A former Fed advisor's case against running QT at one speed

Andrew T. Levin, a Dartmouth economics professor and former Federal Reserve advisor, is arguing that quantitative tightening should be run as a series of case-specific decisions rather than one pace applied to everything ^[1]. That matters because the position being unwound is not marginal: the Fed holds roughly 30% of all outstanding Treasury notes and bonds, and its share of agency mortgage-backed securities exceeds 40% ^{[4] [5]}.

Concentration is the entire argument. The Fed's balance sheet passed $8 trillion at its peak, and roughly $4.6 trillion of Treasuries and agency MBS were bought after 2020 alone ^{[2] [3]}. On those two figures, post-2020 purchases account for something close to three-fifths of the peak book ^[15]. A seller that large does not get to be price-insensitive on the way out.

The mechanism Levin points at is mundane and that is why it bites. When a central bank stops reinvesting the proceeds of maturing bonds, that supply returns to private markets; if it returns too fast, prices fall, yields rise, and financial conditions tighten more than policymakers intended ^[8]. Letting older, lower-yield securities roll off gradually allows the remaining portfolio to adjust and produces smaller cumulative market losses than active sales ^[9]. Levin, who has also advised central banks in Sweden and Norway and co-authored research on the costs and benefits of quantitative easing, reports a consistent finding: gradualism reduces disruption risk, while rapid or uniform tightening can trigger yield spikes and balance sheet losses that outweigh the short-term benefit of normalising faster ^{[6] [7]}.

The cross-border point is sharper than the usual comparative-central-banking filler. Levin's framework holds that each institution faces its own portfolio composition, maturity profile and domestic market conditions, and that the Fed's double concentration in Treasuries and MBS is a different problem from the ECB's book of sovereign debt spanning eurozone issuers with varying credit profiles ^[12]. Same instrument, different transmission.

Two caveats on the reporting, which comes via a single write-up in Crypto Briefing. It dates both the Fed's and the ECB's reduced-reinvestment programmes to mid-2023 and projects Fed normalisation continuing through around 2025 ^{[10] [11]}. It also attaches no as-of date to the 30% and 40%-plus ownership shares ^{[4] [5]}, and those shares move as the runoff proceeds. Treat the levels as illustrative of scale rather than a current reading.

What to watch is the language, not the level. The practical instruction in the piece is to read central bank communications for changes in the pace of runoff or hints of active sales, on the reasoning that when the central bank owns nearly a third of outstanding supply, small strategy changes produce outsized effects on yields and liquidity ^[14]. A $60 billion monthly passive roll-off is about $720 billion a year of supply handed back to private buyers ^[16], and the open question is whether pension funds, insurers and other institutional demand absorbs it at current prices ^[13]. Any shift from passive runoff toward outright sales would be the signal that the balance sheet has stopped being a background variable and started setting the long end.