• The yen was the first non-Western currency to become a key FX reserve currency in around 1980. Its share in the global FX reserve peaked at 8.7% in 1991 and had fallen to 3.4% by 2009.

• The yen has been quietly making a comeback in the last 10 years. Its share has risen to 5.8% in 2025, despite its persistent depreciation. Reserve managers have been buying into yen weakness.

• The dollar’s share fell from 64% in 2015 to 56% in 2025. With the Euro crisis restraining the shift into the euro, reserve managers have been raising weights of other currencies such as the Canadian dollar. The yen’s gain is by far the biggest.

The first non-Western reserve currency

The yen earned its place as a global currency through the 1970s and 1980s. The details of its rise are perhaps a topic for another report, but at its core it was a diversification move by reserve managers in the aftermath of the Nixon shock. By 1980, the Japanese economy was already bigger than the UK or Germany and had become an export powerhouse, accounting for 6.5% of global exports. Reserve managers bought into the yen as Japan opened up its capital market in the late 1970s. The yen’s share of world FX reserves rose from 4.6% in 1980 to a peak of 8.7% in 1991. This surge was followed by a long slide though. Between 1991 and 2009, the yen’s share in world FX reserves declined almost without interruption, bottoming at 3.4% in 2009.

Reserve managers buying through falling yen

You may expect a continuation of the decline of the yen as a global currency between 2015-25. A pleasant surprise here, at least for Japan. Since 2015, the yen has been quietly making a comeback. Its share has recovered to 5.8% in 2025. What makes the recovery remarkable is that the yen’s share in FX reserve rose while the yen depreciated, from around 120 per dollar in 2015 to 160 per dollar by early 2026. In my estimation, reserve managers bought yen to the tune of 3% of the global FX reserve between 2015 and 2025. Otherwise, the decline in the yen’s value would have reduced the share by 1%. I will explain the factors behind the rise in the next section.

chart 1

Please click the link below for interactive charts.
https://takujiokubo.github.io/jma-data/2026-08-04-fx-reserve-jpy/

Why the yen: Part of diversification away from USD

The resurgence of the yen as an FX reserve currency should be considered in a global context. USD has been gradually losing its dominance over the last few decades. Its share of the global reserves fell from 79% in 1975 to 56% in 2025. The decline is partly due to the depreciation of USD. In the 12 months after the Plaza Accord in 1985, USD lost 35% of its value against the yen. While the Plaza Accord seems to have revived the competitiveness of USD and paradoxically the credibility of USD as FX reserve currency for a while, USD started to show a marked slide again recently from 65% in 2016 to 56% in 2025. Explaining why USD started to slide again in 2010s is a topic of another report. See chart below for USD’s share in FX reserve, expressed in market exchange rate terms, as well as in Constant Exchange Rate (CER) terms that aim to net out the effects of currency depreciation. Please see Box at the end of this report for details on the statistics, including CER.

Chart 2

The euro and gold could not absorb the retreat from USD

As the world attempted to reduce its exposure to USD in the 2010s, the euro was initially the obvious alternative. The euro’s share in the FX reserve rose from 18% at its debut to a peak of 24% in 2009. Then the euro crisis hit. While the euro’s share halted since 2015, it has been hovering around 20% in the last 10 years. Aside from the euro, gold is an obvious choice, and gold’s share of world reserves has indeed jumped from 9% in 2015 to roughly a quarter today. But, the role gold can play as an international reserve is limited in my view, due to its limited supply. In quantity, official gold holdings grew only 11% in tonnes over the decade while the gold price nearly quadrupled.

Chart 3

The RMB is not the immediate answer

So, if neither the euro nor gold, where could global FX reserve managers park their wealth? The Chinese RMB was once thought to be the natural destination. The role RMB may play in the world is an exciting topic, but for now, the RMB’s share in FX reserve stopped rising lately. I attribute the pause to China’s reluctance to reduce its control of the cross-border flow of capital.

The world chose diversification, the yen among them

In the end, the world decided to rely on the golden rule in finance, diversification. Between 2015 and 2025, the share of world reserves held outside the big four reserve currencies rose from about 7% to nearly 13%. Part of that sits in currencies the IMF names, such as the Canadian and Australian dollars, and part in a residual of smaller currencies the IMF does not break down. The return of the yen as a global currency in the last 10 years should be interpreted as a diversification move away from USD.

Chart 4

Japan’s FX reserve has been overweight USD

This brings us to the topic of FX intervention Japan has been conducting in the last few days. As I discussed in my report Japan Cashes In Its National Carry Trade, Japan has a vast war chest of FX reserves, estimated to be close to 1 trillion USD, even after the massive selloff of USD in the last few days. The Japanese government does not publish the currency composition of its reserves, but it is probably 80-90% in USD if not more. So, Japan has not been following the global trend of diversifying away from USD, nor has it bought much gold. Its gold holdings have been essentially unchanged since the early 1970s. As I commented in the aforementioned report, the selloff of USD in its FX reserve, a.k.a. FX intervention, can be considered as Japan finally catching up with the world in reducing its FX reserve weight on USD. In my view, the US government is chipping in the intervention, partly to offset the impact of the second-largest FX reserve holder, Japan, selling USD off. There is a mutual benefit between the Japanese and the US government. It is not the intention of the Japanese government to start a panic in USD and US’s support does add valuable credibility to Japan’s FX intervention. So, all is well.

One clarification before I close this report. It is arguable some part of the rise in the yen share in FX reserves may not represent a permanent rise. Some FX reserve managers are said to be swapping their USD reserves out to hold Japanese treasury bills to earn an extra premium. My research suggests its impact is likely to be minor. One piece of anecdotal evidence is the 2023-24 episode. When the premium on the dollar-yen swap disappeared in 2023-24 and the money left Japanese T-bills, the yen share in the global FX reserve stayed mostly unchanged.

A note on the statistics

The box below provides a concise methodology I used in producing the full 50-year story of the FX reserve. My sincere admiration to the IMF for producing detailed statistics on FX reserve, and managing to publish the currency decomposition while respecting the privacy of countries reluctant to reveal their makeup of their currency compositions.

A thank-you, and an announcement of subscription price raise

I would like to thank you for reading this far. This research has been fun for me. By the way, you may have noticed that I introduced a paid subscription tier two weeks ago, and a few of you have been kind enough to take the offer.

You are welcome to stay as a free subscriber. I will most likely keep publishing my views for free subscribers. Economic research is such a fun for me and it is nice to be read and hopefully get some feedback, both positive and negative. The paid tier is mostly for professional financial market participants who may enjoy a better access to me and the proprietary data I produce. In my humble opinon, my JGB yield curve model should be valuable for serious bond market participants, as there are serious misunderstanding and mis-estimation of the term premia in JGB yields. My latest yield curve model and forecast with data-down load functions is behind paywall, but here is a one from two weeks ago for you to see what is behind the paywall.
https://takujiokubo.github.io/jma-data/jgb-yield-curve-model/

Paid subscribers also get priority when I answer questions, though I will try to respond to feedback from free subscribers when I can.

I plan to raise the monthly subscription price to $10 on August 16 at 10pm JST. If you subscribe before then, your current price will be grandfathered for as long as you remain subscribed.

Box: 50-year FX reserve-share series

My contribution here piggybacks on IMF’s data. The IMF publishes detailed statistics on FX reserves, Currency Composition of Official Foreign Exchange Reserves (COFER), with an estimate of the currency composition of the global FX reserve total. But COFER only starts in 1995, and it has a “currency unknown” part of FX reserves in 1995-1999 that the IMF only starts to allocate into currencies from 2000. My series extends the statistics back to 1975, drawing mainly on the IMF’s own Annual Reports, supplemented by BIS statistics and national central-bank disclosures.

There are two key adjustments I made. First, before the euro, European central banks held ECUs, claims backed by gold and US dollars. I count the dollar-backed part of the ECU as USD and drop the gold-backed part, as gold is not foreign exchange. Second, exchange rates distort shares. When the yen falls against the dollar, the yen’s share in global FX reserves shrinks even if nobody sells a single yen off reserves. I therefore compute shares of currencies with holdings valued at end-2025 exchange rates. The ECB publishes the euro’s share of global FX reserves in the same manner in its annual report, The International Role of the Euro (ECB, 2026). On this constant-exchange-rate (CER) basis, movements reflect changes in the amounts held, not in their valuation.

Similar logic works when we want to observe change year by year. For each year I estimate how the change in exchange rates would move the share, and read the difference from the actual change as what reserve managers added or shed. In 2022, for example, the dollar’s appreciation alone would have raised its share by 1.8 points, but the share fell 0.5 points, implying managers shed about 2.2% of global FX reserves worth of USD that year. These estimates strip out exchange-rate effects only; reinvested interest income stays inside them. My year-by-year decomposition follows the approach of Arslanalp, Eichengreen and Simpson-Bell (2022).

