity transmission is much more dispersed and can take up to two years.
Contracts, regulated tariffs and taxes change the speed. More renewable generation weakens the link between gas and electricity prices, with different effects across countries. A single timetable for every energy bill would conceal precisely what determines when households feel the shock.

Inflation expectations also warrant attention. The ECB’s August survey recorded a rise in expected annual inflation three years ahead, from 2.7% to 2.9%. This is the median, the midpoint of the ordered responses. That is not realised inflation, but shows how persistent households expect the price increases to be.
September’s subsequent check is less comfortable: headline inflation reached 3.8% and core 2.5%. August’s easing in core did not become a continuous downward trend. We still need to distinguish direct energy inflation from a broader, sustained acceleration in other prices.
Slower inflation does not restore old prices either. It means the next increase is smaller, while households continue paying the level already reached. Real consumption depends on income catching up with cumulative price increases. Lower core inflation and weak purchases can therefore coexist without contradicting each other.

AI, orders and the price of money

Europe’s AI sector is usually discussed as a story of falling behind. The more useful question here is how the same global boom can support European growth and make it harder to sustain.
The positive connection runs through investment, digital services and equipment suppliers. On 13 October, ECB Chief Economist Philip Lane said AI accounted for roughly a quarter of recent growth in lending to firms. That is a share of credit growth, not of all outstanding loans, and is not a separate August statistical series.

The negative connection runs through the global cost of long-term finance. Large investment requirements and expectations of future productivity gains are plausibly contributing to higher global interest rates. European bond markets feel the movement even though the domestic investment boom is smaller.
A European company outside the technology supply chain can face more expensive finance without receiving additional orders. A supplier of data centre equipment may tolerate the same higher rate because it expects more revenue. Both operate in the same economy, but the cost of money affects them differently.
This does not establish that AI explains the entire rise in yields. Inflation expectations, government budgets and perceived risk also matter. It is a contributing mechanism, rather than the sole cause of every market movement.

The implication for the ECB is practical. If long-term finance is becoming more expensive because of external forces, some cooling of demand arrives without another change in the deposit rate. That belongs in the assessment of the next decision.
Technology investment can pull the economy forward while the financing of the global boom raises the bill for everyone else. Strong investment segments therefore do not guarantee an even recovery.

What this means for the ECB, credit and households

August’s picture leaves the ECB balancing resilient activity against the energy shock’s effect on income. There are engines capable of sustaining demand. At the same time, a larger imported energy bill takes money away from other spending. The same supply shock raises prices and weakens parts of demand.
The deposit rate was 2.25% in August. On 10 September, the ECB raised it to 2.50%, effective 16 September. Weak August volumes cannot therefore be a consequence of that decision. They describe the economy on which it begins to act.

Our reading favours cautious waiting after September’s increase, provided price pressure remains limited outside energy and expectations do not deteriorate further. This is a conditional analytical judgement, not a promise of a pause. Broader acceleration in prices and wages would strengthen the case for further tightening.
In her 17 October statement, Christine Lagarde emphasised that higher long-term rates were already tightening conditions, while energy transmission beyond that built into the projections remained contained. Policy must account for the cooling that markets are already delivering.
Monetary quantities do not send an unambiguous easing signal either.

M3 includes money and closely related short-term instruments. Faster growth in that aggregate can coexist with slowing corporate lending and more expensive loans. Banks’ reserves at the central bank are a third, distinct quantity. A decline in reserves alone does not establish a mechanical constraint on lending: loan demand, risk and bank capital matter too.
A bank can have enough funding and still decline a loan if it expects a customer’s sales to weaken. A firm can have access to credit and choose not to use it because the project no longer covers the higher interest cost. Money available within the financial system is not money equally accessible to every borrower.
The composite cost of new corporate borrowing was 3.77% in August. Behind the average sit different businesses and needs. One company borrows to expand capacity; another finances deliveries before receiving the proceeds from sales. Similar credit growth can accompany very different economic strength.

September’s Credit Pulse found European credit markets calm at its observation date. That does not mean cheap loans. A credit spread is the risk premium over the underlying yield. The final borrowing cost can rise even if that premium remains small.
October’s liquidity review likewise showed that the different channels were not moving together. Those thematic signals have their own dates and markets. They help explain why difficulties can emerge first among particular borrowers before there is a general credit shortage.
For households, the effects meet in the monthly budget. Energy absorbs more income, a new mortgage costs more and purchases on credit become harder to afford. For firms, the issue is whether additional sales cover both energy and finance. Growth can continue even as more participants find that less money remains after the bills.

What could change this reading

The strongest case for a better picture is that services already support broader growth than industrial data reveal. If official services volumes confirm the surveys and investment orders spread across more activities, our description of weak growth will prove too cautious.
Budget execution can also surprise positively. Faster infrastructure projects and investment funded internally could allow activity to withstand higher rates. Cheaper energy would release income for other purchases without requiring immediate ECB easing.

The less comfortable scenario is persistent cost pressure alongside weaker orders and consumption. More credit growth would then pay rising bills instead of financing future output. A weaker labour market would remove one of households’ main buffers.
Broader energy transmission is another risk. It would limit the ECB’s freedom to wait even as activity weakened. Conversely, persistently soft demand could restrain selling prices, at the cost of lower profits and postponed investment.
Our article “Rate hikes don’t open Hormuz” explains this conflict, including the differences from 2011. The historical comparison is useful as a warning about a mechanism, rather than a prediction that a debt crisis must repeat.
The next volume releases must show which risk dominates. If only the strongest segments keep expanding, the gap between the aggregate result and households’ experience will remain wide.

What to watch

On 27 October, the third-quarter bank lending survey and September monetary data will show whether credit access tightens and corporate lending slows further. The ECB’s 29 October decision will test our case for waiting and the weight it gives to long-term rates.
Unemployment on 30 October will show whether the rounded rate moves above 6.4%. Producer prices on 5 November provide the next energy cost check; retail sales on 6 November will show whether purchases recover more of their earlier losses.
Official August services output on 9 November will test the strongest support for our current reading. The next industrial production release follows on 16 November.

For a reassessment towards broad acceleration, we retain our declared test: monthly industrial growth above 0.3%, positive retail sales and a PMI above 50, supplemented by confirmation in official services output. Weakness spreading into services would strengthen the case for stagnation or contraction. This is a pre-announced check, rather than a promise that one month automatically defines the entire cycle.

Growth has to pay the bill

The eurozone is moving forward unevenly. Services, technology investment and public spending provide support, while industry and household purchases remain fragile. Energy and finance raise the cost of that progress.
We do not yet know whether the growth engines can broaden the recovery enough to offset pressure elsewhere. One strong segment and one positive lending rate do not answer that question. The answer must come from broader real activity and incomes that leave more money after the bills. Until then, the ECB’s caution also depends on how much cooling prices themselves are already causing.

Sources

• S&P Global, final August composite, services and manufacturing PMI releases (1); S&P Global, final August composite, services and manufacturing PMI releases (2)

• Eurostat, August industrial production, 15 October

• Eurostat, August retail trade, 6 October

• Eurostat, final August inflation, 17 September

• Eurostat, August industrial producer prices, 5 October

• Eurostat, August unemployment, 1 October

• ECB, key interest rate history and 10 September decision (1); ECB, key interest rate history and 10 September decision (2)

• Eurostat, July services output, 7 October

• Eurostat, current real GDP series; ECB, September projections and Ireland adjustment (1); Eurostat, current real GDP series; ECB, September projections and Ireland adjustment (2)

• Eurostat, employment; ECB, negotiated wages (1); Eurostat, employment; ECB, negotiated wages (2)

• Philip Lane, Diagnostic Challenges for ECB Monetary Policy, 5 October

• Philip Lane, The AI opportunity for Europe, 13 October

• ECB, How firms are planning to finance AI adoption, 2 October

• Eurostat, August international trade in goods, 16 October

• ECB, second-quarter household and corporate sector accounts, 2 October

• ECB, The uneven journey of wholesale gas and electricity prices to consumer bills, Economic Bulletin 6/2026

• ECB, August Consumer Expectations Survey, 18 September

• Eurostat, final September inflation, 16 October

• ECB, monetary developments in August, 25 September

• ECB, August bank interest rates, 1 October

• Christine Lagarde, IMFC Statement, 17 October

• ECB, statistical release and monetary policy meeting calendars (1); ECB, statistical release and monetary policy meeting calendars (2)

Not financial advice

Liquidity Desk | liquiditydesk.org

After the narrow market

September showed a narrow market. Capital paid for two sources of profit, energy and Asian semiconductors, and punished almost everything that depends on broad economic acceleration.

Between the two issues, the tide went out. October’s liquidity analysis, The Ebb Tide, described it as water pulling back from every shore at once: the ECB, the Fed and the Bank of Japan raised rates within eight days. The dollar strengthened. The long end in the United States kept rising.

This month the market is not looking for a new theme. It is returning to companies and sectors whose earnings are already proven and do not depend on cheap credit. US and European technology, US health care and Brazil gain. Banks, real estate, carmakers and most emerging markets lose.

The question this month is why banks are falling just as the rates that supposedly feed them are rising, and why South Korea, September’s leader, is among the weakest.

Macro backdrop

The rate environment

Three hikes in eight days. This is what is new in October’s picture, and it sits behind almost every move below.

The Federal Reserve raised its target range by 25 basis points, to 3.75-4.00%, on 16 September, by a unanimous vote. The effective rate is 3.88%. The ten-year yield kept rising, to 5.22% on 8 October from 4.95% on 10 September, with a peak of 5.31% on 5 October. The gap between the two is 1.34 percentage points, almost the same as in September. The market still demands compensation for long-term risk, even after the Fed has acted. The minutes of the meeting point to a new reason as well: heavy debt financing of artificial intelligence infrastructure is competing with the government for the same capital. ECB chief economist Philip Lane said the same in an interview on 6 October.

The European Central Bank raised its deposit rate to 2.50%, effective 16 September. The ten-year German Bund, however, did not keep rising: 3.53% on 9 October, against 3.56% on 11 September. The peak was 3.69% on 28 September, the highest level since April 2011 (see the separate analysis). The reason for the hikes is still energy. According to Philip Lane, euro area inflation is 3.8%, with energy at 18.8% and everything else at only 2.3%.

The Bank of Japan raised its rate from 1.00% to 1.25% on 18 September, by a 7-2 vote, effective 24 September. The average yield at the 30-year Japanese government bond auction on 8 October is 4.109%, against 4.079% on 3 September. The gap to the policy rate narrows to 2.859 percentage points, because the short end is catching up with the long end. This directly affects the carry trade, the practice of borrowing cheaply in yen and investing the money in higher-yielding assets elsewhere. The more expensive the yen loan, the thinner the profit from the trade.

The next test is packed into four days: the Fed on 27 and 28 October, the ECB on 29 October and the BOJ on 29 and 30 October.

The broad dollar

The Federal Reserve’s Broad Dollar Index stood at 121.3848 on 2 October, the latest available date, against 118.0732 on 4 September. That is a rise of 2.80%, after a 0.93% decline in the previous issue.

The direction has reversed, and so has the mechanism. A stronger dollar reduces the dollar value of foreign revenue for US companies and makes dollar debt more expensive for emerging economies. The second channel is clearly visible in Latin America and India. Brazil is the exception, explained in the Latin America section.

Risk appetite

The VIX stood at 15.41 on 8 October, against 17.84 on 10 September. The VIX measures the price of protection against short-term moves in the US equity market. Below 20 signals calm, above 20 greater nervousness and above 30 active fear. Between the two issues it did not close above 20 on a single day. The high was 17.71 on 16 September, the day of the Fed decision.

Three rate hikes, a stronger dollar and oil above $120 do not produce panic. That is the frame for everything below. The declines in banks, real estate and emerging markets are an orderly repositioning, not a flight.

Oil

Oil has two prices, and this month they are further apart than ever. Brent for physical delivery, according to EIA data, was $125.44 on 6 October, against $118.06 on 11 September, with a peak of $135.51 on 2 October. Brent futures, the price quoted in the news, settled at $102.31 on 1 October and at $100.32 on 5 October. The gap between the real barrel and the paper barrel is more than $30.

The reason is that crude is now getting through, but fuel is not enough. According to Reuters, Middle East crude exports rose above pre-war levels on four of the seven days in the last week of September, even though three tankers were struck in the Strait of Hormuz on 29 September. Diesel and other fuels, however, remain in short supply after damage to refineries in the Gulf and in Russia. Russia has halted diesel exports until the end of October, and on 1 October Chinese refiners suspended fuel exports beyond Hong Kong and Macau. The same day brought reports that the United States is sending a third carrier strike group to the region, and that President Trump is weighing new strikes on Iran after the midterm elections. Futures jumped 4.37% in a day. On 2 October the G7 agreed to release 100 million barrels of diesel and crude from emergency stocks over four months.

US WTI moved the other way: $96.24 on 6 October, against $101.27. The shock is no longer in the volume of crude, but in fuel and in the risk of renewed escalation.

United States: Technology returns

Technology leads again. XLK gains 5.81% for the month after only 0.72% in September. Its six-month return is 40.29%. September’s cooling turned out to be a pause, not a reversal. On an ebb tide, capital returns to the sector whose earnings growth exceeds the yield on bonds.

Health care is second with 2.20%, after minus 1.83% in September. XLV is up 14.12% over six months. The sector has structural support that does not depend on interest rates: ageing populations in the United States, Europe and Japan mean demand for treatment that is locked in for years ahead. When credit becomes more expensive, predictable demand becomes more valuable.

Energy stalls. XLE adds only 0.48% after 8.24% in September, even though physical Brent hit a new high. Its three-month return remains 17.05%. Producer shares no longer follow every new high in the price of the barrel. This is not a reversal. The sector keeps its accumulated strength, but the market no longer pays extra for it.

Financials are the surprise in the bottom half. XLF loses 4.49% even though the Fed raised rates. By the textbook, a higher rate widens banks’ lending margin. But when the long-term rate rises together with the short-term rate, and credit becomes more expensive for everyone, the risks to loan quality rise too. The market is selling banks as lenders exposed to the economy, not as collectors of interest.

Rate-sensitive sectors close the table. Utilities lose 4.70%, materials 4.83% and real estate 5.22%. Utilities and real estate are valued as bond substitutes. With the ten-year yield above 5%, the bond itself offers the yield, without the risk of the share.

RRG: The slower signal

The horizontal axis shows relative strength against SPY, while the vertical axis shows the change in that strength. The upper-right Leading quadrant combines strength with positive momentum. The lower-right Weakening quadrant means a sector remains strong but is losing speed. The upper-left Improving quadrant shows recovery from a weak base, while the lower-left Lagging quadrant combines a weak position with weak momentum.

The map confirms the table. XLK and XLE are in Leading, and both are moving up and to the right. In September technology was losing momentum; now it is gaining it again. XLK remains the sector with the highest relative strength on the map. Energy keeps its direction even though its monthly return is close to zero, because the monthly map tracks a smoothed trend, not a single month.

XLV is the only sector in Improving and is moving to the right, towards Leading. If that direction holds for another month or two, health care will become the third durable leader.

XLI drops from Weakening into Lagging. Industry is losing the last of its relative strength against the market. All other sectors are in Lagging. For XLC, XLP, XLU and XLY the latest step is upward, meaning their momentum has stopped deteriorating, but their relative strength is still weakening. This is an early sign, not a turn.

The RRG does not measure actual money flows. It shows relative performance and the change in that performance. The chart is therefore a map of market preference, not proof that money leaving banks has moved into technology. It is a transparent RRG-style approximation built from the final available weekly adjusted price in each month, not the official proprietary JdK model. The last point is the incomplete October observation through 9 October 2026.

Europe: Technology leads, banks fall

Europe repeats the US reversal, and does so more sharply. Technology swings from minus 4.09% in September to plus 4.03% and leads the European table. EXV3 is up 25.27% over six months. At its core are ASML, SAP and Infineon, directly tied to the global semiconductor and artificial intelligence cycle. European and US technology move together, while South Korea pulls back. There is also a structural reason the sector withstands the hikes. According to an analysis on the ECB blog, firms in the euro area’s most AI-intensive sectors carry less debt and fund themselves more through equity and bonds than through bank loans. The policy rate therefore hits them less than firms that live on bank credit.

Oil and gas are second with 3.75% and stay positive for another month. Here European producers gain more than US ones, reversing September’s picture. EXH1 is up 14.32% over three months. A more expensive Brent matters more for European companies, whose main crude grades are linked to it, than for US ones, which depend more on the cheaper WTI.

Banks are the heaviest signal. EXV1 loses 8.75% after plus 0.44% in September. This is the second-worst result in the table, in exactly the month the ECB raised rates. The six-month return remains 10.05%, because until now higher rates widened banks’ margins. This month the ghost of 2011 returns. The German Bund became a safe haven again, and the yield gap between French and Italian debt and the Bund widened sharply: for France from about 0.85 to 1.27 percentage points between 11 September and 9 October, with a peak of 1.41 on 2 October, and for Italy from 0.85 to 1.11 points (in detail in the separate analysis). European banks hold large amounts of their own governments’ debt. When government debt loses value, banks lose with it. The market is selling not just the margin, but the balance sheet.

The bottom of the table is entirely cyclical and rate-sensitive capital. Carmakers lose 8.97% and remain negative over six and twelve months. Higher consumer credit costs now add to the structural pressure from Chinese competition. Construction falls 8.47% and real estate 6.97%, the second consecutive month with a decline of almost 7%.

Basic resources and telecoms reverse their September gains. EXV6 loses 6.70%, but its three-month return remains 12.60%. This is a pullback after a strong run. EXV2 loses 6.75%, so September’s rebound proved short-lived.

Asia: Korea gives back part of its gain

In Asia, we use country funds rather than sectors. Currencies, regulation and economic cycles differ too widely to fit into one common sector index.

South Korea falls 6.67% after a 15.69% gain in September. This is the weakest result in the region. But EWY remains up 26.54% over six months and 110.52% over one year. The Korean fund is extremely volatile, because almost half of it is Samsung Electronics and SK Hynix, effectively a single position in memory for artificial intelligence. A pullback after a strong run is not the same as structural weakness.

Taiwan stays positive with 2.48% and keeps a six-month return of 49.70%. TSMC and its surrounding supply chain continue to hold the most visible role in capital spending on artificial intelligence.

Japan is almost flat at minus 0.10%. The BOJ’s higher rate narrows the gap to the 30-year yield and makes the carry trade thinner. The Japanese market is not falling, but it is not receiving the new capital it received over the summer. Domestic demand does not help either: the BOJ’s consumption activity index falls 0.6% in the third quarter, based on the July and August averages.

China falls less than in September. FXI loses 2.17% and MCHI 2.60%. On 29 September the People’s Bank of China cut the rate on its one-year collateralised lending to the state development banks from 1.75% to 1.5%, expanded targeted lending quotas and introduced a mortgage subsidy. The measures target specific industries, not the main policy rates. The market reads them as a brake on the decline, not a turn.

India loses 6.23% and is now down 12.54% over one year. A stronger dollar hits hardest the markets that import oil and pay for it in dollars. India is exactly such a market.

Latin America: Brazil against the dollar

The dollar is the main variable for the region. It works through two channels: it reduces the value of commodity revenue in local currency and makes dollar debt more expensive to service. This month the broad dollar rose 2.80%, and the region reacted almost by the textbook. With one big exception.

Brazil gains 12.77%, after 8.53% in September. EWZ is now up 26.53% over three months. Three things set it apart from the region. First, a cut in the Selic policy rate from 14.00% to 13.75% took effect on 17 September. Brazil is easing while the Fed, the ECB and the BOJ tighten. Second, the real strengthened against the dollar, even though the dollar strengthened against almost everything else. The rate was 4.9892 reais per dollar on 9 October, against 5.0918 on 11 September. The biggest move came on 5 October, the first trading day after the first round of the presidential election.

The election is the third and most important explanation. According to the official results of the Superior Electoral Court (TSE), opposition candidate Flávio Bolsonaro finished first with 47.03% of valid votes, ahead of President Lula with 45.16%. The two go to a runoff on 25 October. The next day the real gained 4.8% against the dollar, the strongest one-day move between the two issues. The market is not waiting for the runoff. It is already pricing the possibility of a change of course. That makes Brazil’s gain fragile: if the runoff reverses the result, part of it could unwind just as quickly.

The regional ILF fund gains 4.46% only because of Brazil’s weight in it. Without Brazil, the picture is clearly negative.

Mexico loses 6.16% and Chile 8.15%. Mexico remains exposed to the US trade cycle, and the stronger dollar adds pressure. Chile is an indirect bet on Chinese industrial demand through copper, and China is not yet signalling a new acceleration.

Argentina loses 7.71% after plus 2.11% in September and is now negative over three and six months. Its reform thesis, built on fiscal consolidation and liberalisation of the currency regime, had so far set it apart from the region. This month it falls with it. A one-year return of 27.90% shows that the accumulated gains are not yet erased.

The rotation signal

Capital is not fleeing. It is returning to proven earnings.

The VIX is 15.41, lower than in September, despite three rate hikes and oil above $120. This is not panic. It is an orderly withdrawal from everything that depends on cheap credit, towards earnings the market has already seen.

The first theme is the return to proven earnings. XLK gains 5.81%, EXV3 4.03% and XLV 2.20%. US and European technology lead together, and health care joins as a second choice. When bonds pay more than 5%, a share with no growth and a dividend below that yield makes no sense. The market pays only for companies whose earnings growth can exceed the safe yield on the bond.

The second theme is that higher rates no longer help banks. XLF loses 4.49% and European banks 8.75%, in the same month the Fed and the ECB raise rates. Until September, higher rates were a driver for banks. When rates rise for long enough, the market starts pricing not the margin, but the borrowers. Real estate, construction and carmakers, the sectors most dependent on credit, fall together with banks on both sides of the Atlantic. In 2011 the same chain from government debt to banks sat at the centre of the crisis.

The third theme is the strong dollar as the regulator of emerging markets. The dollar rises 2.80%, and India, Mexico, Chile and Argentina fall by between 6% and 8%. This is the familiar mechanism in its pure form: more expensive debt and cheaper local revenue.

The quiet signal is Brazil. The only large economy in the analysis that cut rates is also its strongest market. EWZ gains 12.77%, and the real strengthens against a strong dollar. If Brazilian easing continues while others tighten, the rate differential will keep attracting capital.

Energy remains at a high level, but stops accelerating in the United States. Physical Brent set a new high, while XLE added only 0.48%. The market has already paid for the energy shock and is not paying twice.

The next week of truth is 27 to 30 October, when the Fed, the ECB and the BOJ decide within four days. If all three tighten again, the market’s choice will become narrower still. If they pause, the first confirmation of a rising tide should come from banks and industry, the two sectors that lost the most this autumn.

Data and sources

• ETF performance: StockAnalysis, retrieved on 10 October 2026. Values are rolling returns to the close of 9 October, as displayed by the source.

• FOMC decision of 16 September 2026: Federal Reserve. Minutes: FOMC Minutes. Calendar: FOMC calendar.

• Federal Reserve effective rate and ten-year yield: FRED DFF and FRED DGS10.

• Broad Dollar Index: FRED DTWEXBGS.

• VIX: FRED VIXCLS.

• Brent and WTI spot, EIA: FRED DCOILBRENTEU and FRED DCOILWTICO.

• Oil market, 1 and 5 October 2026: Reuters, Oil jumps 4% on reports China halts fuel exports, US troops head to Middle East and Oil slips as Middle East crude exports rise, G7 to release stocks.

• ECB decision of 10 September 2026: European Central Bank. Calendar: ECB.

• Philip Lane, 5 and 6 October 2026: speech and interview with ANSA.

• ECB blog, 6 October 2026: Funding the AI revolution: evidence from euro area sectors.

• Ten-year German government bond yield: Deutsche Bundesbank.

• Spreads against the Bund: France approximated as the Banque de France TEC10 yield minus the Bundesbank curve; Italy from ANSA daily closes (8 October via SoldiOnline). In detail: Rate Hikes Don’t Open Hormuz.

• Bank of Japan decision of 18 September 2026: BOJ. Calendar: BOJ meeting calendar. Consumption activity index: BOJ.

• 30-year Japanese government bond auctions of 3 September and 8 October 2026: Japan Ministry of Finance and Japan Ministry of Finance.

• People’s Bank of China measures of 29 September 2026: PBOC.

• Selic and USD/BRL: Banco Central do Brasil, SGS 432 and SGS 1.

• Brazil presidential election, first round of 4 October 2026: Tribunal Superior Eleitoral, official results.

• Monthly RRG-style methodology and numerical verification: Liquidity Desk calculations using the final available weekly adjusted price in each month for SPY and the 11 US sector ETFs.

Liquidity Desk | liquiditydesk.org

This is our August review of the US economy. All the main data for the month have now been released, and the last one included is the trade balance published on October 6. The annual update of the national accounts on September 30 rewrote part of the history. Every comparison here therefore uses the current versions of the series, not the numbers printed in our July issue.

Where the economy stands

The phase remains Acceleration. The car from our July issue is still picking up speed, but no longer in the same way. What is changing is what drives it.

The engine has moved from production to the consumer. Real consumer spending rose 0.6% in the month, retail sales 1.1%, and business activity in services reached 61.7 on the ISM survey. Industrial production, however, stalled at 0.0%, and manufacturing output shrank 0.3% after seven consecutive monthly gains.

The brake is in incomes. Real disposable income, meaning income after taxes and inflation, did not change, while the saving rate fell from 4.6% to 4.1%. Americans are increasing their spending faster than their incomes are growing and covering the difference out of savings. Hiring remains narrow, and housing is squeezed by mortgage rates around 6.67%.

Prices leave the Federal Reserve no room. Headline inflation is 3.4%, core PCE 3.0%, and producer prices accelerated to 5.4%. On September 16 the Fed raised its rate to 3.75-4.00%, unanimously.

Here is the paradox of the month. The Fed is pressing the brake, but the banks are hitting the gas. Bank loans to businesses grew by $46.7 billion in August alone. The bias toward tightening will stay until that gap closes.

Core indicators

The trajectory

Three indicators carry the story of the month.

The first is the saving rate. It is at its lowest level of the year, down from 5.6% in January to 4.1% in August. This is the line that explains the acceleration in spending. Real household spending is 2.6% above a year ago. Real disposable income is only 1.3% higher. The gap between those two numbers is covered by savings.

The second is credit to business. The annual growth rate of commercial and industrial loans accelerated from 4.2% in February to 9.5% in August, the fastest of the year. This is the indicator that runs against the Fed. More on it below, in the detail being underestimated.

The third is producer prices. They are slower than at the May peak but accelerating again against July, and they remain 2.0 points above consumer prices. The trajectory rule classifies this as a turn down. The direction is not reassuring, though, because the July relief did not last even a month.

The turns in retail sales and industrial production have the same shape. Against February they are stronger, against May weaker. That is not a decline. It is an economy that accelerated sharply in the spring and is now driving at a steady, if lower, speed.

Payrolls show the biggest move on paper, from a loss of 10,000 to a gain of 133,000. Here the mechanical rule misleads, because February was an abnormally weak month. The average gain for June to August is 51,000 a month, and for the past 12 months 45,000. More on this in the brake.

The engine

The engine in August has two parts. The bigger one is the consumer. The more durable one is investment.

Consumer spending is about two thirds of the US economy, so every move in it matters. In August real household spending rose 0.6%, the strongest month of the year. Nominal spending increased 0.9%. The gain was broad: both goods and services contributed. Retail sales rose 1.1% against an expected 0.8%, and business activity in services on the ISM survey jumped from 59.1 to 61.7, with new orders at 60.9.

How big is the engine really? In terms of pace, smaller than it looks. July was weak, with real spending up only 0.1%. The average for the two months is about 0.3% a month, slightly above the 0.24% average for the first half of the year. August is mainly catching up, not a new gear. But it also means something else. The consumer has not given up, despite mortgage rates around 6.7%, gasoline up 3.9% in a month, and rising interest rates.

Is it sustainable? This is the weak point, so it is covered under the brake. In short: spending is growing 2.6% a year with income growing 1.3%. That can continue only as long as there is something to draw on.

The second part is investment, and it is more durable. Orders for core capital goods, meaning machinery and equipment excluding defense and aircraft, rose 1.6% to $87.6 billion. They are 14.1% above last August. Imports of capital goods increased by $6.2 billion, of which $2.4 billion were semiconductors. Output of business equipment is 7.1% above a year ago, although it fell 0.5% in the month.

That equipment also leaves a trace in trade. The trade deficit widened from $92.8 billion to $105.6 billion, more than the expected $102 billion. Imports rose 4.3% and exports 1.4%. Part of the difference is equipment for the investment cycle. Another part is raw materials, including $3.3 billion of oil and $3.1 billion of nonmonetary gold. Gold passes through trade without being consumption or investment in the usual sense, and the statistics adjust for it when calculating GDP. The deficit is once again the invoice for investment, but this time it also contains gold. These nominal flows do not directly give the contribution of trade to growth in the third quarter.

The BEA annual update strengthened this part of the story. Real final sales to private domestic purchasers, which combine consumer spending and private investment, grew at a 4.6% annualized rate in the second quarter. Before the revision the figure was 4.2%. Headline GDP is 2.2%. The private machine is running more than twice as fast as the headline number.

Corporate profits supply the fuel. Profits from current production rose by $384 billion at an annual rate in the second quarter, to $4.71 trillion. Companies with cash flows like these can invest even at higher interest rates. And they are not alone: the banks are lending them more.

What would stop the engine? For the consumer, the answer is exhausted savings or a weaker labor market. For investment, the answer is a repricing. Fed Governor Michael Barr put it this way on September 29: if investors do not see the expected returns from building AI infrastructure, the readjustment of investment will hit growth both directly and through wealth. So far the orders say the opposite.

The brake

The brake has three parts: income, labor and housing. The first matters most, because it decides how long the engine can keep running.

Nominal personal income rose 0.2% and disposable income 0.3%. Prices ate the entire gain. Households covered the difference by setting aside less. This is growth borrowed from the months ahead. Not through new credit, but through savings that would otherwise have been kept for later.

Is this brake structural or cyclical? It is cyclical. The saving rate cannot fall forever. At 4.1% it is 1.5 points below January. When it stops falling, spending will have to grow at the pace of income. That is roughly half the current pace.

There is also a distribution that the aggregates hide. In the minutes of its September meeting, the Fed notes that rising stock prices support spending mainly among higher-income households. Lower- and middle-income households feel energy prices more, and energy is up 16.3% over the year. The average blends the two groups.

The second element is labor. August payrolls were revised from 162,000 to 133,000. That is still above the expected 56,000. The average for June to August, however, is 51,000 a month, and for the past 12 months 45,000. The composition remains narrow. Restaurants and local government education together added 83,000, or 62.5% of the monthly gain. The information sector lost 18,000 jobs, and professional and business services 5,000.

There is good news too. Labor force participation rose from 61.4% to 61.6%, and the labor force grew by 683,000 people. Unemployment remained at 4.1%. That weakens our July concern that unemployment was falling only because people were leaving the labor market.

The August JOLTS data, the official survey of job openings and labor turnover, add weight. Job openings fell from 7.34 million to 7.08 million. They are now almost exactly equal to the number of unemployed, 7.03 million. One open job for every person looking for work. The market is no longer overheated, but it is not yet weak. The hiring rate is 3.3%, the quits rate 1.9%, and the layoffs rate 1.0%. This is a market with low hiring and low firing. People are not losing their jobs, but they also find it hard to get a new one.

