Consumer Pulse.
A monthly macro and market update on Japan's consumer economy.
Last month we set three conditions for restoring a constructive view. July answered one of them. The income line strengthened again — real wages rose 2.4%, the best print of the cycle. Household spending fell 3.6% in real terms, a second consecutive decline. The conversion gap has stopped being a hypothesis. Two new pressures arrived alongside it: the BoJ took the policy rate to 1.25% on 18 September, and August inbound arrivals fell 9.6%. What did improve is the micro, where a handful of operators now grow traffic and ticket together.
Seven months in, last month's question has an answer, and it is mostly the unwelcome one. In August we named three conditions for returning to a genuine overweight: real household spending back at or above zero, comparable traffic positive across several independent formats, and a retail figure consistent with the wage line once deflated. One of the three arrived. Comparable traffic did return — Ryohin Keikaku and Nitori both grew visits and ticket in August, which is the first clean evidence of that kind in this series. The other two did not. Real household spending fell 3.6% in July after the 3.3% decline in June. And the retail print that looked like the third condition, METI at +4.0%, dissolves on inspection into automobiles and price. Underneath all of it the income line strengthened again, to +2.4% real. A household earning more and buying less for two consecutive months is no longer a transmission lag. It is the base case, and the work this month is figuring out where it still leaves room to be long.
The income side improved on every line. The MHLW July release, published 8 September, printed real cash earnings at +2.4% YoY, with total cash earnings at +4.7%, scheduled cash earnings at +4.1% and special cash at +6.3%. June was revised favourably in the same publication. This is the strongest real wage reading of the cycle, and the +4.1% scheduled line matters more than the headline: it is base pay rather than bonus timing, which means the Shunto settlement is now sitting in the recurring wage bill rather than in a one-off distribution.
The spending side moved the other way.
One weak household survey is noise. Two, arriving as the income line accelerates, is a pattern. In August we wrote that one month could not separate a transmission lag from a behavioural change, and that July's figures would. They have, and they point at behaviour. The household has more money in the payslip than at any point in this cycle and is spending less of it in real terms than a year ago.
A measurement caveat belongs here, because the two numbers are being quoted against each other all over the market and they do not subtract. The MHLW survey covers salaried employment. The FIES covers households of two people or more. METI measures merchant revenue. The universes, weights and price adjustments are different, so there is no arithmetic gap between +2.4% and −3.6% and anyone presenting one is inventing it. What is meaningful is direction: two series that should broadly move together have now moved apart for two consecutive months, in the same direction each time.
Two explanations remain live and they carry very different consequences. Either households are rebuilding precautionary savings, in which case the money returns to the economy when confidence in the income path is established and the delay is the whole story. Or the price elasticity is sharper than we assumed, and households are buying fewer units rather than trading down within the basket, in which case volume does not recover until relative prices do. The August confidence and Economy Watchers readings, both improving, are marginally consistent with the first. The FIES is consistent with either.
Last month's correction imposed a fixed reading sequence on company data: same-store sales, then traffic, then basket, then units and mix, then margin. August is the first month in which that sequence does real work, and it sorts four operators that a sector screen would have grouped very differently.
Ryohin Keikaku grew like-for-like sales including e-commerce by 9.6% on traffic up 2.7% and ticket up 6.8%. Nitori grew comparable sales 4.4% on traffic up 3.3% and ticket up 1.1%. Both are growing because more people are walking in and each of them is spending more. PPIH improved to +1.4% same-store sales with traffic at −0.2% and basket at +1.6% — a genuine recovery from the corrected June figures, and traffic is now close to flat rather than clearly negative, but the growth is still coming from the ticket. Uniqlo Japan printed −1.3% including online, with traffic at −9.1% offset by ticket at +8.6%.
Four operators, one month, one macro environment, four different qualities of growth. The category label has stopped organizing the outcome: Nitori sells home furnishing into a rate-hike month and grows visits, while Uniqlo sells apparel into the same month and loses nearly a tenth of its footfall. The variable that separates them is execution rather than exposure, and on the standing method only the first two count as demand evidence.
The Uniqlo print needs one qualification before it is used as an apparel read. The company attributes part of the decline to weather and to four extended store closures, three of them following the Kumamoto earthquake and one following heavy rain in Chiba. That does not repair a 9.1% traffic decline, but it does mean the number is contaminated as a measure of underlying apparel demand, and it should not be annualized or used as the sector proxy without the September print alongside it.
