IOR or what to do with excess reserves?

In recent days, discussions have intensified around the need for the Federal Reserve to stop paying interest on reserves (IOR) (same discussion in UK, by the way). And with the Fed set to pay over $176 billion in interest to financial institutions this year - funds that would otherwise help reduce the federal deficit - political calls to eliminate IOR are gaining traction, especially from Ted Cruz, but also Jeremy Siegel. But in practice, such a move would destabilize the foundation of the current monetary policy regime and even the USD market itself.

In the years since the 2008 financial crisis, the Federal Reserve has dramatically reshaped the mechanics of US monetary policy. At the core of this transformation is the payment of Interest on Excess Reserves (IOER), a power granted to the Fed by Congress in 2006 and hastily implemented in October 2008 as financial markets spiraled into crisis. Combined with the overnight reverse repo facility (RRP), IOER has become a cornerstone of interest rate control in an environment dominated by ample reserves and an oversized central bank balance sheet (from March 2020 it is called IOR, so this is the term I am going to use from now on).

Source: Nordea

In more detail, before 2008, the Fed operated under a ‘scarce reserves’ framework (similar to that of most central banks), where small adjustments to the supply of reserves could influence the federal funds rate, the rate at which banks lend to each other overnight. But this model broke down once the Fed began large-scale quantitative easing (QE). Specifically, the Fed implemented the QE in response to the 2008 financial crisis because its traditional tool - lowering the federal funds rate - was no longer effective. Once short-term interest rates hit zero (the zero lower bound), the Fed needed a new way to stimulate the economy (so they say!).

So, QE involved the Fed buying large quantities of longer-term securities, mainly long-term US Treasuries and mortgage-backed securities (MBS), to lower long-term interest rate (encouraging borrowing and investment), inject liquidity to stabilize stressed markets, support asset prices (wealth effect) and to signal a long-term commitment to accommodative policy (anchoring expectations). As such, by expanding its balance sheet, the Fed replaced risky or illiquid assets in the private sector with safe reserves. These purchases expanded the Fed’s balance sheet from under $1 trillion in 2007 to nearly $9 trillion by 2022 (during the Covid-19 crisis).

Source: Federal Reserve

The consequence of QE was a banking system flooded with (excess) reserves. In such an environment, controlling interest rates through reserve scarcity was no longer feasible. Instead, the Fed turned to a system of “administered rates”, directly setting the return banks could earn on their reserves (IOR), and offering money market funds a comparable risk-free return via the RRP facility (as a lower bound limit, the primary investment destination for money market funds or MMFs, as well as government-sponsored enterprises or GSEs, among others).

Currently, the Fed pays 4.4% on reserves held by depository institutions and 4.25% via the RRP facility, which accepts cash from non-bank entities like money market funds. These two rates form a corridor designed to anchor the federal funds rate within the target range set by the Federal Open Market Committee (FOMC). Banks have no incentive to lend reserves at a lower rate than IOR, and non-banks will not accept less than the RRP rate for their cash. Together, they establish a de facto floor under short-term rates.

Source: Federal Reserve

Now that we have a clearer picture of how the Federal Reserve’s monetary policy operates today, we can discuss the implications of a decision to stop paying interest on excess reserves.

How do these reserves interact with the key regulations, namely the Liquidity Coverage Ratio (LCR) and the Supplementary Leverage Ratio (SLR)? Under Basel III’s LCR, banks are required to hold enough High-Quality Liquid Assets (HQLAs) to withstand a 30-day stress period. And here, reserves shine. They’re considered Level 1 HQLA, if you want, the gold standard of this post-regulation era (more about HQLAs and their factor here). Why is that? There is no haircut, no cap, fully liquid, and immediately available to meet obligations. Holding reserves is a sure way to pass the LCR test, especially in times of uncertainty. Treasuries also count as Level 1, but reserves are even more pristine: they don’t carry market risk and settle instantly. So, from the LCR’s perspective, more reserves = better liquidity.

But there’s a twist. Under another the SLR, reserves are… a problem.