Is this part structural? Partly. Average hourly earnings growing 3.1% a year with inflation at 3.4% show that labor is not a source of inflation. The Fed minutes describe unusually low labor market churn and strong demand for AI-related specialists. That gap between the workers in demand and everyone else is more structural than cyclical.

The third element is housing. Housing starts fell 2.6% and permits 2.1%. New home sales rose 6.4%, but the uncertainty around that estimate is plus or minus 19.5 points, so there is no turn to speak of. The average 30-year mortgage rate was 6.67% in August. Private residential construction spending rose 1.1%, but the statistical uncertainty is large here as well. The new homes for sale would now run out in 8.5 months at the current sales pace, compared with 9.0 months in July. The market is not collapsing. It is standing still and waiting for cheaper money.

This brake is cyclical and sensitive to interest rates. It is also where the Fed’s tightening shows most clearly. After the end of the month it tightened further: on October 1 the mortgage rate was already 7.28%.

Prices

Prices in August say three different things depending on where you look.

Headline inflation is standing still. CPI is 3.4% for a second month, but the monthly increase of 0.4% is the largest in three months. Gasoline rose 3.9% in the month, and energy overall is 16.3% above a year ago. Without energy and food the picture is calmer: core CPI slowed to 2.4%.

The Fed’s preferred measure looks better than in July, but because of a revision, not because of August. The BEA annual update changed how some services are measured, among them software and portfolio management fees. July core PCE was cut from 3.3% to 3.0%. August is also 3.0%. Before the revision, Fed staff estimated that the new methodology alone takes about 0.2 points off the August core rate. Most of the improvement is in the ruler we measure with, not in inflation itself.

The three-month pace looks even better. Core PCE for June to August, annualized, is about 2.0%. Fed officials themselves warn in the minutes, however, that this measure usually understates inflation in the second half of the year.

The third thing is further up the chain. Producer prices accelerated from 4.8% to 5.4%. Final demand goods rose 1.1% in the month, energy goods 4.2%. The price indexes in the ISM surveys are 71.1 in manufacturing and 72.6 in services. A reading above 50 means more companies are paying higher prices than lower ones. Above 70 it is close to unanimous. Import prices, which do not include tariffs, are 7.0% higher than a year ago, and 5.5% higher excluding fuel.

The gap between producer and consumer prices is 2.0 points. It does not translate mechanically into future inflation, because the two indexes have different baskets. But it shows pressure that someone will absorb. Either companies through their margins, or customers through prices. The Fed minutes give a hint: some participants note that companies are passing costs on to consumers more successfully than before.

That is why the Fed has little freedom. Core PCE stands at 3%, headline inflation is above 3%, and pressure is coming from upstream.

The detail being underestimated

The Fed is tightening. The banks are loosening. Both are happening at the same time, and they are rarely put side by side.

IMAGE: us_august_2026_money_credit.png

Commercial and industrial loans, meaning bank loans to businesses, grew by $46.7 billion in August alone. Their annual growth rate accelerated to 9.5%, the fastest of the year. Over the same four weeks, bank reserves at the Fed fell by $59.6 billion. Money at the base of the system is shrinking, while banks are piling credit on top.

This continues the theme of our last two Global Liquidity Snapshot issues. In September we called bank credit the detour by which liquidity reaches the economy while central banks shrink their balance sheets. At the time the July monthly series showed corporate lending weakening. The benchmarking of bank statistics to bank regulatory reports on October 2 rewrote that story. The August data show that the detour is not only open, it is widening. The October issue already sees the first sign of a slowdown. In the four weeks to September 23, total bank loans and leases grew by $51.1 billion, against $108.4 billion in the previous four weeks.

Why is this happening? The minutes of the September meeting answer the question themselves. Several participants consider the current rate “not restrictive or only mildly restrictive”. Many participants describe financial conditions as supportive of growth. Several point out that credit is widely available and that banks have eased their standards. Our own calculation shows why. In August the upper bound of the policy rate minus core PCE gives a real rate of about 0.75 points. The middle of the range gives about 0.6. That is the price of money after inflation. It is lower than the headline rate suggests.

Credit also arrives by other routes. Large technology companies are financing the construction of AI data centers with bonds. According to the ECB, their capital spending through 2028 exceeds $1 trillion, about 3% of US GDP, and can no longer be covered by their own cash flows. They are issuing in euros as well. The Fed minutes note that this supply of debt competes with Treasuries and pushes long-term yields up. The Bank of Japan describes in a separate analysis the rapid expansion of private credit in the United States.

The Fed is holding the brake pedal. The banks, the bond market and private credit are holding the gas pedal. Tightening works where credit passes through the interest rate: mortgages and small business. Where companies have profits, a credit rating and access to markets, it is barely felt.

One qualification against our own reading. In July business loans fell by $9.4 billion. The average for July and August is about $19 billion a month, below the pace of the first half. The annual acceleration is real, but the monthly impulse is no stronger than in the spring. The signal is in the direction, not in a single month.

What it means

The August data confirm that growth is not weakening and inflation is not slowing. Our latest quarterly map put the United States in a regime of strengthening growth and accelerating inflation. Nothing in August moves the economy out of it.

The Fed’s answer came on September 16: a 25 basis point increase to 3.75-4.00%, by a vote of 12 to none. In July the three dissenters wanted exactly this. This time the whole committee joined them. According to the minutes, most participants expect one more increase before the end of the year. The direction is clear: tightening. The only question is how much.

Why is the Fed tightening when core PCE is 3.0% and not 3.3%? Because it is looking forward, not back. The minutes list energy, AI-related demand, and the danger that after more than five years of inflation above 2%, expectations could come unanchored. The increase is insurance. The Fed is not reacting to the latest number. It is reacting to the fact that credit and spending are not slowing.

For liquidity conditions the conclusion is twofold. Short-term rates and yields are rising: 10-year Treasuries averaged 4.68% in August and passed 5.3% in early October. Part of the rise is the term premium, the extra yield investors demand for holding long bonds, and it is linked to the hyperscaler debt described above. Market liquidity, however, remains ample. The high-yield bond spread, the premium risky companies pay over the government, averaged 2.70 points. That is low. Behind that average, though, there is a split. The gap between the weakest risky companies (CCC) and the sounder ones (BB) passed 10 points on October 1, according to our October issue. Rates are high, but most companies are still finding financing. For the weakest it is already very expensive: CCC-rated bonds were paying about 17% a year as of October 1. For companies with profits, this is an environment in which they can invest.

The Fed’s own balance sheet is barely moving. Between July 29 and August 26 it shrank by $7.3 billion. According to the minutes, the New York Fed has also stopped the purchases of short-term bills it used to keep reserves ample, because it sees no need for them. The reverse repo facility, the buffer that absorbed swings for years, is practically empty. That means every further refill of the Treasury’s account at the Fed comes directly out of bank reserves. For now, tightening comes from the price of money, not its quantity.

Ordinary people feel the other side. The mortgage rate has passed 7%. Gasoline is more expensive. Wages are growing 3.1%, slower than prices.

What the data already confirm: consumer spending and investment are holding up growth, prices further up the chain are accelerating, and bank credit is growing against the Fed. What remains an assumption: that savings will last until incomes catch up, and that tightening will reach credit before it reaches housing and hiring.

What we said last time

In July we said the economy was in Acceleration, with an investment and manufacturing engine and a transmission delivering power unevenly. The phase has held. The threshold we set for changing it, manufacturing new orders below 50 and a three-month average payroll gain near zero, was not reached. We said the Fed would hold rates with a readiness to raise them. It raised them on September 16. We said the July drop in retail sales might be calendar noise. The August gain of 1.1% confirmed that.

The case against our reading

The strongest number against Acceleration is industrial production. Zero for the month, a decline in manufacturing, and new orders down three points. Add the 29,000 new jobs in September and unemployment rising to 4.2%, and a different story emerges. The peak pace is behind us. The economy is entering Growth with slowing momentum, and August’s spending is a one-off catch-up after a weak July.

The second line of attack is against the brake. We read the fall in the saving rate as borrowing from the future. Another mechanism explains the same data. Barr describes it in his September 29 speech: when households expect higher future incomes and their wealth is rising, they quite rationally save less. Stocks have risen strongly this year. If that is the cause, low savings are not exhaustion but a new normal, and spending could hold up longer than we argue.

What would have to happen for us to be wrong about the phase? The first estimate of third-quarter GDP on October 29 would need to show private domestic demand below 3%, and real spending in September would need to fall. What would have to happen for us to be wrong about the brake? The saving rate would need to hold around 4% for several months without spending slowing.

Risks in both directions

The upside risk is that the investment cycle keeps surprising. The Fed minutes note that the scale and pace of AI infrastructure construction keep exceeding expectations. Orders for capital goods, 14% above last year, say the same. If the new equipment starts to raise productivity, growth can continue without the same inflation cost. Productivity is already growing 2.2% a year.

A second upside risk would be energy, but for now it is unlikely. As long as the conflict around Hormuz is active, a lasting drop in fuel prices is not realistic. If a lasting peace does come, headline inflation will fall quickly, real income will recover, and households will be able to spend without drawing on their reserve.

The downside risk is twofold. The first is savings running out at the same time as hiring weakens. The 29,000 jobs in September are the first signal in that direction. If incomes do not catch up, spending will have to slow abruptly. The second is that the Fed keeps tightening until it reaches business credit. Tightening hits first where credit is most sensitive to interest rates: housing and small business. They are already weak.

The most unpleasant scenario is a repricing of AI expectations just as savings are exhausted. Then both engines stop at the same time, while inflation from energy and imports stays and does not allow quick relief.

What to watch

On October 14, September CPI will show whether core stays around 0.3% a month and how much of the September rise in fuel prices reaches consumer prices. Another number like that would reinforce expectations of a hike in October. On October 15, retail sales will test whether the August gain is a lasting level or a catch-up. A drop below zero would confirm our reading of spending on borrowed savings. The same day, producer prices will show whether the gap with consumer prices is widening beyond 2 points.

On October 27, new home sales will show how the market is coping with mortgage rates above 7%. On October 27 and 28, the Fed decides whether to raise rates again. On October 29 come the first estimate of third-quarter GDP and September PCE. A saving rate below 4% would mean the reserve is still being drained.

The threshold for changing the phase is manufacturing new orders on the ISM survey below 50 together with a three-month average payroll gain near zero. Both together would move our assessment from Acceleration to Growth with slowing momentum. The average for July to September is 51,000.

Closing

The US economy keeps accelerating, but the fuel for that acceleration increasingly comes from household savings and bank credit rather than from incomes. The Fed is pressing the brake. The banks are still hitting the gas, while the market is already choosing whom it finances.

We do not know how long the reserve will last. Nor do we know which will prove stronger: the Fed’s brake or the banks’ gas.

Sources

• BEA, Personal Income and Outlays, August 2026

• BEA, GDP third estimate, corporate profits and annual update, Q2 2026

• BEA and Census Bureau, U.S. International Trade in Goods and Services, August 2026

• BLS, Employment Situation, August 2026

• BLS, Employment Situation, September 2026, with August revision

• BLS, Employment Situation, Table B-1

• BLS, Job Openings and Labor Turnover, August 2026

• BLS, Consumer Price Index, August 2026

• BLS, Producer Price Index, August 2026

• BLS, Import and Export Prices, August 2026

• BLS, Productivity and Costs, Q2 2026

• Census Bureau, Retail Sales, time series API

• Census Bureau, Manufacturers’ Shipments, Inventories and Orders, August 2026

• Census Bureau, New Residential Construction, August 2026

• Census Bureau, Building Permits Survey

• Census Bureau, New Residential Sales, August 2026

• Federal Reserve, Industrial Production and Capacity Utilization, G.17, September 18, 2026

• Federal Reserve, H.8 Assets and Liabilities of Commercial Banks, October 2, 2026

• Federal Reserve, H.8 Notice to Users

• Federal Reserve, H.6 Money Stock Measures, September 22, 2026

• Federal Reserve, July 2026 Senior Loan Officer Opinion Survey

• Federal Reserve, FOMC statement, July 29, 2026

• Federal Reserve, FOMC statement, September 16, 2026

• Federal Reserve, Minutes of the FOMC, September 15-16, 2026

• Federal Reserve, Governor Barr, Economic Conditions and Monetary Policy, September 29, 2026

• ISM, Manufacturing PMI, August 2026

• ISM, Services PMI, August 2026

• ECB Blog, Big tech, big debt, August 31, 2026

• Bank of Japan, The Expansion and Diversification of Private Credit Funds, September 25, 2026

• FRED, C&I loans, BUSLOANS

• FRED, Bank credit, TOTBKCR

• FRED, Reserve balances, WRESBAL

• FRED, 30-year mortgage rate, Freddie Mac PMMS

• FRED, 10-year Treasury yield

• FRED, ICE BofA US High Yield OAS

• Trading Economics, US economic calendar, market consensus

Not financial advice.

This is our second monthly Macro Pulse for China. All the main August releases are now out. The last to arrive were industrial profits, on 29 September. This month we watch two doors. The factory gate: production, exports and loans to companies. And the front door of the home: what families buy, the housing market and household borrowing. In August the first one opens. The second stays almost shut.

Where the economy stands

The phase in August remains growth with slowing momentum. August repairs July but does not erase it. Industrial production accelerated to 5.2% from a year earlier, from 4.5% in July. Dollar exports grew 25.0%. The official composite PMI, however, is 49.5, below 50 for a second month running. The PMI is a survey of purchasing managers in which a reading below 50 means more firms report deterioration than improvement. This is a rebound within the phase, not a new phase.

The engine is the same as in July: electronics and cars, sold abroad. High-tech manufacturing grew 16.7%. What is new is that companies are taking out long-term loans again, after repaying them in July.

The brake is also the same: the home. Retail sales grew 0.4%. Households repaid about CNY 203bn more than they borrowed. Property investment is 19.9% lower over eight months. The brake is structural, and one good month in the factories does not release it.

The central bank, the PBoC, has no reason to tighten. Consumer inflation is 0.8% and core inflation 1.0%. The rate stays put and easing remains targeted. The real cost of money even fell slightly in August, not because of the PBoC but because inflation rose.

Core indicators

The PBoC’s August credit data are preliminary. NBS and RatingDog are two different surveys. NBS covers more large and state-owned enterprises, RatingDog more private and export-oriented firms, so the two are not compared directly.

The trajectory

For production, retail sales and trade, the February column is the combined January and February figure, which is how Chinese agencies publish them. For CPI and core CPI it is the official average of the two months, because Chinese New Year distorts each of them on its own.

Three lines carry the story.

The first is industry. It is above May but below the start of the year. What matters is that the rebound is not just a convenient base effect. Seasonally adjusted, production grew 0.54% from July, against just 0.11% a month earlier. Factories really are producing more.

The second is retail sales. The table shows them turning, because they are above a negative May. The monthly number says otherwise: seasonally adjusted, August sales were 0.13% lower than in July. Annual growth is positive, but spending is not rising. Factories are speeding up, shops are standing still.

The third is credit. Total social financing, the broadest measure of money flowing to the economy, slows at every point in the table: 8.2%, 7.7%, 7.4%, 7.2%. M2, the broad money in the system, fell from 9.0% to 7.5%, and bank loans in yuan from 6.0% to 4.9%. All three lines show the same slope. This is the same six-month slowdown we described in July. Under the headline number, though, something new happened in August. We come back to it in the section on the overlooked detail.

The surveys need a caveat. RatingDog improved from July in both factories and services, but remains below February and May. That is why its direction is “slowing”. Recovering lost speed is not the same as a new trajectory. Spring was stronger than August.

On prices, the big jump is already behind us. PPI, the prices at which factories sell their output, was -0.9% in February and 3.9% in May. The whole rise happened in spring. From May to August the index barely moved and now stands at 3.8%. Consumer prices made no such jump. CPI is 0.8%, the same as the January-February average, and core CPI is 1.0%, below the winter 1.3%. Over six months, factory prices went up and shop prices did not.

The only lines accelerating against both earlier points are exports and imports. That is August’s strength, so we start there.

The engine: chips, cars and the neighbours

China’s factory is growing at two speeds. Industry as a whole added 5.2%. High-tech manufacturing added 16.7%, computers and electronic equipment 17.2%, and output of industrial robots was 34.6% higher than a year earlier. At the other end, cement fell 11.7% and oil refining 6.9%. The factories that sell to the world and to other companies are growing. The ones that sell to building sites are shrinking.

How big is the engine? In August China exported goods worth USD 401.4bn and imported USD 282.4bn. The surplus was USD 119.1bn in a single month. From January to August, just three groups, integrated circuits, computers and their parts, and cars, brought in USD 582.5bn of exports. That is about a fifth of all goods exports. It is not a share of GDP, because every exported chip contains imported components. But it is enough to move the entire trade statistics.

Chips repeat July’s story, only more strongly. Customs publishes both the value and the quantity of exports for its main products. The value of exported integrated circuits from January to August was USD 256.8bn, 103.9% more than a year earlier. The number of chips for the same eight months was only 4.1% higher. The average value of an exported chip is about 96% higher. In July the gap was 88%. Part of this is a shift to more expensive products rather than the same chip at double the price. China is still exporting dearer chips, not many more chips.

Cars are the opposite case. Car exports from January to August were USD 129.1bn, up 53.2%, and the number of cars exported rose 51.2%. Here the growth is almost entirely volume. At home the picture is different. Car output in August was 2.7% lower, and output of saloon cars 22.7% lower. New energy vehicles grew 21.9%. Value added in the car industry still rose 8.7%. China is making fewer cars, but with more value and more of them electric, and it is selling more and more of them abroad.

The currency adds a caveat. In yuan, August exports grew 18.6%; in dollars, 25.0%. The difference comes from a stronger yuan than a year ago. Neither number is growth in the units shipped.

The geography has not changed. From January to August, exports to the United States grew 6.2% and make up 10.3% of the total. Exports to Hong Kong grew 50.4%, to South Korea 35.3% and to Southeast Asia 25.8%. China is selling more and more to its neighbours, who are part of the same supply chain for servers and electronics.

Is the engine sustainable? The surveys say “for now, yes”. The official PMI also asks factories whether they are receiving more or fewer new orders from their customers than a month earlier. In August this index jumped 2.1 points to 50.6, meaning more factories reported rising orders than falling ones. Orders from abroad alone returned above 50, to 50.1. Profits are more cautious. Cumulative growth over eight months is 15.7%, but in August alone it was 4.2%, after 11.2% in July. Profits of electronics makers more than doubled from January to August, up 110%. The engine is real, but its profit is concentrated in a narrow circle of companies.

There is also something the data cannot show. For chips and cars we know how much of the growth is price and how much is volume. For exports as a whole we do not. The overall price and volume indices, which showed in July that two thirds of export growth was price, have not yet been published for August. Nor do we know whether some orders are front-loading, where a customer stocks up now so as not to buy later.

What could stop the engine? A fall in chip prices would cut the value of exports quickly, without China losing a single customer. Measures against Chinese cars in the markets they are entering. And a stronger yuan, which makes Chinese goods dearer for foreign buyers.

The brake: the home

Retail sales grew 0.4% from a year earlier. Excluding cars, growth was 2.5%. Car sales at larger retailers fell 18.5%. One item drags the total down, but it does not explain everything.

The breakdown shows exactly where Chinese households are not spending. Furniture sales were 7.9% lower and building materials 11.8% lower. Telecom equipment grew 27.3%, supported by state subsidies for trading in old appliances for new ones. Sales of services from January to August grew 4.9%, against 1.0% for goods. Chinese households buy phones and pay for services, but save on everything that goes into a new home.

Property remains the deepest part of the brake. From January to August, property investment was 19.9% lower, new construction starts 24.8% lower and floor space sold 12.1% lower. These are cumulative figures for eight months, not a fall in August alone. But all three are slightly worse than the January to July figures. The bottom of construction is not yet in sight.

Home prices give the first better signal. In the four largest cities, new and second-hand homes rose 0.1% from July. In Shanghai, new homes are 3.0% dearer than a year earlier. Outside the first tier, however, prices keep falling every month. From a year earlier, new homes in second-tier cities are 2.7% cheaper and in third-tier cities 4.1% cheaper. For second-hand homes the falls are 4.9% and 5.6%. Only 21 of the 70 cities tracked show a monthly rise or no change in new home prices, two fewer than in July, and only 8 for second-hand homes, the same as in July. The bottom is visible in Shanghai, not in the provinces.

Credit confirms the hesitation. Households repaid a net CNY 203bn or so in August, less than CNY 460bn in July, but still more than they borrowed. Their short-term loans, which often finance purchases, fell by about CNY 122bn. Long-term loans, which include mortgages, fell by about CNY 82bn. When the price of a home is falling, a household does not take out a mortgage, even at a low rate.

The weakness is wider than property. Total fixed-asset investment over eight months is 7.2% lower. Excluding property, the fall is 4.2%. Private investment fell 10.1%, and 6.4% excluding property. Infrastructure is 4.0% lower. The exception, again, is equipment, up 9.3%. Companies are replacing their machines, not expanding.

The labour market is getting slightly heavier. Urban unemployment is 5.3%, 0.1 points above July. The employment index for services and construction in the official PMI is 45.4, well below 50.

Is the brake structural? Yes. In July we could blame part of the weakness on floods and heatwaves. August has no such excuse. In services and construction the picture is the reverse of the factories. The new orders index there is 44.1: many more firms are receiving fewer orders from their customers than more. In construction it is just 42.4, and construction activity itself is 46.9, almost the same as in July. The rain stopped, but the new contracts did not come back. The brake will loosen when families believe their home will not lose more value. Shanghai is the first place where that may be happening.

Prices: factories get dearer, shops barely move

Consumer inflation returned to 0.8% from a year earlier, from 0.5% in July. In the month alone, prices rose 0.4%. Part of the jump is seasonal: fresh vegetables rose 5.5% from July. From a year earlier, food remains 1.4% cheaper and pork 11.8% cheaper. Excluding food, prices are 1.2% higher.

Core CPI, excluding food and energy, is 1.0%, after 0.9% in July. It is the better measure of domestic demand. It is lower than the winter average of 1.3%. A one-month rise does not reverse the medium-term picture. One item stands out again: the “other goods and services” group, which includes jewellery, is 7.3% dearer. That is more the price of gold than pressure from demand.

PPI grew 3.8% from a year earlier and 0.4% in the month. Mining prices are 17.8% higher and raw materials 6.7% higher. The prices factories pay for their inputs rose 5.8%. On consumer goods, however, factories are selling 0.5% cheaper. Costs climb up the chain but do not reach the shelf.

The gap between PPI and CPI remains 3.0 points, as in July. This gap shows where the pressure is: on producers of goods for the domestic market. The two indices measure different baskets, though, and the gap is not a direct measure of margins.

The surveys show that in August some factories began to pass costs on. In the official PMI, the index of raw material prices rose from 53.2 to 56.6. The index of selling prices jumped from 47.8 to 50.4, above 50 for the first time since May. The private RatingDog survey says the opposite: costs are rising, and firms cut their selling prices for the first time in 2026. The two surveys do not contradict each other. Larger firms can raise prices. Smaller private and export firms cannot.

Profits confirm that the pressure is sector by sector, not everywhere. Industry’s overall profit margin from January to August is 5.66%, 0.44 points higher than a year earlier. Electronics makers earn 110% more. Carmakers earn 16.0% less, power producers 15.1% less, and makers of cement, glass and other non-metallic minerals 46.7% less.

The conclusion on prices is the same as in July, but milder. Domestic inflation is low and does not constrain the PBoC. Factory inflation is high, but larger firms are starting to pass it on.

The overlooked detail: companies are borrowing long again

The credit headline for August is dull. Bank loans in yuan grew by about CNY 60bn in the month, almost nothing for an economy the size of China’s. Their annual growth slowed to 4.9%. Under that near-zero number, however, are two opposite flows.

The first is at companies. Long-term loans to enterprises and organisations rose by about CNY 320bn. In July they fell by CNY 230bn. Short-term corporate loans fell by about CNY 160bn, and bill financing, where a bank pays a company early for a receivable from a trade deal, added CNY 100bn, after CNY 376bn in July. Companies are swapping money for day-to-day payments for money that lasts years.

The second flow is at households, which repaid about CNY 203bn. The two flows almost cancel out in the total. That is why the headline misses it.

The same shift shows up in total social financing. In August it added CNY 1.66trn. CNY 1.01trn of that was government bonds, or 61%. In July the government’s share was 94%. Bank loans to the real economy in this measure turned positive at CNY 55bn, after minus CNY 590bn in July. Our September liquidity piece described July’s credit as a broken detour: the state was creating financing, and private credit was not following. August shows the first stretch where the route reconnects.

Why is this overlooked? Because the total has not changed, but its composition has. A long-term loan is a loan for a machine, a plant or a project. A short-term loan or a bill is money for wages and suppliers.

There are four caveats. The monthly flows are differences between the PBoC’s rounded cumulative figures, and the data are preliminary. The corporate category also includes organisations such as public institutions, so not all the new debt goes to private companies. A long-term loan can refinance old debt instead of buying a new machine. And one month does not make a credit cycle. If the improvement repeats in September and reaches households, we will have to change our reading of the brake.

What this means

The August data give the PBoC a mixed but readable picture. Growth is recovering in the factories and staying weak at home. Consumer inflation is low, factory inflation high. The last confirmed quarterly position was Q2: growth weakening and consumer inflation easing, with higher factory prices. August does not contradict it. The lean remains towards easing, but targeted.

The rate is not moving. The one-year LPR, China’s benchmark lending rate, has been 3.0% since December 2025. The five-year rate, the reference for mortgages, is 3.5%. The real rate, the LPR minus current inflation, fell from 2.5 points in July to 2.2 in August. July’s passive tightening has stopped. The reason is higher inflation, not action by the PBoC.

Quantities look tighter than they are. In August the PBoC drained a net CNY 409bn or so through its operations. Bank reserves at the central bank fell by CNY 311bn. This time the state did not park large sums in its account at the PBoC; its deposits grew by only CNY 30bn. The interbank rate, however, fell from 1.40% to 1.38%. If banks needed money, they would pay more to borrow it from one another. Banks have money. The problem is who wants it.

After the end of the month, the PBoC acted along the same lines. On 29 September it cut the rate on PSL, its targeted loans to the state development banks, from 1.75% to 1.50%. It raised the credit line for technology by CNY 200bn, to CNY 1.4trn, and raised the share of banks’ loans it refinances under that line from 60% to 100%. It added CNY 500bn for agriculture and small businesses, of which CNY 300bn is for private firms. These are limits, not money spent. They become a flow when banks use them.

Closer to home is the mortgage subsidy. From 1 October the state covers 1 percentage point a year of the interest on a new loan for a first home, for up to five years. The home must be no larger than 120 square metres and cost no more than CNY 1.5mn, and the subsidy covers a loan of up to CNY 1mn. On a CNY 1mn loan that is up to CNY 10,000 a year. For now the programme accepts new loans for one year. For a family that already wants to buy, this is a big discount: close to a third of the five-year benchmark. For a family afraid that its home will lose more value, it may not be enough.

What does this mean for liquidity? It is not missing in the banks. It is starting to reach companies, but not yet households. The mortgage subsidy is the first measure aimed squarely at that link. The first test will be October credit, published in November. September’s data cannot show an effect yet.

For an ordinary person in China, August looks much like July. Food is cheaper than last year. Their home, unless it is in Shanghai, is losing value. Work is slightly harder to find. Credit is getting cheaper, but they still do not want it.

What is confirmed? That factories are producing more, that companies are borrowing long-term again and that households keep repaying debt. What remains an assumption? Whether the corporate loans will become investment. Whether the subsidy will bring families back to mortgages. And how much of export strength is price.

Risks in both directions

The strongest argument against our phase comes from September. The official manufacturing PMI is 50.1, the composite 50.7, and new factory orders 50.5. In our Q2 review we set a composite PMI above 50 and new orders above 50 as the threshold for a change. The September surveys meet it. RatingDog manufacturing is 52.1. The RatingDog composite rose to 52.4, and services to 51.6. Add corporate credit, the first price rises in the big cities and the new mortgage subsidy, and the path to acceleration is clear. If the September hard data confirm the surveys, August will turn out to be the start of a new phase that this piece reads cautiously.

The downside risk has four sources. The first is chip prices. If they fall, China’s strongest number, exports, will lose speed with no warning in the unit counts. The second is front-loading. If some of August’s orders were stock-building, the autumn will give it back. The third is costs. In our September Commodity Snapshot, Brent crude is above USD 100. That hits exactly the RatingDog firms that cannot raise their prices. The fourth is credit. If the new corporate loans turn out to be refinancing, and households keep repaying debt, the factory rebound will stay separate from the home.

What to watch

On 14 October, September CPI and PPI will show whether core CPI holds at 1.0% or above. If it falls below 0.8%, pressure for a rate cut grows. September trade comes out the same day. It matters whether customs also publishes the August price and volume indices.

Around mid-October the PBoC publishes September credit data. The test has two parts: a positive household flow, and a second month of growth in long-term corporate loans.

On 19 October come Q3 GDP, the September monthly data and home prices. The threshold for a shift to acceleration stays as set in our Q2 review: GDP growth of at least 4.5% with stronger consumption. If retail sales stay below 1%, the threshold will not be met, whatever the surveys say. On housing, we watch whether the number of cities with no fall in new home prices rises above 21.

On 20 October the LPR is set. A cut would show the PBoC moving from targeted measures to cheaper money across the board.

On 31 October, the official October PMIs. The threshold for stagnation is a composite PMI below 50 for two consecutive months, together with production growth below 4%. July and August were two months below 50, but production stayed above 4%, so the threshold is not met.

Closing

China is leaving July’s weakness behind through the factory, not through the home. Exports, high-tech production and new long-term corporate loans make the rebound real. Property and household debt keep it narrow. The factory gate is open. The front door is ajar only in Shanghai. We do not know whether the corporate loans will become investment, or whether the mortgage subsidy will bring families back to buying. The answer starts to arrive with the credit data in November.

Sources

• NBS, Purchasing Managers’ Index for February, May, August and September (Chinese) 2026

• S&P Global, final RatingDog China General Manufacturing PMI for February, May, August and September 2026

• S&P Global, final RatingDog China General Services and Composite PMI for February, May, August and September 2026

• NBS, National Economy in August 2026

• NBS, Industrial Production Operation for January-February, May, July and August 2026

• NBS, Total Retail Sales of Consumer Goods for January-February, May, July and August 2026

• NBS, Consumer Price Index for February, May, July and August 2026

• NBS, Producer Price Index for February, May, July and August 2026

• NBS, Investment in Fixed Assets, January-August 2026

• NBS, Investment in Real Estate Development and Sales, January-August 2026

• NBS, Sales Prices of Residential Buildings in 70 Cities, August 2026: tables and commentary by city tier (Chinese). The count of cities with no monthly fall is our own tally from the tables.

• NBS, Profits of Industrial Enterprises, January-August 2026

• NBS, GDP for Q2 2026 and Industrial Capacity Utilization for Q2 2026

• GACC, total trade in USD for January-February, May, July and August 2026, and total trade in CNY for August

• GACC, Major Export Commodities, August 2026. The average value per exported chip is our own calculation.

• GACC, Trade by Country and Region, August 2026

• GACC, Trade Indices list, checked on 2 October 2026

• PBoC, Financial Statistics Report for July and August 2026 (Chinese). Monthly flows by sector are our own differences between rounded cumulative figures.

