The Yen Paradox Full Research Note

Japan should, by most conventional logic, have a strong currency. It doesn’t. Working out exactly why turns out to be one of the more useful lessons in macroeconomics available today — for currency traders, commodity desks, and anyone doing business with or in Japan.

Japan is the world’s fourth-largest economy, a leader in advanced manufacturing and semiconductor equipment, and the destination behind a record 42.7 million tourist visits in 2025. Yet the yen sits near 40-year lows against the dollar, and by one key measure of real purchasing power, hasn’t been this weak since the early 1970s. This note goes through why, driver by driver, with the numbers behind each one.

01 — Interest RatesThe Gap That Never Closed

For decades, the Bank of Japan held interest rates near zero, or below it, to fight deflation. That only began to change in March 2024, when the BOJ delivered its first hike in 17 years and finally exited its negative-rate regime. Markets widely expected the yen to strengthen. It kept falling.

The reason lives in real interest rates — the nominal rate minus inflation — rather than the headline number. A negative real rate erodes the purchasing power of anyone holding cash; a positive one rewards them for it. And far from closing, that gap actually widened through 2025, because Japanese inflation climbed faster than the BOJ’s cautious rate hikes could keep pace with.

That gap — north of 3.5 percentage points by December 2025 — is more than enough to sustain the classic “yen carry trade”: borrow cheap yen, convert to dollars, buy higher-yielding US assets, repeat. It also explains why rate hikes haven’t reliably strengthened the yen the way the textbooks suggest. The December 2025 hike, for instance, had been priced in with near-100% certainty beforehand, so when it finally landed, traders who’d already positioned for yen strength simply banked their profits, pushing the currency down rather than up. BOJ Governor Kazuo Ueda didn’t help matters, describing the milestone as having “no special meaning” — about as dovish a way as possible to announce a historic rate rise.

What a carry-trade unwind actually looks like

In August 2024, unwinding carry-trade positions sent the yen up more than 12% in three weeks and dragged the Nikkei down 12.4% in a single session — its worst one-day drop since 1987. The trigger was a one-two punch: weaker-than-expected US labor data landing right after an earlier-than-expected BOJ hike, together narrowing the rate gap faster than markets had priced in. The volatility spike that followed technically overshot what US markets alone would justify, which only amplified the reaction across leveraged positions worldwide.

The yen-carry positions everyone blamed — foreign speculators shorting yen through futures markets — were relatively modest and had mostly unwound by year-end. The larger, slower-moving exposure sat with domestic Japanese life insurers holding unhedged foreign bonds.

That’s a meaningful twist. The carry trade propping up yen weakness today looks less like a hedge-fund phenomenon and more like a homegrown one, closely tied to the household and corporate outflows covered in Sections 02 and 03. And it isn’t a new pattern, either. IMF research modeling Japan’s previous sudden yen appreciations — in 2006, 2007–08, and 2015–16 — found carry-trade reversals routinely explained 40–60%+ of the peak move in each episode, amplifying whatever triggered it: a shift in risk appetite, a Fed surprise, a spike in volatility. A near-identical, smaller-scale version played out in the January 2019 “flash crash,” when thin holiday liquidity and a wave of stop-loss triggers sent the yen up several percent against multiple currencies within a single hour.

It’s worth pausing on the mirror image of all this: the last time the yen made this much news, it was for the opposite reason. The steep yen appreciation of the mid-1990s is widely blamed for pushing Japanese manufacturers to relocate production across Asia, a shift visible in the wave of outward Japanese investment through the late 1990s. That earlier bout of yen strength is arguably the origin point of the “hollowing out” the next section describes.

02 — Current AccountA Surplus That Never Comes Home

Here’s where even professional forecasters get tripped up: Japan posted record current account surpluses in back-to-back years — ¥29.3 trillion in 2024, higher still in FY2025 — and the yen kept falling anyway. A surplus is supposed to mean more yen demand from abroad, not less.