Chart 5

References: ECB, The International Role of the Euro, 2026;

S. Arslanalp, B. Eichengreen and C. Simpson-Bell, “The Stealth Erosion of Dollar Dominance”, IMF Working Paper 22/58, 2022.

Please click the link below for interactive charts from this report.
https://takujiokubo.github.io/jma-data/2026-08-04-fx-reserve-jpy/

For paid subscribers, immediately below the line:

• Global FX Reserve Currency Share — A fuller presetation of my estimates of FX reserve market share for USD and JPY between 1980-2025. For EUR, it spans between 2000 and 2025, with data download button. I will be adding GBP and CHF once I have time to scrutize if my estimates are valid.
  

• Interactive version of charts I used in this report with data download button.

Read more

• I add a Warsh factor to our JGB yield curve forecast, the market’s repricing of the Fed’s inflation-fighting credibility. It will add 10 to 15bp to our 10-year and 30-year JGB forecasts at end-2026.

• My main BoJ call is unchanged. The next hike to 1.25% comes on October 29-30 and the terminal rate is 1.5% in March 2027. The 30-year JGB peaks near 4.5% around year-end and then declines toward 3.8%.

• BoJ is dialing up its inflation upside risk and dialing down economic downside risk from the Middle East. A September hike is now a distinct possibility.

• July Summary of Opinions, due August 10, and August 27 Himino speech should be important for BoJ watchers.

Introducing the Warsh risk premia on US bond yield

I am introducing what I call the Warsh factor in my JGB yield curve model and forecast. My preference is to stick with the analysis of the Japanese economy and the BoJ. But after listening to Warsh press conference and seeing the 30 year US treasury yield rose above 5.2%, it seems less sensible to ignore the factor in my JGB yield forecast model. I will explain the Warsh factor in the later sections, but it raises the term premium in the US yield curve toward end-November 2026. As a result, the Warsh factor adds 10-15bp to our 10-year and 30-year JGB forecasts at the end of 2026.

The Warsh factor: Extra risk premium on bond yields

The Warsh factor, as I call it, is the extra term premium investors demand on US bonds while the new Fed chair obfuscates the Fed’s policy path. We can already see it in action. According to the New York Fed, the 10-year term premium estimate rose 26bp in July, accounting for essentially all of the month’s rise in the 10-year yield, with none from expected policy rates. After the July 29 FOMC meeting, the 30-year yield rose to 5.2%, a 19-year high, even as the 2-year fell. Former New York Fed president Bill Dudley named what the market is now charging: “Higher risk premium in the long end.” Economists at Bank of America and Apollo’s Torsten Slok read the move the same way.

How it enters our model: one s.d. shock, half of it sticks

My yield curve model uses the New York Fed’s term premium estimate as an input to JGB term premia, so incorporating the Warsh factor is straightforward. I assume that a risk premium worth one standard deviation has been building up since May, when Warsh took office as the Fed chair. The risk premium will keep building until November, then recede by half as the rest of the FOMC hopefully reins him in. In short, I treat the Warsh episode as a run-of-the-mill term premium event, of which half sticks for the rest of his term. It adds about 10-15bp to our 10-year and 30-year JGB forecasts at end-2026, fading to a few bp by 2029. How the Warsh factor plays out is quite uncertain, so I will be making adjustments in the coming months. For details on my yield curve model, please see here.

What’s unchanged: two BoJ hikes to 1.5%

My main BoJ policy calls are unchanged, two more hikes in October 2026 and in March 2027 to reach the terminal rate of 1.5%. My assumption for the terminal rate has been 1.5% since early 2026. In my view, the BoJ is understandably wary not to tighten its policy too early or too much. The Bank is still unsure how well-anchored inflation expectations are. Moreover, while the risk of Japan returning to deflation may seem distant, it is probably still a scarier scenario for the BoJ than an inflation overshoot. Thus, the BoJ is likely to prefer to pause at 1.5% to ascertain whether a higher rate is appropriate for Japan. Having said that, as I wrote in my report, the BoJ has made a conscious switch in its policy posture in the last 2 months and it is eager to bolster its credibility as an inflation-fighting central bank. Moving the goalpost higher than 1.5% has to be in the BoJ’s toolbox now. In addition to my main scenario, I am producing an alternative “faster and further” rate hike scenario, assuming three hikes to 1.75%, one in September 2026, another in March 2027, and the last one in December 2027.

The main BoJ calls: October hike, 1.5% terminal

In additon to the Warsh factor, my main forecast calls for two more hikes to 1.5% by March 2027. The next hike to 1.25% comes at the October 29-30 meeting, followed by a hike to 1.5% in March 2027, where the policy rate stays through our forecast horizon. On this path, combined with the Warsh factor, the 10-year JGB yield rises to about 3.3% by mid-2027 and then settles near 3.1%. The super-long end peaks first and comes down further. I see the 30-year topping out near 4.5% around the end of this year, then declining to about 3.8% by mid-2028 as domestic term premia normalize and the Warsh factor partially unwinds. The 20-year and 40-year trace the same arc, ending near 3.7% and 3.9%.

The alternative: September, and 1.75%

My alternative forecast moves the next hike to the September 17-18 meeting and adds a third hike in December 2027, taking the terminal rate to 1.75%. In my view, the BoJ would rather wait until October to evaluate the accumulation of data. The September Tankan, due for publication on October 1, is one key statistic the BoJ would want to see before feelling comfortable raising the policy rate. However, the key issue now for the BoJ is the yen. If the yen starts weakening visibly toward the end of August, it could push the BoJ to raise its policy rate in September. Vice Governor Himino is scheduled to speak on August 27, so if we see the yen slide in the coming weeks, we should look for a hint in his speech.

The July MPM was hawkish, but no clear hints for a September hike

There are some details worth noting in the July policy statement and Governor Ueda’s press conference on July 31, but I will spare you the details. The key messages in them were 1) upside inflation risk is the key concern for the BoJ and 2) downside risk to the economy from the Middle East has receded somewhat. More hawkish stance, but no clear hints for a September hike.

As for the 1.75% terminal, the BoJ would prefer to test 1.5% through most of 2027, and if the Bank feels additional hikes are required to restrain the upside inflation risk, the Bank will prepare us to modify our expectation upward and deliver a hike. For the yield curve, the extra hike matters most at the front and the belly. The 10-year ends about 15bp higher than in our main forecast, the 30-year less than 10bp higher.

What we watch: the yen more than the data

The minutes of the BoJ’s June meeting and the Summary of Opinions from the July meeting are due on August 5 and August 10. They will give us more color on where each BoJ member stands. Remarks from members in the official BoJ minutes and Summary of Opinions are anonymous, but an analytical frame work I produced earlier let us attribute remarks to specific members with different shades of confidence. Please see my report “Behind the Votes — Introducing the BoJ Policy Stance Indicator”. Otherwise, the Economy Watchers’ survey on August 10 and April-June GDP on August 17 are the key releases. Financial market movements, especially the exchange rate, are perhaps more important for the BoJ than economic data when it comes to the September hike decision.

Please click the link below for interactive charts and csv download functions.

https://takujiokubo.github.io/jma-data/2026-08-03-jgb-warsh/

For paid subscribers, immediately below the line:

• The Main Forecast page — the full projection behind this report: spot yields, risk-neutral rates, and term premia for the 2- to 40-year tenors, monthly from 2002 through 2029, in 15 interactive charts and a year-end forecast table. Every chart downloads as CSV.

• The Alternative Forecast page — the identical exhibit set under the faster-and-further scenario: a September hike and a 1.75% terminal rate, so you can difference the two paths yourself.

Read more

The Japanese government intervened in the foreign exchange in New York hours on July 30, buying yen and selling dollars. The dollar fell below 157 briefly, but it is already back above 160 as of 1pm Tokyo time on July 31. The Ministry of Finance has not confirmed it, but we know it was an FX intervention. The size of it is yet unconfirmed, but from what I gathered, my guess is around $80–100bn. That is a bit bigger than the last intervention in April–May 2026, at $74bn.

Washington gave its blessing

US authorities seem supportive. Treasury Secretary Scott Bessent said overnight that the yen “has substantially overshot its equilibrium price and is very undervalued,” adding that he is “not concerned about the yen strengthening against the dollar,” according to Fox Business. Selling dollars when the US Treasury market is wobbly seems somewhat risky. But MoF apparently had him on board.　

Together with the April–May intervention, Japan shed roughly $150–175bn. It is not a small figure, so what is the impact on its FX reserve?

Japan is still the second largest FX reserve holder, even after last night

Japan held $1,180bn of FX reserves at end-2025, second only to China’s $3,358bn. Switzerland is third at $915bn. At end-June 2026, Japan’s reserve was down to $1,090bn, so if the intervention overnight was $100bn, that takes the reserve to around $990bn. Still bigger than Switzerland’s.