• PBoC, Financial Statistics Report for February and May 2026, full text on the official financial portal of Heilongjiang province, citing the PBoC (Chinese)

• PBoC, total social financing, monthly flow and stock

• PBoC, Balance Sheet of Monetary Authority, to August 2026 and liquidity operations in August 2026 (Chinese). The net sum across instruments is our own calculation.

• ChinaMoney, LPR of 20 August 2026 and LPR history (Chinese)

• PBoC, targeted policy tools of 29 September 2026 (Chinese)

• Ministry of Finance, PBoC and NFRA, mortgage interest subsidy from 1 October 2026 (Chinese)

• NBS, 2026 release calendar and MOFCOM, official copy of the 2026 customs release calendar (Chinese)

For information and analysis only. Not financial advice.

Liquidity Desk | liquiditydesk.org

Introduction

In September the Nasdaq 100 rose 3.23%. The Russell 2000, the index of smaller US companies, lost 5.40%. That is a gap of 8.6 percentage points in a single month. The Hang Seng fell 3.73%, the Shanghai Composite 3.61% and the broad European STOXX Europe 600 2.49%. This was not a crash. The S&P 500 ended the month almost unchanged. But the market split into a small group that kept moving forward and everyone else, who lost ground.

The cause starts with three decisions. On 10 September the ECB raised its deposit rate by 25 basis points. On 16 September the Fed did the same. On 18 September the Bank of Japan followed. Three hikes in eight days. Warren Buffett has an old line for this: only when the tide goes out do you discover who has been swimming naked.

The tide is already going out. Excess liquidity in the euro area fell by €33.0 billion during the month. Japan’s monetary base shrank by ¥10.2 trillion between August and September alone. The balance sheet of China’s central bank was CNY 350.1 billion smaller in August. And after month end, the gap between the weakest and the stronger high-yield bonds in the US moved above 10 percentage points. The question this month is whether the water is pulling back evenly or only from some shores. And which boats are already touching bottom.

Central Banks

Three of the four major central banks raised rates within eight days. The quantities beneath those decisions are not moving in the same way. The Fed is tightening almost entirely through the price of money. The Eurosystem and the Bank of Japan are tightening through both price and quantity. China’s central bank is moving the other way on price, but not on quantity.

Federal Reserve

The Fed balance sheet stands at $6.743 trillion as of 30 September, against $6.731 trillion on 26 August. The increase is $12.1 billion. Treasury securities within it rise by $18.0 billion, because the Fed rolls over maturing Treasuries and reinvests principal from mortgage-backed securities into Treasury bills. This is not new QE. The balance sheet is effectively neutral.

On 16 September the Fed raised the target range by 25 basis points to 3.75 to 4.00%. The decision was unanimous, 12 to 0. At the previous meeting in July, three members voted against holding because they wanted a hike. This time there was not a single dissent. Interest on reserve balances, the rate the Fed pays banks on their reserves, is 3.90% from 17 September. Quantities barely move. Net Fed Liquidity rises by $14.3 billion over the month (details in the Net Fed Liquidity section). In the US, the ebb comes from the price of money, not from a shrinking balance sheet.

Regime: tightening through price, neutral on quantity. The Fed is making money more expensive without withdrawing assets.

European Central Bank

The Eurosystem balance sheet falls from €5.915 trillion to €5.897 trillion between 28 August and 25 September. Of the €18.1 billion decline, €19.8 billion comes from securities held for monetary policy purposes. The APP and PEPP portfolios continue to shrink because maturing principal is no longer reinvested. No revaluation distorts this comparison. The latest published weekly balance sheet is as of 25 September, before the quarterly revaluation of gold and foreign currency at the end of September. That revaluation will show up in the first October balance sheet.

Excess liquidity confirms the direction. It falls from €2.142 trillion on 31 August to €2.109 trillion on 30 September, a decline of €33.0 billion (-1.54%). Last month it was almost stationary. Now it is moving. On 10 September the ECB raised all three rates by 25 basis points, effective 16 September. The deposit facility rate is 2.50%, the main refinancing rate 2.65% and the marginal lending rate 2.90%. The published decision does not give a numerical vote.

Regime: tightening. A higher price of money and excess liquidity that is genuinely falling.

Bank of Japan

The Bank of Japan balance sheet falls from ¥644.662 trillion on 31 August to ¥641.281 trillion on 20 September, a decline of ¥3.381 trillion. The full month-end figure for 30 September is published on 7 October. What matters is what sits beneath the balance sheet. Holdings of government bonds actually rise slightly, by ¥0.375 trillion. But banks’ current account deposits at the central bank fall by ¥14.069 trillion over the same three weeks. Current account deposits are bank reserves, the most liquid money in the system. Banks use them to settle payments with each other and to buy bonds. The credit they extend also rests on them. In Japan they make up more than three quarters of the entire monetary base: ¥416.6 trillion out of ¥535.4 trillion at the end of September. When they fall, banks have less free money, even if the central bank’s total balance sheet looks almost unchanged. The balance sheet also shows where the money went. About ¥7.3 trillion moved into the government’s account at the Bank of Japan, the equivalent of the US Treasury General Account. Banks repaid another ¥3.4 trillion under the BoJ programme that gave them cheap loans to support lending. About ¥3.0 trillion was absorbed through repo operations, in which the central bank hands over bonds against cash and commits to buy them back. Part of the ebb is temporary, part is not. Money in the government’s account can return to banks when the state spends it. The repo is short term. The repaid programme loans, however, do not come back on their own.

The monetary base, the independent check, confirms the move for the full month. Its average level falls from ¥543.007 trillion to ¥532.846 trillion, a drop of ¥10.161 trillion. The annual decline is 15.2%, against 15.7% in August. A smaller annual decline is not an improvement. The base keeps shrinking from month to month, and the end-September level is ¥8.344 trillion below the end-August level. On 18 September the policy rate was raised by 25 basis points to 1.25%, effective 24 September. The vote was 7 to 2. The two dissenters, Toichiro Asada and Ayano Sato, opposed the hike.

Regime: tightening. The rate is rising, and the monetary base is shrinking on both the monthly average and the month-end level.

People’s Bank of China

The latest published PBoC balance sheet is for August. Assets fall by CNY 350.1 billion to CNY 49.857 trillion. The main reason is a CNY 508.1 billion drop in claims on banks, meaning the central bank has pulled back part of the loans it had extended to them. This time the balance sheet does not mislead. Reserve money falls by CNY 308.9 billion and bank deposits at the PBoC by CNY 310.6 billion. Government deposits rise by only CNY 30.5 billion. The contraction is real and reaches bank reserves. It matches the net CNY 408.6 billion withdrawal through August operations reported in the previous issue.

In September the direction on price reverses. On 29 September the PBoC cut the rate on its one-year pledged supplementary lending, its targeted lending to selected sectors, by 25 basis points to 1.50%. The quota for technology relending rises by CNY 200 billion to CNY 1.4 trillion. The quota for agriculture and small business rises by CNY 500 billion to CNY 4.85 trillion. Of that, CNY 300 billion is reserved for private firms. The new capacity totals CNY 700 billion. That is a quota, not credit already extended. Whether it is used will show up in the September and October credit data. The one-year loan prime rate remains at 3.0% and the five-year rate at 3.5%.

Regime: quantitative tightening in August, targeted easing in September. The price of targeted funding is falling, but the total quantity of liquidity in the latest measured month is shrinking.

Central Bank Synthesis

Money and Credit Transmission

In the previous issue, bank credit was the detour that kept the system moving without help from central banks. The monthly money and credit data are for August, before the three hikes, and they show that this route is still open. For September there are so far only the weekly US bank data through 23 September, which cover one week after the Fed decision. The September monthly data arrive during October, first in Japan on 13 and 14 October, then in the euro area and the US on 27 October. They will be the first real test.

United States

US M2 reaches $23.343 trillion in August, after a monthly increase of $124.9 billion. That is a rise of 0.54% for the month and 5.66% over the year. Commercial and industrial loans rebound strongly. They grow 1.62% in August to $2.935 trillion, the strongest monthly gain in the last five months. The annual rate is 9.52%.

The July decline has been revised. Last month it was -0.20% at a level of $2.899 trillion. Now it is -0.33% at $2.888 trillion. The revision comes from benchmarking the weekly data against banks’ quarterly reports for June, included in the 2 October release. The weekly data run to 23 September. Between 26 August and 23 September, bank loans and leases rise by $51.1 billion and deposits by $75.5 billion. Total bank credit adds only $18.6 billion, because banks reduce their securities holdings by $32.5 billion. The pattern is the same as in August: more loans, fewer bonds. But the pace is half of August’s $108.4 billion.

Euro Area

M3 rises by €31.4 billion in August to €17.646 trillion. The annual rate accelerates from 3.4% to 3.5%. Credit to companies is slowing. Adjusted loans to non-financial corporations grow 4.2% over the year, against 4.4% in July. Household loans hold at 3.1%. The adjusted series remove the effect of loans sold and securitised and match the growth rates the ECB publishes.

The bond channel is closing. Net debt issuance by non-financial companies is €9.6 billion in June, €7.2 billion in July and -€0.2 billion in August, meaning companies repaid more than they issued. Bank credit is still growing, but it is now almost the only open channel. This does not prove that companies have swapped bonds for bank loans. It shows that the second channel has stalled. The latest ECB bank lending survey covers the second quarter. A net 7% of banks tightened standards for companies, and a net 9% for mortgages. The next survey is published on 27 October and will be the first after the hike.

China

China’s M2 recovers after the July decline. It rises by CNY 1.30 trillion in August to CNY 356.81 trillion, or 0.37% for the month. The annual rate keeps slowing, however, from 7.7% to 7.5%. Total social financing, the broadest measure of credit reaching the real economy, adds CNY 1.658 trillion in August, against CNY 1.407 trillion in July. Government bonds provide CNY 1.010 trillion, or 60.9% of the flow. In July their share was 93.7%. August credit is broader than July’s, but the state remains the main channel. Renminbi loans to the real economy add only CNY 55.2 billion. Year to date, total social financing is CNY 23.91 trillion, CNY 2.64 trillion less than a year earlier.

Renminbi bank loans grow by around CNY 60 billion for the month. Beneath that figure, the two private sectors move in opposite directions. Companies and organisations add around CNY 260 billion, while households reduce their borrowing by around CNY 203 billion. Annual growth in bank loans slows from 5.1% to 4.9%. Households keep paying down debt while the state borrows. The new targeted quotas from 29 September are trying to change exactly that, but the August data predate them.

Japan

Japan’s M2 stands at ¥1,296.4 trillion in August, up 2.0% over the year. M3 is ¥1,640.0 trillion, up 1.2%. Without seasonal adjustment, M2 falls by ¥0.2 trillion against a revised July. On seasonally adjusted levels it rises by around 0.11%. On either measure, broad money is effectively flat. Lending by major, regional and shinkin banks reaches ¥680.3 trillion, up 5.4% over the year. At the major and regional banks alone, growth is 5.8%.

Japan remains the clearest example of the offset. The monetary base is shrinking 15.2% over the year, while bank lending grows 5.4%. The central bank is draining the water, and commercial banks are still putting it back. The September lending data arrive on 13 October and will show whether this survived the hike.

Money and Credit Synthesis

Net Fed Liquidity and Its Relationship to M2

Net Fed Liquidity is the Fed balance sheet minus the Treasury General Account and overnight reverse repo. It shows how much dollar liquidity is not locked in the government’s account or parked back at the central bank.

Net Fed Liquidity

As of 30 September, Net Fed Liquidity is $5.794 trillion, against $5.780 trillion on 26 August. The increase is $14.3 billion, or 0.25%.

Almost the entire increase comes from the Fed balance sheet. The Treasury account ends the month almost where it started. Reverse repo is practically at zero and has nothing more to give back. The monthly comparison hides a large move in the middle of the month. Net Fed Liquidity reaches $5.869 trillion on 16 September, the day of the Fed decision. Then quarterly corporate tax payments in mid-month lift the weekly average of the Treasury account from $877.0 billion to $977.1 billion in a single week. Net Fed Liquidity falls to $5.770 trillion on 23 September, before recovering to $5.794 trillion at month end.

The reverse repo buffer remains exhausted. When the Treasury account grew by $100 billion in one week, there was no parked cash to absorb the blow. With reverse repo empty, a build-up like that is paid for out of bank reserves.

M2 vs Net Fed Liquidity

The comparison covers the same twelve-month window, August 2025 to August 2026.

The annual decomposition of Net Fed Liquidity: the Fed balance sheet adds $127.5 billion, the drain of reverse repo adds another $31.5 billion, and the higher Treasury account removes $360.7 billion. The three parts sum exactly to -$201.7 billion. Over one year, the Treasury has drawn almost three times as much out of the system as the Fed balance sheet has added. This is not an accounting measure of privately created money. M2 is a monthly average, Net Fed Liquidity is a weekly snapshot, and the two series measure different things. The divergence has a narrower meaning. Broad money is growing, while the federal liquidity plumbing is smaller than it was a year ago.

Read more

On 7 July 2011 the European Central Bank raised its main interest rate for the second time in three months, from 1.25% to 1.50%. The reason was oil. Brent averaged $117 a barrel, energy in consumer prices was up almost 12% on the year, and headline inflation in the euro area was 2.6%. Core inflation, meaning prices excluding energy and food, was only 1.2%.

In March, Italian 10-year bonds paid on average 1.66 percentage points more than German ones. That gap is called the spread, and the German bond, the Bund, serves as the yardstick because the market sees it as the safest in the euro area. By November the Italian spread had reached 5.19 points. On 12 November Silvio Berlusconi resigned. The new ECB president, Mario Draghi, cut rates at his very first meeting, and by December had taken them back to where they started.

Fifteen years later the ECB is again raising rates because of oil. In June and September 2026 the deposit rate went from 2.00% to 2.50%. This piece asks whether history is repeating itself, and how far the resemblance goes.

A key for a different lock

This year’s shock comes from supply. The Strait of Hormuz has been almost closed since the end of February. Brent as quoted in the news (the front-month futures contract) averaged $102 a barrel in September and reached $108.75 on 15 September. A physical cargo for immediate delivery cost far more: on EIA spot prices it reached $130.80 the same day, $22 higher. Paper oil and the real barrel parted ways, because the real barrel cannot get through the strait. According to Eurostat’s flash estimate, energy in consumer prices is up 18.8% on the year and headline inflation is 3.8%.

But rate hikes don’t open Hormuz. They work on demand: they make credit dearer, cool spending, and so slow the rise in the price of everything else. On the price of oil that cannot get through the strait, they do nothing. The ECB is trying to unlock a supply shock with a key made for demand. The key does not fit the lock. Turned hard enough, it can break the door at its weakest point.

The ECB itself admits that second-round effects are absent for now. Its statement of 10 September says that wages show no material response to the energy shock at this stage. The data agree. According to Eurostat, wages in the euro area grew 3.0% on the year in the second quarter of 2026, after 3.4% in the first and 4.1% a year earlier. Wage growth is slowing during the oil shock, not speeding up.

The Governing Council nevertheless voted unanimously to raise rates. Lagarde called the decision a “no-brainer”. Asked whether a 2.50% deposit rate was not already at the top of the neutral range (the level at which rates neither stimulate nor hold back the economy), she answered that the ECB does not attach great importance to the neutral rate in the current circumstances. That leaves the door open for the bank to move into territory where rates start holding the economy back, while the shock comes entirely from outside.

September in two phases

The hike of 10 September caused no panic. The month, however, ended in a way that looks more like 2011 than like any year since.

The first phase, from 4 to 28 September, was a general rise in yields. The German 10-year yield rose 0.31 points to 3.64%, and the French and Italian yields by about half a point each. In the two days around the ECB decision alone, 9 and 10 September, the Bund rose 0.14 points and France and Italy about 0.20 each. Spreads widened, but slowly. This was a rate shock for the whole market.

The second phase began on 29 September. The Bund yield turned down, while French and Italian yields kept rising. When a bond’s yield falls, its price rises: investors were buying German debt and selling French and Italian. Over three sessions the French spread jumped from 1.13 to 1.41 points and the Italian from 0.96 to 1.20. This is the signature of 2011: the Bund stops falling in price along with the rest and becomes a safe haven.

On the Bundesbank’s yield curve the Bund reached 3.69% on 28 September, the highest level since April 2011. The French 10-year constant-maturity yield (the Banque de France’s TEC10) reached 4.90% on 1 October. That is the highest since the series began in 2004, above the 4.86% peak of 3 July 2008. On that same day the ECB raised rates for the last time before the crisis.

Spain, so far, is standing aside. According to Banco de España data, the Spanish 10-year yield rose in September almost exactly as much as the German one, so its spread has not moved: against the Bundesbank curve it stayed between 0.35 and 0.47 points through the month. Neither the ECB decision nor the oil peak shifted it. There is no official data yet for 1 and 2 October.

2008 and 2011: the precedent

Twice in the history of the euro the ECB has raised rates against oil-driven inflation. Both times it reversed course within months.

The resemblance to 2011 lies in the mechanism, not the scale. The Italian spread is now 1.20 points. In November 2011 it was above five. But in 2011 too, the widening did not begin with panic. In March, a month before the first hike, the spread was 1.66 points. In July, the month of the second, it averaged 2.72. The jump to 5.19 came over the following four months.

How a rate hike reaches the spread

ECB rates do not directly set the spread between France and Germany. There are three ways in which a hike still gets there.

The first is the interest bill. A heavily indebted state refinances part of its debt every year at the new, higher rates. If the average interest rate on the debt grows faster than the nominal growth of the economy, debt relative to GDP rises on its own, without any new deficit. The bigger the debt, the sooner a rise in rates turns into a budget problem. For Germany, with its low debt, that is a matter of years; for France and Italy it is a matter of the next budget. How exactly this arithmetic works is covered in an earlier piece in this series, The Price of Money and the Weight of Debt, using the United States as the example.

That piece also contains a detail that matters for today’s story. After the Second World War, France and Italy shrank their debt through what is known as financial repression: interest on government debt was held below inflation for years on end, and savers paid the bill. According to research by Reinhart and Sbrancia, as cited by the Federal Reserve Bank of Richmond, between 1945 and 1980 bondholders lost an average of 6.6% a year in real terms in France and 4.6% in Italy. In the euro area that exit is closed: neither Paris nor Rome has its own central bank to hold its interest rates down. That is why the bill shows up in the spread, and why TPI is the only thing that resembles the old exit.

The second is who is buying. Between 2015 and 2022 the largest buyer of European government debt was the Eurosystem itself. Today it is shrinking its portfolio (quantitative tightening). Its assets stand at €5.90 trillion, against a peak of €8.84 trillion in June 2022. The money banks hold in its deposit facility (a rough measure of excess liquidity) has fallen from €2.63 trillion a year ago to €1.96 trillion. New bonds have to find private buyers, and a private buyer demands a higher yield when risk is rising.

The third is the flight to safety. When investors get nervous, they sell the riskier government bonds and buy German ones. That is the second phase of September. In it, the ECB rate is not the trigger but the backdrop: it has made debt dearer for everyone, and then the market starts to distinguish who can afford it.

What weakens the thesis

The parallel with 2011 is strong, but it is not proof. Six things need to be said against it.

Core inflation was above target even before the shock. In 2011 it was 1.2-1.6%, meaning the ECB was raising rates while inflation, excluding oil, was below target. Today core inflation is 2.4-2.5%, and it was already 2.4% in February, before the war. The ECB’s projection has it at 2.6% in 2027. This is the bank’s strongest argument, and it has a logic to it: a second inflation shock within five years is more dangerous to the bank’s credibility than a single shock was in 2011. The hikes are insurance. The premium, however, is paid by the heavily indebted states.

The big widening came while oil was getting cheaper. Brent peaked on 15 September and by 29 September had fallen: the futures contract from $108.75 to $102.59, the physical cargo from $130.80 to $113.96. Spreads widened most after 22 September. If the chain were simply “oil, then the ECB, then spreads”, the move should have come earlier. The final push came from budgets: on 1 October France tabled its draft budget for 2027, and ANSA described the day as one of fear over public finances, “and not only Italy’s”.

Yields are rising around the world. Lagarde said so plainly on 10 September: this is not a euro-area problem alone. Part of the rise in European yields comes from the United States and Japan, and from the enormous financing needs tied to artificial intelligence. That part the ECB can neither cause nor stop.

The levels are far from 2011. A spread of 1.20 points for Italy and 1.41 for France is uncomfortable, but it is not a crisis. Italy today is around the levels it was at in early 2025.

Spain is not taking part. In 2011 Spain was at the centre of the crisis. Today its spread is not moving. That means the market is not punishing the periphery as a whole, but specific states with specific budget problems.

The safeguards are different. In 2011 the ECB had no instrument to intervene against an unwarranted widening of spreads. Today it does: the Transmission Protection Instrument (TPI), created in July 2022. The question is whether it can be used where it is needed. That is the next section.

France: the door that gives first

France pays more than Italy for 10-year debt. Its spread is 1.41 points. On Banque de France and Bundesbank data, which allow a comparison going back in time, it is the widest since May 2012. The state that until recently was seen as the second pillar of the euro is now priced above Italy.

The draft budget for 2027 was presented on 1 October. The spread widened by almost 0.15 points that day. France has been in the excessive deficit procedure since July 2024. The presidential election is in spring 2027. At the 10 September press conference a journalist asked Lagarde about a political proposal to freeze or cancel part of France’s debt. She replied that it would be a breach of the Treaty. On the widening of spreads itself, she said she would not comment.

This is where TPI comes in. Under the 2022 criteria, a state can receive support if it complies with the EU fiscal framework: it must not be in an excessive deficit procedure or, if it is, must not have been assessed as failing to act on the Council’s recommendations. France is in the procedure. So support depends on whether Brussels judges that it is following the recommendations. The instrument exists on paper. Its first real test may be France, and in the middle of a presidential campaign.

In one sense this is a harder situation than 2011. Back then the ECB had to invent an instrument. Now it has to decide whether to use it for the second-largest economy in the euro area, when its own rules give it grounds to say no.

What to watch

For now the market is discriminating. It is punishing France and Italy, not Spain, and not the whole periphery. This is 2011 in embryo, not in full swing. The decisive moment will be whether the ECB keeps raising rates while the Bund rallies as a safe haven. At that point the key is no longer cooling demand. It is breaking the door.

This is analysis, not a trade recommendation. If the thesis is right, though, the mistake will show first in expectations of further hikes: as in the autumn of 2011, they would have to be priced out.

Conclusion

In 2011 the ECB raised rates against oil it could not control, and between March and November the Italian spread tripled. In 2026 it is doing the same, with higher core inflation and more safeguards. Those differences are real, and they make today’s decision easier to defend than the one back then.

The mechanism, however, is the same. Rate hikes do not bring down the price of oil. They make debt dearer for everyone, and then the market chooses whom to punish. In the last days of September and the first days of October, it started choosing.

Rate hikes don’t open Hormuz. But they can open spreads.

Sources

1. ECB, key interest rates, deposit facility and main refinancing operations, daily series FM.D.U2.EUR.4F.KR.DFR.LEV and FM.D.U2.EUR.4F.KR.MRR_FR.LEV: https://data.ecb.europa.eu/data/datasets/FM

2. ECB, euro area HICP (headline, excluding energy and food, energy), annual rate of change, to December 2025: https://data.ecb.europa.eu/data/datasets/ICP

3. Eurostat, HICP prc_hicp_minr, EA21, to the September 2026 flash estimate (updated 2 October 2026): https://ec.europa.eu/eurostat/databrowser/view/prc_hicp_minr/default/table

4. Eurostat, labour cost index lc_lci_r2_q, wages and salaries, EA21 (updated 16 September 2026): https://ec.europa.eu/eurostat/databrowser/view/lc_lci_r2_q/default/table

5. ECB, negotiated wage indicator STS.Q.U2.N.INWR.000000.3.ANR: https://data.ecb.europa.eu/data/datasets/STS

6. ECB, monthly long-term interest rates by country IRS.M.{cc}.L.L40.CI.0000.EUR.N.Z: https://data.ecb.europa.eu/data/datasets/IRS

7. ECB, Eurosystem weekly financial statement (total assets, deposit facility), ILM dataset: https://data.ecb.europa.eu/data/datasets/ILM

8. ECB, Monetary policy statement and press conference, 10 September 2026: https://www.ecb.europa.eu/press/press_conference/monetary-policy-statement/2026/html/ecb.is260910~6a45359cfc.en.html

9. ECB, Monetary policy statement and press conference, 11 June 2026: https://www.ecb.europa.eu/press/press_conference/monetary-policy-statement/2026/html/ecb.is260611~372040d313.en.html

10. ECB, “The Transmission Protection Instrument”, press release, 21 July 2022: https://www.ecb.europa.eu/press/pr/date/2022/html/ecb.pr220721~973e6e7273.en.html

11. EIA via FRED, Brent spot price, daily DCOILBRENTEU: https://fred.stlouisfed.org/series/DCOILBRENTEU

12. Reuters/LSEG, ICE Brent Crude Energy Future, front month LCOc1, one-year table viewed 3 October 2026: https://www.reuters.com/markets/quote/LCOc1/

13. Reuters/LSEG, 10-year yields FR10YT=RR, IT10YT=RR, DE10YT=RR, one-month tables viewed 3 October 2026: https://www.reuters.com/markets/quote/FR10YT=RR/

14. Deutsche Bundesbank, yield on the federal securities curve, 10 years, daily BBSIS/D.I.ZST.ZI.EUR.S1311.B.A604.R10XX.R.A.A._Z._Z.A: https://www.bundesbank.de/en/statistics/time-series-databases

15. Banque de France, Webstat, TEC10 FM.D.FR.EUR.FR2.BB.FRMOYTEC10.HSTA and monthly benchmark yields by country:

https://webstat.banque-france.fr/

1. Banco de España, table ti_1_3, 10-year bonds on the secondary market, series D_G0B1F0ZP: https://www.bde.es/webbe/es/estadisticas/compartido/datos/csv/ti_1_3.csv

2. ANSA, daily reports on the BTP-Bund spread, 2 September to 2 October 2026, e.g. 1 October 2026: https://www.ansa.it/sito/notizie/economia/2026/10/01/lo-spread-btp-bund-conclude-la-giornata-in-aumento-a-1185-punti_d1f1e887-e2b4-4e76-8174-cf9c9824f29f.html

3. Ministère de l’Économie, press conference presenting the PLF and PLFSS 2027, 1 October 2026: https://presse.economie.gouv.fr/ip-conference-de-presse-presentation-du-plf-et-plfss-2027-jeudi-1er-octobre-2026/

4. Liquidity Desk, “The Price of Money and the Weight of Debt”, 30 June 2026: https://liquiditydesk.org/p/the-price-of-money-and-the-weight

5. Federal Reserve Bank of Richmond, “A Look Back at Financial Repression”, Econ Focus, 2021 (citing Reinhart and Sbrancia)

6. ECB, Meetings of the Governing Council (monetary policy meeting of 28-29 October 2026): https://www.ecb.europa.eu/press/calendars/mgcgc/html/index.en.html

Liquidity Desk | liquiditydesk.org

This is the first monthly Macro Pulse for China after our second-quarter review. All the main July releases are now out. The last to arrive were industrial profits, on 28 August. In this issue we separate the shop window from the warehouse. The shop window is exports: record sums that look like a boom. The warehouse is what people in China buy, build and borrow at home.

Where the economy stands

The phase in July remains growth with slowing momentum. July is the weakest month of the year in the business surveys. The official composite PMI from the National Bureau of Statistics (NBS) fell to 49.3. The PMI is a survey of purchasing managers in which a reading below 50 means more firms report deterioration than improvement. The hard data, however, did not collapse. Industrial production grew 4.5% from a year earlier, and dollar exports grew 23.9%. This is a dip within the phase, not a change of phase.

The engine is exports linked to artificial intelligence: chips, servers and computers, plus cars. It is narrower than it looks. Customs price indices show that July exports grew more through price than through volume.

The brake is domestic demand. Bank loans in yuan fell by CNY 340bn in a single month, and household loans by CNY 460bn. Retail sales grew 0.6%. The brake is structural, because it rests on the property sector and household balance sheets. In July, floods and heatwaves added a cyclical layer on top.

The central bank, the PBoC, is on hold on its rate and eases only in targeted ways. Real conditions are tightening on their own, though. Inflation is falling while the rate stays put, so the real cost of money is rising. August and September send a better signal than July, but the hard data for the third quarter still have to confirm it.

Core indicators

All July values are final. The monthly growth rate of retail sales was revised from 0.06% to 0.01% in the August release. NBS and RatingDog are two different surveys. NBS covers more large and state-owned enterprises, RatingDog more private and export-oriented firms, so the two are not compared directly.

The trajectory

For production, retail sales and exports, the January column is the combined January and February figure, which is how Chinese agencies publish them. For CPI and core CPI it is the official average of the two months. Chinese New Year fell on 17 February in 2026 and on 29 January a year earlier. That makes January on its own look artificially low and February artificially high.

Three lines carry the story. The first is credit. M2, the broad money in the system, and bank loans have slowed every month since the start of the year. This is not a July shock. It is a slope that has lasted six months.

The second line is prices. PPI, the index of prices at which factories sell their output, is the only indicator that is clearly accelerating. Consumer prices are slowing. The gap between them is not new, but July holds it at 3 percentage points.

The third line is exports. They are growing faster than in April and faster than at the start of the year. This is the strongest number in the table and so it needs the most careful reading. We return to it in the section on the overlooked detail.

Two lines show a turn and deserve a note. Production and retail sales are above their April levels but below the start of the year. The gap with April is under half a percentage point, however, which is within normal monthly noise. The more honest reading is that both fell to a lower level in spring and have stayed there since.

RatingDog manufacturing also turns, but downwards. It is above January and below April. Private factories have lost most of their spring acceleration.

RatingDog services fell harder than any other line: from 54.1 in June to 50.4 in July. This survey covers more private firms in trade, transport and consumer services. That is where July’s floods and weak spending hit first. The official survey shows the same from another angle: new orders in services fell to 45.2 and services selling prices to 47.9. Service firms are not just selling less. They are selling cheaper to keep their customers.

August and September serve as a check on direction. The official composite PMI rose to 49.5 in August and to 50.7 in September. Industrial production accelerated to 5.2% in August and exports to 25.0%. Retail sales, however, slowed to 0.4% and M2 to 7.5%. The surveys and the factories are recovering. Credit and consumption are not following.

The engine: chips and cars

China’s engine in July has two names: electronics for artificial intelligence, and cars.

Industry shows the first. Output of computers and electronic equipment grew 19.1% from a year earlier. High-tech manufacturing grew 16.9%, faster than in June. Integrated circuit output was 20.7% higher than a year earlier, and industrial robots 30.2%. According to the NBS, new growth drivers accounted for 50.9% of industrial growth in the first seven months.

Customs data show how big the engine is. From January to July, exports were USD 393.8bn higher than a year earlier. USD 107.7bn of that came from integrated circuits and USD 51.8bn from computers and their parts. Together the two groups account for 41% of all additional exports. High-tech products as a whole grew 40.7%. Capital is moving the same way. Foreign direct investment into China fell 6.2% from January to July, but investment in high-tech industries rose 32.7% and now makes up 41.6% of the total. Domestic investment in equipment grew 9.0%, while total investment fell.

