Dovish Data
AI Duration, Fed Reform and the Long End. The flawed target, benchmark and model. Bank equities are pointing to more constructive monetary policy in the US and Japan.
In this week’s note:
The Fed’s inflation target, benchmark and model are all flawed. The 2% PCED target was a crisis-era commitment that became permanent policy without adequate consideration of second-order effects, while the reliance on revision-prone PCED and a monetary-policy-centric inflation model ignores the fiscal origins of the postwar, 1970s and pandemic inflation shocks. Chairman Warsh understands the institutional problem, but reforming the Fed’s framework will require moving carefully against a deeply entrenched status quo.
Mother Market got the inflation and policy setup right. July CPI, PPI and retail sales were consistent with cooling inflation momentum, lower odds of a September rate hike and a clean market forecast of disinflation that does not depend on additional Fed restraint. The end of reserve management purchases is a step toward balance sheet reform, but it remains modest relative to the Fed’s ongoing footprint in longer maturity Treasuries.
AI infrastructure spending is rate sensitive, just not primarily to the Fed funds rate. The Big Spenders’ capex plans are already responding to changes in the cost of equity, credit spreads and long real rates, while AI-related duration supply is likely to remain an important pressure point for the Treasury market. If policymakers respond with policy-rate hikes rather than balance sheet restraint, they risk strengthening the AI impulse while tightening financial conditions for the rest of the economy.
Government interest expense is not yet the Treasury market’s central problem. Deficits remain large, but outlay growth is running below nominal GDP and swap spreads suggest the rise in long real rates is not primarily a Treasury supply story. The risk is that higher short rates and another turn toward fiscal expansion after the election push the government closer to funding interest expense with still more debt.
The market data have improved, but the long end has not yet cooperated. Payrolls, CPI and retail sales were all bond-friendly, yet the back end of the Treasury market failed to rally, likely reflecting ongoing AI-related supply concerns and investors waiting for post-Labor Day issuance. We are not giving up on longer maturity USTs, while financials—especially regional banks—remain a favored expression of a bull steepening curve and coming regulatory relief. Japanese equities are likely to respond favorably to rate hikes, banks are the key tell.
Note: We use Copilot to help us with the summary, but the 3663 words and 17 charts and tables that follow the paywall is all us, authored the old-fashioned way.