Selling off USD may be a good diversification move

Should Japan be worried about the reduction? I do not think so. There has been unofficial criticism over the years from the IMF that Japan holds an excessive level of reserves. Offloading dollars may also be a good diversification move. Since mid-2010s, the world’s reserve managers have been shying away from the dollar. The USD share of allocated reserves was about 70% in 1999, but it has fallen below 60% since 2020. Japan’s reserve has probably been more heavily weighted in USD than the global average, so this is perhaps good timing for Japan to follow the trend.

Stabilizing FX rate and making a large profit

Japan used to intervene frequently in the FX market, but since 2005 it has intervened rather sparingly. The last big episode of USD purchase was in 2011 when Japan bought $184bn of USD at around 80 yen per USD. Overall, we estimate that the book cost of USD on Japan’s FX reserve is around 114 yen, with a range of 110–120 yen.

Assuming a ¥114 book, the Japanese government likely realized a profit of around $50bn, or ¥8.1trn, on this year's USD selling operation. And we should not forget that Japan has been enjoying a fat interest rate spread of USD over near zero yen carry since late 1990s.

Bonus: A political cover for a consumption tax cut

Incidentally, Prime Minister Takaichi proposed yesterday to cut the food consumption tax from 8% to 1% starting in April 2027, ostensibly just for two years. It will cost roughly ¥10trn. She has pledged that the tax cut will not be financed by debt issuance, but she has not clarified how the tax cut would be financed.

Thus, the seeming profit from selling off USD may come in handy for her. She will probably argue that the profit the Japanese government made on FX reserves can help finance the tax cut. Her argument is flawed, but it does give her some political cover. We will discuss our assessment on Japan’s gross and net debt in another report.

Whose problem is the shrinking dollar share?

All in all, my view is that the FX intervention last night was a tolerable policy move. While the intervention alone is unlikely to stop the yen from depreciating further, at least it slows the pace and allows the BoJ to raise its policy rate on a sensible schedule. In my view, the reduction in FX reserves from 1,180 billion USD to 990 billion USD, or even down to 900 billion USD if Japan is to intervene again this year, is not necessarily a cost for Japan. At 900 billion USD Japan would slip below Switzerland and lose second place. If this were the football “World Cup”, it may matter. Since it is not, it is just a portfolio management. In a way, Japan’s FX reserve has been a profitable carry trade. With the interest rate spread between US and Japan gradually narrowing and the yen at a historical low, it actually makes sense to unwind its position a little. In my view, it is actually the US government who should be worried about the erosion in the USD share of global FX reserves, but apparently Scott Bessent is ok with it. So, all is well.

Please click the link below for interactive charts and csv download functions.


https://takujiokubo.github.io/jma-data/2026-07-31-fx-carry-unwind/

• JGB yields experienced a historic rise between 2022-2026. Most of the rise came from a change in long-run inflation expectations and a normalization of term premia (TP) as the BoJ lifted its unconventional easing.

• TP across the yield curve have largely finished its normalization process. TP at the longer end of the curve are fully back to normal. TP at 10-year has a moderate room to expand further.

• We think the hiking cycle in Japan is already in a late phase. Typically, the longer end of the yield curve start to decline in a late cycle.

JGB yields have been on a Long Climb since 2022

If you are a financial professional in Japan and under the age of 50, chances are you have never “felt” a Japanese government bond yielding more than 4%. By the time I joined the industry in 1997, the 20-year JGB, the longest maturity bond at the time, had already fallen below 4%, and it kept falling into the 2000s. I remember discussing with a friendly senior pension fund manager in the early 2000s who complained about the lack of yield. As a young economist already accustomed to a deflationary mind-set, I felt a 20-year JGB with more than 2% was already generous.

Today the longest maturity JGB is the 40-year, and its yield has been rising persistently from below 1% in 2022 to around 4% now and 20 year yield is also not far behind. When I first saw the news that a zero coupon bond yield, or spot yield, of 40-year bond has crossed the 4% mark in late 2025, I remember feeling a sense of amazement in my mind.

Chart 1

In retrospect, the rise from 1% to 4% was a transition process from a low yield environment to a normal yield environment. And it was a passage other major global bond markets already went through in 2022-23. Remember the “secular stagnation” theory? While the post-covid inflation returned the “secular stagnation” back in a history back in a relatively short span of time, it took a longer time for Japan to realize that the deflationary time is over.

Perhaps it was fortunate or even deliberate that it took a longer time. The transition back to “normal” yield environment was not necessarily a painless process in some part of the world. The UK suffered a mini fiscal crisis in 2022 and its bond market had to be rescued by its central bank. Nothing as dramatic happened in the US, but it suffered the hiccup of the third-largest bank failure in US history, Silicon Valley Bank, in 2023.

Chart 2

Seen in this light, the long climb in JGB yields from 2022 to 2026 can be seen as a slow and controlled process. Japan did have a few scary moments, such as the Ueda shock of 2024 when the BoJ surprised the market by delivering an unexpected hike and the yen briefly surged as carry trades rushed for the exit. The super-long selloff of mid-2025 was a scary moment too. But in both circumstances, policy makers were able to calm the market. All in all, the long climb so far has been a rather drawn-out affair over 4 years. The question now is: how long does the climb continue?

Our answer: The Long Climb is Nearly Over

Our answer: the climb is likely nearly over. The 40-year yield is increasingly showing signs of stabilization in the last 6 months. According to the daily par-yield curve published by the MoF, it peaked at 4.04% on May 19 and has fallen to 3.8% as of July 16. Given the Middle East uncertainty, or more succinctly, the Donald Trump uncertainty, it is probably too early to say that 40-year has stopped rising. As we discuss later, there are future scenarios in which the longer end of the yield curve may resume the climb in earnest. But, the inevitable unpredictability of the future aside, the transition from the deflationary, low-yield era to a “normal” yield era appears nearly complete, in my view. According to a model I will explain later, the fundamental forces that were driving the Long Climb have mostly run their courses.

Chart 3

In the next two sections, I will explain what I think drove the Long Climb and why I think it is likely nearly over.

What drove the Long Climb: deflation’s end, and the BoJ’s exit

In our analysis, broadly two forces drove the Long Climb since 2022. One is economic, and the other is the historic change in Japan’s monetary policy.

Force one: the end of deflation

The economic force comes mostly from the rise in inflation expectations and additionally from the rise in Japan’s real interest rate. According to the BoJ’s Composite Index of Inflation Expectations (BoJ-CIE), 10-year inflation expectations sat well under 1% for most of the 2000s and 2010s (see chart below). The BoJ’s extraordinary easing from 2013 helped lift the inflation expectation, but the big push came when the post-Covid global inflation wave reached Japan in 2022. On our extension of the BoJ-CIE indices, 10-year inflation expectations rose from 0.9% in 2021 to around 1.7% today. The real interest rate has risen too. There are a number of competing estimates, but on our weighted average of those compiled by the BoJ, it sat below zero for most of the past two decades, turned marginally positive in 2024, and stands at around +0.1% today. In my model analysis, the economic forces contributed to the rise in Japan’s natural nominal interest rate by roughly one percentage point, from below 1% to about 1.9%.

Chart 4

Force two: the BoJ's exit and the term premium

The other force is the historic change in monetary policy. The BoJ’s unconventional easing in 2010s, the vast JGB purchases and the yield-curve control (YCC) policy, lowered nominal yields by compressing what we bond analysts call the term premium, the part of the yield that compensate investors for holding the long duration risk. As the BoJ gradually relaxed YCC policy through 2023 and began to reduce its JGB purchases in 2024, the compressed term premium has been returning to its historical norm.

The level of term premium differs depending on the maturity of bonds. In the case of 10 years as an example, we estimate these forces account for about 2.4 percentage points of the rise in the 10-year JGB yield between 2022 and 2026, according to our model that I will explain in the next section. The 10-year JGB spot yield rose from 0.1% at the beginning of 2022 and stands at 2.7% as of mid-July this year, a rise of 2.6 percentage points. So, the two forces nearly explain the entirety of the rise.

Our model: ACM with rising stars

Our yield-curve model is a variant of the Adrian-Crump-Moench (ACM) model, which splits a bond yield into a risk-neutral (RN) component, the expected path of short-term rates, and a term premium (TP), compensation for bearing duration risk and other risks. We make one Japan-specific modification. A standard ACM assumes the economy’s long-run interest rate to be stable. That assumption does not fit a Japan whose inflation expectations have clearly risen. Following Bauer and Rudebusch’s 2020 paper, “Interest Rates Under Falling Stars,” we let the underlying stars, trend inflation and the equilibrium real rate, move over time. In Japan’s case, they are rising, not falling. For more details on our model, please see the appendix.

The chart below shows what our adjustment to the ACM model does. A standard ACM now puts the 10-year TP around 1.2%, well above its 0.8% historical norm. Our TP reads 0.6% to 0.7%. If you apply the standard ACM model to Japanese yield curve, it generally attributes the rise to TP and TP shoot up when 10 year yield rises. Worse, recent Japanese data has made the standard estimate genuinely unstable. It depends on the length of data set one uses, but if the sample starts around 2000, it would contain twenty years of yields pinned near zero followed by four years of climb, making the estimation of the long-run-rate unstable. I have seen some estimates of a surging TP circulating in the market. They most likely come from standard ACM models, and in my judgement they significantly overestimate TP in the last few years as it does not account for the change in long-run interest rate that took place in the last few years.