The second engine is cars. Car exports from January to July reached USD 110.8bn, up 54.9%. Here the growth is genuine volume: the number of cars exported is 53.7% higher. In July, the customs index shows the volume of cars and parts exported up 40%, at a slightly lower price. Output of new energy vehicles grew 29.9%. Total car output is flat, and domestic car sales fell 17%. China is making the cars it cannot sell at home and selling them abroad.

Is the engine sustainable? Partly. Profits show it is real. Profits of electronics makers more than doubled from January to July, up 110%. Total industrial profits grew 17.6%. But the monthly pace is fading. Profits grew 11.2% in July alone and 4.2% in August. Inventories of finished goods were 10.8% higher at the end of July. Factories are producing slightly faster than they sell.

Geography shows the limits. Exports to the United States grew only 2.6% from January to July and are now just 10.2% of the total. Exports to Southeast Asia grew 25.1%, to Hong Kong 48.2% and to South Korea 33.3%. China is not exporting more to its biggest customer. It is exporting more to its neighbours, who are part of the same chip and server supply chain.

What could stop the engine? Three things. The first is a fall in chip prices. If the global cycle of data-centre investment slows, prices fall and the value of exports drops quickly, even without fewer units shipped. The second is measures against Chinese cars. Export growth of more than 50% provokes a response in the markets the cars are entering. The third is the yuan. Between December 2025 and September 2026 the yuan gained about 5% against the dollar, which makes Chinese goods dearer for foreign buyers.

The brake: households are not borrowing, firms are not building

The brake sits in credit and in property. July is the month in which bank loans in yuan fell by CNY 340bn. In April they stood still. In July they clearly fell.

Households repaid CNY 460bn more than they borrowed. Their short-term loans, which often finance purchases, fell by CNY 340bn. Their long-term loans, which include mortgages, fell by CNY 120bn. Companies also cut their debt by CNY 130bn, and long-term corporate loans fell by CNY 230bn. The only growing channel was bill financing, in which a bank pays a company early for a receivable from a trade deal. It added CNY 376bn. That is money for day-to-day payments, not for new plants.

Total social financing, the broadest measure of credit to the economy, added CNY 1.41 trillion. CNY 1.32 trillion of that was government bonds. Without the state, the private flow was close to zero. Our September liquidity issue described this as a broken bypass. The state can still create financing. Private credit is not following it.

Property shows why. Investment in real estate from January to July was 19.2% lower, new construction starts 24.0% lower and floor space sold 11.8% lower. New home prices fell from a year earlier in every tier of cities: by 1.1% in the four largest, 2.8% in the second tier and 4.2% in the third. Prices of existing homes fell further, by 3.7% to 5.8%. Only 8 of 70 cities reported existing home prices rising or holding steady on the month. When the price of a home is falling, a household does not take out a mortgage, even at a low interest rate.

Consumption follows property. Retail sales grew 0.6%, and 2.5% excluding cars. Household appliances fell 1.9%, furniture 8.8% and building materials 14.2%. These are purchases that go together with a new home. The only strong item was telecom equipment, up 20.4%, supported by state subsidies for replacing old devices with new ones. Services are holding up better than goods. The index of services output grew 4.3% in July, and sales of services 5.0% from January to July, against 1.1% for goods. Restaurants, though, grew only 1.4%. Chinese consumers are spending on phones and travel, but saving on everything to do with the home.

The labour market is getting heavier. Urban unemployment is 5.2%, 0.2 points above June. Youth unemployment for 16 to 24 year olds, excluding students, jumped from 14.9% in June to 17.9% in July. Part of the jump is seasonal, as graduates enter the market. In August, however, it reached 18.9%, the highest in a year.

Business investment is not helping either. Total fixed asset investment was 6.7% lower and private investment 9.4% lower. The exception is equipment, up 9.0%, supported by state renewal programmes.

Is the brake structural? Mostly yes. The property downturn has lasted since 2021, and households are holding on to their savings. July adds a cyclical layer too. The NBS points to extreme heat, torrential rain and floods that temporarily halted work on building sites. That part recovers quickly: construction activity in the PMI survey fell to 47.0 in July, but was back at 50.3 by September. New construction orders tell a different story. They fell to 40.1 in July and were only 45.7 in September, far below 50. Work on site comes back when the rain stops. New contracts do not, because they depend not on the weather but on demand for housing. This is the structural part, and it needs a change in expectations for home prices, which is not yet visible.

Prices: petrol pulls down, chips push up

Consumer inflation fell to 0.5% from a year earlier, from 1.0% in June. The main reason is petrol. According to the NBS, petrol prices fell 10.7% in a single month and pulled the headline index down by about 0.35 points. On the year, petrol was only 1% dearer, after 17% in June. Food was 1.5% cheaper and pork 13.3% cheaper.

Core CPI, excluding food and energy, is 0.9%. It is the better measure of domestic demand, and it is falling too: from an average of 1.3% in January and February to 0.9% in July. One item stands out. Gold jewellery is 24.6% dearer than a year ago. That is not demand pressure but a reflection of the gold price.

PPI is 3.5%, after 4.1% in June. Prices in mining rose 16.4% and for raw materials 6.1%, while factories sold consumer goods 0.8% cheaper. The gap between PPI and CPI is 3.0 points. It shows where the pressure sits. Factories pay more for energy, metals and chemicals, but cannot pass it on to buyers at home. Margins of producers selling into the domestic market are shrinking. Profits confirm it: carmakers earned 20.4% less, and makers of cement and glass 48.2% less.

The PMI survey shows the same squeeze inside factories. In July, the index of prices factories pay for inputs was 53.2, meaning more firms paid more. The index of the prices at which they sell was 47.8, meaning more firms cut their prices. Factories were buying dearer and selling cheaper. In September both indices jumped, to 60.8 and 54.0. For the first time since spring, factories are starting to pass some of the cost on.

The check after July matters. In August CPI returned to 0.8%, core to 1.0% and PPI to 3.8%. In September input prices in the PMI survey jumped to 60.8, close to the March peak of 63.9. In our September Commodity Snapshot, Brent crude was back above USD 100, because of the war with Iran and restricted flows through the Strait of Hormuz. July’s cheaper petrol was a pause, not a turn.

The conclusion on prices is twofold. Domestic inflation is weak and does not constrain the PBoC. Imported inflation is strong and squeezes factories from within.

The overlooked detail: chips are getting dearer, not selling more

China’s exports look like a boom. In July they grew 23.9% in dollars. Over seven months the surplus reached USD 687.4bn. Headlines speak of trade strength. The customs price indices tell another story.

China’s customs agency publishes separately how much the price and how much the volume of exported goods has changed. In July the value of exports in yuan was 17.8% above last year. The price was 11.6% higher and the volume only 5.5% higher. Two thirds of the growth is price.

In electrical equipment, where chips sit, the price was 28.2% higher and the volume 3.7% higher. In machinery, where servers and computers sit, the price was 26.8% higher and the volume 0.8% lower. China is exporting almost the same volume of machines but selling them for more.

It is clearest in the chips themselves. From January to July the value of exported integrated circuits was USD 216.0bn, almost double a year earlier. The number of chips exported was only 6.2% higher. The average price of an exported chip was 88% higher. China is not exporting many more chips. It is exporting dearer ones.

Imports show the same. In July their value in yuan was 21.2% higher, but their volume only 0.4% higher. Machinery imports were 74.7% dearer and 11.3% smaller in volume. China is buying the same chips it later exports inside finished products, and paying much more for them. Chip imports from January to July reached USD 361.8bn, up 58.3%.

Why is this overlooked? Because the value of trade looks like demand. It is closer to price. Factories assembling servers are getting orders, but part of each order is a dearer component rather than more work. Import volume, the best measure of domestic need for raw materials and components, is almost flat.

This has two consequences. The first is for the engine. If chip prices fall, the value of exports will fall quickly without Chinese factories losing a single customer. The second is for the brake. Strong trade numbers do not bring strength to households. They bring profit to a narrow circle of electronics companies.

What this means

The July data give the PBoC grounds for easing, not tightening. Growth is slowing in the surveys, credit is contracting and core inflation has fallen below 1%. The last confirmed quarterly position was Q2: growth weakening and consumer inflation easing, with higher factory prices. July does not contradict it.

The PBoC, however, is on hold on its rate. The one-year Loan Prime Rate (LPR), the benchmark lending rate in China, has stood at 3.0% since December 2025. The five-year rate, the benchmark for mortgages, stands at 3.5%. Real conditions are tightening on their own. The real interest rate, meaning the LPR minus inflation, rose from 1.8 points in April to 2.5 points in July. Without the PBoC moving anything, money is becoming dearer in real terms.

Quantities give a mixed picture. In July the PBoC injected a net CNY 688bn through its operations. Bank reserves at the central bank, however, fell by CNY 104bn. The reason is that the state raised money through bonds and parked it in its account at the PBoC faster than it spent it. Government deposits rose by CNY 736bn. In August the PBoC itself withdrew a net CNY 409bn. The interbank rate did not rise and stayed around 1.4%. Banks have money. They have no one to lend it to.

On 29 September the PBoC responded, but in a targeted way. It cut the rate on PSL, its targeted loans to the state development banks, by 0.25 points to 1.50%. It raised the credit lines for technology by CNY 200bn and for agriculture and small business by CNY 500bn. From 1 October, the finance ministry, the PBoC and the banking regulator are introducing a subsidy of 1 percentage point a year on the interest on new mortgages for first homes. This is more about steering credit than about making money cheaper across the board.

In the terms of our framework, this is a hold, with passive tightening in real terms. The PBoC’s monetary policy committee, meeting on 19 September, spoke of “moderately loose” policy and “stronger counter-cyclical adjustment”. For now, the words are ahead of the actions.

What does this mean for liquidity? Liquidity in China is not missing from the banks. It is not reaching households and private firms. New money comes from government bonds rather than bank loans. That is a weaker engine, because the state spends slowly. Budget spending from January to July grew only 1.3%, and local governments’ revenue from land sales fell 30.8%.

For global markets China remains a two-sided factor. Strong exports of chips and cars are positive for supply chains in Asia. Shrinking import volumes of raw materials are negative for exporters of copper, oil and iron ore. Crude oil imports from January to July were 13.2% lower by volume.

For the ordinary person in China the picture is quiet. Shop prices barely move. Their home is losing value. Pay is rising, but work is harder to find, especially for the young. Credit is cheap, but they do not want it.

What is confirmed? That domestic credit is contracting, that core inflation is falling and that exports are growing more through price than volume. What remains an assumption? Whether the mortgage subsidy will turn housing credit around. Whether chip prices will stay high. Whether September’s recovery in the surveys will show up in the hard data.

Risks in both directions

The strongest argument against our phase is that we are underestimating the recovery. The September surveys look much better than July’s. The official composite PMI is 50.7 and new factory orders are 50.5. In our Q2 review we set exactly this as the threshold for a change: a composite PMI above 50 with new orders above 50. The September surveys meet it. RatingDog manufacturing is 52.1, the highest in five months, and construction is back above 50. If the hard data for September confirm the surveys, July will turn out to have been a temporary dip caused by the weather and oil.

The downside risk has three sources. The first is the price of chips. If it falls, the strongest number in China’s economy, exports, will lose speed with no warning in unit volumes. The second is oil. Brent above USD 100 raises factory costs just when factories cannot raise prices at home. The third is credit. If households keep repaying loans through the autumn, neither the rate nor the subsidies will reach consumption.

The risk for the PBoC is asymmetric. If it waits too long, falling inflation will push the real rate ever higher. If it eases sharply, it will pour money into a system where the problem is demand for credit, not its price.

What to watch

On 14 October, September CPI and PPI will show whether core CPI stays below 1%. If it falls below 0.8%, the pressure for a rate cut will grow. September trade data come out the same day. What matters is whether exports grow in volume, not only in value.

Around mid-October comes the PBoC’s September credit data. The key line is household loans. A positive flow would be the first sign that the mortgage subsidy is working.

On 19 October come Q3 GDP and the September monthly data. The threshold for a change of phase remains the one from our Q2 review: GDP growth of at least 4.5% with stronger consumption. If retail sales stay below 1% in September, the threshold will not be met, whatever the surveys say. The threshold for a slide into stagnation is a composite PMI below 50 for two consecutive months, together with production growth below 4%. July and August were already two months below 50, but production stayed above 4%, so the threshold has not been met.

On 20 October the LPR is set. A cut would show that the PBoC is moving from targeted measures to making money cheaper across the board.

On 31 October the official October PMIs will show whether September’s recovery is lasting.

Closing

China is selling dearer chips and more cars to the world, while at home it is repaying loans and not buying homes. The shop window is lit, but the warehouse is half empty. The engine is real, but narrow and dependent on the price of one product. The brake is broad and sits in household balance sheets. We do not yet know whether September’s recovery in the surveys is the start of a new phase or just a rebound after bad July weather.

Sources

• NBS, Purchasing Managers’ Index for July 2026

• NBS, Purchasing Managers’ Index for August 2026

• NBS, Purchasing Managers’ Index for September 2026 (Chinese)

• NBS, Purchasing Managers’ Index for January 2026, April 2026, June 2026

• S&P Global, RatingDog China General Manufacturing PMI, August 2026, with the final July reading

• S&P Global, RatingDog China General Services PMI, August 2026, with final July readings

• S&P Global, RatingDog China General Manufacturing PMI, September 2026

• S&P Global, RatingDog China General Manufacturing PMI, February 2026

• S&P Global, RatingDog China General Services PMI, February 2026

• NBS, Industrial Production, July 2026 and August 2026

• NBS, Retail Sales, July 2026 and August 2026

• NBS, National Economy in the First Seven Months of 2026

• NBS, press conference on July 2026 data (Chinese)

• NBS, CPI, July 2026 and August 2026

• NBS, PPI, July 2026 and August 2026

• NBS, CPI, February 2026, with the January-February average

• NBS, Fixed Asset Investment, January to July 2026

• NBS, Real Estate Investment, January to July 2026

• NBS, home prices in 70 cities, July 2026, official commentary (Chinese)

• NBS, Industrial Profits, January to July 2026 and January to August 2026

• NBS, GDP, Q2 2026

• NBS, Capacity Utilisation, Q2 2026

• NBS, 2026 release calendar

• GACC, total trade in USD, July 2026

• GACC, major exports, July 2026

• GACC, major imports, July 2026

• GACC, trade by country, July 2026

• GACC, export price and volume indices, July 2026

• GACC, import price and volume indices, July 2026

• PBoC, financial statistics, July 2026 (Chinese)

• PBoC, financial statistics, August 2026 (Chinese)

• PBoC, financial statistics, January 2026 (Chinese)

• PBoC, financial statistics, April 2026, as reprinted by People’s Daily (Chinese)

• PBoC, financial statistics, H1 2026, official mirror (Chinese)

• PBoC, liquidity operations, July 2026 and August 2026 (Chinese)

• PBoC, balance sheet of monetary authority

• PBoC, adjustments to monetary policy tools, 29 September 2026 (Chinese)

• Ministry of Finance, PBoC and financial regulator, mortgage interest subsidy (Chinese)

• PBoC, Monetary Policy Committee, Q3 2026 meeting (Chinese)

• ChinaMoney, LPR history

• Ministry of Finance, fiscal data, January to July 2026 (Chinese)

• MOFCOM, foreign direct investment, January to July 2026 (Chinese)

• FRED, Chinese yuan to US dollar, DEXCHUS

• Trading Economics, China youth unemployment rate

• Trading Economics, China balance of trade calendar

Not financial advice.

Liquidity Desk | liquiditydesk.org

It is late September, and Iowa is bringing in the corn. The combine, the tractor, the truck to the silo: everything runs on diesel. A gallon of diesel in the US (3.8 litres) now costs an average of $6.53, a record. A year ago it cost $3.75. The price has jumped 74% in twelve months, right in the middle of the harvest and five weeks before the midterm elections on 3 November.

On 22 September Donald Trump offered a solution that looks obvious to anyone filling a tank: “I’ve said let’s not send out the diesel.” America, he added, makes a lot of diesel. On Sunday, 27 September, he said he was looking at a ban “very seriously” and added: “we may do it”.

At first glance the arithmetic is simple. America exports huge volumes of diesel. Keep it at home, and there is more to go around, so the price falls.

But a refinery is not a tap you turn on and off at will. It is more like a river. The ban is a wall built downstream. For the first few weeks the water rises behind the wall and diesel really does get cheaper. Then the reservoir fills. To keep it from overflowing, you have to cut the flow upstream, at the source. And that source does not carry only diesel. It also carries the gasoline for cars and the fuel for planes.

This article follows the river: from the American refinery, through crude oil prices, to the European haulier.

One week, three directions

The last seven days show how unsettled things are.

On 22 September Trump said he backed limiting exports, and Treasury Secretary Scott Bessent confirmed the administration was examining whether a full or partial ban was possible. The next day Politico reported that a 90-day ban was being prepared, and shares in Valero, Marathon Petroleum and Phillips 66 fell. Energy Secretary Chris Wright pushed back at once. Under a full ban, he argued, refineries would process less and gasoline would get more expensive. Instead of a ban, he opened talks with refiners about cutting exports voluntarily. On 25 September Bloomberg reported that the White House had assured Senator Ted Cruz there would be no full ban.

On 27 September Trump changed direction again. According to Bloomberg, National Economic Council director Kevin Hassett, Bessent and Trade Representative Jamieson Greer spent the past week analysing what a short-term ban would do. The oil companies are proposing something else: a temporary suspension of the federal excise tax on diesel.

So far there is no signed order and no regulation. There is a president who wants a ban, an energy secretary who opposes it, and a market that is already starting to price it in.

How America became the world’s filling station

To see how much is at stake, we need to go back.

In 1975 the US exported an average of about 1,000 barrels a day of diesel and other middle distillates (a barrel is 159 litres). That is a rounding error, not a market. The country imported oil, its own output was falling, and its refineries worked for the home market.

Exports began to climb in earnest around 2008. By 2012 they had passed a million barrels a day. In 2025 they averaged 1.25 million barrels a day, 1,250 times the 1975 level.

Then came 2026. The war with Iran began on 28 February, and the Strait of Hormuz is still almost closed. From mid-July Russia dropped out too: after Ukrainian strikes on its refineries, it stopped exporting fuel. American refineries were left as the last large supplier the world could count on.

One clarification. These are gross exports. Net exports (exports minus imports) are smaller, but even so this is a volume the country cannot burn through on its own in a few weeks.

1975: the same idea, the other end of the pipe

Today’s situation is most often compared with the law of 1975. After the Arab oil embargo, Congress passed the Energy Policy and Conservation Act (EPCA), which required the president to ban exports of crude oil produced in the US. The ban was lifted only in December 2015.

The Congressional Research Service stresses one important detail. Finished fuels, including gasoline and diesel, were never covered by it.

A common simplification needs correcting here. The 1975 ban did not keep American oil cheap for forty years. Most of the time it had no real effect, because the US was producing less and importing more. There was nothing to export. The ban started to bite only around 2011, when the shale boom flooded the country with light crude. Domestic prices then fell almost $30 a barrel below international ones. (Part of that gap in 2011-2013 also came from a lack of pipelines from the oil hub at Cushing, Oklahoma, to the coast, not from the ban alone.)

In exactly those years American refiners found themselves in a very comfortable position. They bought cheap local crude and sold diesel freely at world prices. Diesel exports had already started rising in 2008, but this is when they accelerated: from about 650,000 barrels a day in 2010 to over a million in 2012.

That is the irony. The 1975 law held back the raw material before the refinery and indirectly sped up the growth of the world’s biggest diesel export machine. A 2026 ban would hold back the finished product after the refinery and block the outlet of that same machine.

The idea is the same, “keep it in the country”. Applied at the two ends of the pipe, it produces opposite results. Hold back the raw material, and the refinery gets cheap crude and profits. Hold back the product, and the refinery has nowhere to sell and must cut output.

The real precedent: Russia, 2023

There is in fact a much closer example, and it is recent.

On 21 September 2023 Russia, at the time the world’s largest seaborne exporter of diesel, banned almost all exports of diesel and gasoline to bring prices down at home. The measure threatened to remove about a million barrels of diesel a day from the world market.

The ban lasted 15 days. On 6 October the government allowed diesel that reaches the ports by pipeline to be exported again, provided producers sold at least half of their output at home. First Deputy Energy Minister Pavel Sorokin explained why: refinery tanks were full, and processing was about to fall.

There is a second lesson. ICE Gasoil, the European benchmark for diesel, rose 4.5% on the day of the ban, but only a week later it was almost back to its pre-ban level. From the start the market assumed the ban would not last. It reacted to how long the ban was expected to run, not to the headlines.

How the reservoir fills

First, diesel gets cheaper. When exports stop, diesel that was meant to leave by tanker stays on the Gulf Coast. There is more of it at home and the price falls. According to Goldman Sachs, as long as there is room in the tanks, each week of a ban would lower the average pump price by about 25 cents a gallon, roughly 4% of the current $6.50. That is why the measure appeals to politicians, and the effect is real.

Then the tanks fill. This is the clock that everything depends on. Wood Mackenzie estimates that under a ban about 700,000 barrels a day of surplus diesel would go into storage. The tanks around the Gulf of Mexico (known as PADD 3, where most export-oriented refineries sit) would fill in just over a month, and those in the rest of the country after roughly five more weeks. Goldman gives a longer window: all US diesel storage would theoretically fill in 9-10 weeks if exports fell by 1.6 million barrels a day. The difference comes mainly from whether one looks only at the Gulf Coast or at the whole country. The conclusion is the same: we are talking about weeks, not months.

Finally, the source is shut. A refinery cannot stop making diesel and carry on making gasoline. One barrel of crude yields gasoline, diesel, jet fuel, propane and fuel oil all at once. The proportions can be shifted somewhat, but not reversed. Once diesel has nowhere to go, the refinery has to process less crude overall.

According to Wood Mackenzie, to keep storage from overflowing, American refiners would have to process more than 2 million barrels a day less crude. That is about 12% of all refining in the country. (The full, paid report puts the minimum estimate at 2.7 million barrels a day.)

Less crude processed also means less gasoline. Goldman calculates that once diesel storage is full, each additional week of a ban would raise the gasoline price by about 30 cents a gallon. At the current price of around $4.48, that is close to 7% a week. Wood Mackenzie reaches the same point by another route: the US would have to import more gasoline, and at a higher price.

Compare the two numbers. While there is room in the tanks, diesel gets about 4% cheaper each week. Once the room runs out, gasoline gets about 7% dearer each week. And in America there are far more gasoline cars than diesel trucks. That is the argument Chris Wright is using against his own president. Trump himself concedes the point: a ban “can often times lead to a little bit of an increase on gasoline for cars”. He says he may do it anyway.

What crude oil is already showing

Brent is the international benchmark for crude oil prices, and WTI is the American one. WTI is usually a few dollars cheaper. On 24 September, during trading, the gap reached $12.02 a barrel, the widest since 6 May.

The logic runs like this. If American refineries process 2 million barrels a day less, they also buy less crude. More oil stays in the country without a buyer, and WTI gets cheaper relative to Brent. Wood Mackenzie adds that the Gulf Coast region imports less than 2 million barrels of crude a day. With a cut of that size, America would have to do more than stop importing. It would also have to export more of its own crude.

In other words, the market is doing almost exactly what it did under the old ban: pushing the price of American crude down relative to the world price. Only the reason is reversed. Then it was the raw material that was trapped. Now it would be the finished product.

Here, though, we need to be precise, because it is tempting to put the whole move down to diesel. The gap is also widening for a separate, independent reason: shipping has become more expensive. A large tanker of crude from the Gulf of Mexico to Asia now costs about $50 million, against $16 million before the war with Iran. According to Bob Yawger of Mizuho, American crude now has to be about $8 cheaper than Brent to cover the freight, double the usual $4. So part of that $12 is down to tankers, not diesel. The signal is real, but it is not clean.

Europe is downstream

If America builds the wall, Europe is where the water stops arriving.

After the EU stopped importing Russian diesel, Europe replaced much of it with American supply. According to Wood Mackenzie, almost half of US diesel exports are going to Europe in September 2026, against an average of 30% in 2025. European refineries are already running at full capacity. There is no spare.

If the ban is introduced, Wood Mackenzie expects the diesel margin in Northwest Europe to jump by 27%. (The margin is the difference between the price of diesel and the price of the crude it is made from. The bigger it is, the more the refinery earns and the sharper the shortage.) Goldman calculates that each week of a ban would raise the wholesale price of diesel in Europe by about $3 a barrel, or just under 2%. If European governments release their strategic reserves onto the market, that would offset about half.

The market reacted to the mere discussion. According to Bloomberg, on 23 September American diesel on the NYMEX exchange fell against European diesel far enough that its premium became the smallest since the end of April, and the one-day move was the largest since 2022. American diesel is already getting cheaper relative to European diesel, without any decision at all.

Who would fill the gap? According to Wood Mackenzie, only China has significant spare capacity, about 300,000 barrels a day. Russia could add about 190,000, but it is restricting its own exports. Analysts note drily that China may well decide that helping is not in its interest.

This is also the link to the wider economy, and the reason this story belongs in Liquidity Desk rather than only in the energy news. Diesel moves trucks, trains, ships and farm machinery. Expensive diesel does not stay at the filling station. It works its way into the price of everything that gets transported. For the European Central Bank, which is already watching energy costs rise because of the war with Iran, an American ban would mean imported inflation. And it would not come from the market. It would come from a decision in Washington.

The two cases side by side

What weakens the argument

The argument so far is strong, but there are five things that could reverse it or soften it.

The ban may never happen. Goldman considers the scenario “very plausible”, but explicitly not the most likely one. The administration has softer options too: voluntary limits by the refiners, quotas, or a temporary suspension of the federal excise tax. Under all of them, less diesel leaves the market, or exports are not touched at all. The full chain of consequences applies only to a complete ban.

A partial limit could be tuned so that the reservoir never fills. If exports are cut only by as much as the country can absorb during the harvest and the start of the heating season, the tanks may not fill. Wood Mackenzie believes the East Coast (PADD 1), which is normally short of diesel, could take some of the surplus. That gives you the first phase without the third. That is probably the idea behind the “voluntary” cuts.

Refiners have some room to manoeuvre. They can make more jet fuel and gasoline at the expense of diesel. Under a full ban this does not solve the problem, but it delays the moment the tanks fill, and with it the cuts.

The election calendar may matter more than the physics. The election is on 3 November. If a ban starts in early October and Wood Mackenzie is right that storage around the Gulf of Mexico will fill in just over a month, gasoline will start getting dearer around mid-November. In other words, cheap diesel arrives before the election and expensive gasoline after it. We are not claiming this is the aim. But investors should keep in mind that a measure can be bad economics and sound politics at the same time. And the Russian example shows that domestic prices really do fall, even if the ban lasts only two weeks.

Perhaps the threat itself is the point. Bloomberg commentator Javier Blas has offered a more cynical explanation. In his view, the talk of a ban is a message to the EU and the UK: release your own reserves, stop bidding prices up while you refill them, and lean on Kyiv over the strikes on Russian refineries. If so, the talk alone may do the job without any real ban. The European Commission responded just a day later with a warning that any disruption would hurt both sides.

And one note on method: the gap between Brent and WTI is not a reliable gauge of the risk of a ban, because, as we saw, much of it is down to freight.

What to watch

If the ban is introduced, four markets will show very quickly whether America is getting cheaper diesel or its refineries are clogging up.

It is also worth adding RBOB gasoline futures, to see whether gasoline starts getting dearer while diesel gets cheaper, and the shares of American refiners (Valero, Marathon Petroleum, Phillips 66) against their European peers. They were the fastest to react to the news on 23 September.

For now the strongest signal is that the market is moving ahead of the decision. American diesel has fallen against European diesel, WTI has moved away from Brent, and refiners’ shares have dropped. The market is already building the dam on paper.

Conclusion

A ban on diesel exports looks like a trade measure. In fact it interferes with how refineries work, and through them with the whole chain: from gasoline in America, through the gap between the two crude benchmarks, to prices and inflation in Europe.

The 1975 law held back the raw material and indirectly sped up the growth of the world’s biggest diesel export machine. A ban in 2026 would hold back the finished product and block the outlet of the same machine. Russia tried exactly that in 2023 and gave up after 15 days.

A reservoir can hold the water back for a few weeks. The river does not stop.

Sources

1. Axios, “Trump backs diesel exports ban, marking a shift for administration”, 22 September 2026: https://www.axios.com/2026/09/22/trump-diesel-exports-ban

2. CNBC, “Oil industry warns a diesel export ban will raise fuel prices as Trump weighs restrictions”, 23 September 2026: https://www.cnbc.com/2026/09/23/trump-diesel-export-ban-gas-prices.html

3. Bloomberg, “Trump Says He’s ‘Very Seriously’ Looking at Diesel Export Ban”, 27 September 2026: https://www.bloomberg.com/news/articles/2026-09-27/trump-says-he-s-very-seriously-looking-at-diesel-export-ban

4. CNBC, “Trump ‘very seriously’ considering diesel export ban as global supply crunch worsens”, 28 September 2026: https://www.cnbc.com/2026/09/28/diesel-oil-trump-export-ban-fuel-prices.html

5. EIA, Gasoline and Diesel Fuel Update, weekly retail prices (diesel and gasoline, week of 21 September 2026): https://www.eia.gov/petroleum/gasdiesel/

6. EIA, U.S. Exports of Distillate Fuel Oil, monthly data: https://www.eia.gov/dnav/pet/hist/LeafHandler.ashx?n=pet&s=mdiexus2&f=m

7. EIA, U.S. Exports of Distillate Fuel Oil, annual data: https://www.eia.gov/dnav/pet/hist/LeafHandler.ashx?n=PET&s=MDIEXUS2&f=A

8. EIA, U.S. Exports of Distillate Fuel Oil, weekly data: https://www.eia.gov/dnav/pet/hist/LeafHandler.ashx?n=PET&s=WDIEXUS2&f=W

9. EIA, Brent and WTI spot prices, daily data: https://www.eia.gov/dnav/pet/pet_pri_spt_s1_d.htm

10. EIA, Refiner and Blender Net Input, weekly data: https://www.eia.gov/dnav/pet/hist/LeafHandler.ashx?n=PET&s=WGIRIUS2&f=W

11. Congressional Research Service, R43442, “U.S. Crude Oil Export Policy: Background and Considerations”: https://www.everycrsreport.com/reports/R43442.html

12. Congressional Research Service, R44403, “Crude Oil Exports and Related Provisions in P.L. 114-113: In Brief”, 22 February 2016: https://www.everycrsreport.com/files/20160222_R44403_3c9330ccdf269b7d812ef9985476c1cb86406a77.html

13. Wood Mackenzie, press release, 24 September 2026: https://www.woodmac.com/press-releases/a-us-diesel-export-ban-would-trigger-global-stock-drawdowns-cut-us-refinery-utilisation-and-risk-driving-gasoline-prices-higher

14. Wood Mackenzie, “US diesel export ban” report (summary): https://www.woodmac.com/reports/refining-and-oil-products-us-diesel-export-ban-150506792

15. Reuters via US News, “Talk of US Export Ban on Diesel Deepens US Crude Futures’ Discount to Global Benchmark”, 25 September 2026: https://money.usnews.com/investing/news/articles/2026-09-25/analysis-talk-of-us-export-ban-on-diesel-deepens-us-crude-futures-discount-to-global-benchmark

16. Goldman Sachs note of 26 September 2026, via ANI: https://www.aninews.in/news/business/us-diesel-export-curbs-could-lower-diesel-prices-initially-lift-gasoline-prices-later-report20260927182117/

17. Goldman Sachs note of 26 September 2026, via investingLive: https://investinglive.com/commodities/goldman-us-diesel-export-ban-would-cut-us-prices-about-4-lift-european-cost-d-uh/

18. Bloomberg, “Europe’s Diesel Prices Race Ahead of US on Export-Ban Risks”, 24 September 2026: https://www.bloomberg.com/news/articles/2026-09-24/europe-s-diesel-prices-race-ahead-of-us-on-export-ban-risks

19. S&P Global, “Russia eases diesel export ban, restores refiners’ subsidy as storage fills”, 6 October 2023: https://www.spglobal.com/commodity-insights/en/news-research/latest-news/crude-oil/100623-russia-eases-diesel-export-ban-restores-refiners-subsidy-as-storage-fills

20. S&P Global, “Russia warns over futile hopes of a quick end to fuel export ban”, 28 September 2023: https://spglobal.com/commodityinsights/en/market-insights/latest-news/oil/092823-russia-warns-over-futile-hopes-of-a-quick-end-to-fuel-export-ban

21. Reuters via gCaptain, “Russia Lifts Ban on Most Diesel Exports”, 6 October 2023: https://gcaptain.com/russia-lifts-ban-on-most-diesel-exports

22. DTN, “ULSD Futures Gain on Russian Export Ban, WTI, Brent Soften”, 21 September 2023: https://www.dtnpf.com/agriculture/web/ag/news/world-policy/article/2023/09/21/wti-slips-usd-gains-ulsd-climbs

23. The Deep Dive (start of the war with Iran, 28 February 2026): https://thedeepdive.ca/goldman-sachs-models-what-a-us-diesel-export-ban-would-actually-cost/

24. Bloomberg via The Straits Times, “Trump says he’s ‘very seriously’ looking at diesel export ban”, 28 September 2026: https://www.straitstimes.com/world/united-states/trump-says-hes-very-seriously-looking-at-diesel-export-ban

The pressure test

When engineers want to test a pipe, they do not wait to see where it leaks. They raise the pressure and watch which joint gives way first.

