Generated by Codex with GPT 5.6 Sol XHigh

The Federal Reserve’s balance-sheet problem resembles a city built around a reservoir. Quantitative easing filled the financial system with liquidity; banks, markets and regulations then adapted to that abundance. Kevin Warsh, the Fed’s new chairman, would like to lower the waterline. The difficulty is that draining it too quickly could damage the system that grew up around it.

Why the Fed wants to shrink

The Fed began buying mortgage-backed securities during the global financial crisis and later added long-dated Treasury bonds to stimulate the wider economy. Further purchases during the pandemic helped expand its assets to about \$6.7trn, nearly 21% of American GDP. Treasuries make up roughly two-thirds of the total.

There are economic and institutional arguments for reversing that expansion. If quantitative easing works by pushing down long-term borrowing costs, then keeping the policy in place also keeps distorting the market for long-term debt. The Fed has another mismatch: it earns fixed yields on bonds bought when rates were very low, while paying commercial banks a floating rate on the reserves it created to finance those purchases.

Warsh’s deeper concern is the Fed’s independence. Even though the central bank buys government bonds in secondary markets, owning trillions of dollars of public debt can make it appear to be financing the Treasury. Today that may be mainly a problem of perception, but leaving the arrangement in place could make political pressure and fiscal dependence more likely.

A system built on abundant reserves

Shrinking assets requires shrinking liabilities as well. About \$2.5trn of the Fed’s liabilities is physical currency, which cannot simply be withdrawn. The Treasury’s deposit account contributes roughly \$800bn. The main adjustable liability is bank reserves, which have climbed from around \$10bn before the financial crisis to about \$3.1trn.

Those reserves are no longer an idle surplus. Before the crisis, banks held little at the Fed and lent spare funds to one another. After quantitative easing flooded the system, the Fed began paying interest on reserves to control short-term rates. New liquidity rules also encouraged banks to maintain large buffers. Meanwhile, bond purchases often created uninsured bank deposits, supporting larger investment portfolios and more lending. Each round of quantitative easing therefore left banks more dependent on plentiful reserves.

The safe minimum is hard to identify. When the Fed reduced reserves between 2017 and 2019, overnight interest rates suddenly spiked in September 2019. Similar strains returned last year. The Fed eventually bought Treasury bills at roughly \$40bn a month for several months, expanding its balance sheet again to replenish reserves. A slow decline can thus seem harmless until the market reaches an invisible threshold.

Engineering a safer exit

Proposals range from paying banks slightly less on reserves, which would encourage interbank lending, to modernising payment software so incoming and outgoing transfers can be offset before settlement. Better netting could let banks operate with smaller buffers, as they do in Britain and the euro area. A harsher approach would allow an overextended institution to fail, forcing banks to prepare for scarcer liquidity, but that would test financial stability directly.

Warsh has acknowledged that a balance sheet built over 18 years cannot be unwound in 18 weeks. The most realistic near-term compromise may be to change its composition rather than its size: as long-dated bonds mature, the Fed could replace them with short-term Treasury bills. That would reduce its influence over long-term borrowing costs without draining reserves. It would not lower the reservoir, but it could limit the distortions and political risks while the Fed searches for a safe way to do so.