Chart 5

The climb at the long end is likely nearly over

Term premia have fully decompressed

Applying our model to today’s curve, we see that the forces that drove the Long Climb are nearly spent. The term premia at the 20-, 30-, and 40-year tenors have all fully decompressed from the abnormally low levels they fell to during the unconventional monetary policy era. Our estimate of the 30-year term premium (TP) stands around 2% as of mid-July, a moderately elevated level compared with its historical norm. The 40-year TP, by contrast, has been coming down lately.

Chart 6

Why is the 30-year TP elevated?

Why is the 30-year TP elevated? Some analysts have attributed it to fiscal fear. I do not think so. A premium driven by fiscal concern should push the 40-year TP up as well. Instead, the 40-year TP seems to have stabilized, if not having started to decline.

As we showed in our previous report “Japan’s Fiscal Vigilantes Are Misreading the Curve,” investors in the longer end of the yield curve seem to be willing to pay a premium for low-coupon vintage bonds issued during the deflation-era. Currently, bonds with 21-25 years and 31-35 years maturity are the deflation era vintages (30-year and 40-year issued in 2016-2020). In my view, preference by investors for low coupon bonds is leading to relative cheapness (higher yield) at 28-30 year sector that are sandwiched by deflation era vintages, leading to higher estimate for 30-year bond yields.

The late hiking cycle points to a peak

Technical relative value aside, I do think there are economic reasons to expect the longer end of the curve to stop rising. As I mentioned, fundamental forces that have been driving the yield to rise are largely spent. In addition, the hiking cycle that started in 2024 is most likely already in a mature phase. The policy rate that started at -0.1% in 2024 has already risen to 1%, and getting closer to the current estimate for a long-run rate of 1.8%. My current projection is for the BoJ to raise its policy rate to 1.5% by March 2027 and pause. Historically, long end of the yield curve tend to start to decline in the late-cycle. The decline we are seeing in 40-year TP may well be an indication of such cyclicality.

An opportunity in the 30-year

There is then an interesting investment opportunity in the 30-year JGB, either as a naked position if your investment horizon and risk tolerance is long enough, or as an arbitrage trade combined with the 40-year. As a macroeconomist, my sense is that the level of superlong interest rates is already fair, given the current inflation and BoJ monetary policy outlook. Below are my yields forecast for the next 3 years.

What could go wrong for the end of Long Climb call

No forecast is complete without an account of what could break it. In my view, there are three major macro uncertainties for the yield curve.

Scenario 1: supply-shock recession (flattening)

The first is a supply-shock (oil shock) led mild global recession. In this scenario, overseas central banks would have to raise policy rates moderately to keep inflation expectations from rising, and the yen would likely depreciate. The BoJ would either pause or continue hiking slowly. The longer end of the curve is unlikely to rise, however, and may even decline, as it incorporates the recession and the eventual policy rate cuts that follow once the short-term inflationary pressure subsides. This is a flattening scenario, with policy rates rising and the long end staying put or drifting modestly lower.

A variation of scenario 1: a triple sell-off of US assets

One interesting variation of this scenario is a triple sell-off of US assets, with the Warsh-Fed dithering over whether to raise rates. In this variation, the yen strengthens against the dollar, while the longer end of the JGB curve may get pulled up as US Treasuries sell off. The BoJ will likely pause its hiking cycle. This would lead to a steepening in the short-medium term.

Scenario 2: demand-shock recession (bull-flattening)

The second macro uncertainty is a demand-shock global recession, perhaps through a bursting of the AI stock-market bubble. That would likely strengthen the yen and put the BoJ on pause. The longer end of the JGB curve would likely come down, a bull-flattening scenario.

Scenario 3: overheating and an inflation spiral (bear-steepening)

The third, an upside uncertainty, is an overheating global expansion leading to a weaker yen and an inflationary spiral in Japan with inflation expectations rapidly rising above 2%. The BoJ will then need to raise its terminal rate to above 2% aggressively. This is a bear-steepening scenario in the short-medium term, and the long end of the curve could rise toward 5%, proving the thesis of this report critically wrong.

Conclusion: the Long Climb is likely nearly over

JGB yields have come a long way from the days when the BoJ explicitly capped the 10-year yield at 0.1%. As we have shown in this report, the Long Climb in JGB yields of 2022–2026 is explained by the decompression of the term premia and the rise in inflation expectations. The passage from an artificially compressed yield environment to a normal one could have been a rough one for policy makers, as the UK and US experiences of 2022–23 show. Japanese yields have managed to resurface to normal yield levels without a major financial incident. In our view, term premia at the longer end are now fully decompressed, with the 30-year sector appearing somewhat elevated.

The BoJ's unfinished balance-sheet job

Heavy challenges remain for the BoJ, aside from dealing with macro uncertainties. The Bank has been steadily unwinding its monetary policy accommodation since 2024. It is on a path to reduce its JGB purchases by two-thirds by 2027. It is currently offloading JGBs from its balance sheet on average by 40 trillion yen, worth 6% to GDP, in each of the next few years. Its ownership share of the 25-to-40-year sector has already fallen from 40% in early 2018 to 14% in 2026. Even so, it will take another eight to ten years for its holdings to return to where they stood before unconventional monetary easing began. We will evaluate BoJ’s quantitative tightening policy and its implications for the JGB market in another report.

What normalcy looks like from here

What happens now that normalcy has returned to the JGB yields? We think the hiking cycle is already at a mature phase. The likely path from here is a flattening, with the short to belly part of the curve moderately rising while the longer end flattens out. The 10-year yield, the flagship of the JGB market, should soon be above 3% as BoJ stays on its hiking cycle but 3% is what normal looks like, with inflation approaching the BoJ’s 2% target and the economy on a steady footing. The major risk to this outlook remains the third scenario in the previous section, an inflationary spiral that forces the BoJ beyond 2%. As I wrote elsewhere, the communication from the BoJ turned decidedly hawkish in June, acknowledging this risk. We did not discuss the risk scenario of Japan returning to deflation, but the risk is still out there and haunts BoJ officials.

This report is part of a series reassessing Japan’s fiscal sustainability. Its contribution to that theme is this. The rise in yields in the last few years should be seen as a transition move from the low yield environment to a “normal” yield environment. I see little sign of a fiscal risk premium embedded in JGB yields and I see an opportunity in 30-year bonds. There are a few more long term themes in the report pipelines and I look forward to finishing them. Thanks for reading this far!

Appendix — The JMA yield-curve model in brief

Any n-year bond yield decomposes as

Yield(n) = RN(n) + TP(n)

where RN, the risk-neutral component, is the average short-term rate expected over the bond’s life, and TP, the term premium, is the residual. Our model estimates this decomposition for JGBs with a variant of Adrian, Crump & Moench (2013). Following Bauer & Rudebusch (2020), we let the long-run level of expected short rates move with its fundamentals, anchoring it to i* = r* + π*, the equilibrium real rate plus trend inflation expectations, each measured for Japan from the sources below. Term premia are benchmarked against their 2004–2012 averages, the last era of free JGB pricing, after the banking crisis and before QQE.

References.

• Adrian, Crump & Moench (2013), “Pricing the Term Structure with Linear Regressions,” Journal of Financial Economics, 2013, vol. 110, issue 1, 110-138. This is the base model. The New York Fed publishes daily ACM term-premium estimates for US Treasuries at its “Treasury Term Premia” data page.

• Bauer & Rudebusch (2020), “Interest Rates Under Falling Stars,” American Economic Review vol. 110, issue 5, 1316-54. It is seminal paper demostrating the legitimacy of time-varying r* and π* as determinants of the yield curve.

• Bank of Japan Monetary Affairs Department, “Developments in the Natural Rate of Interest and the Assessment of the Degree of Monetary Accommodation”. BoJ Review 2026-E-4 — the six natural-rate estimates our r* draws on.

• Osada & Nakazawa (BoJ, 2024), “Assessing Measures of Inflation Expectations:

  A Term Structure and Forecasting Power Perspective”, BoJ Review 2024-E-4. The Composite Index of Inflation Expectations (BoJ-CIE), the basis of our π*, was published in 2024. It has not been officially updated, although we have seen BoJ officials use an updated image charts.

  
  
  Please click the link below for interactive charts and csv download functions.
  
  https://takujiokubo.github.io/jma-data/2026-07-20-long-climb/
  
  
  Paid subscribers have an access to our Yield Curve Model estimates, including historical term premia decomposition as well as our forecast that are periodically updated. Paid subscribers will also have Excel workbook with the underlying data for every chart in this report.
  