September was exactly that kind of test for credit markets. Three central banks tightened almost at once, oil jumped, and long US government bonds reached levels the market had not seen in almost two decades.

Most joints held. One is already leaking, and another is standing at the edge of what is acceptable.

Which ones they are, why Europe looks calmest during its biggest rise in yields in months, and which number will decide October: the full analysis is for subscribers.

Read more

This is the first monthly Macro Pulse for Japan after our second-quarter review. All the main July releases are now out. Two things that were missing in August have arrived as well: the second estimate of Q2 GDP and the Ministry of Finance’s corporate statistics. Both change the picture. In this issue we separate the money that flows into companies’ tills from the money households actually spend.

Where the economy stands

The phase in July remains growth with slowing momentum. July brought no fresh slowdown. It also did not meet the threshold we set in our Q2 review for a shift to acceleration: a positive contribution from both consumption and business investment at the same time. Japan’s economy looks like a shop with a full till and empty trolleys. Companies are earning more than ever. Households are buying less than a year ago.

The engine is global demand for chips and equipment for artificial intelligence. Export volumes are 5.2% higher than a year earlier, the fastest pace since January. Corporate profits in the second quarter were the highest since the statistics began in 1954.

The brake is households. Their real spending is 3.6% lower than in July last year, even though wages are growing at the fastest pace of the year. The brake is mainly cyclical: expensive energy after the war in the Middle East and a weak yen are eating into income. Beneath it sits a structural layer, the ageing population.

Prices are turning up and leave the Bank of Japan little room for patience. Japan’s standard core CPI, the overall index excluding fresh food, rose from 1.6% to 1.8%. The bank waited in July, the government defended the yen with intervention, and on 18 September the BoJ raised its rate to 1.25%. The bias is towards further tightening.

Core indicators

The trajectory

One month cannot decide the phase on its own. A six-month window shows whether July is a continuation or a turn. The direction compares July with January and April, following the rule set in our template.

The composite PMI has moved by less than one point across the whole window. The private sector is growing at almost the same pace as at the start of the year, despite the war in the Middle East.

Three lines carry the story. The first is wages. Total cash earnings rose 4.7% from a year earlier, the fastest this year. Real wages, meaning wages adjusted for inflation, are 2.4% above last year. June was also revised up. In our Q2 review we cited 3.4% nominal and 1.6% real growth. The final data show 4.0% and 2.2%. Workers’ income base is sturdier than we thought.

The second line is household spending. It moves the other way. The annual decline deepened for a second month in a row. The monthly gain of 0.5% after June’s 6.3% slump does not change the direction. The gap between wages and spending is the most important contradiction in Japan’s data, and we return to it in the section on the brake.

The third line is prices. All three inflation measures are turning up at the same time. Consumer prices are rising faster than in April and June. Producer prices are rising faster still. This is no longer the second-quarter picture, when consumer inflation was cooling while the pressure sat further up the chain. Now it is starting to move down.

Industrial production deserves one note. The index is almost at its June level, which was the highest of the year. The small July decline is not a reversal of June’s jump. Factory shipments rose 2.1% on the month, and capacity utilisation reached its highest level of the year. Factories are not producing for the warehouse. They are selling.

The engine: chips fill the till

Japan’s engine in July comes from abroad and has a specific name: demand linked to artificial intelligence. It shows up in exports, in orders, in production and in profits.

Exports rose 23.2% from a year earlier in yen terms. Much of that is price, because a weak yen makes every good sold in dollars more expensive in yen. This time, however, volume is moving too. The physical quantity of goods exported is 5.2% above last year, after almost zero growth in May and June. The composition is clear. Exports of semiconductors and electronic components rose 49.1% by value. Exports of chipmaking equipment rose 40.7% by value and 36.2% by volume. By comparison, cars, Japan’s traditional export, rose 19.4% by value but only 0.6% by volume. Cars bring in more yen. Chips bring in more work.

Industrial production shows the same mix. In July, production machinery grew fastest, up 5.1% on the month, led by semiconductor equipment. Information and communication machinery rose 10.2%, electronic components 2.2%. The declines were elsewhere: transport equipment excluding cars, business machinery and metal products.

Orders confirm this is not a one-off. Core machinery orders, meaning private-sector orders for machinery excluding volatile orders for ships and from power utilities, fell 3.7% on the month. That is the usual monthly swing of the series. From a year earlier they are 11.2% higher than in July 2025. Firms keep ordering equipment.

How big is the engine? Here we need to be careful. Manufacturing and exports are a smaller part of Japan’s economy than household consumption. One sector can fill companies’ tills without lifting the whole economy. That is exactly what the GDP data say. The second estimate revised Q2 growth up from 0.3% to 0.4% on the previous quarter. But most of the contribution still comes from net exports and government consumption. Domestic demand as a whole still subtracts from growth.

Is the engine durable? Probably yes, in the short run. The August manufacturing PMI, a survey of purchasing managers in which a reading above 50 signals improvement, rose from 54.5 to 54.9. New orders at factories grew at the fastest pace since January 2018. The companies themselves name semiconductors and AI-related products as the reason. Factory hiring was the fastest since February 2018. This does not look like demand pulled forward, as we suspected in our Q2 review. It looks like a cycle.

What could stop it? Three things. The first is a reversal in global investment demand for AI infrastructure. Japanese equipment suppliers sit at the start of the chain and are the first to feel it if large buyers delay projects. The second is a strong yen. If the yen appreciates sharply, the same sales will bring in fewer yen and profits will shrink. The third is energy. Japan’s factories run on imported fuel, and the war in the Middle East remains an open risk to supplies.

The brake: the household wallet

The brake is in the household wallet. Real spending by households of two or more people fell 3.6% from July last year. In nominal yen the decline is 1.5%. So households are not only paying more for the same basket. They are buying less.

Retail sales look better. Sales rose 2.1% on the month and 3.7% from a year earlier. But these are nominal values. Goods in the consumer basket cost 2.7% more than a year ago. Once price increases are removed, the real growth in sales is much more modest. July’s rebound also follows a 3.9% drop in June. Taken together, the two months show a plateau, not a recovery.

Why are households spending less while wages grow 4.7%? The answer lies in who receives the wages. The wage survey measures pay for employees at firms with five or more staff. The spending survey covers all households of two or more people, including those in which nobody works. In ageing Japan, many of them live on pensions. Pensions do not rise at the pace of wages, while higher food and energy prices hit them in full. So strong wage growth and weak consumption do not contradict each other. They describe different people.

Sentiment is improving, but slowly. Consumer confidence, a Cabinet Office index of household expectations, rose from 33.8 in June to 34.9 in July and 35.5 in August. The direction is right. The level, however, remains far below February’s 39.7. In March, after the war in the Middle East broke out, the index fell 6.4 points. The survey of shop staff, taxi drivers and others in direct contact with customers, known as Economy Watchers, improved from 44.0 to 45.7. A reading below 50 means more of them see deterioration than improvement.

July also brought a local shock. On 28 July an earthquake of magnitude 7.1 on the Japanese scale struck Kumamoto prefecture. Economy Watchers cite it as a factor weighing on expectations. Its effect on national data cannot be isolated from the July figures, but for the region it is real.

The labour market shows exactly where the brake sits. Unemployment fell to 2.4%. The ratio of job openings to applicants remained at 1.18. But new job postings are 0.8% below last year. The decline is concentrated in hotels and restaurants, down 10.7%, and in trade, down 7.3%. In manufacturing, postings rose 6.8%. Employers are showing where they see demand. Factories are hiring. Sectors that live on household spending are cutting their postings.

Is the brake cyclical or structural? Mostly cyclical. The energy shock, the weak yen and dearer food can recede. Food excluding fresh produce is 3.0% more expensive than a year ago, against 5.6% in January. The pressure there is already easing. If energy and the yen stabilise, real incomes will recover and consumption has room to catch up. But beneath the cyclical layer lies a structural base. An ageing population and the dependence of many households on fixed income make Japanese consumption slower to respond to good news from the labour market. That part will not loosen after one good quarter.

The second brake sits inside companies themselves. They are earning record profits but not investing in proportion. We devote the section on the overlooked detail to it.

Prices: the pressure moves down the chain

In the second quarter consumer prices were cooling while the pressure stayed with producers. In July that picture is changing. The pressure is starting to reach consumers.

The overall consumer price index rose 1.9% from a year earlier, after 1.6% in June. Japan’s standard core CPI, excluding fresh food, stands at 1.8% after 1.6%. The measure excluding fresh food and energy, sometimes called core-core, is 1.9% after 1.7%. All three are turning up at the same time. Energy prices rose on the year for the first time since the start of 2026, despite government measures to reduce household bills. Goods prices rose 2.7%. Services remain much calmer, at 1.2%.

Producer prices tell of stronger pressure. The CGPI, the Bank of Japan’s index of prices at which firms sell goods to one another, rose 7.7% from a year earlier. Here we need to correct our own reading. In our Q2 review we wrote that the CGPI had almost lost monthly momentum in July, with growth of only 0.1%. The revised data from September show a 0.4% monthly rise and 7.7% annual growth instead of 7.2%. July was not a pause. It was the high point of the year.

Import prices in yen are 29.3% above last year. The average price of crude oil imported in July was $114.35 a barrel, 60.2% more than a year earlier. In yen the increase is 78.0%, because a weaker yen adds to the dearer oil.

The gap between producer prices and core CPI is 5.9 percentage points. That does not mean consumer inflation will reach 7%. The two indices cover different things. But it shows how much cost is still waiting to be passed on. The survey of service companies suggests the pass-through is already under way. In August they raised their selling prices at the second-fastest pace in the history of the series.

The check after July is mixed. In August headline CPI stayed at 1.9%, core CPI eased slightly to 1.7%, and core-core held at 1.9%. Energy slipped back into negative territory. The CGPI fell 0.2% on the month, mainly because electricity, gas and petroleum products became cheaper. Import prices dropped 3.0%. So the peak of energy pressure may have passed. But prices excluding food and energy are not retreating, and wages are growing at the fastest pace of the year. That is exactly the combination the Bank of Japan has been waiting for over years: inflation that holds even when energy gets cheaper.

The overlooked detail: record profits, almost no investment

In the second quarter Japanese companies earned ¥44.67 trillion in ordinary profit, meaning profit from core business plus financial income such as dividends and interest. That is the highest quarterly level since the Ministry of Finance statistics began in 1954. Growth was 24.6% from a year earlier. Operating profit, from core business alone, rose 23.7%. So the record is not only an accounting effect of the weak yen on foreign earnings.

Capital spending by the same companies rose 1.6%. In manufacturing it fell 3.7%. The arithmetic is simple. Compared with a year earlier, quarterly profit is about ¥8.8 trillion higher. Investment is about ¥0.2 trillion higher. Of every extra ¥100 of profit, about ¥2 goes into new machinery, buildings and software.

The gap is sharpest exactly where the engine is. Makers of information and communication machinery increased profit by 171.2% and cut investment by 29.3%. Makers of transport equipment, mostly carmakers, increased profit by 50.6% and cut investment by 17.0%.

Why is this overlooked? Because record profit looks like proof of a strong economy. It is proof of strong companies. The difference lies in where the money goes. Profit that stays in the till turns into dividends, share buybacks or deposits. Profit that is invested turns into orders for other firms, jobs and higher productivity.

There are several honest explanations. Construction costs are rising and make projects more expensive. The shortage of workers limits how fast anything can be built. Uncertainty around the Middle East and world trade makes companies wait. And investment generally lags profit. In June, companies in the BoJ’s Tankan survey planned 6.8% more investment for the fiscal year that began in April. If those plans materialise in the second half, Q2 will turn out to be a simple delay.

For now, however, the GDP data confirm the picture. Real business investment fell 0.9% in Q2. The till is filling. The machines are not being bought. Until that changes, the strength of Japan’s companies will not become the strength of Japan’s economy.

What this means

July’s data give the Bank of Japan more arguments for tightening than for waiting. Growth continues and the corporate sector is accelerating. Inflation is turning up precisely in the measures the bank watches most closely: prices excluding food and energy, and wages. Consumer prices are still below the 2% target, but the BoJ itself expects them to move clearly above it from the second half of the fiscal year. Our last confirmed quarterly position, for Q2, was the most uncomfortable one: growth weakening while inflationary pressure rises. July does not refute it, but it tips the balance towards the inflation side.

The bank is reacting gradually. On 31 July it left the rate at 1.0% by a vote of 8 to 1. The dissenting member proposed 1.25%. On 18 September the BoJ did exactly that, by a vote of 7 to 2. The two dissenters argued that the economy had not accelerated enough and that core CPI was still below 2%. The bank says it will keep raising rates. That decision came after the period, so it is not a July observation. But it shows how the BoJ read July.

Between the two meetings the government acted through the currency. The average rate in July was 162.45 yen per dollar, a yen 10.8% weaker than a year earlier. Between 30 July and 26 August the Ministry of Finance sold dollars and bought yen worth ¥15.4 trillion. According to Reuters and AP, the US Treasury also took part in the intervention. The effect was limited: in August the yen averaged 158.81 per dollar. Intervention buys time but does not change the direction. A lasting shift can only come from interest rates.

What does this mean for liquidity? Conditions are tightening, but not abruptly. The monetary base, the money the BoJ creates directly, was 15.7% below last year in August. Banks, however, keep lending: loans grew 5.4% from a year earlier. The real interest rate, meaning the rate minus inflation, remains negative even after the September increase. With a 1.25% rate and core CPI at 1.7% in August, it is roughly minus half a percentage point. Monetary conditions still support the economy. The BoJ says so itself.

The real tightening is coming from the bond market. The yield on 10-year Japanese government bonds (JGBs) was 2.69% at the end of June and 2.80% at the end of July. On 24 September it stood at 3.07%. The two-year yield, which tracks expectations for the BoJ rate, jumped from 1.38% to 1.91% over the same period. The market is already pricing more increases. The part of credit that depends on long-term rates is becoming more expensive faster than the BoJ rate itself.

For ordinary people the picture is split in two. An employee at a large company is getting the biggest raise in years, and work is easy to find. A pensioner or a household on fixed income pays for dearer food and energy without compensation. Higher rates will help savers in yen but will weigh on borrowers with floating-rate mortgages.

What is confirmed? That the corporate half of the economy is accelerating, that wages are rising and that inflation is turning up. What remains assumption? Whether record profits will become investment. Whether higher wages will reach spending. Whether the energy peak has already passed.

Risks in both directions

The strongest argument against our phase is that we are too cautious. The PMI is above 52, machinery orders are 11% higher than a year ago, exports are growing in volume, wages are accelerating and companies are earning record profits. August data reinforce this picture. The composite PMI jumped to 53.5. According to S&P Global, this is the fastest private-sector growth since before the war broke out in February. If record profits start turning into investment and wages into purchases, Japan could move into acceleration faster than we assumed. Today’s caution would then look like a delay.

The downside risk has three sources. The first is consumption. If real household spending stays negative, growth will depend entirely on exports, which are exposed to any turn in the global AI cycle. The second is the interest rate. If the BoJ tightens faster while households are still cutting back, domestic demand could break. The third is energy. A new escalation in the Middle East would hit factory margins and household budgets at the same time.

The risk for the BoJ is asymmetric. If it moves too late, the weak yen and wages could push inflation durably above target. If it moves too early, it could stop exactly the shift from wages to spending it has been waiting for over years.

What to watch

On 30 September, preliminary industrial production and retail sales for August will show whether factory shipments keep growing. On 1 October the September Tankan will show whether companies keep their plans for 6.8% more investment. If plans are cut while profits are at a record, the detail on profits and investment becomes the main story. On 2 October, Tokyo CPI for September gives the first price signal for the month. August wages follow on 7 October and August household spending on 9 October. Together they will show whether the gap between pay and spending starts to close. On 23 October, national CPI for September will show whether core CPI is returning towards 2%.

On 30 October the BoJ decides on rates and publishes its new Outlook Report. The question is whether the bank will raise rates again before the end of the year.

The threshold for shifting the phase to acceleration stays the same as in our Q2 review: a positive contribution to GDP from both private consumption and business investment in Q3, with the composite PMI above 52. The first estimate of Q3 GDP comes out on 16 November. If real household spending remains negative on the year in August and September as well, the threshold will not be met. The threshold for a move down to stagnation is a composite PMI below 50 for two consecutive months, together with a new decline in production.

Closing

Japanese companies have never earned more, while Japanese households are spending less than a year ago. The engine is global demand for chips. The brake is expensive energy, a weak yen and an ageing wallet. The BoJ has already started tightening. The coming months will show whether money from the till reaches the machines and the wallet. For now we do not know which will come first: companies starting to invest, or rates rising far enough to put them off.

Sources

• S&P Global, Japan Manufacturing PMI, August 2026, with final July comparisons

• S&P Global, Japan Services and Composite PMI, August 2026, with final July comparisons

• Trading Economics, Japan Composite PMI, final January 2026 reading

• METI, industrial production, final July 2026 data

• METI, capacity utilisation, July 2026

• METI, retail trade, July 2026

• Cabinet Office, machinery orders, July 2026

• Cabinet Office, Q2 2026 GDP, second preliminary estimate, 8 September

• Cabinet Office, consumer confidence

• Cabinet Office, Economy Watchers Survey, July 2026

• Statistics Bureau of Japan, CPI, 2025 base, e-Stat

• Statistics Bureau of Japan, household spending, July 2026

• Statistics Bureau of Japan, Labour Force Survey, July 2026

• MHLW, Monthly Labour Survey, July 2026, preliminary

• MHLW, job openings to applicants ratio, July 2026

• Ministry of Finance, trade statistics, July 2026

• Ministry of Finance, seasonally adjusted trade statistics

• Ministry of Finance, balance of payments, July 2026

• Ministry of Finance, corporate statistics, April to June 2026

• Ministry of Finance, foreign exchange intervention, 30 July to 26 August 2026

• Ministry of Finance, JGB yields

• Bank of Japan, CGPI, August 2026, with revised July figures

• Bank of Japan, policy decision of 31 July 2026

• Bank of Japan, Outlook Report, July 2026

• Bank of Japan, policy decision of 18 September 2026

• Bank of Japan, Time-Series Data Search: monetary base, M2, lending, USD/JPY

• Japan Meteorological Agency, The 2026 Kumamoto Earthquake

• Reuters, the US Treasury explanation of the joint intervention

• Associated Press, confirmation of US and Japanese participation

Not financial advice.

Liquidity Desk | liquiditydesk.org

America has not run out of crude oil. WTI is at $100.30 a barrel, refineries are operating close to their physical limit, and measured distillate demand is lower than it was a year ago. Yet the average highway diesel price reached $6.285 a gallon in the week of September 14. This is the first move above $6 and the highest nominal reading in the EIA weekly series since 1994. The cause is downstream of the refinery. Distillate inventories are 13% below their seasonal norm, Russian exports are restricted, Ukrainian strikes are reducing Russian refining capacity, Hormuz remains unreliable, and US plants have little spare room to produce more fuel. The shock has arrived at the worst possible moment. The corn and soybean harvests are just beginning, and diesel powers the combine, the truck, and the refrigerated trailer. The August PPI already showed a 24.1% monthly jump in diesel. The September record has not yet entered the official inflation data.

Crude oil is the commodity. Diesel is the bill.

Energy: the shock is now downstream of the refinery

The US economy does not run on crude oil. It runs on gasoline, diesel, jet fuel, heating fuel, and chemical feedstocks. The distinction sounds obvious, but in September it explains the entire market.

WTI ended September 18 at $100.30 a barrel and Brent at $103.87. Over one month, the two benchmarks are up 18.85% and 13.37%, respectively. This is a new return to triple digits after the summer decline that pushed WTI below $70 in early July.

Finished fuels, however, are moving even more aggressively over a 12-month horizon. RBOB Gasoline, the US wholesale gasoline futures benchmark, is at $3.5276 a gallon and has risen 79.24% over the past year. NYMEX HO, the New York Harbor Ultra-Low Sulfur Diesel contract, is at $5.0578 a gallon. That is a 120.01% annual increase. The contract retains its historical Heating Oil name, but the product traded today is low-sulfur ULSD.

This is the difference between an oil shock and a distillate shock. WTI is up 60.74% over one year. ULSD is up 120.01%. If crude alone were driving the price, the two moves would be much closer. The gap shows that the market is paying a separate premium to turn a barrel of crude into usable fuel.

The EIA confirms the physical strain. In the week through September 11, US refineries processed 17.3 million barrels a day while operating at 96.8% of capacity. Distillate production was 5.2 million barrels a day. Inventories rose by 1.6 million barrels over the week but remained 13% below their five-year seasonal average. Gasoline inventories were 5% below normal.

This is the central paradox. Four-week distillate consumption is 3.3% lower than a year earlier. The price is not rising because Americans suddenly use much more diesel. It is rising because even weaker demand has not been enough to rebuild inventories while the system is already operating close to its limit.

The global market is intensifying the pressure. The war with Iran continues to disrupt normal flows through Hormuz. Ukrainian strikes on Russian refineries have led to domestic restrictions and a Russian diesel export ban. The US does not import Russian diesel directly, but the connection runs through the global market. When Russia removes fuel from international trade, buyers seek replacement cargoes, including from US refiners. US net distillate exports have been near or above their five-year high since February, keeping domestic inventories low. The Russian shock reaches America not through imports, but by pulling American diesel into the international market. China is also restricting fuel exports. On September 10, the US diesel crack spread, the difference between the value of diesel and crude oil, reached a record $112.17 a barrel, according to LSEG data cited by Reuters.

Local disruptions now carry more weight because the buffer is small. ExxonMobil’s 275,000-barrel-a-day refinery in Joliet, Illinois, was shut after a power outage. It was still offline on September 17, and on September 18 ExxonMobil said that the restart of individual units was under way. Gasoline prices in Chicago rose after the initial outage, and analysts warned of broader pressure across the Midwest. Joliet did not cause the national diesel record, which was reached before the outage. The incident temporarily removed capacity for about 11 million gallons of gasoline and diesel a day when the buffer was already inadequate. It shows how sensitive the system has become to every additional problem.

The July US edition set three conditions for a new energy impulse: WTI above $90, ULSD above $4, and CRB above 500 points. All three have been met. Scenario A was our base case in July, and it did more than materialise. It moved from geopolitical risk to a measurable shortage of finished fuel.

The US commodity dashboard

Prices and changes from Trading Economics at the end of the final trading session on September 18, 2026. These are indicative market series. The EIA weekly diesel price and official macroeconomic data are discussed separately.

The dashboard shows not one rally but three connected strains. The first is crude oil above $100. The second is the shortage of finished fuels, with ULSD rising twice as fast as WTI over the past year. The third is agriculture, where corn, wheat, and soybeans are climbing just as the harvest begins to reach the market.

Not every commodity confirms the same inflation wave. Gold is down 2.97% over one month and silver 1.03%, while copper is up only 1.97%. Steel has gained 2.34%. This does not look like indiscriminate buying of every real asset. The move is concentrated in energy, refined products, transport, and parts of agriculture.

We show the Containerized Freight Index because it completes the diesel story. Diesel raises the cost of moving goods inside the US, while the container index captures pressure on the international transport of finished products. An imported product can therefore arrive at a US port at a higher cost and then face a more expensive truck journey inland. The index does not prove the diesel shock and it is not the Baltic Dry Index. The BDI tracks dry bulk cargoes such as grain, ore, and coal.

The transmission chain: from the barrel to the checkout

Crude oil to the pump

US drivers are already seeing part of the new move, but not the full September increase. RBOB at $3.5276 is a wholesale price. The pump price includes the crude feedstock, refining, blending, taxes, distribution, and the retail margin. Pass-through comes with a lag and does not follow the futures contract one for one.

The EIA weekly average for all grades of gasoline reached $4.455 a gallon in the week of September 14. The politically sensitive $4 threshold has already been crossed. That hurts households, but diesel remains the more important macroeconomic channel. A driver can postpone a trip. A combine cannot postpone ripe corn, and a refrigerated truck cannot leave milk in a parking lot.

ULSD to transport and producer prices

Diesel is a cost embedded in every stage of the physical economy. It powers trucks, locomotives, construction equipment, part of shipping, and backup generators. When its price doubles in one year, the effect does not remain inside the energy category.

The BLS can already see the first wave. In August, the producer price for diesel fuel jumped 24.1% in one month. Headline PPI rose 0.4% over the month and 5.4% over the year. Final-demand goods energy increased 4.2%, transportation and warehousing 2.3%, and truck transportation 2.0%.

That was before the EIA reported $6.285 a gallon. The August data contain the beginning of the move, not the September peak. If wholesale ULSD remains around $5, the September and October PPI reports will carry a stronger diesel imprint.

Pass-through to the end consumer does not occur at one speed. Large carriers use fuel surcharges and can update contracts relatively quickly. Some retailers absorb the cost temporarily. Others change prices immediately. Perishable food is most exposed because it moves frequently and in refrigerated trailers.

Diesel and urea to the harvest

In the spring, the main agricultural question was how much nitrogen would go into the soil. In the autumn, it is how much diesel the machine will burn to take the crop out of the field and deliver it to the silo.

Urea has returned to $459.50 a ton, up 12.07% over one month. It remains below the $480 to $500 zone that we monitor, but it is already close enough to bring the input-cost risk back into the next season. Henry Hub remains cheap relative to global gas markets, but urea is an internationally traded product and the US farmer does not automatically receive the full benefit of cheap domestic gas.

Diesel is the more immediate problem. Reuters describes one farm where a single combine uses about 300 gallons, and another producer who expects to spend as much as $1,500 a day to run just one machine. Calculations by a Purdue agricultural economist put the additional fuel cost from a year earlier at about $11 an acre for corn and $7 for soybeans.

The farmer does not have the same choice as an ordinary consumer. If the crop is ready, it must be harvested. That makes diesel demand less price-sensitive during the most strained weeks.

Harvest to food and interest rates

More expensive diesel does not mean that all food prices rise by the same amount. Fuel is only one part of the final price. Processing, labour, packaging, advertising, and retail margins carry more weight for packaged products. Transport matters more directly for fresh fruit, vegetables, meat, and dairy products.

Reuters reports that produce transport prices from parts of California are 40% to 120% above last year’s levels. Refrigerated trailer rates from the Yakima Valley have reached a four-year high. This is the point where the diesel record starts turning into food inflation, but with a lag.

The August CPI does not yet show the full effect. Food prices rose 2.7% over the year, while energy was up 16.3% and gasoline 27.4%. The diesel shock first appears in PPI, fuel surcharges, and farmers’ margins. It reaches the shelf later.

The harvest began in the most expensive diesel week

September usually brings seasonal downward pressure on grain prices. The new crop starts entering the market, storage fills, and uncertainty about yields declines. This year, corn has risen 11.52% over one month just as the harvest begins. Wheat has moved above $7 a bushel. Soybeans are also higher.

Corn has the strongest fundamental case. The USDA estimates the crop at 15.80 billion bushels with an average yield of 178.5 bushels an acre. Production is 7.2% below last year and the yield is 4.3% lower. The September forecast cuts US corn output by 5.4 million tonnes from August and is the main reason the global coarse-grain estimate was reduced.

Crop condition is also weaker. As of September 13, 57% of corn was rated good or excellent, down from 67% a year earlier. Only 8% had been harvested, slightly above the five-year average of 6%. Most of the harvest cost still lies ahead.

This makes corn at $5.28 a bushel more than an energy derivative. The price reflects a lower physical yield, a more expensive harvest, and higher transport costs at the same time. Ethanol is down 0.74% over one month, according to Trading Economics, while corn is up 11.52%. The biofuel channel is not the main driver of the current move.

Soybeans offer the strongest counterargument to the thesis of a broad shortage. The USDA raised its average yield forecast to 52.8 bushels an acre and expects production of 4.535 billion bushels, 6.4% above last year. Ending stocks are projected at 310 million bushels. As of September 13, 6% of the crop had been harvested, twice the five-year average for that date.

More expensive diesel therefore does not mean that physical soybeans will be scarce. It means that a larger crop can be produced at a weaker margin. The distinction matters. The futures price and the producer’s financial position can move in different directions.

Wheat remains a separate story. US planted area is the smallest in the USDA series dating back to 1919, while exports are forecast at a three-year low. Global trade is also under pressure from problems in the Black Sea. The break above $7 confirmed the signal from the July edition that grain has its own reason to remain expensive, regardless of WTI’s daily move.

The three crops are therefore saying different things. Corn points to weaker supply and higher costs. Soybeans point to good supply but a squeezed producer. Wheat points to a long-term constraint in acreage and trade flows. Their common denominator is not scarcity. Their common denominator is diesel.

CRB above 540: a broad move, but not everything is rising

The CRB Index reached 543.93 points, up 5.92% over one month and 46.71% over one year. That is above the 500 threshold monitored in July and above the previous peak of about 516 points from the most intense phase of the Hormuz shock.

The index is not a pure oil measure. Energy carries roughly 39% of its weight, agriculture 41%, and metals account for the rest. The current break therefore carries more information than a standalone WTI spike. Energy and grain are moving at the same time.

Yet this is not a fully synchronised rally. Precious metals are correcting over the month. Copper and steel are rising moderately. Lithium is down 11.44% and cobalt hydroxide 19.80%. This weakens the argument that the market is simply trading a general debasement of the dollar or uncontrolled financial liquidity.

A more precise description is a concentrated shock to physical supply chains. Oil is rising because of geopolitical risk. Diesel adds a refining premium. Container freight reflects the higher cost of moving finished goods. Corn carries a lower production forecast. Wheat carries constrained acreage and problems in global trade.