The catch is what that surplus is actually made of. Two decades ago it came overwhelmingly from trade: cars and electronics sold abroad, converted back into yen. Today the trade balance is frequently in deficit, and the surplus comes almost entirely from primary income instead — interest and dividends on Japan’s overseas investments, which reached a net international investment position north of ¥561 trillion (roughly $3.5 trillion) by the end of 2025, the largest of any country on earth.

Most of that income, though, never makes it home. Japanese multinationals have spent decades building factories abroad instead of exporting from Japan — the so-called “hollowing out” of domestic manufacturing — and the profits those overseas subsidiaries generate are largely reinvested locally rather than repatriated. Roughly half of the dividends paid to Japanese firms from overseas subsidiaries are reportedly retained as reserves outside the country. Some analysts have proposed a “cash-flow adjusted” current account that strips out this reinvested income; under that measure, the surplus — and the yen support it’s supposed to provide — looks dramatically smaller.

The IMF’s own 2024 external-sector assessment reached a more benign official conclusion, judging the surplus “broadly in line with” fundamentals. But even that assessment conceded the underlying issue: only about 20% of Japan’s foreign portfolio assets were yen-denominated, versus 56% in dollars, meaning much of that investment income is earned, and likely reinvested, in dollars rather than ever cycling back through yen markets. The export data backs this up. Despite a roughly 20% real depreciation of the yen since early 2022, Japan’s share of global exports still fell in 2024 to a level below 2019, with subdued manufacturing exports a key culprit, tied to the same outward investment relocating production abroad. It’s the same story told from the trade side: capacity that once generated yen-converting export receipts now sits offshore generating yen-retaining investment income instead.

03 — Household SavingsHouseholds Are Leaving Too

If Japanese companies are keeping their earnings abroad, households have started doing something similar with their own savings. The 2024 overhaul of Japan’s tax-free retail investment scheme, NISA — higher contribution limits, no time restrictions — triggered a surge into foreign stocks and funds: roughly ¥10.4 trillion (about $66 billion) in 2024 alone, the highest since 2015. Nomura estimates this shift accounted for around half of the dollar’s 2024 rise against the yen.

The split wasn’t quite as one-sided as “capital flight” suggests, at least at the outset. By December 2024, cumulative NISA holdings stood at ¥52.7 trillion, with roughly 40% still allocated to domestic equities according to preliminary survey data. Even so, a genuinely large sum did leave the country — and the pool of capital still available to leave is enormous.

That imbalance has a longer history than the 2024 reform suggests. A decade earlier, Japanese households kept over half their financial assets in cash and deposits, versus roughly 15% for comparable US households, with total US household financial assets more than double Japan’s in absolute terms. The original NISA scheme, introduced in the mid-2010s, was explicitly designed to move savers out of idle cash and into markets to support domestic capital formation. The 2024 overhaul achieved half of that goal — savers really have moved out of cash — but a large share of the money went abroad rather than staying home, a partial rather than total fulfillment of the reform’s original intent.

Notably, Japan’s much larger pool of pension and insurance assets has stayed comparatively still throughout this shift. Unlike the US, where individuals choose their own 401(k) allocations, Japanese public pension funds have traditionally run on a pay-as-you-go basis, with the government rather than the saver deciding how funds are invested. That helps explain why today’s outbound pressure is concentrated in the retail NISA channel rather than in Japan’s much larger institutional pension system.

There’s a further, quieter leak in the same direction: Japan’s growing deficit in digital services — payments to foreign, mostly US, technology and cloud companies — which has widened as the economy digitizes without a comparable base of internationally competitive Japanese software firms to offset it. It’s a reminder that “leading in technology” isn’t uniform across subsectors. Japan remains dominant in semiconductor materials, precision equipment, and hardware, but far less competitive in software, where the money increasingly flows the other way.

04 — The Demographics MythBlame Productivity, Not Population

The most popular explanation for Japan’s stagnation, and by extension its weak currency, is its aging, shrinking population. It’s a real challenge — the population peaked in 2010, the working-age population even earlier in 1998, and Japan went from the youngest G7 country by old-age dependency in the 1960s to the oldest by the late 2000s. Older research on Japan’s stagnation leaned heavily on this story. But more recent work suggests demography was never really the main driver, and the distinction changes how you should think about Japan’s long-term trajectory, currency included.