  

Read more

• The super-long JGB curve inverts beyond 28 years — an odd behavior if fiscal fear is the primary driver.

• The strange two-hump shape in the yield curve is a coupon effect: low-coupon bonds issued in the deflation era trade at a premium pulling down the curve.

• Investors are not demanding extra yield to hold Japan’s longest duration. On the contrary, they are paying a premium for high-duration bonds.

Since 2024, the Japanese Government Bond (JGB) market has been going through a dramatic transformation. At the beginning of 2024, as the BoJ was about to embark on its current hiking cycle, the 30-year JGB par yield was 1.7%. Two and a half years on, 30 year par yield is close to 4% as of mid-July 2026. In zero-coupon spot-yield terms, it has breached the 4% line, yielding 4.1% as of July 14 — the highest ever, since the Japanese government first issued 30-year bonds in 1999.

Global commentators have voiced concerns over Japan’s fiscal sustainability since the late 1990s, when it first lost its Triple-A rating from the global rating agencies. Fiscal vigilantes have warned of a looming fiscal crisis in Japan ever since. Naturally, they are at it again, seizing on the sharp rise in Japanese bond yields. When Sanae Takaichi became Japan’s first female prime minister, she was also fortunate enough to be compared to Liz Truss, the UK premier whose short tenure ended after she triggered a mini-fiscal crisis for her country.

Japan did experience some fiscal concern episodes in the last 20-plus years, but it has never developed into anything as serious as the UK’s mini-crisis in 2022, nor, of course, as other full-blown crises such as the Euro-crisis. Despite the rise in long term bond yields, I do not think Japan is any closer to a fiscal crisis either. In fact, I think the long end of the JGB yields currently offers investment opportunities. I am preparing a pipeline of reports on Japan’s fiscal sustainability, and this report is the first of the series. In this report, I will show why I think fiscal vigilantes are mis-reading what is taking place on the JGB yield curve.

Closer look at the curve casts doubt on the fiscal-concern story

Rising bond yields could be mistaken for a sign of fiscal concern — but look closer, and details are off. The chart below is the par-yield curve published by Japan’s Ministry of Finance. If higher yields were driven by fiscal concern, yields should rise as maturity lengthens. Instead, as the chart shows, the curve climbs to 3.85% at 25 years and then bends modestly back down: the 30-year sits at 3.79%, the 40-year at 3.75%. While the inversion is subtle, it does pose a challenge to a fiscal concern story. A 40-year bond is far riskier than a 25-year if you fear for fiscal sustainability, so it should reward you with a higher yield, not a lower one. We will next look at the spot-yield curve that reveals an even more profound story.

Inversion is more profound on spot yield curve

Spot yields are the interest rate on bonds that pay no coupon until maturity. So it is a purer form of the price the market charges the Japanese government for the length of time it is asking to borrow for. Again, if you are worried about Japan’s fiscal sustainability, you would charge a higher yield to compensate for the risk.

The chart below shows an estimate of the spot-yield curve from one of the key wholesale fixed-income brokers, Japan Bond Trading. It has two humps. It rises smoothly to 4.20% at 28 years, dips to 3.78% at a trough around 34 years, then climbs back to 4.18% at 40 years. The inversion is far clearer here: the 34-year zero-coupon bond charges Japan roughly 0.4% less than the 28-year — even though its holder must wait six more years to be repaid. A fiscal story cannot produce this. Fiscal risk accumulates with time. It does not suddenly ease off after year 28, only to switch back on after year 34. Whatever is bending this curve is following something other than the passage of time.

Puzzle solved: Coupon factors explain the shape, not fiscal risk

Here is that same spot curve again — but with a second panel beneath it, plotting the average coupon of the coupon bonds at each maturity. Let me explain. As I said, spot yields are for bonds without coupons. But since the vast majority of JGBs are issued with coupons, economists and bond analysts compute spot yields by stripping the coupons out of ordinary bonds. If you want to know how, ask your friendly neighborhood AI. A typical method is called boot-strapping.

Line the two up, and the mystery dissolves. The trough in the yield curve falls exactly where the coupons are lowest. Out at 32 to 34 years, the bonds carry coupons of around 0.5% — a fraction of the 2–3% coupons on either side. Why do 32-to-34-year bonds carry such low coupons? Because they were originally 40-year bonds issued in deflationary times. A 40-year bond issued in 2019 with a 0.5% coupon becomes a 32-year bond in 2026, same coupon. You get the idea.

So the conclusion is this: the dip in yield at 32 to 34 years is not a dip in the price of time. It is just a dip in coupons, showing up in the spot rates.

Thank you for your patience if bond math is not your idea of fun. Let me stop dragging you through technicalities and restate the takeaway. If investors were genuinely worried about fiscal sustainability, they would charge more to hold longer bonds — and the longest, most exposed bonds would yield the most. The inversion tells you fiscal fear is not what is shaping the long end of the curve. At this point a fiscal vigilante might ask: why, then, have super-long yields — the 30- to 40-year — been rising so fast these past few years? My short answer is the return of inflation and the end of the BoJ’s extraordinary policy — but the full story is the subject of another report I hope to finish soon.

Investors seem hungry for the duration risk

There is one more piece of evidence I like to show why the bond market is not particularly concerned about fiscal risk. What you will see below is that in fact, rather than shunning the long maturity bonds, bond investors actually hunger for yield, paying a premium to own high-duration bonds. Let me explain.

What you see are the yields on bonds issued between 2016 and 2021 with low coupons when the BoJ was actively suppressing yields through its yield-curve control (YCC) policy. Because their coupons are low, every one of these bonds carries high duration and is therefore risky — highly sensitive to bond-market volatility. Yet they command a higher price, and a lower yield, than the fair level we estimate. This may be too technical for some, but I think it is another solid piece of evidence that investors are not behaving like people worried about Japan’s fiscal sustainability. They are not demanding extra yield to hold Japan’s duration risk — they are paying a premium for it. So, hungry for duration, and happy to pay for it. Convexity enthusiasts may note these bonds are also the most convex — true, but that is just another way of wanting duration, and it cannot explain why the premium tracks the coupon vintage bond by bond.

You may have noticed that we used the JMA fair value yield curve in the chart above. So, a short section to explain what they are and how they differ from other spot yields estimates by brokers. The broker’s estimate is fitted to the actual bonds trading at each maturity — the right tool for pricing and trading them. But imagine the 33-to-35-year sector happened to be populated by high-coupon bonds instead. The estimated curve would take a different shape, even with identical policy expectations, inflation expectations, and fiscal outlook. A measure of the price of time should not change when only the packaging of the cashflows changes. Our estimate is built to satisfy that requirement.

Here it is. Adjusted for the coupon effect, the curve rises smoothly to a single peak around 30 years and eases gently beyond — as the government's own par curve also does. Two facts say this is not a self-serving construction: its implied par yields match MoF's official curve to about 2 basis points, although MoF data plays no role in our estimation; and JGB strips — actual traded zero-coupon bonds — sit within about 5 basis points of it. If the strips also dipped at 33-to-35 years, our reading would fail. They don't.

Conclusion

Let me conclude. The rapid rise in super-long JGB yields through 2024–2026 may make you think the fiscal vigilantes are right about a looming crisis. Do not fall for it. In this report we made our case through the lens of the yield curve. There were some bond-market technicalities along the way, but the message is simple: if fiscal fear were driving the long end, every additional year of exposure would demand more compensation, and the longest, most duration-heavy bonds would be the cheapest of all. Instead, duration-heavy bonds command a premium — investors pay up to hold them.

Why yields have risen at all — and whether they have risen too far — is a question about the level, and the subject I will address in the next report. My short answer for you for now is that this mostly reflects Japan finally leaving the deflationary era behind. I thank you for reaching the end of this report, as I admit it was a bit technical. I promise the next report will be more plain. Fiscal sustainability is something of a holy grail for macroeconomists covering Japan, myself included, and I look forward to writing about it in the coming weeks.

It closed at the end of February this year, when the conflict in the Middle East sent oil price higher and confronted the Bank with two fears pulling in opposite directions. The oil shock threatened to choke off Japan’s growth — a reason to wait. However, that same shock, feeding through to prices, threatened to push inflation above the 2% target — a reason to speed up rate hikes.

The severity of the grip of the vice became clear in April, when the board split 6 to 3, with Nakagawa Junko — until then one of the doves — broke ranks from the BoJ executives to demand a hike. Governor Ueda commented in the following press conference that he, as the chair of the board, feels he needs to treat the fact that as many as three dissented as a serious matter.

The Bank did raise its policy rate in June, to 1.0%. But the decision must have been a hard one for BoJ executives who understand the severity of the conundrum. As we wrote in the report at the time, while tightening a policy rate may seem like a natural response to an oil shock, the risk of bringing Japan back to deflation remains a terrifying thought for BoJ executives.

Today, mercy arrived.

Two data releases landed on the same morning — the June Tankan and May’s industrial production — seem to loosen the vice on BoJ’s neck.

