The Price of the Floor

For most of the past century, Congress refused to let the Federal Reserve pay interest to banks on their reserve balances. Legislators from both parties called it a handout to financial interests. In 1977, Senator William Proxmire, chair of the Senate Banking Committee, put it bluntly: “How could I support this new program to hand out hundreds of millions of dollars of money from the Treasury to the banks?”

Today, the Fed pays banks tens of billions of dollars a year in interest on reserves, and the prevailing view among policymakers is that this arrangement is essentially neutral—a technocratic detail with no meaningful distributional consequences. In a new paper I co-authored with Josh Younger, published by the Hutchins Center at Brookings, we argue that this view is wrong. The Fed’s current system of monetary policy implementation—known as the “ample reserves” framework—creates significant and quantifiable value for commercial banks, in some years peaking as high as $200 billion. Understanding the mechanics and scale of that value transfer is a prerequisite to any serious debate about whether the current system is the best one available.

The debate over whether the Fed should pay interest on reserve balances (IORB) is as old as the Fed itself. When the Federal Reserve Act was signed in 1913, the question was left deliberately ambiguous. For the next century, bankers pushed for the authority, arguing that forcing them to park funds at the central bank without compensation amounted to a tax. Legislators, by and large, resisted.

The dynamics shifted in 2006, when Congress finally authorized the Fed to pay IORB, scheduled to take effect in 2011. The Global Financial Crisis intervened, and in October 2008, Congress accelerated the timeline. Within days, the Fed began paying interest on reserves for the first time in its history.

What was initially conceived as a modest compensation for small reserve holdings rapidly became something far more consequential: the Fed’s primary tool for controlling short-term interest rates. Combined with a historically large balance sheet swollen by rounds of quantitative easing, IORB became the backbone of the “ample reserves” framework that the Fed informally adopted in 2014 and announced in 2019.

Here’s the basic mechanics. When the Fed conducts quantitative easing, it purchases long-term Treasury securities and mortgage-backed securities, typically from non-bank financial institutions. Those institutions receive deposits at commercial banks, and the banks in turn receive reserve balances at the Fed. Bank balance sheets expand on both sides—reserves as assets, deposits as liabilities.

Under the ample reserves framework, banks earn the risk-free policy rate on those reserve assets. Today that rate is 3.65%. But they pay their depositors substantially less. Are you making over 3% on your deposit account at JP Morgan Chase? It’s not just you; sophisticated wholesale depositors aren’t either, as we show in the report.

The spread between what banks earn from the Fed and what they pay depositors is determined by what’s known as the “deposit beta”—the fraction of the policy rate that banks pass through to depositors. And that beta, as extensive empirical evidence shows, is well below one.

This creates what we call the deposit franchise channel of value creation. It’s the same basic mechanism that has always made banking profitable—earning more on assets than you pay on liabilities—but the ample reserves framework has turbocharged it by massively expanding the volume of reserves in the system. Before the crisis, total reserves in the banking system averaged about $35 billion. Today, the system requires well over $3 trillion to function. We estimate that the value created through this channel peaked at roughly $200 billion—or about 1% of total banking assets—in 2023.

The second channel is more indirect but no less important. We call it the maturity transformation channel. When the Fed buys long-term securities through QE, it absorbs duration risk that would otherwise sit on private sector balance sheets. When interest rates rise—as they did dramatically in 2022-2023—the Fed takes the mark-to-market losses. Those losses ultimately reduce remittances to the Treasury, representing a transfer from taxpayers to the private sector.

In theory, this channel is symmetric: the Fed’s portfolio could just as easily generate gains. But in practice, QE programs are launched precisely when rates are near zero, meaning the Fed systematically buys securities at historically low yields. When rates normalize—often by more than markets anticipated—the losses are concentrated and substantial. The Fed’s “deferred asset” from accumulated losses now stands at historic levels.

The official defense of ample reserves rests on two pillars. The first, rooted in Milton Friedman’s famous argument, holds that because the central bank can create reserves at no social cost, the efficient policy is to eliminate the private opportunity cost of holding them—which means paying a market rate of interest. The second treats the Fed and Treasury balance sheets as a single consolidated entity, in which case QE is merely a debt management exercise with minimal fiscal impact.

Both arguments contain important truths, but neither grapples seriously with the distributional consequences. Friedman was writing about a world of modest reserve holdings and a pre-QE central bank. The consolidated balance sheet view has a different limitation. It treats the Fed and Treasury as a single entity and concludes that swapping long-term bonds for short-term reserves is simply a debt management choice with a minor fiscal impact. But the Fed is an independent institution. The securities it purchases during QE, the timing of those purchases, and the duration risk it absorbs all reflect choices that may diverge from what Treasury officials would prefer. Those choices create specific beneficiaries in ways that a consolidated accounting framework is designed to obscure, not illuminate. The fact that large banks voluntarily hold reserves far in excess of strict regulatory requirements—even when alternative money market instruments offer more attractive returns—is itself evidence that the current arrangement is economically advantageous for them.

Our paper does not advocate for a specific alternative, but we outline several frameworks that could preserve the Fed’s ability to control interest rates while curbing the implicit transfers embedded in the current system.

The simplest reform would be to shrink the balance sheet significantly while maintaining the payment of interest on reserves. Regulatory changes—particularly to resolution planning and liquidity stress testing—could allow banks to substitute short-term Treasuries for reserves, reducing the equilibrium balance sheet without sacrificing rate control.

A more ambitious approach would be tiered remuneration: paying interest on reserves only on a minimal level of required reserves, with excess reserves earning nothing. This would force banks to evaluate their true liquidity needs and create market-driven price signals about reserve demand—a significant improvement over the current survey-based system, which tolerates wide variation in how banks estimate their own requirements.

The Fed could also stop paying interest on reserves altogether, relying instead on the overnight reverse repo facility to maintain a floor under money market rates. This would route more cash through money market funds—intermediaries whose expense ratios of 5 to 20 basis points are far smaller than banks’ typical deposit spreads.

Finally, there’s the option of returning to the pre-crisis corridor system, in which the Fed manages rates through open market operations with a much smaller balance sheet. The pre-2008 system was not perfect, but the Fed lost control of interest rates only in rare and extreme circumstances, and some of those same disruptions have occurred under ample reserves as well.

This is not an abstract debate about central bank plumbing. The choices embedded in how the Fed implements monetary policy have real distributional consequences that have been obscured by a technocratic consensus that treats them as trivial.

Every system of monetary policy implementation brings costs and benefits. The ample reserves framework delivers excellent interest rate control, but it also generates significant new revenue for commercial banks, funded ultimately by the public through reduced Fed remittances to the Treasury and the mechanics of deposit pricing. Whether those transfers are appropriate given the benefits is a judgment call. But it’s one that should be made with open eyes and rigorous evidence, not waved away as a matter of “optics.”

Read the full paper here at Brookings.