That reframing was already taking shape inside the IMF’s own analysis before it became an academic consensus. Staff decomposed Japan’s productivity growth into an “innovation” component and an “allocative efficiency” component — essentially, how well capital and labor are matched to the economy’s most productive firms — and found that allocative efficiency declined 21% between 2000 and 2019, dragging down average annual productivity growth by roughly a full percentage point. The Fund pinned this on rising market frictions, not population decline, and even suggested Japan’s own ultra-low rates may have let unviable firms survive longer than they should have, delaying the restructuring the economy needed.

A formal growth-accounting exercise makes the same point with harder numbers. During the 1955–75 boom, when Japan grew 7–8.6% a year, demographic factors — population growth plus rising participation — contributed only about one percentage point of that; the other six to seven points came from productivity. The recent stagnation tells the same story in reverse: demographic drag has been minimal to nonexistent in some recent decades (rising participation from women and older workers has actually offset population decline since 2015), while productivity growth is what collapsed, from around 7% a year in the 1950s–60s to essentially flat or negative by 2015–2024.

The thing that made Japan grow fast, and the thing that made it stop, is mostly the same variable in both cases — productivity, not population.

Separate, more granular fiscal projections back this up. The pace of elderly-population growth is itself decelerating sharply: the 65-and-over population grew 22.2% from 2010 to 2020 but is projected to grow only 2.5% from 2020 to 2030, since the postwar baby-boom cohort had already finished turning 65 by 2014. Over 2010–2020, even as Japan’s working-age population and total hours worked declined, labor productivity rose an annualized 1.3%, producing real GDP growth of about 0.9% for the decade — productivity gains more than offset the shrinking workforce. Projected forward, Japan’s combined medical and long-term-care burden is expected to rise only modestly, from 6.9% to about 7.3% of GDP by 2030, versus 8.0% if GDP simply moved in strict proportion to the shrinking working-age population.

There’s a real regional wrinkle worth flagging, though. More than 60% of Japan’s roughly 1,741 municipalities are projected to see their elderly populations actually shrink between 2020 and 2030 — a reversal from the 2010s — and pension income declines alone could cut local taxable income by 5% or more in dozens of these municipalities. It’s a genuine, if localized, fiscal risk sitting underneath the more reassuring national picture.

The same research complicates the popular “Tokyo is swallowing the country” narrative, too. Net migration into Tokyo has risen in recent years, but not because more people are moving in — it’s because fewer people are moving out. Gross migration, movement in both directions, has collapsed nationwide since the 1970s, and Japan is now one of the least geographically mobile advanced economies on record. That decline in overall mobility, not concentration in Tokyo, correlates with the slowdown in regional and national growth, suggesting Japan’s deeper problem is an economy where labor, capital, and firms have simply stopped reallocating toward more productive uses.

05 — Debt & RatesThe Fiscal Constraint

None of this makes demographics irrelevant. Aging may itself drag on productivity indirectly, through less risk-taking and slower technology adoption, a link researchers are still working out. But it reframes the floor under the yen: not a hard demographic ceiling Japan is powerless to change, but a productivity and mobility problem — arguably more fixable, even if it hasn’t been fixed yet.

Layered on top is Japan’s government debt, and this constraint comes with hard numbers attached. Gross government debt sat at roughly 236.7% of GDP in 2024, projected to dip modestly to around 227.6% by 2029 as nominal growth outpaces the effective interest rate, before climbing again as aging-related costs and a growing interest bill take over. Interest payments on that debt are projected to double by 2030 and quadruple by 2036, reaching around 13% of government expenditure. Ten-year JGB yields were already expected to push past 200 basis points by 2026, up from under 20 basis points as recently as 2018–2021. Higher policy rates raise debt-servicing costs sharply, so the BOJ has every incentive to move cautiously even when inflation argues for faster tightening — what some analysts call “fiscal dominance” of monetary policy.