Take first the risk the Bank uniquely fears most, that the Middle East conflict and the oil shock leading to a depressed economic activity, a recession and disinflation. Today's data undercuts the premise. Despite the supply chain disruption, industrial production managed to rise in May and the outlook for June points to further expansion ahead.

The June Tankan brought even brighter news. The headline business-conditions index for manufacturers rose sharply to +22 from +17 in March, when the market was bracing for a slip to +14. The Japanese economy seems quite resilient despite the shock emanating from the Middle East.

Much of that resilience traces to a single source — the global AI build-out. Japan makes the picks and shovels for it: the semiconductor equipment, the memory chips, the optical fiber. Our AI hardware index, built from the production data, is up 14.5% over the past year while the broader manufacturing base drifts sideways. It is a narrow lifeline, but it is bearing weight, and the sectors leading the Tankan’s improvement — production machinery, nonferrous metals — are the ones plugged into that demand.

Now the other jaw — inflation. Here the news was benign, mostly confirming what many economists, including those inside BoJ were expecting, continued rise in corporate inflation expectation. Large manufacturers now put five-year-ahead general price rises at 2.2% a year, non-manufacturers at 2.1%. A continuation of the ongoing trend in the past 2 years. But importantly, they both cross the 2% target for the first time since the survey began asking in 2014.

What today’s Tankan delivered was a public confirmation that the BoJ is achieving its 2% inflation target and the lesser need for monetary easing. BoJ Tankan included more details such as the rise in sales price/input price DI, but the message is the same. Trend inflation in Japan is on the rise and deflationary norm has been broken. The economic data today suggest that BoJ is on course to raise its policy rate again in October and in March 2027 to reach 1.5%, in our forecast.

Of course, none of this frees the Bank entirely. The vice has loosened, but it is still there. The Middle East conflict can flare up again and the AI bubble could burst any day. Even if everything goes well, the BoJ needs to deal with Takaichi government who seems to think they have some ground to meddle with the monetary policy.

However, compared with how economic data looked before BoJ Tankan and Industrial production, the vice seems much less life-threatening. Governor Ueda just came back from a surgery that necessitated 10 days of hospitalization. He can use some stress releasing news and I hope BoJ Tankan was exactly it for him.

So this sentence below, published as an anonymous remark from the policy meeting BoJ held on June 15-16, caught my eye.

Import prices have also been driven up by exchange rate developments. Such price increases are considered to place a burden on the business of a considerable number of firms, including small and micro firms. Against this backdrop, it has become more appropriate than before to adjust the degree of monetary accommodation.

Source: Bank of Japan Summary of Opinions from June 15-16 2026 Meeting

“Adjust the degree of monetary accommodation” is BoJ-speak for “rate hike.” So, in plain English, it says the yen is weak, it’s pushing up import costs and squeezing small firms — so BoJ should raise rates. A board member just drew a straight line from the exchange rate to the policy rate, albeit anonymously.

BoJ is Escalating Its Verbal Yen-Intervention

To be honest, this admission is not a total surprise. The Bank of Japan has in fact been modifying its communication on the exchange rate for the last 6 months, and above average BoJ Watcher, such as myself, must have been picking up on the subtle change in BoJ’s tone on the exchange rate. There was a noticeable shift in how Ueda answered the FX question in December 2025, when he admitted that board members had discussed the implications of yen weakness for inflation during the policy meeting. The June remark I quoted above goes a large step further, admitting that leaning against yen weakness is part of what a policy rate hike is meant to achieve.

The remark’s inclusion in the June Opinions is of course deliberate — an escalation in the BoJ’s verbal yen-intervention. Faced with a weak yen and the risk that inflation expectations drift above 2%, the BoJ apparently has decided it would rather break the taboo of linking the policy rate to the exchange rate.

Bank of Japan as “wanna-be Hawk”

Since early June, we see the Bank of Japan making a conscious effort to step up its hawkishness in order to boost its credentials as an inflation-fighting central bank. Leaning against yen weakness is a part of the policy transmission it is now pursuing.

As part of the effort to signal and clarify its more hawkish posture, I think the BoJ will speed up the pace of its rate hikes. Since March 2024, the BoJ has raised its policy rate roughly twice a year. Having just lifted the rate to 1.0% at its June meeting, the market — judging by the OIS-implied path as of June 24 — is pricing in about 0.57 of a hike by October and 0.91 by December 2026, or nearly a full hike.

In my assessment, the market may be pricing in the next hike in the next few months, but I do not think it has really understood the change in the BoJ’s posture. But a hike in October, if it does happen, should be an obvious message that the BoJ is stepping up. I am penciling in a hike to 1.25% in October 2026 and then another hike to 1.5% in March 2027 — ahead of a market that has barely begun to price the second move. What happens to the yen also depends on the Fed, of course. But over the next 12 months, I think there is a good chance the BoJ raises rates by more than the Fed does and if so, exchange rate market should reflect the change in BoJ’s posture.

It seems that the centre has now given way in the last few weeks. To borrow from Yeats, the deflationary anchor has fallen apart, and the centre can no longer hold, it seems. As recently as this March, when a fresh oil shock hit, their first instinct was to wait to see how the damage would feed through Japan’s economy. But something changed when they faced the rebellion of Nakagawa Junko, until then one of the board’s doves.

The rebellion of Nakagawa must have been a shocker. It made them realize that the Bank of Japan could be on the verge of being seen as indecisive — a dangerous trait when inflation expectations are de-anchoring. So the executives seem to have decided on a big makeover: to shed the “wait and see” image and recast themselves as hawks. Governor Ueda’s June 3 speech was the decisive break, explicitly naming the “behind the curve” risk and pointing to the critical importance of maintaining credibility. In our view, the June 16 statement was its follow-through. It promised, without qualification, to keep raising rates and named, for the first time, the risk that trend inflation could run above 2%. BoJ subtly implied that it may need to raise its policy rate beyond the neutral rate.

The board map shows the repositioning at a glance.

The executive bloc, sitting at “hold” a year ago, has moved decisively toward the hawks — carrying the swing voters Koeda and Masu with it. The centre was no longer able to hold against the hawks.

Let me be honest. I do not think the BoJ executives are yet ready to become true hawks. I think they still want to be careful about raising rates. While the “behind the curve” risk is real, the terror of potentially seeing Japan return to deflation is even more haunting. But they now know they must be perceived as hawks to maintain or worse, to recover the BoJ’s credibility as an inflation fighter.

Economists have a name for this bind: a credibility problem, rooted in what the textbooks call time inconsistency. The logic is simple. A central bank with sufficient credibility as an inflation fighter barely has to act to hit its target — trust anchors expectations on its own. On the other hand, a central bank that does not have credibility is often forced to act more aggressively. Credibility is a substitute for tightening.

Mario Draghi, amidst Euro crisis, is the famous case. In 2012 he promised the ECB would do “whatever it takes” to save the euro, and calmed the markets without having to conduct a bailout. It may look effortless now, but he had committed the ECB to potentially unlimited purchases, with no guarantee the markets would believe him and no political cover if they did not. Had the market called the bluff, the ECB would have had to make good on it. The words of Draghi worked because the resolve behind them was probably real.

The stakes may not be as high for the BoJ now as they were for the ECB in 2012. But the problem it now faces is structurally the same. If the BoJ can convince the market it will defend price stability, expectations re-anchor and it may not need to raise rates as far as would otherwise be necessary. By appearing to be a hawk, it can avoid having to be one. But the appearance only works if the resolve is genuine. After three years of patience, the talk alone may not be sufficient for the BoJ. To gain credibility with the market, it may have to hike, and keep hiking, accepting the very risk it dreads — a slide back toward deflation. Credibility cannot be announced. It must be bought, and the price is real. It is unfortunate that Governor Ueda has health concerns at this critical juncture, but I wish him a timely recovery — the BoJ needs a credible face of the highest order if it is to pull a rabbit out of the hat on this time-inconsistency conundrum and I do believe Governor Ueda has the potential.

Let me tell you why.

Quantitative tightening is about the size of the BoJ's balance sheet — the stock of bonds it holds. The monthly purchase amount is a flow. ZeroHedge, and others, watched the BoJ slow the pace at which it trims that monthly flow, and concluded QT was ending. But slowing the flow of purchases is not the same as halting the shrinkage of the stock. The two move quite differently.

And the stock is shrinking faster, not slower. Look at how much the BoJ's holdings have actually fallen since QT formally began in August 2024. In the first twelve months, to July 2025, they fell by about ¥20 trillion. In the next twelve months, to July 2026, it fell by about ¥45 trillion — more than double. On our projection, they fall by a further ¥47 trillion in the twelve months to July 2027. The taper was shallow at the start, and only bites fully as purchases approach the ¥2 trillion-a-month floor next April. The squeeze is not easing; it is intensifying.