The historical parallel with 2008 remains incomplete. The broad commodity peak then was followed by a credit collapse and a demand crash. Today, the US economy is still growing, unemployment is 4.1%, and the Fed has even raised interest rates. A sharp correction is possible, but it requires either a physical recovery in fuel supplies or a sufficiently strong contraction in demand. A higher interest rate alone does not produce diesel.

The Fed raised rates after diesel raised the bill

On September 16, the Federal Reserve raised the target range by 25 basis points to 3.75% to 4.00%. The decision was unanimous. The official explanation is straightforward: economic activity remains solid, the labour market is resilient, and inflation is too high.

The decision confirms the conclusion from our second-quarter US analysis. We placed the economy in the quadrant where growth is strengthening and inflation is accelerating, creating a bias towards tightening. On July 29, the Fed was still holding rates steady even though three members voted for an increase. By September 16, the bias had become action.

The August CPI supports that caution. Headline inflation is 3.4% over the year and 0.4% over the month. The core index excluding food and energy is 2.4% annually and 0.3% monthly. Gasoline rose 3.9% in August and 27.4% over the year. The energy component is up 16.3%.

PPI is the more important signal for the coming months. Headline producer prices are 5.4% above last year’s level. Diesel jumped 24.1% in one month, truck transportation 2.0%, and transportation and warehousing 2.3%. The shock is already embedded in business costs before it has appeared fully in the consumer basket.

The Fed’s preferred PCE measure was 3.7% in July and core PCE 3.3%. The central bank’s new projections put PCE inflation at 3.7% in 2026, core inflation at 3.4%, and the median appropriate policy rate at 4.1% at year-end. This implies one more small increase or at least an extended hold near the current level.

The classic supply-shock trap returns here. A higher interest rate can cool consumption and investment. It cannot open Hormuz, rebuild a Russian refinery, or refill US distillate tanks. If it is high enough, it can reduce diesel demand through a weaker economy. The price is less production, less transport, and more pressure on farmers and independent carriers.

The White House sees the same constraint. Reuters reports that the administration is considering the use of the Defense Production Act to expand refining capacity. Refiners themselves prefer improvements and expansions at existing plants because a new refinery would cost more and take years. That may improve the system’s resilience. It cannot lower prices before the end of the current harvest or before the November midterm elections.

The Fed and the White House are therefore working with different tools on the same problem. One can suppress demand. The other can finance future capacity. Neither can produce a spare gallon this week.

What we are watching over the next 30 days

The current market no longer needs a new geopolitical shock to remain expensive. Existing constraints only need to persist while refineries enter seasonal maintenance and the harvest raises domestic demand. A one-day fall in WTI is not enough for normalisation. The physical distillate balance must improve.

Confirmation signals: the distillate shock persists

• The EIA average remains above $6 for four consecutive weeks. This would show that the record is not a one-off price peak but a new fuel-surcharge regime.

• NYMEX HO / ULSD remains above $5 through mid-October. This threshold would sustain pressure on PPI, freight transport, and farm margins.

• Distillate inventories remain at least 10% below their five-year average at the end of September. Refinery maintenance would then begin without an adequate buffer.

• Corn remains above $5.25 a bushel until harvest passes 25%. Physical supply normally pushes prices lower. Holding above this threshold would confirm that the weaker yield and cost shock are stronger than seasonal pressure.

• The next PPI report shows another monthly increase of more than 1% in truck transportation. That would be direct evidence that September diesel is passing beyond the energy category.

Signals of controlled normalisation

• NYMEX HO / ULSD falls below $4.25 and the EIA weekly price below $5.50. The combination matters. A lower futures price without relief at the pump does not reduce current bills.

• The distillate inventory deficit relative to the five-year average narrows below 5%. This would indicate that production and imports are finally outpacing demand.

• The Joliet refinery returns to sustained operation and no new large unplanned outages occur. One plant does not solve the global problem, but its recovery reduces regional risk for the Midwest.

• There is a measurable recovery in fuel flows from Russia and through Hormuz. A political announcement without physical barrels is not enough.

• The CRB Index moves back below 500 and corn below $4.75. That would mean the correction has broadened and is no longer limited to crude oil.

Scenario framework

Scenario A remains our base case, but this does not mean an uninterrupted rise in price. WTI can move sharply in either direction on every headline from Iran, Russia, or Ukraine. The diesel balance changes more slowly. Inventories are low, refineries are operating close to their limit, and seasonal maintenance lies ahead.

The strongest counterargument should not be ignored. The current $6.285 is a nominal record, not an inflation-adjusted one. In 2026 dollars, the 2022 peak around $5.82 would be approximately $6.56, while the 2008 peak around $4.74 would be close to $7.20. Measured distillate consumption is also lower and the soybean crop is larger. If physical flows recover, the price can fall quickly.

The problem is timing. The farmer pays for fuel now. The carrier pays for fuel now. The inflation data will show it later.

Oil can fall on one headline. The harvest cannot wait for the next one.

Sources

• Trading Economics, Commodities: market prices and monthly changes as of September 18, 2026; annual changes come from the screenshots supplied by the author.

• U.S. Energy Information Administration, Weekly Retail Gasoline and Diesel Prices: the $6.285 highway diesel price for the week of September 14 and the series history since 1994.

• U.S. Energy Information Administration, Weekly U.S. All Grades All Formulations Retail Gasoline Prices: the $4.455 gasoline price for the week of September 14, 2026.

• U.S. Energy Information Administration, Weekly Petroleum Status Report: refinery inputs, capacity utilisation, production, inventories, imports, and consumption for the week through September 11, 2026.

• U.S. Energy Information Administration, What goes into diesel prices?: the international shortage, US net distillate exports, and the connection to low domestic inventories.

• The White House, Executive Order 14384: the continuing US ban on imports of Russian oil and petroleum products.

• Associated Press, U.S. diesel prices soar past $6: comparison with the inflation-adjusted peaks in 2008 and 2022.

• Reuters, U.S. diesel passes $6 for the first time: global drivers of the shortage, US inventories, and the crack spread, September 10, 2026.

• Reuters, record diesel squeezes farmers and food transport: combine fuel use, the additional cost per acre, and food transport, September 18, 2026.

• Reuters, Joliet refinery remains offline: the refinery’s capacity and status on September 17, 2026.

• Reuters, Joliet refinery unit restart is ongoing: the restart of individual units on September 18, 2026.

• CBS News Chicago, gas prices rise after Joliet refinery outage: the regional price effect after the shutdown.

• Reuters, White House weighs use of the Defense Production Act: measures under consideration to expand refining capacity, September 11, 2026.

• U.S. Bureau of Labor Statistics, CPI August 2026: headline and core inflation, energy, gasoline, and food.

• U.S. Bureau of Labor Statistics, Employment Situation: the August 2026 unemployment rate.

• U.S. Bureau of Labor Statistics, PPI August 2026: headline PPI, diesel, energy, transportation, and truck transportation.

• U.S. Bureau of Economic Analysis, Personal Income and Outlays, July 2026: PCE and core PCE.

• FTSE Russell, CoreCommodity CRB Index Methodology: composition and weights in the CRB Index.

• Federal Reserve, FOMC statement, September 16, 2026 and Summary of Economic Projections: the rate decision and the inflation and policy projections.

• USDA NASS, September Crop Production briefing: corn and soybean acreage, yields, and production.

• USDA NASS, Crop Progress: harvest progress and crop condition as of September 13, 2026.

• USDA ERS, Corn and Other Feed Grains Outlook: the September change in the US corn forecast.

• USDA ERS, Soybeans and Oil Crops Outlook: soybean production, yield, ending stocks, and seasonal price.

• USDA ERS, Wheat Outlook: wheat acreage, exports, and global trade.

This publication is for informational and educational purposes only. It does not constitute investment advice, a recommendation, or a solicitation to buy or sell any financial instrument. Past performance is not indicative of future results. Every investor should conduct independent research and assess their personal tolerance for risk.

Liquidity Desk | liquiditydesk.org

This is the first monthly Macro Pulse for the eurozone after our second-quarter review. The main July releases are now available. Industrial production, published on 16 September, was the last essential piece. It did not confirm the strength suggested by business surveys. The final August inflation reading, published on 17 September, is included only as a check on the direction of prices after July. We need to separate what firms expected to do from what they actually produced and what households bought.

Where the economy stands

The phase in July is stagnation with improving leading signals. We called the previous quarter an acceleration. We now know more about what drove that quarter’s growth, and we have the first full set of real-economy data for the following month. Together, those checks do not allow us to repeat the earlier label as an established fact.

The business dashboard is turning green. The composite PMI, a survey of purchasing managers in which a reading above 50 signals expansion, rose from 50.0 in June to 52.0 in July. Services moved from 49.4 to 51.7, while the manufacturing PMI climbed from 51.4 to 51.9. The economic sentiment indicator recovered to 97.1 from its spring low of 93.7. Goods exports rose 1.2% from June. Corporate lending grew faster too.

The wheels are still spinning, however. Industrial production fell 0.1% in July after a revised 0.1% decline in June. It was unchanged from a year earlier. Retail sales volume dropped 0.6% on the month. Unemployment held at 6.4%, a cushion against a deeper downturn but not proof of renewed acceleration. The eurozone has no monthly GDP release, and statistics on services arrive later.

The engine is recovering business activity and some external demand. The brake is the household sector. Purchases are weakening as energy becomes more expensive and credit harder to obtain. That brake looks mainly cyclical. Higher bills and borrowing costs can recede without permanent damage to productive capacity. There is also a longer-term risk to export competitiveness that one better month will not solve.

Prices make the ECB’s position uncomfortable. Headline inflation rose from 2.8% in June to 2.9% in July, while producer prices accelerated again. The final August reading reached 3.2%. The ECB has already raised its deposit rate from 2.25% to 2.50% in September. From here the choice is harder. Further tightening could cool purchases that are already weak, while waiting could allow the price shock to last longer.

Core indicators

The trajectory

One month cannot settle the phase on its own. The six-month view makes July look more like a crossroads. Some indicators have recovered from the spring shock. Others have not regained their earlier pace. The final column compares July with January and April under the template’s direction rule. A small difference is called small rather than dressed up as a strong trend.

The most important divergence is between the PMI and actual output. The composite PMI is now higher than in both January and April. Industrial production, however, is level with a year ago, and each of the last two months brought a 0.1% decline. The survey asks firms whether activity improved from the previous month. Official statistics measure the volume produced. They are different windows onto the same economy, not interchangeable readings. For now, the expectations window is brighter than the output record.

The second divergence is between firms and households. Economic sentiment rose from 93.7 in April to 97.1 in July. Industrial and services confidence improved within it. Consumer confidence deteriorated from -2.9 in April to -5.2 in July. Annual retail-sales growth slowed from 2.2% in January to 0.6% in July. People have not stopped buying. But the direction of their purchases runs against the direction of business surveys.

The third divergence is in prices. Headline inflation fell from a May peak of 3.2% to 2.8% in June, then edged back up to 2.9% in July. Core inflation, which excludes energy and food, moved much more slowly, from 2.2% at the start of the year to 2.5%. Producer prices made a sharper turn. Their annual increase reached 5.8%, from 4.6% in June. The price risk does not come from runaway domestic demand. It comes mainly from another rise in energy costs and the possibility that those costs gradually reach other goods and services.

The engine: business wakes up, but cannot yet pull every carriage

The recovery begins with firms. Services crossed back above the PMI’s expansion line of 50 in July after falling below it in June. Manufacturing remained in expansion. That is a meaningful change from the second quarter, when the factory survey looked stronger than the services survey. If both sectors keep expanding, the economy has a chance to turn a one-month signal into broader growth.

We cannot measure the engine by a survey alone. Industrial production fell 0.1% in July. The detail beneath the headline is better, but mixed. Output of materials and parts used by other factories rose 0.3%. Capital goods rose 0.5%, energy 0.9%, and durable consumer goods 0.9%. A 1.6% fall in non-durable consumer goods outweighed those gains in the aggregate. The weak headline is not a uniform factory collapse. Nor does it justify calling the whole sector an accelerating one. Four groups improved, but total output still declined.

External trade offers more tangible support. Seasonally adjusted goods exports rose from EUR 257.7 billion in June to EUR 260.8 billion in July, an increase of 1.2%. Imports fell 0.4% to EUR 255.7 billion. The monthly surplus therefore grew from a revised EUR 1.0 billion to EUR 5.0 billion. These are flows of real goods, but the published trade values are nominal. On their own, they do not tell us how much the physical volume increased. That is why we do not automatically turn the larger surplus into a contribution to real GDP in the third quarter.

Durability is in doubt here too. The unadjusted trade surplus for January through July was EUR 17.0 billion, against EUR 92.8 billion in the same period of 2025. Exports over those seven months grew 1.2% from a year earlier; imports grew 5.8%. One better July does not erase that gap. Some of the monthly improvement may reflect the timing of shipments rather than a lasting recovery in competitiveness. The ECB’s September projections also warn that more expensive energy, the euro’s appreciation since late 2024, and US tariffs are weighing on exports.

Credit adds fuel, but distributes it unevenly. Broad money M3 grew 3.4% from a year earlier in July, after 3.3% in June. Loans to non-financial corporations, adjusted for transfers, accelerated from 4.0% to 4.4%. Household lending moved less, from 3.0% to 3.1%. More borrowing does not necessarily mean more productive investment. The ECB’s bank lending survey shows that some business demand is for inventories and working capital, while some is for large-company investment and refinancing old debt. Those motives have different shelf lives.

Firms are borrowing more despite stricter rules. In the second quarter, the share of banks tightening their approval standards for business loans exceeded the share easing them by 7 percentage points. For loan demand, banks reporting an increase outnumbered those reporting a decline by only 3 points. This is not a credit boom. Some firms have investment plans and liquidity needs, but banks have not opened the floodgates. Whether the better surveys become real growth over the next few months may be decided here.

The brake: households feel the cost before the recovery

The brake is household spending. Retail-sales volume fell 0.6% in July from June. Food, drinks and tobacco sales rose 0.4%, but non-food goods excluding fuel fell 1.4%, and automotive fuel sales declined 0.8%. Annual growth for the total was only 0.6%. Consumers are cutting back on purchases they have more freedom to postpone. The detail shows where they are saving.

The labour market is still preventing a deeper slide. Unemployment was 6.4% in July, the same as in revised June data. Eurostat estimates 11.264 million unemployed people in the eurozone. Employment increased 0.1% in the second quarter to 176.4 million. But a quarterly gain of 0.1% is not a strong new hiring wave. It is a stabiliser. If jobs hold, households can absorb some of the price shock. If hiring weakens, today’s caution could become a longer decline in purchases.

People’s own expectations explain why low unemployment is not enough. In the ECB’s July survey, the median household expected its nominal income to rise 1.0% over the next twelve months, but its spending to rise 3.6%. This is not an ECB forecast of actual income or expenditure. It measures how consumers see their own prospects. The gap says they are preparing for a tighter budget. Consumer confidence deteriorated to -5.2 points in July even as business sentiment recovered.

Banks reinforce the brake. For housing loans, the share of banks tightening approval standards exceeded the share easing them by 9 percentage points. For consumer credit, the difference was 12 points. The demand picture runs the other way. Banks reporting weaker demand outnumbered those reporting stronger demand by 15 points for mortgages and 2 points for consumer credit. The direction matters more than the size of those gaps. Households are pulling back from debt just as firms are seeking more finance. The same interest rate feels different to a company with orders and to a family deciding whether to buy a home.

For now, this looks mainly cyclical. Energy bills raise costs, interest rates make credit-funded purchases more expensive, and uncertainty encourages delay. If energy prices calm down and real incomes recover, the brake can ease. The more persistent problem is the eurozone’s exposure to imported energy and external trade shocks. One ECB decision cannot make either disappear. The distinction between temporary pressure on household budgets and a longer-term vulnerability matters.

Prices: the shock has not travelled through the whole chain

Headline eurozone inflation was 2.9% in July from a year earlier, against 2.8% in June. Core inflation, excluding energy and food, was 2.5% after 2.4%. Both rose, but not by the same amount or for the same reason. Energy prices were up 10.3% from a year earlier, after 8.5% in June in Eurostat’s published table. Services inflation edged from 3.2% to 3.3%. Non-energy industrial goods moved from 0.7% to 0.9%, while food, alcohol and tobacco eased from 1.5% to 1.2%.

This is not broad-based overheating. Energy is the hottest point. It also touches almost every firm and household. A company facing a higher bill for transport, electricity or heating may initially accept a lower margin. If the shock persists, it becomes more likely to raise its final prices. That is why producer prices deserve a separate look. We examined why gas can become more expensive for Europe even without a physical shortage, and how that cost reaches industry and households, in our August Commodity Snapshot.

Industrial producer prices rose 1.6% in July alone and 5.8% from a year earlier. In June the monthly change had been -0.3% and the annual rise 4.6%. The energy component jumped 5.6% on the month and 12.9% on the year. Excluding energy, producer prices were unchanged from June and 3.1% higher than a year earlier. The distance between PPI at 5.8% and consumer inflation at 2.9% is 2.9 percentage points. It does not mean consumer inflation must reach 5.8%. The two indices cover different things. It does show where firms meet cost pressure before consumers see a final price.

The check after July is uncomfortable, though not one-sided. Final headline inflation for August was 3.2%, below the initial 3.3% estimate. Energy accelerated to 14.3%, but services cooled to 3.0% and core inflation slipped to 2.4%. The renewed rise in the headline rate is still mainly an energy story. Softer core inflation is a real argument against the claim that the shock has already spread everywhere. If services keep cooling, the ECB may pause after its September increase. If prices outside energy rise with a lag, its room to wait will narrow.

The ECB’s September projections put average headline inflation at 3.0% in 2026, 2.5% in 2027 and 2.1% in 2028. That is a scenario, not an outcome already recorded. It assumes the energy shock gradually fades. The bank also describes worse scenarios if the conflict in the Middle East lasts longer and costs feed more strongly into wages and prices. For the reader, the important line is between temporarily expensive energy and persistently higher inflation outside energy.

The overlooked detail: what the second-quarter revision actually changed

Our published second-quarter analysis described faster GDP growth supported by inventories, with the external sector holding it back. That causal story no longer stands. The breakdown of GDP by component had not been released when we wrote it. Eurostat published the figures on 7 September, and they point the other way. Changes in inventories subtracted 0.5 percentage points from quarterly growth. Net exports added 0.9 points. Household consumption contributed 0.2 points, while government consumption and fixed investment contributed little.

This is not a minor statistical adjustment. The earlier article saw factories producing for stock as the main engine and trade as a headwind. The final account says inventories were a brake and net exports carried growth. We cannot change the published article. We can say clearly here what we learned after publication and why it changes the next reading.

At the same time, headline GDP growth was revised up from 0.4% to 0.6% from the first quarter. That looks like stronger acceleration until we examine Ireland. Its GDP jumped 10.2% in the quarter amid multinational-company activity that says little about local demand. The ECB substitutes a measure of Irish domestic activity that excludes large, one-off asset transactions for Irish GDP. On that adjusted basis, the eurozone grew 0.3% in the second quarter, exactly as it did in the first. The headline accelerated. The underlying pace did not.

This is why our July classification is stricter. Better surveys are a promise. Stronger exports are support. But once the second quarter shows no acceleration on the ECB’s cleaner measure, and July’s industrial and consumer volumes contract, we cannot call that promise fulfilled. This judgement may prove too cautious. At least it does not repeat a convenient story that the new data have already overturned.

What this means for the ECB and liquidity

The July data alone give no clean signal for either another rate increase or easing. Headline inflation accelerated, but industry and retail sales weakened. The ECB held its deposit rate at 2.25% on 23 July. On 10 September it raised the rate to 2.50%, effective 16 September. That is a decision already taken, not proof that the July economy accelerated. If energy pressure remains high and spreads into other prices, the next move may again be tightening. If core inflation continues to cool while real volumes stay weak, a pause becomes more likely. Headline inflation does not yet support rapid easing.

Liquidity conditions are tightening through two channels. The first is the price of money. A higher deposit rate influences short-term market rates and gradually the cost of new loans. The second is bank behaviour. Banks are already reporting tighter approval standards for business loans, mortgages and consumer credit. The higher policy rate compounds that caution. Even if M3 and lending are growing from a year earlier, financing is not becoming easier for every new project or purchase.

For a household, the effect is straightforward. More expensive credit makes a mortgage or an instalment purchase harder to carry while higher energy bills take another slice of the budget. For a firm, the outcome depends on orders and its balance sheet. A company with enough sales may keep investing despite the rate. Another may borrow only to cover working capital or refinance an old obligation. That is why corporate loan growth of 4.4% does not equal acceleration across the whole economy.

The biggest unknown is how far the energy shock will travel. For now, August’s cooling in services and core inflation argues for waiting rather than an automatic further increase. Producer prices show that the risk has not passed. In every scenario, someone bears part of the cost.

Risks in both directions

The strongest case against our stagnation label is that surveys sometimes see a turn before official statistics do. July’s PMI was above 50 in both manufacturing and services. Four of the five main industrial groups increased output on the month despite the weak aggregate. Exports rose, and corporate lending accelerated. If total industrial production and retail sales return to growth in August, July’s wheelspin may look like a brief interruption in a recovery already under way. The ECB also expects support from public spending and investment in infrastructure, defence and technology. This is the strongest honest version of the optimistic case.

The opposite risk is that energy stays expensive for longer. Business costs would keep rising while household income struggles to catch up. Further credit tightening could postpone home purchases and investment. July’s nominal trade surplus may also mislead if import prices and the timing of shipments drive the balance more than durable export strength. In that scenario, the surveys will have signalled a false start.

Neither direction is certain. Stagnation in this article is not a verdict on the entire third quarter. It is the most economical name for what July recorded while we wait to see whether recovering confidence turns into goods produced, services delivered and purchases made. If all three move together, the phase must change. If prices rise while volumes fall again, the problem is more serious than a temporary pause.

What to watch

Eurostat publishes July construction output on 18 September. It will show whether stronger construction confidence has begun to appear in real activity. August unemployment follows on 1 October and the September flash inflation reading on 2 October. If headline inflation keeps rising while core remains near 2.4%, the debate will be about how long the energy shock lasts. If core moves back above 2.5% too, the risk of broader inflation increases.

August producer prices are due on 5 October, retail sales on 6 October, industrial production on 15 October and goods trade on 16 October. The threshold for changing our phase is testable. If August industrial output rises by more than 0.3% on the month, retail sales turn positive and the PMI stays above 50, there will be a case to move from stagnation to acceleration. If volumes keep falling despite strong surveys, the current classification stands.

Closing

July looks like a car whose engine is revving again while its wheels have yet to grip the road. Business surveys, exports and credit give the eurozone a chance to accelerate. Production and household purchases do not yet confirm it. The new breakdown of the second quarter also demands more humility than our earlier account. Inventories did not drive growth, and net exports did not hold it back. The next real volumes, not the next promises, will decide whether the economy has left stagnation. Meanwhile the ECB must contain inflation without stalling a recovery that has barely begun.

Sources

• S&P Global, final eurozone composite PMI for August 2026, including final July comparison

• S&P Global, final eurozone manufacturing PMI for August 2026, including final July comparison

• Eurostat, July industrial production, 16 September 2026

• Eurostat, July retail trade, 4 September 2026

• Eurostat, final July inflation, 19 August 2026

• Eurostat, final August inflation, 17 September 2026

• Eurostat, July industrial producer prices, 3 September 2026

• Eurostat, July unemployment, 1 September 2026

• Eurostat, economic sentiment indicator, official series

• Eurostat, sectoral confidence, official series

• Eurostat, July international trade in goods, 15 September 2026

• Eurostat, second-quarter GDP and employment, third estimate, 7 September 2026

• ECB, September macroeconomic projections and Ireland-adjusted activity measure

• ECB, monetary developments in July 2026, 27 August

• ECB, July bank lending survey, 21 July

• ECB, July Consumer Expectations Survey, 21 August

• ECB, monetary policy decision, 23 July 2026

• ECB, monetary policy decision, 10 September 2026

Liquidity Desk | liquiditydesk.org

After the detour

August showed a market that could still rise even as central banks withdrew part of the underlying liquidity. Bank credit kept the detour open, large technology companies led, and Asia looked like the cleanest place to take risk.

That route is not closing in September, but it is getting narrower. The market is choosing specific sources of profit: energy, semiconductors and individual companies with resilient cash flow. Technology remains strong over longer horizons, but the move is concentrating in Taiwan and South Korea. Energy continues to lead in both the United States and Europe. At the same time, industry, materials, consumption, real estate and much of the European market are falling together.

The question this month is therefore not whether there is an appetite for risk. There is, but it is selective. The question is why investors are paying for two narrow themes while abandoning the sectors that should benefit from broad economic acceleration.

Macro backdrop

The rate environment

The long end of the bond market is moving higher in both the United States and Europe. This is the most direct pressure on sectors with distant earnings, heavy debt or strong dependence on consumer credit.

The Federal Reserve kept its target range at 3.50% to 3.75% on 29 July, while the effective rate remains 3.63%. The ten-year US Treasury yield, however, reached 4.95% on 10 September, up from 4.63% on 13 August. The gap between the ten-year yield and the effective federal funds rate is now 1.32 percentage points. The market is demanding more compensation for long-term risk even though the Fed has not changed its short-term rate.

The European Central Bank took the next step. On 10 September, it raised all three policy rates by 25 basis points. From 16 September, the deposit facility rate rises to 2.50%, the main refinancing rate to 2.65% and the marginal lending rate to 2.90%. The reason is energy-driven inflation pressure from the conflict in the Middle East. The ECB now forecasts inflation of 3.0% in 2026, 2.5% in 2027 and 2.1% in 2028. The ten-year German government bond yield has reached 3.56%, up from 3.22% on 14 August.

The Bank of Japan is holding its policy rate at 1.00%, but its long end is becoming more expensive again. The average yield at the 30-year Japanese government bond auction on 3 September was 4.079%, up from 3.937% in August. The gap to the policy rate is 3.079 percentage points. This matters for the carry trade, the practice of borrowing cheaply in yen and investing the money in higher-yielding assets elsewhere. As long-term funding in Japan becomes more expensive, that trade becomes less attractive. The bank’s next meeting is on 17 and 18 September.

The broad dollar

The Federal Reserve’s Broad Dollar Index stood at 118.0732 on 4 September, the latest available date, down from 119.1790 on 11 August. That is a decline of 0.93%.

A weaker dollar works through two channels. It raises the dollar value of foreign revenue for US companies and reduces the burden of dollar debt for emerging economies. The second channel is visible this month in Brazil and the broad Latin American fund. Mexico and Chile remain negative, however. The dollar matters again, but it is not enough to explain the whole region.

Risk appetite

The VIX stood at 17.84 on 10 September, up from 14.63 on 13 August. The VIX measures the price of protection against short-term moves in the US equity market. Below 20 signals calm, above 20 greater nervousness and above 30 active fear.

The market is more tense, but it is not in panic. That distinction matters. The declines in industry, consumption and real estate are not a forced flight to cash. They are a repricing under pressure from higher long-term rates and a narrower choice of profitable themes.

United States: One leader, broad retreat

Energy is the only convincing US leader. XLE gains 8.24% for the month after 9.11% in August. This is not a one-off jump. The fund is up 11.83% over three months, 17.16% over six and 46.20% over one year. Continued uncertainty around energy flows through the Middle East is keeping a premium in the sector, while the ECB now treats the energy shock as persistent enough to justify higher rates.

Technology looks resilient only when viewed beyond the latest month. XLK is up 0.72%, but its six-month return remains 34.28% and its one-year return 38.94%. This is cooling after a powerful cycle, not a collapse of the thesis. The difference is that the US fund is no longer accelerating, while the Asian markets that manufacture the key chips and memory are.

Communication Services is the second positive sector, but it gains only 0.69% and remains negative over one year. This is an early recovery attempt from a weak base, not established leadership.

The rest of the market is retreating. Industrials fall 6.63%, consumer discretionary 5.61% and materials 4.19%. All three were positive in August. The reversal is too synchronised to be read as a collection of isolated corporate problems. A higher ten-year yield makes investment, housing, cars and credit-funded consumption more expensive at the same time.

Financials and health care are also negative for the month, but their longer periods remain strong. XLF is up 9.61% over three months and XLV 8.18%. This is why the relative rotation graph keeps them in Leading despite the negative month. They still carry accumulated relative strength against the S&P 500, but their momentum is fading.

RRG: The slower signal

The horizontal axis shows relative strength against SPY, while the vertical axis shows the change in that strength. The upper-right Leading quadrant combines strength with positive momentum. The lower-right Weakening quadrant means a sector remains strong but is losing speed. The upper-left Improving quadrant shows recovery from a weak base, while the lower-left Lagging quadrant combines a weak position with weak momentum.

The monthly horizon filters out weekly noise and shows the more persistent direction. XLE is the clean signal. Energy is in Leading and moving up and to the right. XLK also remains in Leading and has the highest relative strength on the map, but its latest step is down and to the right. Technology is preserving its leadership while momentum gradually cools.

XLI is now in Weakening and moving down and to the left. This is the clearest confirmation that the broad cyclical trade is losing strength. Materials, financials and health care sit in Improving, but their latest direction does not show a common acceleration. XLV is moving to the right, while XLB and XLF are losing momentum. The quadrant itself changes for materials and financials under 8-, 10- and 12-month smoothing, so those signals remain less certain.

Communication Services, consumer staples, real estate, utilities and consumer discretionary remain in Lagging. XLC is beginning to lift its momentum, but its relative strength is still weakening. The rest of the group is moving down and to the left. The monthly map does not show a broad sector turn. It shows two persistent themes, energy and technology, against a market that is still searching for direction.

The RRG does not measure actual money flows. It shows relative performance and the change in that performance. The chart is therefore a map of market preference, not proof that every dollar leaving industry has moved into energy. It is a transparent RRG-style approximation built from the final available weekly adjusted price in each month, not the official proprietary JdK model. The last point is the incomplete September observation through 11 September 2026.

Europe: Energy holds, breadth disappears

Europe repeats the US energy story, but with a heavier decline below the surface. Oil and gas lead with a 5.04% gain. US XLE is stronger at 8.24%, again illustrating the difference between a producer that benefits directly from a higher commodity price and a region that absorbs more of the inflation cost of imported energy.

Telecommunications are the surprise. EXV2 reverses from a 1.45% decline in August to a 3.54% gain. Its three-month return remains negative at 6.37%, however, so this is stabilisation from a low base. Automobiles also move slightly above zero, but their six-month and one-year returns remain negative. The structural pressure from Chinese competition and Europe’s slow domestic transition has not disappeared.

Banks retain Europe’s strongest established trend. The monthly gain is only 0.44%, but the three-month return is 17.64%, the six-month return 25.13% and the one-year return 41.72%. Higher rates support income from lending margins as long as the economy avoids recession. The ECB’s latest increase extends that support, but it also raises the risk to credit quality.

Basic resources also deserve a separate place. EXV6 adds 1.02% in September after gaining 5.15% in August and remains among the few European sectors in positive territory. It is the clear one-year leader at 62.70%, ahead of oil and gas and banks. This is a continuation of the commodity cycle, not a signal of broad acceleration in Europe. While industry, chemicals and construction fall, mining and metals companies preserve their accumulated premium.

The real story sits in the bottom half of the table. Technology reverses from a 4.34% gain to a 4.09% decline. Industrial goods and services swing from a 5.70% gain to a 5.03% loss. Real estate falls 6.96%, construction 6.88%, health care 7.14% and retail 8.47%.