The reason Japan hasn’t faced a Greek-style bond crisis despite that debt load comes down to who actually holds it. As of 2011, over 90% of Japanese government bonds were held domestically — banks, postal savings, life insurers, pension funds — versus roughly 70% foreign ownership for Greek government debt at the time of Greece’s crisis. Domestic institutions have kept holding JGBs even as supply keeps expanding, partly because bank capital rules treat government bonds as essentially risk-free. That captive buyer base is a big part of why Japan’s borrowing costs have stayed low even as its debt has climbed. That structure has eroded only gradually, not broken down: as of September 2024, the BOJ alone still held over half of all outstanding JGBs, with foreign and pension-fund ownership rising only slowly as the BOJ’s own balance-sheet reduction proceeds at a pace deliberately more cautious than the Fed’s or ECB’s.

06 — ValuationHow Cheap Is the Yen, Really?

By the Bank for International Settlements’ real effective exchange rate measure, which adjusts for relative price levels across trading partners, the yen was measured at roughly 65.9 in mid-2026, down from 141.8 in December 1986. That’s real purchasing power abroad cut by more than half in forty years, back to levels last seen around 1972, under the old fixed-rate regime.

How much this matters depends heavily on which model is doing the asking. The IMF’s own 2024 assessment makes the point well: staff’s preferred approach, which backs out a currency gap from the current-account gap, produced a fairly modest implied misalignment of around -4.8%. But the Fund’s own standard valuation models, run on the same data, produced dramatically larger undervaluation estimates in the -35% to -39% range, roughly in line with Big Mac-style purchasing-power comparisons. A sevenfold spread between two legitimate institutional approaches to the same question is itself worth noting: even the standard toolkits disagree by an order of magnitude on exactly how undervalued the yen really is.

Some economists argue the gap partly reflects a genuine “Balassa-Samuelson” effect — Japan’s productivity growth in export industries has lagged the US, which over time should show up in relative prices — while others say this is overstated, since Japan’s terms of trade have actually worsened in a way the theory doesn’t fully predict. What isn’t in dispute is the scale of the move: one of the largest sustained currency devaluations of any major economy in the post-war era.

07 — The UpsideTourism & Semiconductors

The record tourism numbers and the TSMC-led rush of new chip fabs into Japan are genuinely happening, and the weak yen is a direct cause of both — not a coincidence running alongside them. Cheap accommodation, meals, and labor costs in dollar terms make Japan an unusually attractive destination for travelers, and an unusually cheap place to build a fab. Real services exports, the category tourism receipts fall into, rose roughly 50% from early 2022 through 2024, riding that same depreciation.

But the scale of these inflows, however record-breaking, keeps getting dwarfed by the structural outflows described above. Even against that tourism boom, Japan’s overall share of global exports still fell in 2024 to a level below 2019, because manufacturing exports stayed subdued. Real, positive, and still not enough to turn the tide.

08 — PoliticsThe Political Wrinkle

The newest complication is political. Prime Minister Sanae Takaichi’s administration pushed a more fiscally expansionary agenda than her predecessors, including a stimulus package expanded to roughly ¥21 trillion. An early July 2026 draft of Japan’s annual fiscal blueprint was written in language markets read as pressure on the BOJ to go slow on further hikes, and the yen and government bonds sold off on the perception alone, before any actual policy changed. The final version, approved later that month, added an explicit footnote reaffirming that “the specific methods of monetary policy are left to the Bank of Japan.”

This didn’t come out of nowhere. The IMF’s own 2025 risk assessment had already flagged “bond market stress from a reassessment of sovereign risk” as a medium-likelihood, high-impact danger, warning that political pressure on a minority government and Japan’s long habit of loosely disciplined supplementary budgets were both eroding the credibility that keeps borrowing costs low. The specific trigger was new; the underlying vulnerability had been sitting in plain sight for over a year. By now, the yen’s story is as much about institutional credibility as it is about arithmetic.