One month into the recovery, the picture has split. The hard data — wages, settlements, inbound arrivals, company same-store sales — has done everything we said it would do in March, and more. The soft data — consumer confidence, sentiment surveys — has collapsed. The split is being driven by a single exogenous shock (oil) that has not yet hit transaction data and may not. What follows is a working framework for trading the gap between what consumers say they fear and what they are actually doing.

In March we identified three forces driving the Japanese consumer back to real income growth: the wage cycle, cooling inflation, and the rotation in inbound demand. All three accelerated through the month. The MHLW February wage print came in at +1.9% real YoY — a meaningful step up from the +1.4% January reading and the second positive print after thirteen months of contraction.

The Shunto settled close to the upper end of where we expected, with Rengo reporting an average wage increase of 5.26% across the third round and SMEs landing at 5.05%. That is the third consecutive year above 5% and — more importantly — the second consecutive year where the SME number is within striking distance of the large-cap number. The transmission to non-regular workers and regional employers, which was the open question coming into April, has now happened. The income base is wider than it was a month ago.

And inbound, which last month's headline data suggested was rolling over, did the opposite. February arrivals printed 3.47 million, a record for the month, up +6.4% YoY despite Chinese arrivals falling another 45.2%. Korea was up +28.2%, Taiwan +36.7%, US +14.7%. The thesis we floated in March — that the inbound story was rotating, not ending — now has clean data behind it.

Then, in late February, oil moved. The Iran-Israel escalation drove WTI to around $120, and the Japanese consumer — primed by two years of imported inflation pain — reacted instantly in the sentiment surveys.

The Economy Watchers Survey moved in lockstep: 42.2 current (from 48.9), 38.7 outlook (the weakest since late 2020). Both surveys are dominated by retail and service workers describing what they observe at the till, and both are historically extremely sensitive to pump prices. This is what a fuel-price shock looks like in Japanese soft data: violent, immediate, and largely psychological.

The dislocation is the entire story this month. The hard data validating the March thesis sits next to soft data that — taken at face value — would suggest the consumer has rolled over. Both are real. They describe different things.

The cleanest evidence that the sentiment collapse has not yet translated into a spending collapse sits in the March same-store sales releases, which all post-date the confidence shock.

Fast Retailing reported +9.2% domestic SSS for March — an acceleration from the +4.6% February print that we flagged in last month's note as weather-driven and potentially one-off. It was not one-off. March temperatures ran +1.9°C above normal across northern Japan, extending the spring sell-through window and allowing full-price clearance on warm-weather product. PPIH printed +4.7% SSS with traffic still positive at +1.6% — the only large-cap retailer that has now maintained both lines green for two consecutive months. Zensho posted +5.6% SSS with traffic at +3.6%, meaning the most affordable casual dining franchise in Japan is gaining footfall, not losing it, in the same month households told pollsters they were terrified about prices.

Two mechanisms explain why the spending base is holding while sentiment is breaking. First, the Shunto settled before the oil shock. Workers know their nominal income for the year. The wage gain is no longer a forecast — it is in the payslip. Second, the categories that are most exposed to a confidence shock are the ones requiring credit decisions: cars, white goods, financed home goods. The categories driving the SSS prints — apparel, daily food, casual dining, discount retail — do not require a 24-month confidence horizon. The consumer is doing exactly what theory would predict: front-loading on small-ticket discretionary while pulling back on big-ticket durables.

This is also where the inbound story matters more than the headline suggests. The 3.47 million February print is not just a tourism statistic — it is exogenous demand showing up in domestic operators' tills on dates that overlap exactly with the confidence shock. Hotels, regional transport, casual restaurants, and discount retail are running on a customer base that does not read the Cabinet Office surveys. PPIH's traffic resilience is not an accident; it is partly Korean, Taiwanese, and Western tourists doing what they were going to do anyway.

The framework we are running is conditional on the oil shock staying contained to sentiment and not transmitting into Core-Core CPI. The transmission channel that matters is the scheduled end of government subsidies on electricity, gas, and gasoline, which start unwinding in April and continue through May. Headline CPI at +1.3% in February is currently sitting below the BOJ target only because of those subsidies. Core-Core is at +2.5%, services inflation is sticky at +1.4%, and rice is running at +17.1% YoY. When the subsidies roll off and oil at $120 hits the same period, headline will move sharply higher. Whether that pushes the next real wage print back into negative territory is the central question for May.

The second-order risk is the BOJ response. If Tokyo reads the wage cycle plus the imported inflation impulse as evidence of overheating and accelerates the rate path, the yen rallies, the inbound math compresses, and the wealth effect supporting domestic premium spending unwinds. The March Tankan was nominally constructive — non-manufacturing DI at 36 — but the outlook component dropped 7 points to 29, which tells you operators are pricing in cost pressure they have not yet seen. Five-year inflation expectations in the same survey ticked up to +2.5%. The BOJ has the cover it needs to move faster if it chooses.

There is also a domestic political risk that did not exist last month. The retail sales print for February came in at −0.2% YoY, the first negative reading in some time. The number is heavily distorted: February 2026 had one fewer Saturday than February 2025, gasoline price effects subtracted ~14.1% from the fuel component, and the underlying non-fuel retail categories grew between +3.8% and +4.0%. But the headline will be cited, and it will compound the soft-data narrative. The risk is not the data; the risk is the policy or corporate response to a narrative that is technically incorrect.