European health care remains weaker than its US counterpart despite the same demographic support from ageing populations. The difference lies in composition and regulation. The European fund is more concentrated in pharmaceutical companies exposed to price controls, while the US fund includes a broader set of insurers, service providers and medical device makers.

Asia: Semiconductors against China

In Asia, we use country funds rather than sectors. Currencies, regulation and economic cycles differ too widely to fit into one common sector index. This month, the country map shows the clearest division anywhere in the world.

South Korea leads with a 15.69% monthly gain. The signal is unambiguous. EWY is up 44.84% over six months and 146.21% over one year. Samsung Electronics and SK Hynix together account for almost half of the fund, making Korea a direct position in memory and artificial intelligence infrastructure.

Taiwan adds 8.54% and remains the structural six-month leader with a 57.45% return. TSMC and its surrounding supply chain give the market something missing from the broad cyclical trade: visible scarcity, strong demand and a specific role in global capital spending on artificial intelligence.

Japan is up 2.61%, but slows from August. The higher long end sends a two-sided signal. It attracts capital towards Japanese bonds and reduces the appeal of the carry trade, while also showing a normalisation of domestic yields. The BOJ meeting on 17 and 18 September will determine which force dominates.

China turns fully downward. FXI falls 5.48% and the broad MCHI fund loses 6.97%. Their six-month returns are negative 5.74% and negative 9.78%, respectively. September’s liquidity analysis already showed the mechanism. China’s social financing in July was supported almost entirely by government bonds, while bank lending and household borrowing contracted. The state can create financing, but private credit impulse has not started.

India is also negative and remains down over both six and twelve months. This continues to contradict the idea that India automatically benefits whenever capital leaves China. Investors now want a specific source of profit and a defensible valuation, not merely an alternative geography.

Latin America: The dollar helps again, but not everyone

The weaker dollar is once again a tailwind for Latin America. It reduces the local cost of dollar debt and supports the dollar value of local assets. The regional ILF fund gains 4.20% after losing 1.50% in August.

Brazil makes the largest reversal. EWZ swings from negative 4.13% to positive 8.53% and is now up 13.06% over three months. Its six-month return of only 1.76% shows that this is a strong tactical recovery, not a completed structural breakout. High domestic real rates and fiscal risk have not disappeared.

Argentina returns to positive territory with a 2.11% gain. Its reform thesis, based on fiscal consolidation and liberalisation of the currency regime, remains separate from the regional cycle. It does not lead this month, but neither is it breaking down.

Mexico and Chile show the limit of the dollar explanation. EWW falls 2.01% and ECH 3.40% despite the weaker dollar. Mexico remains exposed to the US trade and political cycle. Chile is an indirect bet on Chinese industrial demand through copper, and weakness in Chinese equities offers no confirmation of a new commodity cycle.

The rotation signal

Capital is not fleeing risk. It is refusing to pay for risk without a visible source of profit.

The first theme is the persistent energy premium. XLE and EXH1 remain at the top for a second consecutive month. The move is symmetric across the United States and Europe, which makes its cause global. The difference in scale again favours US producers. Europe gains through its companies but loses through its economy because it depends more heavily on imported energy. The ECB’s decision has converted that cost into a higher interest rate.

The second theme is the geographic split in technology. Korea and Taiwan accelerate, US XLK keeps its strength but loses momentum, and European technology returns to negative territory. The market is not buying the word AI everywhere. It pays most where scarcity in chips, memory and production capacity provides the clearest visibility into future earnings.

The third theme is the discipline of the long end. The US ten-year yield, the German Bund and the 30-year Japanese government bond yield are rising at the same time. The result appears across regions. US industrial and consumer sectors fall, European real estate, construction and retail sit among the weakest, and the Japanese market slows before the next BOJ decision.

The quiet signal is that defensive sectors are not providing defence. Consumer staples, utilities and health care fall in the United States, while their European counterparts are even weaker. If this were panic, the VIX would be above 20 and capital would be seeking predictability at any price. Instead, the VIX is 17.84. The market does not pay for a defensive label when higher bond yields offer an alternative.

September’s map is narrower than August’s, but it is no less rational. Energy has physical scarcity and a geopolitical premium. Korea and Taiwan have a specific role in the global AI cycle. Brazil is receiving help from a weaker dollar. Every other asset must prove why its earnings can withstand more expensive long-term capital.

The next test comes immediately, with Fed and BOJ decisions in the same week. If the long end continues higher, the narrow market will become narrower still. If yields settle, the first real confirmation of broader leadership should come from industry and consumption, not from another peak in energy alone.

Data and sources

• ETF performance: StockAnalysis, retrieved on 12 September 2026. Values are rolling returns as displayed by the source.

• Federal Reserve effective rate and ten-year yield: FRED DFF and FRED DGS10.

• FOMC decision of 29 July 2026: Federal Reserve.

• Broad Dollar Index: FRED DTWEXBGS.

• VIX: FRED VIXCLS.

• ECB decision of 10 September 2026: European Central Bank.

• September ECB macroeconomic projections: ECB staff projections.

• Ten-year German government bond yield: Deutsche Bundesbank.

• Bank of Japan decision and calendar: BOJ statements and BOJ meeting calendar.

• 30-year Japanese government bond auction of 3 September 2026: Japan Ministry of Finance.

• Monthly RRG-style methodology and numerical verification: Liquidity Desk calculations using the final available weekly adjusted price in each month for SPY and the 11 US sector ETFs.

Liquidity Desk | liquiditydesk.org

Introduction

August ended with the Nasdaq 100 up 4.18%, the Shanghai Composite up 4.02%, KOSPI up 3.40% and the Nikkei 225 up 3.03%. The S&P 500 added 2.62%. On the surface, the month looked like a straightforward expansion in risk appetite.

But the Russell 2000 gained only 0.86%, while the Hang Seng lost 1.23%. Large technology companies and parts of Asia led. Smaller US companies and Hong Kong lagged. The move was higher, but it was not broad.

Liquidity has not travelled only along the main road for a long time.

Net Fed Liquidity fell by $46.6 billion. The Eurosystem continued to run down its portfolios. Japan’s monetary base contracted by 15.7% over the year. The People’s Bank of China withdrew a net CNY 408.6 billion through its listed instruments in August.

Markets kept moving because the long-established detour through bank credit still works. US banks increased loans and leases by $108.4 billion between 29 July and 26 August. Corporate and household credit in the euro area is also growing. Japanese bank lending is up 5.4% over the year despite the contraction in the monetary base.

The question this month is how long the banking channel can offset the contraction at central banks. And what happens when a central bank balance sheet grows but credit does not follow. China already provides the answer.

Central Banks

The headline balance sheets are misleading this month. The Fed and the Eurosystem look almost stationary. The balance sheets of Japan and China are growing. Underneath them, though, reserves, the monetary base and operating flows point to less base liquidity, not more.

Federal Reserve

The Fed balance sheet stands at $6.731 trillion as of 27 August, against $6.738 trillion on 30 July. The decline is only $7.3 billion. With quantitative tightening over, this is not a renewed run-off programme. The balance sheet is effectively neutral.

The Treasury account is moving in the opposite direction. Its weekly average rises from $910.8 billion to $950.7 billion. Only $456 million remains in overnight reverse repo. Almost the entire monthly decline in net liquidity therefore comes from the Treasury rebuilding its cash position, not from the Fed balance sheet.

The target range remains 3.50 to 3.75% after the 29 July decision. The vote was 9 to 3, with all three dissenters preferring a 25 basis point increase.

Regime: neutral balance sheet, tightening plumbing. The Fed is not actively running down assets, but the Treasury is withdrawing more liquidity than the balance sheet returns.

European Central Bank

The Eurosystem balance sheet falls from €5.941 trillion to €5.915 trillion between 31 July and 28 August. Of the total €25.9 billion decline, €24.7 billion comes from securities held for monetary policy purposes. Lending operations fall by a further €3.8 billion.

There is no large accounting revaluation to explain the move this time. Revaluation accounts remain at €1.208 trillion and the value of gold changes by only €18 million.

Excess liquidity does not confirm sharp tightening, however. It stands at €2.149 trillion on 28 August, almost unchanged from 31 July. The reading for 31 August is €2.142 trillion, a monthly decline of €6.1 billion. The choice of end date matters more than the move itself.

The deposit facility rate remains 2.25%, the main refinancing rate 2.40% and the marginal lending rate 2.65%. APP and PEPP portfolios continue to decline as principal payments are no longer reinvested.

Regime: gradual tightening. The portfolios are running off, but available excess liquidity is still almost stationary.

Bank of Japan

The Bank of Japan balance sheet grows by ¥366.3 billion in August to ¥644.662 trillion. That looks like easing until the layers beneath the headline total are examined.

Banks’ current account deposits at the central bank fall by ¥13.690 trillion. The average monetary base declines from ¥554.926 trillion to ¥543.007 trillion, while its annual contraction deepens from 13.8% to 15.7%. The increase in assets is not reaching the system’s core reserve layer.

The policy rate remains at 1.0% after an 8 to 1 vote. The regime is therefore defined by quantities, not by the unchanged rate.

Regime: tightening. The balance sheet is larger, but the monetary base and bank deposits at the central bank are smaller.

People’s Bank of China

The latest published PBoC balance sheet is for July. Assets rise by CNY 773.3 billion to CNY 50.207 trillion. The main reason is a CNY 620.4 billion increase in claims on deposit-taking institutions.

The headline balance sheet misleads here too. Reserve money falls by CNY 104.2 billion, while deposits of financial institutions within reserve money decline by CNY 219.7 billion. Government deposits rise by CNY 735.6 billion. More central bank assets do not automatically mean more usable bank reserves.

August operations complete the picture. The medium-term lending facility and reverse repo at other maturities provide a net CNY 100 billion. Government bond purchases add CNY 50 billion and treasury cash management adds another CNY 70 billion. Against that stand a CNY 641.5 billion withdrawal through overnight and seven-day reverse repo, CNY 100 billion through the medium-term lending facility and CNY 52.1 billion through pledged supplementary lending. The sum of all listed instruments is a net withdrawal of CNY 408.6 billion.

The one-year loan prime rate remains at 3.0% and the five-year rate at 3.5%.

Regime: selective support, but net tightening across the listed operations. The PBoC is extending the maturity of some liquidity without increasing the total monthly flow.

Central Bank Synthesis

Money and Credit Transmission

The main route from central banks is narrowing. The banking channel still works in three of the four systems. The contrast is clearest between the United States and China.

United States

US M2 reaches $23.218 trillion in July, after a monthly increase of $102.8 billion. That is a rise of 0.44% for the month and 5.41% over the year.

Commercial and industrial loans give a weaker signal. They fall by 0.20% in July to $2.899 trillion after four consecutive months of slowing growth. The monthly rate declines from 1.36% in March to 1.28% in April, 1.01% in May, 0.58% in June and -0.20% in July.

The weekly picture for August is stronger. Between 29 July and 26 August, bank loans and leases rise by $108.4 billion, while deposits increase by $191.2 billion. Total bank credit grows by $62.1 billion because banks simultaneously reduce their securities holdings by $46.3 billion.

The banking offset has not disappeared. Its composition has changed. Credit to businesses weakens in the monthly series, but total loans accelerate in August, funded by stronger deposits and a smaller securities portfolio.

Euro Area

M3 is effectively stationary. The nominal stock falls by €1.0 billion in July to €17.614 trillion, while the adjusted monthly rate is positive by 0.005%. The annual rate still accelerates to 3.4%.

Credit is growing faster than the money supply. Loans to non-financial corporations add €25.5 billion in July to reach €5.456 trillion and accelerate to 4.4% over the year. Household loans increase by €18.4 billion to €6.990 trillion and grow by 3.1% over the year.

Companies continue to raise money through bonds as well. Net issuance is positive by €5.9 billion in July, though below the €11.0 billion recorded in June. This does not prove that bank credit is replacing bonds. It shows that both channels remain open.

The ECB bank lending survey supplies the constraint. A net 7% of banks tightened standards for companies, while net demand for loans increased by 3%. Credit is expanding, but the entrance to it is becoming slightly narrower.

China

China is where the detour breaks down. M2 falls by CNY 1.20 trillion in July to CNY 355.51 trillion, while its annual rate slows from 8.0% to 7.7%.

Total social financing, the broadest measure of credit reaching the economy, adds approximately CNY 1.41 trillion. Government bonds account for CNY 1.32 trillion of that amount. Without this component, the remaining net flow is only around CNY 90 billion.

Renminbi bank loans fall by CNY 340 billion during the month. Household loans decline by CNY 460.3 billion. Annual growth in outstanding bank loans slows from 5.2% to 5.1%.

The state can still create financing. The private credit impulse is not following. This is the difference between money recorded inside the system and money that starts a new cycle of spending and investment.

Japan

Japan’s M2 rises from ¥1,296.1 trillion to ¥1,297.0 trillion in July, while M3 increases from ¥1,639.8 trillion to ¥1,641.2 trillion. Based on the seasonally adjusted levels, simple monthly growth is about 0.16% for M2 and 0.04% for M3. The annual rates are 2.2% and 1.4%.

Loans by major, regional and shinkin banks reach ¥679.2 trillion, up 5.4% over the year. The rate slows from 5.7% in June but remains strong compared with broad money.

This is the cleanest example of the offset. The monetary base contracts by 15.7% over the year while bank lending grows by 5.4%. The central bank is narrowing the main road. Banks are still carrying traffic along the second route.

Money and Credit Synthesis

Net Fed Liquidity and Its Relationship to M2

Net Fed Liquidity is the Fed balance sheet minus the Treasury account and overnight reverse repo. It shows how much dollar liquidity is not locked in the government account or parked back at the central bank.

As of 27 August, NFL stands at $5.780 trillion, against $5.826 trillion on 30 July. The decline is $46.6 billion, or 0.80%.

The breakdown matters. The Fed balance sheet removes $7.3 billion. The larger Treasury account removes another $40.0 billion. The depletion of reverse repo returns only $620 million. The three parts add exactly to the total $46.6 billion decline.

The reverse repo buffer is now effectively exhausted. The next increase in the Treasury account will not be softened by tens or hundreds of billions of parked cash returning to the system. The offset must come from the Fed balance sheet, bank reserves and deposits, or later government spending from the Treasury account.

The annual divergence is even clearer. Between July 2025 and July 2026, M2 grows by $1.193 trillion, while NFL falls by $231.3 billion. The arithmetic gap between them widens by $1.424 trillion.

This is not an accounting measure of privately created money. M2 is a monthly average, NFL is a weekly snapshot and the two series measure different things. The divergence has a narrower meaning. Broad money is growing while the federal liquidity infrastructure is contracting.

Credit Stress

The full breakdown is in Credit Pulse, August 2026, with most data through 28 August and ICE spreads through 27 August. Only the regime calls and the changes that matter are carried over here.

The United States falls from WARNING to NEUTRAL. Europe remains RELAXED. Asia remains NEUTRAL. The calm is not uniform. The US credit premium layer remains in warning because the weakest borrowers continue to lag the rest of the market.

MOVE falls from 83.02 to 70.97 and removes July’s warning in rates and volatility. Corporate funding becomes cheaper relative to SOFR. Appetite for high yield debt remains strong.

The problem is concentrated in the tail. The gap between CCC and BB reaches 8.78 percentage points, the 99th percentile of its twelve-month window and 101 basis points above the end of May. But the broad high yield spread is only 2.63%, while investment grade stands at 0.79%.

This is the strongest counterargument to the case for a liquidity shortage. If the decline in quantities were already squeezing the entire system, funding, investment grade and the broad high yield market would show it. They do not.

The signal is not a broad credit accident. It is selection. The detour remains open to good risk and is almost closed to the weakest borrowers.

Transmission into Risk Assets

Equities

August confirms the message from credit. The market is buying risk, but it is choosing carefully.

The two US extremes are the Nasdaq 100 and the Russell 2000. Large technology companies gain 4.18%, while smaller companies rise by less than 1%. This is an advance, but liquidity is not being distributed evenly.

In Asia, the Shanghai Composite rises by 4.02% while the Hang Seng falls by 1.23%. The two indices look toward the same economy but receive different flows. Chinese state support reaches the mainland market more easily than international Hong Kong. That is consistent with financing dominated by government bonds, but it does not prove the connection by itself.

Europe remains positive but weak. The EURO STOXX 50 gains 0.98%, behind the United States, Japan, Korea and mainland China.

The Next 30 to 60 Days

The base regime remains selectively risk-on. Bank credit in the United States, the euro area and Japan is still offsetting the contraction in base liquidity. Credit markets do not show broad stress, while the large equity indices continue to rise.

The condition is that the detour remains open. In the United States, weekly growth in loans and deposits must continue after the July decline in business credit. In the euro area, corporate credit must withstand tighter standards. In Japan, bank lending must keep growing while the monetary base contracts.

The first risk is the US Treasury. Only $456 million remains in reverse repo. A further increase in the Treasury account no longer has a large buffer to neutralise it. If bank credit slows while more liquidity moves into the government account, the offset will disappear from both sides.

The second risk is China. July credit is almost entirely state-driven, while the PBoC’s August operations are negative on net. For the regime to change, the next release must show not just larger total social financing, but stronger bank and household credit.

The dates that will test the regime are 10 September for the ECB, 16 September for the Fed and 18 September for the Bank of Japan. The next monthly money and credit releases will show whether the banking offset continued after July.

Key Data Changes

This table preserves the memory of the publication. The previous column contains the figures published in the August issue, not necessarily values for the same calendar date. Comparable monthly moves are explained in the text above.

Final Liquidity Verdict

Regime: selectively risk-on, taking the detour.

The August argument was that liquidity was wearing thin underneath. The September issue adds the important second half. The system has not stopped because banks are carrying part of the movement that central banks no longer provide.

The four central banks are not delivering a common expansionary impulse. The Fed balance sheet is almost neutral, but the Treasury account removes $40 billion. The Eurosystem is running down its portfolios. Japan’s balance sheet grows while its monetary base falls by 15.7% over the year. China’s balance sheet also grows, but reserve money declines and August operations are negative on net.

The headline size of a balance sheet is no longer enough. In Japan and China, assets are rising while the usable reserve layer is falling. That is the central lesson of the month.

The offset comes from commercial banks. US loans and leases add $108.4 billion in August. Corporate and household credit in the euro area grows faster than an almost stationary M3. Japanese bank lending rises by 5.4% over the year while the monetary base contracts sharply.

The United States also shows the limit of this offset. NFL falls by $46.6 billion in August and by $231.3 billion between July 2025 and July 2026. Over the same annual window, M2 grows by $1.193 trillion. Broad money is expanding, but the federal liquidity infrastructure is not supporting it.

China shows what happens when the second route fails. Almost all of July’s total social financing comes from government bonds. Bank loans and household loans decline. The central bank can change the scale and maturity of its operations, but it cannot force private credit to move.

Credit markets do not yet see a broad accident. MOVE is calmer, funding is cheaper, and investment grade and the broad high yield market remain relaxed. The warning is confined to the gap between CCC and BB. The market is not rejecting risk. It is rejecting the weakest risk.

That allows the regime to remain positive but narrow. If bank credit keeps growing, markets can continue for some time without help from central banks. If it slows just as the Treasury account withdraws more liquidity and Chinese private credit remains negative, the detour will close.

The movement continues. The main road is no longer carrying it.

Sources

• Federal Reserve H.8, 4 September 2026

• FRED: WALCL, WTREGEN, RRPONTSYD, M2SL, BUSLOANS, TOTBKCR, TOTLL, SBCACBW027SBOG, DPSACBW027SBOG

• ECB: weekly balance sheet as of 31 July, weekly balance sheet as of 28 August, monetary developments for July, Bank Lending Survey

• Bank of Japan: balance sheet as of 31 July, balance sheet as of 31 August, monetary base for August, money stock for July, bank lending for July

• PBoC: balance sheet of the monetary authority, operations in July, operations in August, money and credit data for July

• ChinaMoney, LPR on 20 August

• Market series: S&P 500, Nasdaq 100, Nikkei 225, Russell 2000, EURO STOXX 50, KOSPI, Shanghai Composite, Hang Seng

Credit stress data in the Credit Stress section comes from Credit Pulse, August 2026, with most data through 28 August and ICE spreads through 27 August.

This issue tracks liquidity. Figures from other areas are covered in detail in the rest of the month’s publications: the economies of the United States, the euro area, Japan and China, Global Sector Rotation, Commodity Snapshot, Sector Deep Dive and Credit Pulse.

For informational and analytical purposes only. Not financial advice.

Liquidity Desk | liquiditydesk.org

This is our July review of the US economy, now that all the main data for the month have been released. The final July piece was the trade balance on September 3. The September 4 employment report serves as a directional check. This is not an economy coming to a stop. It has a powerful investment engine, but its transmission is delivering power unevenly to households, labor and housing.

Where the economy stands

The phase in July remains Acceleration, but the acceleration is becoming narrower and more uneven. The US economy resembles a car with a strong engine and a transmission that engages intermittently. Businesses continue to order, produce and invest. Households are spending more cautiously, housing is feeling the weight of high interest rates, and stronger demand only began to translate into more jobs in August.

The engine is visible in surveys and in hard orders. The ISM Manufacturing PMI rose to 55.6, industrial production increased 0.2%, and factory orders grew 0.9%. The second estimate of GDP strengthened the thesis from our previous Macro Pulse. Real final sales to private domestic purchasers, the cleanest measure of consumption and private fixed investment, were revised up to a 4.2% annualized rate in the second quarter.

The brake is broad enough to stop us from calling this an unqualified boom. Real consumption was virtually unchanged in July. Retail sales fell 0.6%, housing starts dropped 12.4%, and new home sales fell 10.5%. July payrolls are now estimated to have risen by 21,000 rather than fallen by 23,000. August’s gain of 162,000 weakens the darkest interpretation, but hiring was concentrated in a few sectors. This is uneven hiring, not a broad new employment wave.

Inflation leaves little room for help from the Federal Reserve. CPI is cooling, but the PCE price index is running at 3.7%, core PCE at 3.3%, and producer prices are 4.7% above a year ago. The central case is a hold with readiness to raise rates if price pressure does not ease. A rate cut would conflict with the combined signal from growth and inflation.

Core indicators

The trajectory

Three rows carry most of the story. The first is the ISM Manufacturing PMI. Its rise from 52.6 in January to 55.6 in July is too large to dismiss as monthly noise. New orders stand at 56.7 and production at 58.5. A reading above 50 means that more companies report improvement than deterioration. The improvement is now broad.

The second is employment. Monthly payroll growth slowed from 160,000 in January to 31,000 in June and 21,000 in July. The latest revisions change an important detail. July’s initially reported loss of 23,000 is now a small gain, while June was raised by 11,000. August added 162,000 jobs. That rejects the idea that the economy has already entered sustained employment contraction, but it does not erase the weak average gain of 31,000 a month over the past year.

The third is labor force participation. It fell from 62.1% in January to 61.4% in July. That is why lower unemployment cannot be read on its own. The labor force contracted by 264,000 in July, the number of employed people in the household survey declined by 87,000, and the number of unemployed fell by 178,000. Unemployment went down because both the numerator and the denominator became smaller.

Retail sales show why a mechanical classification is not enough. The July level remains above January and April, so the strict table rule still points upward. Compared with June, however, sales fell 0.6%. The longer trend remains positive. The short-term impulse is negative. That is exactly what a slipping transmission looks like.

The August data confirm the split without changing the July classification. Manufacturing ISM eased to 54.6 but remained in expansion. Services accelerated to 55.4 and new orders reached 60.9. Employment components in the surveys stayed below 50, but official payroll employment rose by 162,000. Demand is holding and hiring is recovering, although it is not yet broad.

The engine

The investment and manufacturing cycles are the engine of the US economy. This is not only a change in managers’ sentiment. Factory orders rose 0.9% in July, durable goods orders increased 1.1%, and orders excluding transportation gained 0.4%. Industrial production added 0.2%. Business equipment output rose 0.8%, while consumer goods output fell 0.4%.

That difference matters. Factories are not accelerating because Americans are buying more everyday goods. They are accelerating because businesses continue to build capacity. Equipment, software and other intellectual property were among the main sources of strength in the second quarter. July offers no evidence that this cycle has ended. Orders and imports instead show that equipment continues to enter the country.

The cleanest confirmation comes from the GDP revision. Headline growth remains at a 1.5% annualized rate in the second quarter. Beneath it, real final sales to private domestic purchasers were revised to 4.2% from the initial 3.9%. This measure combines consumption and private fixed investment while excluding government spending, inventories and foreign trade. It shows how quickly the private domestic machine is moving. The difference from 1.7% in the first quarter is too large to dismiss as rounding.

Corporate profits add fuel. They increased by $400.9 billion in Q2 after a $74.4 billion increase in Q1. Higher profits do not guarantee more investment, but they give companies internal resources to fund it. That matters when interest rates are high. A business with strong cash flow depends less on new credit and can continue a project that a weaker competitor might postpone.

Productivity is also moving in the right direction. It increased at a 1.4% annualized rate in Q2 and by 2.2% from a year earlier. Unit labor costs rose at only a 1.2% annualized rate during the quarter, according to the final estimate, revised from 1.3%. Companies are producing more per hour without labor costs moving out of control. Part of the gap between strong activity and weak hiring probably comes from here. Businesses are getting more from their existing staff and new equipment.

How durable is the engine? The evidence says it is more than a one-off jump, but it provides no guarantee. ISM has been rising since the start of the year, orders are growing, capital goods imports continue, and private domestic demand has been revised higher. These are four different windows onto the same process.

The weakness is concentration. If investment depends too heavily on data centers, computers and related equipment, a downturn in one major theme could slow the whole machine quickly. Cost is the second risk. Imported capital goods are becoming more expensive, producer prices remain high, and interest rates are not falling. The engine is running, but its fuel is becoming more expensive.

The brake

The brake has three parts. Consumption is losing monthly momentum, the labor market is creating jobs slowly and unevenly, and housing remains constrained by credit. None of these proves a recession on its own. Together, they explain why July’s acceleration does not feel evenly distributed.

Nominal consumer spending rose 0.2% in July, but after accounting for prices the real change was effectively zero. Spending on services increased by $86.2 billion, while spending on goods fell by $49.9 billion. This is an economy where more expensive and less avoidable services keep adding to the bill while goods purchases are postponed.

Retail sales confirm the pause. They declined 0.6% during the month, although they remained 5.0% above July 2025. Christopher Waller offers an important alternative interpretation. Some of the weakness may be calendar-related because promotional purchases were pulled forward into June. That is possible. It may explain part of the monthly decline, but it does not change the fact that total real consumption did not grow in July.

There is a buffer. Real disposable income rose 0.4%, while the saving rate increased from 2.6% to 3.0%. Households are not necessarily being forced to cut spending. They may have started rebuilding an unusually thin stock of savings. If so, the weakness is a cyclical pause. If incomes slow alongside employment, the pause will deepen.

The labor market remains a brake, but not in the way the first July estimate suggested. July payrolls were revised from a loss of 23,000 to a gain of 21,000, while June was revised from 20,000 to 31,000. August then added another 162,000 jobs. Unemployment held at 4.1%, participation rose to 61.6%, and employment in the household survey increased by 569,000. That is a genuine improvement from July’s picture.

The improvement is not broad enough to remove the brake. Food services and drinking places added 59,000 jobs, while local government education added 42,000. Together they accounted for 101,000, or about 62%, of the entire August gain. Manufacturing added 16,000, but health care grew more slowly than usual and information lost 23,000. Average payroll growth over the past 12 months was only 31,000 a month.

July’s JOLTS data also remain soft. Job openings recovered slightly to 7.271 million, but hires fell to 5.1 million and the hiring rate to 3.2%. Voluntary quits totaled 3.056 million. Workers do not see enough better opportunities to change jobs, and many companies still do not see enough reason to expand their teams.

This remains a low-hire, low-fire regime, not a mass-layoff regime. Layoffs and discharges totaled 1.666 million, less than in June. The classic recession mechanism, in which job losses reduce incomes, spending and then more jobs, is not yet present. The danger is different. If weak hiring persists for long enough, it will gradually reduce household income and confidence even without a wave of layoffs.

Housing shows the effect of interest rates most directly. Housing starts fell 12.4% to an annualized 1.239 million. Single-family starts declined 9.9% to 808,000. New home sales dropped 10.5% to 607,000, while supply reached 9.6 months of sales. The average 30-year mortgage rate rose from 6.43% at the start of July to 6.66% at the end.

The housing brake is mainly cyclical and rate-sensitive. Building permits increased 5.0%, including a 2.5% gain for single-family homes. That is an early sign that builders still see demand at the right price and financing terms. It does not cancel the drop in starts, but it shows what could release the brake. Lower mortgage rates or slower home-price growth could bring buyers back relatively quickly. For now, the Federal Reserve cannot easily deliver the first.

Prices

July inflation tells two stories. Consumer prices offer grounds for moderate optimism. Headline CPI rose 0.1% during the month and slowed to 3.4% year over year from 3.5%. Core CPI, which excludes food and energy, increased 0.2% and slowed to 2.5% from 2.6%. Shelter contributed about two-thirds of the monthly increase, while energy prices fell 1.5%.

The Federal Reserve’s preferred measure is less comfortable. The PCE price index is 3.7% above a year ago, while core PCE is at 3.3%. Both indexes rose 0.2% during the month. The difference from CPI comes from the weights and methods used by the two indexes. PCE matters more for policy. It remains clearly above the 2% target.

Producer prices are more troubling. The PPI for final demand was unchanged in July but stood 4.7% above a year earlier. The broader core measure, which excludes food, energy and trade services, rose 0.4% during the month and also 4.7% over the year. Goods prices declined 0.7%, but services increased 0.2% and construction prices jumped 2.2%.

The gap between producer and consumer prices has two possible outcomes. If companies pass higher costs on to customers, CPI and PCE will face renewed pressure in the coming months. If they cannot, margins will narrow. Strong corporate profit growth shows that the sector has a buffer for now. That buffer is not unlimited.

Import prices complete the picture. The headline index fell 0.4% because fuel prices dropped 7.2%. Excluding fuel, prices increased 0.4% during the month and 4.5% over the year. Imports from China became 0.8% more expensive in July, the largest monthly increase since July 2008, and 2.7% more expensive over the year. Capital goods prices rose 0.9%, including computers, peripherals and semiconductors.

Although conditions around the Strait of Hormuz remain fluid, they have not yet materially changed the price of gasoline at the pump, which held near $4.21 per gallon at the end of August.

The August ISM data warn that the pressure has not ended. The manufacturing price index remains at 71.1. In services, it rose to 72.6, the highest level since August 2022. This is a directional check, not part of the July table. The direction is unfavorable. Headline inflation may cool because of energy while price pressure in services and imported equipment remains strong.

The detail being underestimated

The July trade deficit looks like unambiguously bad news. It widened by $17.4 billion to $88.6 billion, or 24.4% in one month. Exports fell 2.1%, imports rose 2.8%, and the real goods deficit increased by 12.7%. Unless other parts of the economy offset that gap, foreign trade will subtract from third-quarter growth.

The composition changes the meaning. Capital goods imports increased by $14.4 billion. Computers added $6.9 billion, computer accessories $6.6 billion, and semiconductors $1.2 billion. Almost the entire import surge came from equipment that companies use for production, data processing and capacity expansion.

The deficit is the invoice for the investment boom. The national accounts record imports with a negative sign because the equipment was not produced in the United States. The company buying it records an investment. The same action can weigh on headline GDP through trade while increasing future production capacity.