What does it mean when BoJ reduces its JGB holdings? Well, somebody else needs to finance the government debt by buying JGB instead. In the next 4 years, BoJ will offload on average 40 trillion yen of JGB per year, equivalent to 6% of Japan’s GDP in EACH of these four years. That is not what a central bank does when it is trying to make life easier for a government that wants to borrow.

Where does the squeeze bite hardest in the coming years? In the belly of the curve. As you can see in the chart below, the BoJ came to own around 70% of all 5–10 year JGBs during the Kuroda years. That share is down to 53% today, and in our projection it falls to 21% by 2030. The 10–25 year sector declines gently; the super-long sector barely moves. The 5–10 year sector will bear by far the largest rise in net supply as the BoJ steps back.

So, no — this is not a central bank quietly ending QT to let the government spend. It is one asking the market to swallow close to 6% of GDP in extra JGBs every year, while raising rates on top. If that is a favor to a free-spending Prime Minister, it is a strange way to grant one.

P.S. To be fair to Tyler Durden, ZeroHedge’s own piece half-conceded the point. Buried below the headline, the author admitted the runoff of maturing JGBs “will proceed steadily” regardless of the taper, having already shaved some 20% off the balance sheet since its 2023 peak. I guess the author could not resist the temptation of having sexy title. Yeah, I know its seduction.

It should not have been. The market's dovish read of Nakagawa was fair if one only read her public speeches — but her last speech before the vote was in November 2025. For five months, a speech-based view of Nakagawa had nothing to update on. And board members only speak in public two or three times a year. This is the structural blind spot in how everyone watches the BoJ.

Now, our new proprietary indicator, BoJ Policy Stance Indicator (BoJ-PSI) aims to fill the gap between speeches. Bank of Japan do publishes remarks by individual members through Summary of Opinions and Minutes. Remarks are anonymous, but in our BoJ-PSI framework, we attribute anonymous remarks to individual members using documented linguistic fingerprints: vocabulary, self-built indicators, characteristic ways of reasoning, all built from named sources. We can then build PSI score for every board member at every meeting, not just when they speak in public. We score members on 3-dimensions, Economy, Inflation and Policy Rate, scored on a 0–10 scale where 5 is neutral.

The chart above shows Nakagawa's Inflation score on BoJ-PSI. As you could see, Nakagawa’s policy stance has been on moderate upward trajectory since September 2025, rising to 6.25 by March 2026. For more details on how we scored her policy stance on Inflation, please see my full report "Behind the Votes — Introducing the BoJ Policy Stance Indicator", no paywall. In essence, our BoJ-PSI enable you to track members policy stance between speeches and it is clear that Nakagawa has been turning hawkish long before market realized it in April 2026.

Nakagawa will be leaving the board at the end of June. She has mostly been a quieter half of board members during her tenure, but I think she managed to use her first and last dissent to change the course of BoJ’s monetary policy. On June 3, Governor Ueda made a speech in which he turned decisive hawkish and I believe it is virtually certain that BoJ will decide to raise rate on June 16, Nakagawa’s last policy meeting.

While we never know what alternative reality would have been, but it is quite possible that BoJ executive may have maintained its wait and see posture for another few months till we see more clarity on the repercussion from the middle east conflict, if Nakagawa did not decide to deal a surprise dissent and shake the confidence of BoJ executives.

In my view, the jury is still out if a rate hike in June is a correct decision for the bank of Japan. The Middle East conflict is far, very far, from over, whatever some President says. From what I can tell, the demands of Israel and the Trump Administration are simply not compatible with how Iran sees its current bargaining position. The disruption in the global supply chain is likely to get worse in the next 6 months before it improves and there is a distinct possibility that a material deterioration in the global supply chain could cause a global recession. On the other hand, it is true that the cost of waiting has been rising for BoJ as yen weakens. As Governor Ueda made decisively clear in his June 3 speech, financial market may have been losing trust in Bank of Japan’s resolve for price stability. As a former CEO and CRO (Chief Risk Officer) of an asset manager, Nakagawa may have been one of the only professionals on the BoJ board who can discuss the danger of losing market credibility.

Whether the June hike proves right or wrong, Nakagawa leaves the board having done what an independent board member is supposed to do: vote her own assessment, even when it surprises. Congratulations on completing five years at the Bank of Japan, and thank you for your service.

This post is a condensed introduction to the BoJ Policy Stance Indicator. The full report goes considerably deeper: how the attribution of anonymous remarks actually works, Nakagawa’s complete scored trajectory; a positioning map of all nine board members on Inflation and Policy Rate; and what to watch in the swing votes of Koeda Junko and Masu Kazuyuki ahead of the next hike. You can read it here — no paywall.

Takuji Okubo
Japan Macro Advisors.

As the chart below shows, the executive bloc has moved decisively into hawkish territory, with Masu and Koeda pulled into their orbit.

One thing to watch on June 16: Hawks may try to stay ahead of the majority. Back in March 2026, one member of the BoJ policy board said that BoJ may consider accelerating its policy hike pace. The remarks in BoJ Minutes are anonymous, but I think it was Tamura.

one member expressed the view that, if the tension over the situation in the Middle East were to become prolonged,….., it would also be appropriate to pay attention to whether it was necessary to accelerate the pace of policy interest rate hikes.

Today’s speech by Ueda partially validates this concern. 0.5% hike would almost certainly be voted down — but it would fuel the discussion in the market that 1.5% may not be the terminal rate in the current hiking cycle.

My revised call for BoJ: 25 basis points to 1.0% on June 16, December 2026 to 1.25%, June 2027 to 1.5%. Same two-hikes-per-year pace, starting one month earlier than I expected.

I track two proprietary indices built from METI’s production data: a narrow AI hardware basket covering five items directly tied to AI infrastructure — semiconductor equipment, logic and memory ICs, measuring instruments, and optical fiber — and a broader AI-adjacent industry basket covering three industries with significant AI exposure. Both are indexed to 2020=100.

The divergence since early 2024 is striking. Overall IIP has actually contracted 2.1% since December 2023. The AI-adjacent industry basket is up 2.6% over the same period. The AI hardware basket is up 21.9%. Japan’s industrial base has been hollowing out for two decades — offshoring, loss of competitiveness, structural decline in sectors like information and communication equipment, which has shrunk by more than half since 2005. Against that backdrop, the AI boom is doing real work. It is not a revival. But it is a lifeline.

On the Middle East: the IP data are reassuring for now. Supply chain disruption is underway and the Strait of Hormuz remains closed, but it is reassuring to see that Japanese business managers remain positive in their production trend.

Which brings me to the BoJ. The IP news is positive, but not enough to change the June calculus. As I wrote after Masu’s May 14 speech and Koeda’s May 21 speech, both swing votes have conditioned their support for a hike on data confirming that global supply chain disruption is not critically affecting Japan. This IP report is a step in the right direction — but one data point is not enough. I do not think either Masu or Koeda will be ready to join the hawks by June 16. My base case remains a hold in June and a hike to 1.0% on July 30–31, contingent on the June Tankan and the Middle East trajectory.

Takuji Okubo, Japan Macro Advisors

My view is that policy rate should be tightened. But at the same time, we need to be conscious that the worst nightmare for Japan and the BoJ still is a return to deflation. While the risk is greatly diminished, but it is still there and we should avoid hastily cooling the economy. I also do not think it makes sense to tighten financial conditions when some businesses are genuinely suffering from severe supply chain disruption. There are many businesses who are currently concerned about their business sustainability, suffering from sharply higher material costs due to the middle east conflict. Behind the curve risk is real for BoJ now and we should not ignore the risks associated with waiting. But I think Japan still enjoys a luxury of stable inflation expectation, anchored by decades of price stability. I think the patience is still the virtue in Japan when it comes to central banking decisions.

.

Following 6-3 split decision in April, hawks need need just one more votes to flip the decision as BoJ governors are likely to want to avoid 5-4 split. As Asada, the newest member and Takaichi-dove, is very unlikely to flip, it makes Koeda as one of two votes hawks need to flip to convince BoJ governors to concede. Judging from today’s speech and press conference, I think Koeda stayed with the center of the BoJ board, data-dependent and balanced. June hike is a real possibility, but I would put my intuitive probability assessment of 25% chance for a hike and 75% for hold decision on June 16.

In her speech, she did show a meaningful shift in her inflation assessment. She now says trend inflation is already at around 2% and she also explicitly names upside risk of trend inflation exceeding 2%. Markets may read this as a hawkish signal for June. I think this misreads the conditionality in her framing though. In the speech, she explicitly stated that the impact of the Middle East conflict has not yet been sufficiently reflected in hard data — on exports, consumption, corporate investment, and prices. Her rate hike language — policy rates at an appropriate pace, with attention to economic trade-offs — is firmly in the executive camp, not the hawk camp. Consumer confidence, which she cited as a key monitoring indicator, has been deteriorating sharply from March. We do have a few data points before the June 16 meeting. But I do not we will see sufficient accumulation of data for BoJ to be able to judge that they have seen enough to resolve the uncertainty Koeda identified today.