The SLR is a non-risk-based leverage rule (more about here). It doesn’t care if you're holding risky junk bonds or Fed cash, all assets count equally against your capital base. This means that holding excess reserves bloats your balance sheet without improving your SLR score. And since failing to meet the SLR can cap your ability to lend, pay dividends, or basically grow, banks are incentivized to minimize “unproductive” assets like low-yielding reserves, especially when those reserves aren’t generating enough return.

That’s why some banks began pushing back against SLR constraints after QE ballooned the Fed’s balance sheet. At one point, the Fed even temporarily exempted reserves and Treasuries from the SLR to prevent market dysfunction, we remember that during the Covid-19 crisis. That exemption ended in 2021, but the underlying issue remains: what counts as liquidity under one regulation is a liability under another.

And this is where the discussion about IOR comes in. Ending IOR would trigger a massive reallocation of liquidity. Banks currently park excess reserves at the Fed because they earn a competitive, risk-free return, which can also be seen as a kind of ‘reward’ for the fact that these reserves impact banks’ financial positions through the SLR. If that incentive disappears, banks would immediately seek yield elsewhere, we might think in Treasury bills and secured repo markets (but not necessarily). This is also confirmed in a Fed survey, where over 29% of respondents argued that any increase in market rates above the IOR would be met with a reduction in current reserve levels ranging from 1% up to 25%. This would flood short-term money markets with excess liquidity, drive rates well below the Fed’s target, and severely impair rate control.

But the most important issue is that depositors might flee banks for higher-yielding money market funds. And here is where the Catch-22 appears. While the total quantity of reserves is fixed, individual banks can choose how much they want to hold as long as someone else holds the rest. In this context, bank-sponsored MMFs would become a crucial outlet for excess liquidity. These funds are typically off-balance-sheet and not subject to the same capital requirements as their sponsoring banks. By sweeping client deposits or idle cash into MMFs, banks can reduce reserve balances without losing customer relationships or access to liquidity. In turn, these MMFs often invest in the Fed’s RRP, earning a yield with minimal credit or duration risk. In other words, they’d just shift from IOR to RRP obligations as money market funds absorb the cash fleeing banks.

Of course, there could also be calls to eliminate the positive rate on the RRP. But then we return to the previous discussion. Removing the floor set by the RRP would cause money market funds and their assets - which could theoretically grow exponentially if IOR is removed first - to begin chasing yields themselves. We can imagine that in such a scenario, control over short-term interest rates would effectively disappear, at least temporarily.

Not to mention that smaller regional banks, like Comerica, would lose an important source of income, as they are more or less dependent on IOR. Of course, I’m not arguing that it’s ideal to have a banking system so closely tied to Federal Reserve liquidity, but I don’t think anyone wants to risk another Silicon Valley Bank collapse. Just not yet.

In this situation, what solution is available? Perhaps the best-known and already successfully implemented approach by the European Central Bank, for example, is the tiering system. Of course, at least in Europe, this tiering system was introduced to shield banks from losses caused by negative rates applied to excess reserves. But the basic principle remains the same, only the policy approach changes :).

Returning, in this tiered system, banks would earn the full interest rate (currently 4.4%) only on a limited portion of their reserve balances. Any excess above a predetermined threshold would earn a lower rate or potentially nothing at all. This would help Fed to manage large reserve balances without incurring unsustainable interest expenses.

For the Fed, tiering offers a powerful middle ground. It preserves the core function of IOR and RRP while easing the burden of paying out tens of billions annually to banks simply for parking cash. A tiered system would reduce this expense by ensuring that only a portion of reserves earn the full rate, encouraging banks to put the rest to work in Treasury bills, repo markets, or other short-term safe assets.

Importantly, tiering could also serve as a fine-tuning tool in the broader machinery of monetary policy. By adjusting the threshold or the differential between tiers, the Fed can subtly steer the incentives for banks. For example, lowering the threshold means more reserves earn less, prompting banks to reallocate excess liquidity, without the Fed needing to shrink its balance sheet through asset sales or abrupt policy shifts. This would also help banks in relation to the SLR. In this way, tiering provides a flexible and precise lever to manage liquidity conditions and interest rate transmission in a system still flush with post-QE reserves.