This explains part of the difference between weak headline GDP growth of 1.5% and much stronger private domestic demand of 4.2%. We should not take the next step automatically. Computer imports prove that money was spent, not that the investment will succeed. Equipment may remain underused, projects may prove unprofitable, and the expected productivity gains may arrive late. We know only one thing for now. The investment cycle continues, and it is large enough to be visible at customs.

What it means

The latest quarterly map placed the United States in a regime of strengthening growth and accelerating inflation. July’s data do not reverse that signal, but they show that growth is becoming more uneven while price pressure remains too strong for a rate cut.

On July 29, the Federal Reserve kept its target range at 3.50% to 3.75%. The decision passed by nine votes to three. Beth Hammack, Neel Kashkari and Lorie Logan preferred a 25-basis-point increase. That is an important change in the balance of risks. The debate is no longer only about how long rates should stay unchanged. Part of the committee thinks the current level is not cooling the economy enough.

August speeches presented three versions of the same concern. Hammack thinks it is time to begin a hiking cycle. Kevin Warsh points to strong capital expenditure, corporate profits and broadly distributed inflation. Lisa Cook recognizes housing weakness and low hiring but sees the risk to prices as larger.

Christopher Waller presents the strongest case for waiting. Slow employment growth, in his view, must be compared with slow labor force growth. When few new workers are entering the economy, it does not need a huge number of new jobs to keep unemployment stable. He also points to the slower short-term pace of core PCE. August’s payroll gain and recovery in participation weaken the case for a cut without forcing an increase on their own. This is an argument for waiting for the next inflation report.

The implication for liquidity conditions is simpler. When growth and inflation accelerate at the same time, the central bank has no reason to add liquidity through lower interest rates. If markets expect rapid easing, July’s data argue against that expectation. Keeping short-term rates high continues to make financing expensive, hold mortgages above 6.5%, and reward companies that can invest from their own cash flow.

This creates an uneven economy. Large and profitable companies can buy equipment and raise productivity. Smaller firms that depend on credit feel the cost of money more acutely. Owners of financial assets see support from profits. A homebuyer sees a monthly payment that remains high. A worker sees fewer openings and less reason to quit. The same regime produces growth in the aggregate figures and a sense of stagnation in daily life.

The data confirm that investment and manufacturing are sustaining the pace, while housing lags and labor is beginning to catch up unevenly. The August report weakens the case for a rate cut because it removes the immediate fear of employment contraction. The assumption is that new capital will raise productivity enough to justify the expense and broaden hiring. That part has not yet been won. The car is accelerating. The transmission is engaging again, but it is not yet delivering power evenly.

Risks in both directions

The positive risk is that the investment cycle broadens. If equipment orders turn into more production, higher productivity and gradually stronger incomes, the current acceleration could continue with a smaller inflation cost. Building permits already offer a weak early sign that housing may find a floor. Lower energy prices would also free income for other purchases.

There is also a chance that the trade deficit will prove less negative than it looks. Some imported equipment may enter service quickly. If it raises domestic production, today’s imports will support tomorrow’s exports or replace future imports. That is a possibility, not a forecast.

The negative risk starts with the breadth of hiring. If August’s gain proves temporary and payroll growth returns close to zero, income and consumption could weaken together. The second risk is that price pressure forces the Federal Reserve to raise rates just as housing and most of the labor market are still cooling.

The most uncomfortable scenario is that the engine loses strength before the transmission recovers. If new orders fall, investment projects are postponed, and households continue to save more, the phase could shift quickly to Growth with slowing momentum. If PPI and PCE remain high at the same time, the central bank will not be able to help easily.

What to watch

On September 11, the August CPI report will show whether headline inflation continues to cool and whether core remains below 3%. Retail sales on September 16 will test whether July’s decline was calendar noise or the start of weaker consumption. Industrial production on September 17 will show whether the August retreat in manufacturing ISM is already visible in actual output.

JOLTS on September 29 should show whether the hiring rate remains near 3.2%. On September 30, BEA will publish the third GDP estimate and its annual update. A revision in private domestic demand below 3.0% would weaken the investment thesis. The August core PCE report on the same day will test whether inflation remains above 3%.

The threshold for changing the phase is clear. Manufacturing new orders below 50, combined with a return of three-month average payroll growth toward zero, would change the assessment from Acceleration to Growth with slowing momentum.

Closing

The US economy enters the third quarter with a stronger investment and manufacturing engine than headline GDP suggests. August payrolls show that part of this strength is beginning to reach labor, but hiring remains concentrated while consumption and housing are weak. That keeps the phase in Acceleration but makes it vulnerable.

The next question is not whether companies are still investing. The data already say they are. The question is whether new capacity will create income and jobs before high prices force the Federal Reserve to press the brake again.

Sources

• BLS, Employment Situation, July 2026

• BLS, Employment Situation, August 2026

• BLS, Consumer Price Index, July 2026

• BLS, Producer Price Index, July 2026

• BLS, Import and Export Prices, July 2026

• BLS, Productivity and Costs, Q2 2026, final estimate

• BLS, Job Openings and Labor Turnover, July 2026

• BEA, Personal Income and Outlays, July 2026

• BEA, GDP second estimate and corporate profits, Q2 2026

• BEA and Census Bureau, US International Trade in Goods and Services, July 2026

• Federal Reserve, Industrial Production, July 2026

• Federal Reserve, July 2026 Senior Loan Officer Opinion Survey

• Federal Reserve, FOMC statement, July 29, 2026

• Census Bureau, Advance Monthly Retail Trade Report, July 2026

• Census Bureau, Manufacturers’ Shipments, Inventories and Orders, July 2026

• Census Bureau, Construction Spending, July 2026

• Census Bureau, New Residential Construction, July 2026

• Census Bureau, New Residential Sales, July 2026

• ISM, Manufacturing PMI, July 2026

• ISM, Services PMI, July 2026

• ISM, Manufacturing PMI, August 2026

• ISM, Services PMI, August 2026

• EIA, US retail gasoline prices

• Freddie Mac, Primary Mortgage Market Survey archive

• FRED, Real final sales to private domestic purchasers

• FRED, PCE price index

• Trading Economics, US economic calendar, market consensus

This edition covers April to June 2026 and includes data released through 31 August. China did not collapse in the second quarter. It continued to grow, but industrial production became stronger as the balance between external and domestic demand weakened. Exports, high technology and industrial policy supported output. Consumers, property and demand for long-term credit remained on the other side of the picture. This is a two-speed economy in which headline growth looks broader than it really is. July and August are used only as directional signals for Q3.

Where the Economy Stands

China is in a slowing-growth phase with a narrow domestic foundation. Real GDP rose by 4.3% from a year earlier and by a seasonally adjusted 0.9% from Q1. The rate slowed from 5.0% year on year and 1.3% quarter on quarter in Q1. This was not a synchronised deterioration. It came from a widening distance between sectors receiving orders, capital and policy support, and those depending on confidence among households and private businesses.

The production side remained the strongest. Industrial value added rose by 4.7% in Q2, manufacturing by 4.8%, and information technology, software and related services by 10.8%. High-tech manufacturing expanded by 12.8% in April, 15.1% in May and 14.1% in June from a year earlier. Exports reached $1.149 trillion during the quarter, while industrial profits in the first half were 18.7% above their year-earlier level. Chinese industrial production did not merely continue to grow. It moved at a very high speed in some technology segments.

The domestic side was much weaker. Retail sales grew by an average of only 0.2% year on year during Q2. Fixed-asset investment contracted by 5.7% in January-June, private investment by 8.5%, and property investment by 18.0%. Household loans fell by CNY 663.5 billion during the quarter. Long-term corporate loans added only CNY 130 billion, while most new corporate lending came from the short-term discounting of commercial bills.

China occupies an unusual position on the growth-inflation map. Growth is slowing and consumer inflation is weakening, which would normally create room for easing. Producer prices, however, accelerated to an average of 3.6% in Q2. This does not show overheated consumption. It shows higher upstream prices and a limited ability among many companies to pass them to the final customer. The main policy problem is therefore not a lack of banking-system liquidity. It is the weak channel from cheap money to home purchases, private investment and consumption.

Macro Verdict

The indicators do not describe an economy on the edge of recession. They describe an economy in which positive growth is concentrated. The official PMI stood almost exactly on the boundary between expansion and contraction, while the private RatingDog survey remained comfortably above it. The difference is not an error. The surveys have different coverage and give more weight to different groups of companies. When the private index is strong but the official one barely remains above 50, the more reasonable conclusion is that certain exporters and technology companies are performing well while the improvement is not evenly distributed across the manufacturing system.

The Trajectory

The most important feature of the trajectory is that actual output accelerated while overall growth slowed. Industrial production moved from 4.1% in April to 5.3% in June, while manufacturing output reached 6.0% at the end of the period. This does not contradict weaker GDP. Industry is only one part of the economy and can raise output even as construction contracts, consumers buy cautiously and capital spending weakens.

The PMI indices show the same divergence from another angle. The official manufacturing PMI remained between 50.0 and 50.3 during Q2. A value above 50 means that more companies report improvement from the previous month than deterioration. A distance of only a few tenths above the threshold signals very little breadth to the expansion. RatingDog was between 51.7 and 52.2, while its services index averaged 53.7. Smaller, private and more export-oriented firms in that sample appear stronger, but the official survey warns that the broad economic foundation is less healthy.

Consumption did not confirm the production impulse. Retail sales were almost unchanged in April, declined in May and recovered to only 1.0% in June. Excluding cars, the picture is slightly better, with average growth of 2.0% in Q2. Car sales, however, contracted by an average of 15.8%, too large a decline to dismiss as noise. Durable purchases are sensitive to confidence in future income, the value of housing and willingness to take on credit. All three channels remain weak.

Prices diverged further in every month. The gap between PPI and CPI was 1.6 percentage points in April, 2.7 in May and 3.1 in June. Chinese industry is paying higher prices for some raw materials, energy and technology components, while the final consumer is not accepting the same increase. Sectors with global demand and pricing power can protect margins. Companies selling mainly into a weak domestic market have less room.

August provides an early test, not a new assessment of Q2. After falling to 49.2 in July, the official manufacturing PMI recovered to 49.8. Output and new orders moved back above 50, but the non-manufacturing PMI remained at 49.0 and the composite index was 49.5. Large companies returned to expansion while medium and small firms remained below the threshold. The signal is better than July, but it is still consistent with narrow growth dependent on selected industries and larger enterprises.

What Pulled Growth Up and What Dragged It Down

China does not publish quarterly GDP contributions by expenditure component in the same detail as some developed economies. We should therefore not invent exact percentage-point contributions from consumption, investment and net exports. The production approach still allows us to see which industries expanded and which ones weighed on growth.

Services, the largest part of the economy, grew by 5.1% and represented 57.4% of nominal GDP in Q2. The differences inside the sector were large. Information technology, software and related services expanded by 10.8%, leasing and business services by 11.6%, and finance by 6.9%. Hotels and restaurants also grew by 5.8%. These are genuine sources of support and show that the domestic economy is not equally weak everywhere.

Industry added 4.7%, while manufacturing grew by 4.8%. High-tech and equipment industries expanded much faster than the overall economy. They benefited from the global computing-infrastructure cycle, the national drive for technological independence and capital spending on automation. Mechanical and electrical products reached 63.5% of Chinese exports in the first half and increased by 20.1%. This is the main link between industrial strength and external demand.

Construction contracted by 4.1%, while value added in real estate fell by 0.2%. These figures look smaller than the decline in property investment because they measure different things, but the direction is the same. New floor space started fell by 23.4%, completed space by 23.7%, and the value of sales by 13.6% in January-June. This removes orders from steel, cement, home appliances, furniture, brokers and local services. It also reduces the land revenue used by local governments to finance part of their investment.

The external sector was strong, but it needs a precise reading. Exports increased by 14.1% in April, 19.4% in May and 27.0% in June. Imports grew even faster, by 25.3%, 27.4% and 36.0%. The Q2 trade surplus reached $315.9 billion. For the entire first half, it was $576.0 billion, slightly below the same period a year earlier. Strong trade is therefore not a story of collapsing imports. It is a story of rapid growth in both flows, probably combining technology demand, higher prices for some imports and shipments brought forward.

The final balance is unusual. Sectors with access to external demand, state priorities and technology capital pull growth higher. Sectors tied to housing wealth, long-term private investment and consumer confidence pull it down. That is why industrial data can look strong while GDP slows.

The Quarter Month by Month

April began with a convincing signal from industry. The RatingDog Manufacturing PMI jumped to 52.2, its strongest Q2 reading, while the official index stood at 50.3. High-tech manufacturing grew by 12.8% and exports by 14.1%. The constraint was already visible. Retail sales added only 0.2%, car sales fell by 15.3%, and the official non-manufacturing index was 49.4. PPI increased by 1.7% during the month and by 2.8% from a year earlier. Production accelerated before the final customer gave the same signal.

The divergence became clearer in May. Industrial production accelerated to 4.5%, high-tech manufacturing to 15.1%, and the private services PMI reached 54.4. Exports and imports grew at double-digit rates. At the same time, retail sales contracted by 0.6%, car sales by 16.1%, and the official manufacturing PMI slipped to 50.0. PPI accelerated to 3.9% while CPI remained at 1.2%. May was the clearest example of a strong corporate and external flow without confirmation from consumption.

June ended the period with the strongest actual production volume. Industrial output grew by 5.3%, manufacturing by 6.0%, and the seasonally adjusted monthly increase reached 0.76%. Export growth accelerated to 27.0% and the trade surplus to $125.6 billion. Retail sales returned to growth of 1.0%, but that was weak relative to the pace of production and trade. Car sales fell again by 16.1%. CPI slowed to 1.0%, core CPI also reached 1.0%, while PPI rose to 4.1%.

July and August provide the first test of whether this structure can continue. Industrial growth slowed to 4.5% in July, retail sales to 0.6%, and the official composite PMI fell to 49.3. The manufacturing PMI recovered in August, but the composite remained below 50. This is not enough to signal recession. It is enough to show that Q2’s exit speed is not guaranteed for the entire economy.

The Engine

The main engine is the combination of technology manufacturing, external demand and industrial policy. The clearest evidence is not only output growth, but also the concentration of profits. In the first half, profits in computers, communications and electronic equipment increased by 96.9%. The rise was 99.4% in non-ferrous metals and 67.8% in chemical raw materials and products. Total industrial profits grew by 18.7%, but the average hides an enormous difference between winning and lagging industries.

The AI cycle is a real part of this engine. Global investment in data centres, power infrastructure, semiconductors and computing equipment creates orders across a long supply chain. China participates through components, electrical equipment, materials and production machinery, not only through final devices. High-tech manufacturing expanded by 13.3% in the first half and equipment manufacturing by 9.3%. Both were well above the overall industrial increase of 5.4%.

The second layer is the policy of technological independence. It directs credit, subsidies, public procurement and research spending towards strategic industries. Investment in intellectual-property products grew by 9.4%, while technology loans increased by 12.6% at the end of June. Utilised foreign capital in high technology reached CNY 170.3 billion and rose by 33.2%, even as total foreign direct investment contracted by 5.0%. This is another example of narrow growth. Capital is not abandoning everything Chinese. It is concentrating in selected sectors.

The third layer is exports. Mechanical and electrical products account for almost two-thirds of merchandise exports. Private enterprises hold 57.0% of total trade and increased their flows by 17.0% in the first half. Some of the most competitive private firms remain able to gain global market share.

There are three constraints. First, part of the nominal growth in technology exports probably comes from prices rather than physical volumes alone. Second, the more China uses foreign markets as an outlet for rapidly expanding capacity, the greater the risk of tariffs, anti-dumping investigations and local-production requirements. Third, manufacturing capacity is not fully used. Overall utilisation fell to 73.0% in Q2 from 73.6% in Q1 and was 1.0 percentage point below its year-earlier level.

Capacity is the critical test of sustainability. Utilisation in computers and electronics was high at 78.7%. It stood at 70.8% in cars and 70.3% in electrical equipment. In non-metallic mineral products, which are closely tied to construction, it was only 59.6%. There is no single form of Chinese excess capacity. There are technology segments with high utilisation and orders, and traditional industries in which weak property and domestic cycles leave machinery and labour underused.

The engine will remain strong if global technology capital spending continues, new export orders hold up and profits turn into productive investment. It will weaken if trade restrictions hit physical volumes, higher raw-material prices compress margins, or companies accumulate inventories faster than sales. Inventories were already growing by 9.5% in June and 10.8% in July. The difference between producing for an actual order and producing for storage will become increasingly important.

The Brake

The brake is the balance sheet of households and the private sector. The property crisis is at the centre of the problem, but its impact is broader than construction. Housing is a major asset for many Chinese families. When its price falls, a household does not necessarily lose current income, but it loses confidence in its wealth. Large purchases become easier to postpone and willingness to take on a new mortgage declines.

The data show that the sector did not reach a bottom in Q2. Property investment contracted by 18.0% in the first half, new floor space started by 23.4%, completed space by 23.7%, and funds raised by developers by 20.2%. Domestic bank lending to the sector fell by 31.7%, deposits and advance payments by 15.8%, and individual mortgages by 24.9%. Residential and commercial floor space available for sale remained at 763.2 million square metres. Construction, sales and financing are contracting together.

Prices also failed to confirm broad stabilisation. New-home prices in tier-one cities rose by 0.1% from May in June but remained 1.3% below their year-earlier level. Annual declines in tier-two and tier-three cities were 3.1% and 4.2%. Existing homes fell more sharply, by 4.9% in tier-one, 5.4% in tier-two and 6.0% in tier-three cities. Only 20 of 70 cities reported a monthly rise in new homes, and only 9 did so for existing properties.

The property decline is passing into local budgets. Local-government revenue from land-use rights fell by 31.5% in January-June. Total government-fund revenue dropped by 21.6% and spending by 16.4%. The general budget looks more stable, with revenue growth of 4.7%, but current spending by local governments increased by only 0.6%. Beijing can announce substantial special-bond quotas without a broad local fiscal impulse appearing immediately in construction, services and private orders.

Consumption shows the second channel. Per-capita income increased by 5.2% in nominal terms and 4.2% in real terms in the first half. Per-capita consumer spending rose more slowly, by 3.7% and 2.7%. Property income added only 1.1%. This does not automatically prove that every household is saving more, but it is consistent with precautionary behaviour. Families receive more income without turning all the additional resources into spending, especially when housing wealth and the employment outlook remain uncertain.

Credit is the third and cleanest channel. Household loans fell by CNY 663.5 billion in Q2. The decline included short-term borrowing, often linked to current consumption, and long-term lending, where mortgages matter greatly. This does not resemble households responding aggressively to low interest rates. The average rate on new mortgages was 3.1% in June, but cheap credit does not compensate for weak expectations about the asset it finances.

The corporate picture is less negative, but it also does not show a new broad investment cycle. Corporate loans increased by CNY 2.53 trillion in Q2. Only CNY 130 billion of that was long term. Investment in equipment rose by 8.1%, but construction and installation investment fell by 8.0%, manufacturing investment by 1.2%, and private investment by 8.5%. Capital is going towards replacement and technology equipment in priority segments without becoming a general expansion of capacity.

The brake is structural because it combines a long property correction, high debt among developers and local entities, adverse demographics and an income model that gives more weight to investment than consumption. There is also a cyclical part. Stronger services, stable employment and recovering home prices could improve confidence. Until sales, mortgages and long-term private loans turn higher, cheap liquidity will have a limited multiplier.

Prices

Consumer inflation is not China’s main threat. CPI was 1.2% in April and May and slowed to 1.0% in June. The Q2 average was 1.13%. Core inflation, excluding food and energy, declined from 1.2% to 1.0%, with an average of 1.10%. That is too little pressure to describe overheated domestic demand. It is more consistent with a cautious consumer and strong competition in final goods and services.

PPI tells another story. Producer prices rose by 2.8% in April, 3.9% in May and 4.1% in June. The average increase of 3.6% is an acceleration, not cooling. Part came from mining, metals, chemical products and energy inputs. Part is linked to global demand from AI and electrical supply chains. High PPI should not be interpreted as evidence that Chinese consumers have started spending more strongly.

The gap between the two indices is more useful than either one alone. When PPI accelerates and CPI slows, companies have three options. They can raise final prices, accept a lower margin or cut other costs. Export technology firms facing scarce supply have more pricing power. Producers of cars, construction materials and mass-market consumer goods face greater pressure. This is visible in profits. The car sector was 19.5% below its year-earlier level and non-metallic mineral products were down 47.8%, while electronics almost doubled profits.

For monetary policy, this combination argues for targeted action rather than mechanical broad easing. Weak CPI creates room for a lower cost of finance. Accelerating PPI warns that part of the problem comes from supply and sector concentration, which a general rate cut does not solve. More importantly, an already low interest rate does not create mortgage or investment demand by itself.

July showed some retreat in producer pressure. PPI slowed to 3.5% and fell by 0.7% from June, while CPI declined to 0.5%. This reduces the risk of immediate pass-through to consumers, but raises the question of whether weak final demand is starting to cool factory-gate prices. The coming months must show whether Q2 marked the peak in PPI or the beginning of a longer price wedge.

The Underappreciated Detail

The most underappreciated detail is the composition of corporate credit. Nearly three-quarters of new corporate lending in Q2 came from the short-term discounting of commercial bills, often described as bill financing. This helps companies fund day-to-day operations, but it is not evidence of a new investment cycle. Long-term corporate loans increased by only CNY 130 billion.

A commercial bill is a promise by one company to pay a set amount to another at a future date. The recipient does not have to wait until maturity. It can take the bill to a bank and receive the money earlier in exchange for interest. A sale made on deferred payment terms becomes liquidity today. The instrument is useful for working capital, supplies and short-term cash flow. It is not the same as a multi-year loan for a new plant, production line or major expansion.

Corporate loans increased by CNY 2.53 trillion in Q2. Bill financing added CNY 1.914 trillion, or 75.7% of the total rise. Short-term corporate loans added another CNY 460 billion, while long-term loans added only CNY 130 billion. The composition shows banks and companies supporting current payments without taking broad long-term risk.

This changes the reading of the headline credit numbers. A large nominal flow towards companies can look like strong investment demand. If most of that flow is short term and backed by trade receivables, the signal is closer to liquidity management. A genuine new capital cycle would require simultaneous acceleration in long-term corporate borrowing, private investment, equipment orders and capacity utilisation.

What It Means

The first implication for the PBoC is that China does not have a classic shortage of bank liquidity. M2 grew by 8.0% in June, interbank rates remained low, and the average rate on new corporate loans was 3.0%. The one-year and five-year loan prime rates remained at 3.0% and 3.5%. The central bank cut rates on some structural tools by 25 basis points and continued to direct credit towards technology, green projects, small businesses and other priority areas.

The second implication is that another general rate cut would probably have less effect than measures changing expected returns and risk. A household will not take a mortgage simply because it is slightly cheaper if it expects the home price to keep falling. A private company will not build new capacity only because the rate is lower if domestic demand is weak and existing facilities are underused. Interest rates are a condition, but not a sufficient engine.

The most effective policy would therefore be a combination. Monetary policy can maintain a low cost of finance and avoid sudden tightening. Fiscal policy needs to stabilise income and consumption more directly, rather than focusing only on investment supply. Property policy needs to reduce uncertainty around unfinished homes, inventories and financing for viable projects. Local governments need a replacement revenue channel because falling land sales limit their ability to maintain services and investment.

Not all stimulus produces the same macroeconomic result. Additional credit to high-tech manufacturing can raise exports, productivity and strategic autonomy. It can also expand capacity faster than domestic demand and intensify trade tension. Transfers to households, social services or more secure home completion have a more direct chance of reducing precautionary saving. The policy choice is between further expansion of industrial capacity and stronger domestic demand.

China remains a two-sided factor for global liquidity. Strong exports of machinery, electronics and industrial goods increase supply to the rest of the world and can constrain final prices. At the same time, high Chinese imports in Q2 supported demand for raw materials, components and energy. If this is sustained volume growth, the effect is positive for foreign suppliers. If much of it reflects prices or orders brought forward, the support will fade later.

The renminbi appreciated by 3.0% against the dollar from the end of 2025 to the end of June, while the CFETS index rose by 4.7%. This gives the PBoC more freedom to maintain easy conditions without immediate depreciation pressure. A stronger currency and trade restrictions can still reduce the price advantage of exports, so policy is likely to remain targeted and attentive to the exchange rate.

The practical conclusion is that headline aggregates reveal less than their composition. What matters is not only total credit, but its maturity and recipient; not only exports, but volumes and product concentration; not only industrial profits, but the sectors creating them; and not only PMI, but the gap between large and small companies. As long as these splits remain polarised, Chinese growth will stay positive but vulnerable to the loss of one of its few strong engines.

What We Said Before

This is the first quarterly Macro Pulse for China, so the review is against our earlier monthly analyses. Our central claim was that China was growing through production, technology and exports while consumption, private investment and property lagged. Q2 confirmed that framework. High-tech manufacturing and electronics profits accelerated, trade remained strong, and retail sales and property investment were weak.

The claim that property is a structural problem rather than a short monthly correction was also confirmed. Sales, construction starts, financing and prices remained under pressure. Monetary transmission into private demand was weak, but the composition of credit provides a better explanation than the label “liquidity trap.” Liquidity is available. There are not enough borrowers willing to use it for long-term risk.

Two earlier formulations need to be softened. First, strong exports cannot be explained only by shipments brought forward ahead of tariffs. This is a plausible factor and a risk for H2, but imports also rose sharply and the technology cycle is real. Second, the data do not show that state-owned companies are the only recipients of stimulus. Their profits are growing faster than those of private firms, but private enterprises hold a large share of trade and the private PMI survey was strong. The more accurate thesis is sectoral and policy concentration, not the complete exclusion of private business.

The Argument Against Our Reading

The strongest argument against the weak-balance thesis is that the economy may be in the early stage of a successful shift towards more productive sectors. GDP is still growing by 4.3%, services by 5.1%, high-tech manufacturing at a double-digit rate, and unemployment was 5.0% in June. Real household income is rising by 4.2%, the private services PMI is above 54, and June retail sales have returned to positive territory. From this perspective, weak property is not proof of broad weakness. It is the painful reallocation of capital from low-productivity construction towards technology and modern services.

This argument has serious evidence behind it. Equipment investment is growing by 8.1%, electronics capacity utilisation is 78.7%, high-tech foreign investment is up 33.2%, and private companies continue gaining foreign markets. If productivity rises enough, China can maintain respectable real growth with less construction and slower credit. A smaller property sector would then reduce financial distortions rather than lead to prolonged stagnation.

The reason not to adopt this optimistic reading as the central case is the lack of broad domestic transmission. Retail sales are barely growing, private investment is contracting, household credit is falling and long-term corporate lending is minimal. Overall industrial capacity utilisation is declining, inventories are rising and the August composite PMI is below 50. Productive rebalancing should gradually create income, spending and private capital appetite beyond the narrow winning sectors. That is not yet visible clearly enough.

What would prove that our reading is too cautious? Three things need to appear together by the end of Q3: retail sales excluding cars holding above 3% year on year; private investment and long-term corporate lending turning convincingly higher; and the official composite PMI remaining above 50 with improvement among medium and small companies. If that happens, the technology engine will have started pulling the rest of the economy.

Risks in Both Directions

The upside risk is that the global technology cycle proves stronger than expected. If orders for electronics, electrical equipment and machinery remain high, profits can finance new investment and jobs. Faster completion of unfinished homes, stabilising prices and more direct support for households would reduce precautionary behaviour. Growth could then move back towards 5% without property returning to its old scale.

The downside risk is that the divergence closes through weakness. New tariffs or lower global AI capital spending would hit the export engine. If that happens before consumption and private investment have recovered, industrial inventories and spare capacity will increase. A further fall in home prices would constrain mortgage demand and local land revenue. The most dangerous combination is weaker exports, a continuing property decline and credit that remains short term.

What to Watch

On 1 September, the RatingDog Manufacturing PMI for August will show whether July’s cooling has reached the private survey. A reading below 50 would be the warning threshold. On 3 September, a composite RatingDog PMI below 50 would confirm a broader contraction. On 9 September, CPI below 0.5% with PPI above 3% would widen the price wedge.

On 15 September, August industrial production, retail sales, fixed-asset investment and property data will test the domestic balance. Retail growth above 3% and a smaller decline in private investment would improve the picture. Credit and total social financing are due around the middle of the month. Sustained positive household lending and long-term corporate borrowing matter more than the total flow.

The phase-change threshold arrives with the official PMI on 30 September. A composite reading above 50 with manufacturing new orders also above 50 would signal renewed breadth. The first Q3 GDP estimate on 19 October provides the harder test. Growth of at least 4.5% with stronger consumption would challenge the scenario of further narrowing.

Closing

China ended Q2 with industrial production strong enough to avoid a sharp slowdown, but without a domestic balance strong enough to produce broad acceleration. Technology, exports and industrial policy are holding up the structure. Property, household credit and long-term private capital remain the weak supports. The next phase will not be decided by whether the system has more liquidity. It will be decided by whether households and companies find a reason to turn it into spending and long-term investment.

Sources

• National Bureau of Statistics of China: Preliminary Accounting Results of GDP for the Second Quarter and the First Half of 2026 - GDP, sector value added and nominal shares.

• National Bureau of Statistics of China: National Economy in the First Half of 2026 - overview, industry, trade, fixed investment, employment and services.

• NBS Manufacturing and Non-Manufacturing PMI, April 2026, May 2026 and June 2026 - official Q2 PMI readings.

• NBS Purchasing Managers’ Index for July 2026 and August 2026 - early Q3 direction.

• S&P Global / RatingDog China General Manufacturing PMI, April 2026 and May 2026 - private manufacturing survey.

• S&P Global / RatingDog China General Services PMI, April 2026, May 2026 and June 2026 via MarketScreener - private services survey.

• Trading Economics: RatingDog China Manufacturing PMI - June reading and quarterly series.

• NBS industrial production: April 2026, May 2026, June 2026 and July 2026.

• NBS retail sales: April 2026, May 2026, June 2026 and July 2026.

• NBS Income and Consumption Expenditure in the First Half of 2026.

• NBS CPI: April 2026, May 2026, June 2026 and July 2026.

• NBS PPI: April 2026, May 2026, June 2026 and July 2026.

• NBS Fixed Asset Investment in the First Half of 2026.

• NBS Investment in Real Estate Development in the First Half of 2026.

• NBS Housing Price Interpretation, June 2026.

• NBS Capacity Utilisation in the Second Quarter of 2026.

• NBS Industrial Profits in the First Half of 2026 and January-July 2026.

• Ministry of Commerce foreign trade releases: April 2026, May 2026 and first half of 2026 via the GACC quick-release mirror.

• People’s Bank of China financial statistics: first half of 2026, official mirror and Q1 2026, official mirror.

• People’s Bank of China: Q2 2026 Monetary Policy Report, official mirror.

• China Money: Loan Prime Rate, June 2026.

• Official rules on commercial bill discounting.

• Ministry of Finance: Fiscal Revenue and Expenditure in the First Half of 2026.

• Ministry of Commerce: Foreign Direct Investment, January-June 2026.

• NBS Regular Press Release Calendar for 2026 and S&P Global PMI release calendar.

Data cutoff: 31 August 2026. July and August observations are used only as directional signals for Q3 and are excluded from Q2 averages and phase classification.

Not financial advice.