The Ultimate Playbook for Reducing The Fed’s Balance Sheet | Professor Darrell Duffie on 4 Tools For Federal Reserve To Shrink Reserve Demand In Banking System

Darrell Duffie argues the correct way to analyze the Federal Reserve's balance sheet is to focus on liabilities rather than assets. The roughly 6.5 trillion dollar balance sheet breaks into approximately 2.5 trillion in paper currency, 3 trillion in commercial bank reserve balances, and 1 trillion in the Treasury General Account. Paper currency cannot be meaningfully reduced because the Fed cannot compel the public to return cash, and cutting the Treasury General Account in half yields only about 500 billion in savings. Reserve balances are the primary lever, and when the Fed sells assets the private sector pays with reserves, extinguishing them dollar for dollar.

The payment system creates enormous structural demand for those reserves. Fedwire processes 4.5 trillion dollars of interbank payments every single day, forcing banks to hold large reserve buffers to execute routine transactions. Duffie notes the asymmetry clearly: excess reserves cause no problem for monetary policy because market rates are guided by the rate paid on reserves rather than the quantity, but too few reserves can cause violent rate spikes. The September 17, 2019 episode confirmed this, with inter-dealer repo rates reaching approximately 1000 basis points above the Fed's deposit rate intraday and broader market repo rates exceeding the interest rate on reserves by over 300 basis points. Duffie attributes the spike to large banks refusing to lend reserves below self-imposed liquidity floors driven by Reg YY and resolution liquidity planning requirements.

Duffie published a paper in Brookings Papers on Economic Activity presenting four tools the Fed can use to reduce reserve demand and thereby allow balance sheet reduction. He is explicit that his work does not advocate for reduction but provides a menu of options in case the Fed chooses or is compelled to pursue it. He considers a reduction from the current roughly 3.07 trillion to around 1.07 trillion by 2028 implausible given the time required to develop and implement the necessary mechanisms.

The first tool is temporary open market operations to offset day-to-day reserve fluctuations caused by the Treasury General Account draining reserves during tax season, foreign banks reducing reserve balances at quarter-end to satisfy capital requirements, and foreign central bank repo operations. Duffie estimates this could reduce the average path of reserve balances by perhaps 100 to 200 billion dollars, with individual operations ranging from adding to removing roughly 300 billion dollars on a given day. Some Fed policymakers believe the Fed should maintain a steady course and act only on large events rather than intervening daily.

The second tool is reforming liquidity regulations, specifically Reg YY and RLAP, which require globally systemically important banks to be self-sufficient in liquidity without relying on the Fed. These rules cause banks to avoid the discount window, the standing repo facility, and reserve account overdrafts even when market repo rates exceed administered rates and those facilities sit essentially unused. Duffie describes a contradiction in which Fed leadership encourages banks to use its facilities when needed while mid-level supervisory staff may send negative signals when banks actually do so. Stigma operates as an equilibrium where if all banks are using Fed facilities any individual bank is willing to do so, but no bank wants to stand out alone as a potential sign of weakness.

The third tool is a liquidity savings mechanism, used by the Bank of Canada, Bank of England, European Central Bank, and Bank of Japan but not by the Fed. An LSM identifies circular payment loops among banks on Fedwire and nets them so reserves are not actually drawn down for offsetting transactions. The Bank of England's CHAPS system has estimated a 20 to 30 percent saving in required reserve balances through this mechanism. The Fed introduced real-time gross settlement before those central banks but did so before liquidity savings mechanisms existed, and since the financial crisis it has held abundant reserves and had no operational reason to economize. Duffie is recommending the Fed investigate potential savings using its own payments data, and whether the result would be larger or smaller than the Bank of England estimate is unknown.

The fourth tool is tiered reserve remuneration, in which the Fed would pay a lower interest rate on reserve balances held above some threshold, incentivizing banks to lend excess reserves into the interbank market rather than hold them passively. The Reserve Bank of New Zealand invented this system in 2007, setting the penalty rate on excess reserves at 100 basis points below the policy rate, and Norges Bank and the South African central bank have implemented similar systems. Duffie declines to publish a quantity estimate but states the reserve reduction would be substantially more than one or two hundred billion dollars. Banks would likely lobby against tiered remuneration because it would reduce their interest income on reserves. Duffie warns that if Congress used balance sheet size as leverage over fiscal policy it could undermine Fed independence, and the Fed currently has no options that would allow a mandated 20 percent balance sheet reduction. Governor Chris Waller has called forcing banks to scramble for reserves massively stupid and inefficient, while Governor Michael Barr acknowledges there are better and worse ways to reduce the balance sheet. Duffie predicts a substantially smaller balance sheet would be more politically palatable and would give the Fed more room to expand dramatically in a future crisis.