The retail aggregate is the month's most quoted number and the least usable. METI July sales rose 4.0% YoY and 2.4% on the month, after June was revised to +0.6%, beating a consensus near +2.5%. The composition explains it: automobiles rose 16.2% while apparel fell 6.6%, and the medicine and toiletry category grew 1.6%. Set against that, Japan Chain Stores Association same-store sales for July grew 0.1%, with food at +0.1% and apparel at −1.8%. The everyday demand picture in the chain-store series and the headline in the METI series are describing the same month, and they do not agree.
The soft data improved for a fourth consecutive reading. Consumer confidence rose to 35.5 in August, up 0.6 point, with willingness to buy durables at 27.5, up 1.9 points. The Economy Watchers current index reached 46.4, up 0.7, and the outlook index 48.3, up 2.5 points. Both remain below 50. Service operators describe conditions improving faster than goods retailers do, which is consistent with where the frequency spending sits. This has now been true for several months without any confirmation in the FIES, and the correct weighting of a leading indicator that has led nowhere yet is lower each month it repeats.
Inbound deteriorated sharply on volume. August arrivals came in at 3.099 million, down 9.6% YoY, after July's 3.442 million and +0.1%. The Chinese cohort accounts for the bulk of the decline, with Korea, Taiwan and other markets holding up better. Our reading since June has been that spend per head compensates for weaker traffic, and Q2 supports it — ¥2.51tn total spend, up 0.2%, on ¥244,000 per visitor, up 3.3%. The problem is that Q3 spending is not published at this cut-off, so the thesis moves from supported to unverified precisely as the traffic decline gets large enough to need a bigger basket to offset it. Accommodation data points the same way: July guest nights of roughly 54.7 million, down 3.5%, with Japanese nights at −3.1% and foreign nights at −4.6%, occupancy at 60.8%, and a survey whose stratification changed in 2026.
The rate decision does not explain any of the data above, all of which predates it. What it changes is the three-to-twelve-month frame. Financed durables, housing-linked consumption and domestic premium spending now face a higher cost of capital on top of a household that is already choosing to save the wage increment. On prices, August headline CPI came in at +1.9%, core at +1.7% and core-core at +1.9% — underlying inflation still close to target, still enough to keep nominal tickets elevated, and no longer eroding the real wage. The index moved to the 2025 base with this release, which means differences of a tenth or two against earlier readings should not be treated as a change in regime.
| Release | Timing | What it confirms or breaks |
|---|---|---|
| Household spending (August) | Early October | The dominant item for a second month. A third consecutive negative real print would make the savings-rate reading the default interpretation rather than one of two live explanations. |
| Real Cash Earnings (August) | Early October | Whether the income leg holds outside bonus timing. The +4.1% scheduled line is the one to read; if base pay holds and spending stays weak, the problem is definitively propensity rather than purchasing power. |
| September company SSS | Early October | The fastest micro test, read decomposed. Whether the Ryohin and Nitori traffic gains extend to a second month, and whether the Uniqlo footfall decline narrows once the closure effects drop out. |
| BoJ Tankan — non-manufacturing | Early October | Not published at this cut-off and now overdue. The last reading of +37 predates both the rate move and two months of weak spending; the September survey is the first corporate read on whether service pricing power and hiring intentions survive them. |
| JNTO September & Q3 inbound spend | Mid-to-late October | Determines whether spend per visitor is still offsetting the arrivals decline. With August at −9.6%, this is the release that either restores or retires the value-over-volume thesis. |
The month's cleanest signal, and the first name in this series to satisfy the full sequence. Japan like-for-like including e-commerce at +9.6%, on traffic +2.7% and ticket +6.8% — visits and pricing contributing together, which is what the standing method is designed to isolate. Two qualifications: the group is international, so Japan same-store sales are not consolidated earnings, and one month of traffic is a datapoint rather than a trend. Watch whether September extends it.
We flagged this last month as the downside expression of the conversion gap. The August print argues the other way: comparable sales +4.4% on traffic +3.3% and ticket +1.1%, a genuine footfall recovery in a category the macro says should be contracting. The macro case has not improved — 1.25% reaches floating-rate mortgages and financed durables — but the execution contradicts it. The fiscal cumulative is considerably less impressive than the single month, promotional timing contributes, and sourcing and currency drive much of the P&L. Conviction here should be moderate in both directions.
