Data and method

Data sources

Source status

Fetched: 2026-09-29 10:29:49

Method

• The yen's real value: BIS narrow real effective exchange rate (2020 = 100). “Historical position” = share of months since 1970 with a lower value.
• Five years later: for every month, the change in the real effective rate 60 months on. Japan's months are split into fifths by level. For 26 economies, the distance from each one's own average (geometric mean since 1964, or — to avoid hindsight — the trailing 20 years).
• Exchange rate vs prices: the log change in the real rate is split into the log change in the nominal effective rate and the rest (relative prices).
• Rate gap: US 2-year Treasury yield (FRED) minus 2-year JGB yield (MoF), monthly averages. Co-movement = correlation of monthly changes in the gap and in log USD/JPY within each period.
• Net yen demand = goods & services + (primary income − reinvested earnings + secondary income) − (direct investment − reinvested earnings) − investment trusts' net purchases of foreign securities. Equivalent to the current account − direct investment − fund purchases. It leaves out other flows (banks, insurers, pension funds, hedging) and so is only part of the story.
• Digital deficit = copyright licence fees + telecommunications, computer and information services + professional and management consulting (the definition commonly used in Japan). Consulting is available from 2014.
• Hedged vs unhedged: US stocks = S&P 500 with dividends (Shiller before 1988, ^SP500TR after); US bonds = a 10-year Treasury bought each month, return approximated by yield plus duration × yield change. Unhedged multiplies by the change in USD/JPY (monthly average). Hedged adds the gap between the Japanese overnight rate and the US 3-month rate. Fees and actual hedging costs are ignored. Terciles use the yen's real value in the starting month.
• Interest bill guide: central-government debt (bonds, FILP bonds and treasury bills in the flow of funds) × the rise in rates. Not a forecast of the budget.

Premise rule

• Premise: “the weak yen will last”. Three checks point toward a stronger yen when: (1) the yen's real value is in the bottom 10% of months since 1970; (2) the US–Japan 2-year gap has narrowed by 0.5 points or more over 12 months; (3) net yen demand over the last 12 months is positive.
• Verdict: USD/JPY (3-month average) 10% or more below its 24-month high → broken. Otherwise two or more of the three checks → wobbling (watch). Otherwise → intact.
• The cheapness check alone rarely changes quickly, so the verdict mostly moves with the rate gap and the flows.

Limits

• Correlation is not causation, and there are few cases as extreme as today. Past averages are not forecasts.
• The balance of payments is revised, and some flows (bank lending, hedging, the BoJ's own operations) are not in net yen demand.
• Pre-1980 hedged returns are hypothetical: capital controls limited foreign investment.

Reference: Rogoff (1996), The Purchasing Power Parity Puzzle, Journal of Economic Literature 34(2)

Updates

Data are fetched and the site rebuilt every morning (Japan time). The balance of payments updates monthly, the flow of funds quarterly, and yields and USD/JPY daily. Text is edited by hand; no AI is used.

Operator

• Operator: After Deal operator (an individual)
• No connection with any organisation, including the operator's employer, and none of its work product or data is used.
• Nothing on this site is investment advice. Data belongs to each publisher and may contain errors or delays.

Corrections and questions: contact form / About After Deal