Seven years of rising yields mark Japan's 2019 long bond down to 31 yen on the hundred

A 30-year JGB bought at the bottom in August 2019 carried a yield of 0.12% ^[3], and if you discount 0.12 a year for thirty years plus 100 at redemption at 4.18% ^[2] you land at roughly 31 yen per 100 of face, a markdown of about 69% ^[3]. That is the crude version of the calculation (flat curve, full term, no accrued), or rather the version that shows the mechanism instead of burying it in a duration figure: 406 basis points ^[1] applied to thirty years of cash flows is most of the principal.

The interesting part is who owns that inventory. Wolf Richter's read is that the ten-year is still this low only because the BOJ's balance sheet, after more than two years of QT, continues to sit on the market ^[7], which means the institution that manufactured a 0.12% long bond is also the largest carrier of the paper that the exit from those policies marks down, and it is now shrinking that book into a market it has stopped supporting ^[6]. Yield-curve control, which pinned the ten-year between slightly negative and slightly positive from mid-2016 to mid-2021 ^[5], is off the table while inflation and the currency both push the other way ^[13].

At 3.02% nominal against July CPI of 1.9% ^[1][8], the real ten-year yield is 1.1% ^[9]; last autumn, with CPI at 3.0%, it was negative ^[9]. Set that against government debt of roughly 248% of GDP, about twice the US ratio and so on the order of 124% there ^[10][4], and ratings of A at Fitch, five notches below AAA, with A+ at S&P and A1 at Moody's four notches below ^[11], and Richter's conclusion is that the yield should be substantially higher ^[12]. A thirty-year high in the price of borrowing is not a high valuation of the risk.

The leg that reaches portfolios outside Japan is the dollar leg. To buy yen you need dollars, and the normal route is selling Treasuries, which was itself a factor pushing Treasury yields up ^[17]; on July 31, with the rate at 164 yen to the dollar, the US instead sold an undisclosed amount of euros to buy yen while Japan sold a record $97 billion ^[15], which by Richter's account was Bessent's way of keeping Japan's problem out of the Treasury market ^[16]. It held for a few days, and two weeks later long-term Treasury yields were above where they had been before the intervention ^[18]. Euro reserves spent, days bought.

There are versions of this where the marking stops. Energy pushed Japanese CPI down earlier this year before the July acceleration to 1.9% ^[8], and if that repeats, the currency steadies and 3.02% starts to look generous. Or the BOJ slows QT to put a soft ceiling back on the long end and lets the currency pay, which is the trade the joint intervention was already arguing about ^[15]. Or yields grind toward what that debt load and an A rating would ordinarily command ^[10][11], with the writedown administered by whoever holds duration rather than announced by anyone. I lean to the third, probably wrong on timing rather than direction, and the test is cheap: if the yen strengthens materially from here without more QT and higher policy rates ^[13], then repression is easier to re-impose than this desk thinks and the arithmetic above is a curiosity. A 1.1% real yield is the number to argue with, and it argues for higher.