Following through on last month's correction. August domestic same-store sales at +1.4% with traffic at −0.2% and basket at +1.6% — a real improvement from the corrected June figures, where traffic was −2.4%, and close enough to flat that the format is no longer losing visits. Growth is still ticket-led, so on our own test this is recovery rather than confirmation. The structural value-seeking case is undamaged; the automatic overweight stays withdrawn until traffic is clearly positive.
Uniqlo Japan at −1.3% including online, with traffic −9.1% against ticket +8.6%. The inverse of the Ryohin configuration and the clearest illustration of why the headline alone is unusable. The company cites weather and four extended closures, three after the Kumamoto earthquake and one after heavy rain in Chiba, which contaminates the print as an apparel-demand read without repairing it. The equity remains a global proxy rather than a domestic one.
The convexity position. Accessible dining is where a conversion catch-up would appear first, and the Economy Watchers readings describe services improving ahead of goods. The underwriting is thinner than it looks: the services Tankan at +37 is a June reading that predates both the rate move and two months of weak spending, and the September survey is not yet published. Labour cost under the settled 5.01% wage bill remains the unresolved margin risk.
The inbound volume shock lands here first. Arrivals fell 9.6% in August with the Chinese cohort worst hit, and the Q3 spend data that would show tax-free revenue per transaction offsetting fewer transactions is not published. Both are mixed domestic and inbound models rather than pure tourism proxies, and the domestic half now faces a higher rate against the same wealth-effect cohort. Two legs weakening at once, with the measurement that would settle it still outstanding.
The constructive path is unchanged from last month and one condition closer. It requires real household spending at or above zero, real wages still positive, and comparable traffic positive across several independent formats. The third condition arrived in August at Ryohin and Nitori; the second has arrived every month since March and strengthened again in July. Only the first is missing, and it is the one that matters most. An August FIES print near flat would establish that June and July were the delay we originally described, and the sentiment improvement already visible in confidence and the Economy Watchers survey would then be the leading indicator it has been claiming to be. We are not positioned against that outcome.
The destructive path has changed shape. It is no longer a second weak month — we have that — but a third, arriving with real wages back below zero. That combination would convert the current problem from propensity into purchasing power, which is a materially worse diagnosis: a household that chooses not to spend can change its mind, while a household that cannot afford to spend needs the price level to move. The rate at 1.25% makes this path more plausible on a three-to-twelve-month view by tightening the financed end of consumption at exactly the moment the unfinanced end is already saving. Watch the scheduled pay line rather than the headline for the first sign of it.
Three measurement limits bound how hard any of this can be pushed. The wage, household and retail series cover different universes, so their divergence is meaningful in direction and not in magnitude. The CPI moved to the 2025 base this month and the accommodation survey restratified earlier in the year, which means small year-on-year differences in either should not be read as regime changes. And three inputs we wanted are absent at this cut-off: Q3 inbound spending, the September services Tankan, and any verifiable contemporary transaction-data reading. They are marked missing rather than estimated, which is the same discipline that produced this month's conclusion — it is why Ryohin and Uniqlo, both growing their ticket, sort into opposite buckets.
The information provided on this website is for informational and educational purposes only and should not be construed as financial, investment, legal, or tax advice. All content reflects the personal opinions, interpretations, and analyses of the author at the time of writing and is subject to change without notice. Nothing contained herein constitutes, or should be interpreted as, a recommendation, solicitation, or offer to buy or sell any securities, financial instruments, or other investment products. The author is not a licensed financial advisor, broker, or investment professional. Any references to specific assets, markets, or strategies are illustrative in nature and do not constitute personalized investment advice. Investing in financial markets involves risk, including the potential loss of capital. Past performance is not indicative of future results. Readers are solely responsible for their own investment decisions and should conduct their own independent research and due diligence before making any financial commitments. You are strongly encouraged to consult with a qualified financial advisor, legal professional, or other relevant specialist before making any investment or financial decisions. By accessing and using this blog, you agree that the author shall not be held liable for any direct or indirect losses, damages, or consequences arising from the use of, or reliance on, the information presented herein. All content is provided "as is" without any warranties of completeness, accuracy, or reliability.