MNI INTERVIEW: Optimal QT Reserve Buffer USD60B-Fed's Haubrich

The Federal Reserve will require only a relatively small buffer of about USD60 billion to cover daily fluctuations in the banking system once its QT program ends, Cleveland Fed economist Joseph Haubrich told MNI, explaining a research finding that suggests the Fed could shrink its balance sheet more than previously estimated.

The Fed has said it intends to reduce its government debt holdings until bank reserves are somewhat above "ample" – a level that ensures it can effectively control its benchmark fed funds and other short-term rates through its administered rates: the interest paid on reserve balances (IORB) and the overnight reverse repo facility (RRP) -- plus an appropriate buffer to cover volatility.

The New York Fed's working assumption is that reserves are ample at around USD2 trillion or 8% of GDP, and when reserves fell below that point in September 2019 the fed funds rate spiked sharply. An appropriate buffer on top of that has yet to be determined, but one paper from the Fed Board in September put it at several hundred billion dollars, consistent with New York Fed projections and market expectations.

Haubrich calibrates the buffer at a much lower level, using a statistical approach that balances the trade-offs of having too large a balance sheet and the financial stress of having too few reserves in the banking system.

"Compared to what they say you need for ample reserves, my numbers don't look that big," he said in an interview. "I was a bit surprised it was as small as it was, particularly since I also show the FOMC is acting as if the cost of too-low reserves was over 20 times as high as the cost of reserves being too high."

USD60 BILLION

Key to determining an appropriate buffer is calculating the costs of unwanted fed funds volatility on the one hand, and those of a large balance sheet on the other, Haubrich said. A dollar of reserve deficiency is more than 20 times as costly as maintaining an extra dollar on the balance sheet, he found.

In September 2019, when reserves dipped below USD1.4 trillion, overnight money market rates spiked to 5% and the effective fed funds rate broke through the top of the target range, forcing the Fed to intervene and to quickly end QT. Afterwards, the FOMC also made technical adjustments to its administered rates, lowering both IORB and the RRP rate by 5bp.

To set appropriate safety parameters, "we can look at what the FOMC has done. A reasonable approach is to say you don't want the effective fed funds rate to move above the IORB, because it's usually below that. A more extreme approach is you don't want the effective fed funds rate above the upper target," Haubrich said.

Using the IORB as a redline, the optimal buffer is USD58 billion, or USD49 billion if the USD1 trillion reverse repo pool is counted with reserves. If the object is to minimize the probability of moving outside the target range, the optimal buffer is USD122 billion, or USD102 billion including reverse repo balances, according to Haubrich's paper.

RANGE OF ESTIMATES

Haubrich's work is part of a growing body of research on implementing monetary policy with a large balance sheet. Analysts both inside the Fed and out have argued the minimal level of reserves required could be much higher than 8% of GDP. (See: MNI: Forks In Road Ahead For Fed's QT Plan -Ex-Staff)

The Fed Board paper, by David Lopez-Salido and Annette Vissing-Jorgensen, found the sum of reserves and ON RRP should be around 15% of GDP, plus a large buffer to cover fluctuations in liabilities including the Treasury General Account, which can move by as much as USD100 billion a day, to avoid rate spikes.

Haubrich's low buffer result is a combination of "looking at when you saw a fed funds rate that's 'too high,' and the variability of the balance sheet. You get the probability and a measure of the probability distribution and together that gives you the quantity," he said. "If you look at the buffer, it will give you information on how far QT can continue."