Markets love gradualism until the dam gives way. What if the most fragile structure in US finance is the assumption that the Fed can shrink a $6.7 trillion balance sheet without moving the ground under it? Kevin Warsh wants to test that assumption. His motive is clean: reduce the blur between monetary and fiscal power. The risk is messier: plumbing and psychology, not ideology, will decide how far he can go.
The balance sheet looks like a dial. Turn it up to ease, turn it down to tighten. But it is not a dial. It is a web of duration, collateral, bank reserves, and expectations. The last decade taught the wrong lesson: that size alone could stabilize the system. In 2013, just a hint of slower bond buying moved yields violently. In 2018, quantitative tightening met a suddenly shallow Treasury market. More recently, a market that prizes liquidity on the screen can vanish in a stress episode, as March 2020 showed. A large central bank portfolio is not a neutral backdrop; it is part of the market’s scaffolding. Remove beams in the wrong order and the structure flexes. Control is often an illusion until it breaks.
The Fed pivoted to an ample reserves system after the crisis. Instead of steering rates by scarce reserves, it pays interest on reserves and uses a floor system. That design needs a lot of reserves to function smoothly. Banks, scarred by 2008 and burdened by regulation, now hold large buffers. In 2019, reserves were still in the trillions when the repo market cracked. Ample was not ample enough. You can lower the balance sheet only if you also lower demand for reserves. How to do that without eroding safety? We could relax liquidity coverage rules or reduce the stigma around using backstops. Both carry costs. The point is simple: this is not a balance sheet problem alone. It is a demand-for-liquidity problem. You cannot drain the pool while swimmers still need the water.
Selling long-dated Treasuries or letting them run off depresses prices and lifts yields. That is the point of tightening. It also lifts mortgage rates and reprices duration risk across portfolios. The transmission is not linear. Mortgage convexity hedging can amplify moves. A 30-year mortgage rate that drifts 50 basis points higher can kill marginal housing demand and freeze supply, a mechanism we already saw. Equities priced off low discount rates are another channel. When multiples are rich, small changes in long-term yields carry outsized effects. Balance sheet runoff that is “market neutral” in theory can be anything but in practice. Investors trained by a decade of backstops assume a gentle slope. History suggests cliffs appear when balance sheets shrink into shallow market depth.
Warsh has said out loud what many imply: a swollen balance sheet blurs monetary and fiscal lines. When the Fed is the dominant holder of Treasuries and mortgages, it takes on distributional choices, whether it intends to or not. Remittances to Treasury have flipped to deferred assets as interest paid on reserves exceeds income from securities. That is a technicality with political heat. Congress hears “losses,” even if accounting is different for a central bank. Who owns the risk of duration in a rising-rate world? Who decides whether the Fed should hold trillions of mortgages? These questions are political because they involve trade-offs. Shrinking the balance sheet may reduce political risk in the long run. The path to get there runs through the very pipes where policy and politics meet.
Think of this as a coordination game. If banks and dealers believe reserves will stay abundant, they lend and intermediate. If they fear scarcity, they hoard. Money market funds can swing between the reverse repo facility and private markets with a small change in rates. Expectations make the equilibrium. In 2019, the belief that reserves were getting tight was as damaging as the actual level. Today, the reverse repo facility still acts as a buffer, but it is not a bottomless sink. As balances fall, the line between ample and scarce moves closer. A clear plan matters. Pre-commitment to slow or pause QT when certain market depth or funding metrics flash red would anchor beliefs. Game theory says common knowledge beats hope. The task is to remove surprises, not just bonds.
The most dangerous path is the one with no alternate routes. Consider a few plausible shocks: a spike in Treasury issuance, a global risk-off episode that demands dollar collateral, or a renewed hit to regional banks from deposit flight and unrealized losses. Each by itself is manageable. Combined with a faster-than-expected drain of reserves, they create a thin margin for error. The 2023 regional bank failures were a reminder that duration losses do not matter until they suddenly do. Treasury market depth can also evaporate under stress, forcing the central bank back in. If the exit from a $6.7 trillion balance sheet requires emergency re-entry, credibility suffers twice. Good engineering starts with failure modes. If the plan cannot handle fat tails, it is not a plan. It is a wish.
If the aim is a smaller, cleaner balance sheet, the system needs to absorb more noise without breaking. Some tools already exist: a standing repo facility to cap spikes in secured funding, and a discount window that works without stigma. Others deserve a fresh look. Countercyclical liquidity buffers that ease during QT. Tiered interest on reserves to reduce the subsidy to passive reserve hoarding. Greater use of Treasury buybacks to smooth the maturity profile the market sees while the Fed reduces its duration footprint. Above all, shift the portfolio mix away from agency MBS. Mortgages sit at the intersection of policy and politics and add convexity risk. An antifragile design accepts shocks and contains them. It does not pretend they will not happen.
The temptation is to fixate on the headline number. Pre-crisis, the balance sheet was about $800 billion. That world is gone. A better metric is whether the market can clear large trades without air pockets, whether repo rates track policy without spasms, and whether banks can meet outflows without forced sales. Publish a reserve demand curve with confidence bands. State a corridor for reserve supply and update it with data, not hope. Announce a taper on any sign that bid-ask spreads in Treasuries widen beyond normal or that turn-of-quarter funding rates gap higher. This is less ideology, more instrumentation. In probability terms, the job is to reduce the variance of outcomes, not just the mean. That is pragmatism with a control system.
Warsh’s task force can make this concrete. Define the objective function: a smaller balance sheet consistent with stable control of short rates and resilient market plumbing. Set guardrails that slow or stop QT when liquidity metrics breach thresholds. Coordinate with Treasury on issuance and buybacks to avoid dumping duration into fragile windows. Prioritize selling or allowing runoff of MBS before Treasuries to depoliticize the portfolio and cut convexity risk. Use tiered interest on reserves to wean banks off excess balances without eroding safety. Communicate the plan in simple terms that align expectations. The paradox is that you shrink faster by moving slower when conditions demand it. Pragmatism, not ideology, is the only way to tame a $7 trillion apparatus without cracking the foundation.