So, my call is June hold. July 30–31 remains the more natural window for the next hike. Obviously, there are lots of uncertainty. Trump keep mentioning striking Iran again. While he cry-wolf has lost all credibility, who knows what he will do any way. If there is any drastic escalation in the middle east conflict in the next few weeks, the BoJ will most likely stay on the sideline in June. In any case, BoJ has another windows for a rate decision in July, and then in September. As I wrote after the April meeting (not here unfortunately as I started my substack only this Monday), the risk of BoJ monetary policy left behind the curve is real. But the worst nightmare for BoJ still remains a return to deflation. So, in my judgement, patience is still the virtue for the Japanese central bank.

However, my purpose here is not to dwell on the old story. After covering deflationary Japan for 20 years between 1997 and 2020, I took on a non-economist role — ironically at a group called The Economist Group in 2020 — and analyzing the Japanese economy had not been my chief occupation since. But I am restarting my economist career now.

A great deal has happened since I stepped back from active research. A global pandemic in 2020, of course, but the global inflation of 2021–2023 completely upended the “secular stagnation” story and we have re-entered an age of inflation and “normal” interest rates. It took a while for the BoJ to join the wave, but Governor Ueda managed to start a hiking cycle in 2024, and here we are with a 0.75% policy rate and a 4% 30-year JGB yield.

Having known deflationary Japan for 20 years, it is astonishing to hear members of the BoJ Policy Board loudly arguing for an immediate rate hike, warning of the risk of an inflationary spiral. Really?

As I contemplate the 30-year JGB yielding 4%, I feel genuine curiosity to understand what has driven it so fast and so high — at least by Japanese standards. Japan, the country I have spent my career studying, is in genuinely new territory. The deflation era that defined everything — monetary policy, fiscal policy, corporate behaviour, household psychology — may be over. That is an extraordinary thing to say, and I do not say it lightly.

It is that sense of being at an inflection point — one I have waited nearly thirty years to see — that brings me back.

What I have been building

Since last October, I have been rebuilding my research capacity. The public-facing part of that work has been a series of reports on Smartkarma, the institutional research platform, where I have been sparingly publishing on BoJ policy decisions. But the larger part of the effort has gone into two proprietary analytical frameworks that I believe will give this publication a distinctive edge.

The first is a systematic BoJ board communication scoring system — a tool that tracks each of the nine Policy Board members across their speeches, press conferences, and anonymous remarks in published meeting minutes and the Summary of Opinions. By attributing the anonymous remarks to individual members, I have about 10 data points to evaluate the policy stances of each member. Otherwise, speech and press conferences only give limited windows into members’ view. For example, Junko Nakagawa who surprised the market by joining the hawk in April, had not spoken publicly since November 2025. By attributing anonymous remarks published in the minutes and opinions, we get a peak into her policy stances in January and March meetings. And there was a sign that she was moving toward the hawk in the March meeting.

The second is a proprietary JGB yield curve decomposition model that separates observed yields into their risk-neutral component and term premia. The model’s answer to what is driving the 30-year yield to 4% is different from the market consensus — and I think that difference matters considerably for how investors should think about positioning in JGBs. I will share the full methodology and back-data for both tools in the coming weeks. In the meantime, their outputs will appear regularly in this publication, and I am genuinely keen to hear pushback from readers.

My first call: no June hike

Markets are currently pricing a meaningful probability of a BoJ rate hike at the June 16 Monetary Policy Meeting. While I would not rule out a June hike and I certainly do not wish to get my first public call wrong, I think the market is overestimating the chance of a hike in June. Let me explain why, using the two speeches that dominated BoJ-watchers’ attention this past week.

Masu: conditional, not hawkish

Board member Masu spoke in Kagoshima last Thursday. The phrase attracting market attention was: “if statistical data do not indicate clear signs of an economic downturn, I believe it is desirable to raise the policy rate at the earliest stage possible.”

Markets read “earliest stage possible” and heard hawkishness. They should have read “if” and heard conditionality. Masu is not signaling an unconditional push for June. He is stating a reaction function: move early, but only once the data confirms the absence of deterioration.

I do not think we will have sufficient accumulation of data by June 16. The next clean read of corporate sentiment is the June Tankan — due on July 1, two weeks after the meeting. End-of-May data releases, including CPI and industrial production, will not provide sufficient clarity on the economic impact of the Middle East conflict and supply chain disruption.

There is a further point. Masu’s inflation assessment — “drawing very close” to 2% but not yet achieved — places him squarely with the BoJ executive bloc, not with the hawks. Compare this with the three April dissenters: Takata believes the target is already achieved, Tamura wants an explicit pace formula for hikes, and Nakagawa fears an inflationary spiral. Masu shares none of these convictions. He is drifting toward the hawks. He has not joined them.

The chart below maps all nine Policy Board members on two dimensions — inflation conviction and rate urgency — as of May 14, using my board communication scoring system.

Source: Japan Macro Advisors BoJ-CSI member profiling system. Positions are analyst estimates as of May 14, 2026. Axes reflect relative stance, not absolute scores.

The picture is clear. Three red dots — Takata, Tamura, Nakagawa — cluster in the top-right. The blue executive bloc sits in the centre. Masu (green) sits close to the executives, marginally more time-sensitive but analytically aligned with Ueda’s framework. The hawks need one more vote for a 5–4 split. Masu is the most likely candidate. He is not there yet.

Himino: No indication that the executive bloc decided to embrace the hawks

Deputy Governor Himino’s speech on Saturday was somewhat disappointing, as his speech did not include any hints for the near term policy inclination. As the most analytically systematic of the BoJ executives, Himino is the member most likely to signal the direction of board consensus ahead of a meeting. His speech was devoted entirely to digital finance and payment system architecture — a deliberate choice of topic that itself communicates: we are not in pre-meeting hawkish signalling mode.

There was, however, something else in the Himino speech worth noting. He characterised US stablecoin strategy with unusual bluntness — describing it as designed to “attract and induce people around the world to finance the US fiscal deficit.” The word “induce” is Himino’s editorial addition to Treasury Secretary Bessent’s original framing. For a senior BoJ official speaking at an academic forum, this is a pointed observation. It is consistent with a broader BoJ wariness about yen weakness — and it suggests that the geopolitical dimension of dollar dominance is now a live concern inside the BoJ, not just a talking point.

But for the June decision: Himino did not indicate any shift in the policy stance of the BoJ executives and that is enough for me to keep thinking that the market is overpricing the chance of a hike in June.

The FX intervention is a message for BoJ to hold the rate hike

Two additional factors reinforce the June hold case. The Japanese government has intervened in currency markets repeatedly over the past two weeks — a signal that the Ministry of Finance is managing yen weakness directly, reducing the BoJ’s implicit pressure to raise rates for FX reasons. And the Federal Reserve is widely expected to hold on June 17, the day after the BoJ meeting — removing another potential catalyst for a surprise BoJ move.

My call: hold at 0.75% on June 16. The next hike is more likely to come at the July 30–31 meeting, conditional on the June Tankan confirming no severe economic deterioration and of course no return to active war in the middle east.

What to watch

The calendar between now and June 16 is dense. In order of importance:

• May 21 — Board member Koeda speech, Fukuoka. Koeda tracks Ueda’s framework closely. A hawkish signal from her would be genuinely surprising and would force me to revisit the June hold call.

• May 23–29 — Month-end data batch, including CPI and industrial production. Unlikely to be decisive, but worth watching for any sharp deterioration.

• June 3 — Governor Ueda speech. Markets will pay attention, but I do not expect a conclusive signal given recent criticism of advance signaling.

• June 15–16 — MPM. Base case: hold at 0.75%.

• July 1 — June Tankan. The first clean post-Iran corporate sentiment data. In my view, this is the data the BoJ is counting on.

• July 30–31 — MPM. The earliest window for the next hike, assuming the Tankan cooperates.

A note on how this publication works

My current plan is to publish technical research reports with data on Smartkarma, where I have an institutional research account. The two pieces summarised above — the Masu note from Thursday and the Himino note from Saturday — are available there free of charge, without a paywall. I mention this purely for transparency; I am not asking you to go elsewhere. I am excited to start my publication at Substack and I also look forward to hearing your comments on this platform.

As this publication grows, I will be introducing the full methodology behind the board scoring system and the JGB yield curve model. Both are close to ready. I want to be confident I can defend every design choice before I publish them formally — these tools are only valuable if they are robust, and I would rather be challenged on solid ground.

If you have questions, reactions, or disagreements — on the June call, the board positioning, the tools, or anything else — I would genuinely like to hear them. You can write to me in English or Japanese.

Welcome, and thank you for reading.

Takuji Okubo Japan Macro Advisors May 2026

Takuji Okubo is a macro economist with nearly 30 years of experience covering the Japanese economy, formerly at Goldman Sachs, Merrill Lynch, and Société Générale, and founder of Japan Macro Advisors. This publication does not constitute investment advice.