Consumer Pulse.

Six months in, the diagnosis changes. Since March this note has rested on one causal claim: rising real income would convert into rising real consumption. The first half of that claim is holding — real cash earnings rose for a sixth consecutive month in June, and the Shunto is fully in the payslip. The second half is not yet visible anywhere in the hard data. Household spending fell 3.3% year-on-year in real terms against a consensus expecting growth. Second-quarter private consumption grew 1.0% in nominal terms and did not grow at all in real terms. June retail sales rose 0.5% in value where the market looked for 3.1%. Three independent series say the same thing, which means this is not a publication artefact in one dataset. The reading moves from "real consolidation of consumption" to income recovery without demonstrated conversion into volume. That is a downgrade, not a reversal — and it comes with an error of ours to correct.

The income line did what we said it would. The MHLW June release printed real cash earnings at +1.6% YoY, a sixth consecutive positive month and an acceleration from +1.4% in May, with total cash earnings at +3.4% nominal and base pay also at +3.4%. On the income side there is nothing to revise. The Shunto has transmitted, the wage bill has widened, and the deflationary anchor on pay is gone.

Then the spending data arrived and went the other way.

A single monthly household survey is volatile and would not, on its own, overturn a six-month framework. The problem is that it does not stand on its own. The Q2 national accounts, published 17 August, show private consumption up 1.0% quarter-on-quarter in nominal terms and −0.0% in real terms, with household final consumption at −0.1% real. Real GDP still grew 0.3%, but the contribution came from net exports at +0.5 point while domestic demand subtracted 0.2 point. And June retail sales rose just 0.5% YoY in value against a +3.1% consensus, down 4.1% on the month, after May was revised down to +5.0%.

Read together, the three series describe one phenomenon with precision. The gap between the nominal and real lines in the national accounts is the entire story: prices are rising, revenue is being collected, and unit volumes are flat. The "nominal renaissance" that has framed this series since March remains a valid description of the macro regime. What we cannot currently claim is that it has become a real expansion of consumption. The household is earning more and buying no more.

Two explanations are available and they have very different consequences. Either the transmission from income to spending is simply slower than we modelled — a lag, which resolves — or households are choosing to save the increment, either from precaution against a 1.0% policy rate reaching their mortgage or from a price elasticity sharper than assumed. One month of data cannot separate these. July's figures can, and that is why the calendar matters more this month than the interpretation.

Last month we described PPIH as the cleanest signal of the month, citing +5.0% domestic same-store sales in June and calling it an operator holding traffic when the weather removed it from everyone else. An audit of the primary disclosure shows that figure was misread. The +5.0% is the fiscal cumulative for the Discount Store Business, not the June monthly comparable. The June column reads differently.

The corrected numbers invert the conclusion. Comparable sales were slightly negative, and the composition is the opposite of what we reported: traffic contracted 2.4% while the basket rose 1.7%. In the discount format specifically, the basket grew 2.9% against traffic down 2.9%. Growth came from price and mix, not from more visits. The all-store line at +1.5% is positive, but that includes new openings and therefore cannot support a claim about underlying demand.

What survives and what does not is worth stating precisely. The structural argument for value-seeking retail is undamaged — it rests on household budget behaviour under inflation, not on any single month. What does not survive is the micro evidence we used to claim that argument was accelerating in the near term. "Discount wins mechanically" was an inference from a number that turned out to be the wrong column, and we are removing it from the automatic overweight.

The methodological consequence is larger than the single name, and it reshapes how the rest of this note reads company data. Published revenue growth, basket growth and comparable traffic growth are three different things, and only the third measures demand intensity. A retailer printing +2% on ticket inflation and store openings with negative comparable traffic carries a completely different cyclical signal from one printing +2% on more visits and more units. From this issue forward the sequence we read is fixed: same-store sales, then traffic, then basket, then units and mix, then gross margin. The same discipline applies to the METI aggregate, which is a value series and never a volume proxy on its own.

Within goods, the hierarchy runs toward proximity and health and away from big-ticket. The June METI format breakdown puts drugstores at +0.6% and convenience at +0.4% in value, supermarkets at −1.3%, and both specialist electronics and home centres at −6.9%. The categories requiring a financing decision or a large discretionary commitment are contracting hardest, which is the expected shape once a policy rate reaches a floating-rate mortgage book. The categories serving recurring need are merely flat. Note that even the positive figures here are nominal — proximity formats are holding revenue, not demonstrably growing volume.

Inbound now separates cleanly into two lines that move in opposite directions, and this is the month it becomes unambiguous.

Arrivals fell 6.8% in June to 3.149 million, extending the decline. Yet Q2 inbound spending still grew 0.2% to ¥2.510tn because spend per visitor rose 3.3% to ¥244,000. Traffic down, basket up — the identical pattern we just corrected in domestic discount, arriving here as a favourable rather than an unfavourable configuration. The investable consequence is sharper than "inbound is positive" or "inbound is negative": assets monetizing the room, the ticket, the meal and the experience are better placed than models needing a rising count of physical transactions or mass tax-free volume.

The soft data turned up, and it turned up before the hard data. Consumer confidence rose to 34.9 in July from 33.8, a third consecutive gain, with the employment component up 1.7 points, income growth up 0.6 and durables up 1.0. The Economy Watchers current index improved to 45.7 from 44.0 and the outlook held at 45.8. The Cabinet Office continues to describe consumption as showing signs of pickup, with restaurants improving gradually while domestic travel stays weak. This is a genuine leading signal and it may be the beginning of the catch-up. It is not yet evidence of one.

Two variables from previous issues have to be demoted. The weather excuse is gone: July ran significantly warmer than normal with low rainfall across much of eastern and western Japan, which removes June's handicap entirely. That means the softness in the June and July macro picture cannot be attributed to the sky, and it also means we cannot yet verify the apparel rebound — the July company same-store prints we wanted for that test were not retrievable at this cut-off, so the seasonal apparel question stays open rather than resolved either way. On prices, national core-core eased to +1.7% in June from +1.8% with core at +1.6%, but Tokyo re-accelerated in July to +1.9% core and +2.0% core-core. There is no clean national disinflation and no uniform re-acceleration. The BoJ held at roughly 1.0% on 31 July by an eight-to-one vote, with the dissenter arguing for 1.25%, which keeps the asymmetry restrictive.