The Federal Reserve should raise the minimum cash reserve requirements on bank deposits. This will have virtually no immediate impact on credit flows, but it will serve as an insurance policy against the growing risk of inflation.

Fed data show that commercial and industrial lending accelerated sharply in the middle of 2011, and on the whole, such loans by U.S. banks grew by 9.7% last year. This isn't unusual. The end of credit crunches like the one we've just gone through can see dramatic and sudden increases in bank lending. After six years of zero loan growth in the banking system from 1933 to 1939, for example, a sudden shift in the economic climate produced a surge in lending by U.S. banks, and from December 1939 to December 1941 lending grew by roughly 20%.

That is precisely the risk the U.S. faces over the next several years. Given the huge amount of reserves held by banks in excess of their legal requirements—excess reserves today stand at roughly $1.5 trillion—there is the potential for an even more sudden increase in credit and money growth today, accelerating the inflation rate.

The Fed has nearly tripled the size of its balance sheet over the past three years. If the assets consisted only of short-term Treasury securities, selling them to reduce so-called high-powered money, thereby sucking cash out of the system and tamping down inflation, would be easily accomplished. But the Fed holds massive quantities of mortgage-backed securities and long-term Treasurys, acquired via two aggressive rounds of "quantitative easing" and "operation twist."

This puts the central bank in a difficult position. As the economy recovers and long-term interest rates rise, the Fed will realize capital losses if it tries to sell long-term securities. If it sells mortgage-backed securities, it would lose again because it intentionally overpaid for them during the subprime crisis to support their issuers and prevent a deeper credit crunch.

Interest from future earnings will more than cover the central bank's losses on the value of its capital. But these losses would matter politically (imagine the headlines, and Ron Paul's use of them). Fear of a political backlash could keep the Fed from selling sufficient amounts of long-term mortgage-backed securities and Treasurys to shrink its balance sheet.

The Fed tells us not to worry; they have other ways to drain money out of the economy without selling securities. For example, Fed officials have pointed to their ability to use reverse repos. In a reverse repo, the Fed lends securities to banks in exchange for cash for a set period. At maturity, the securities are returned to the Fed, and the cash goes back to the primary dealers. By doing this repeatedly, the Fed can contract the money supply. But the reliability of reverse repos on a scale of hundreds of billions of dollars is untested and therefore unreliable.

The Fed could encourage banks not to create loans and deposits by paying higher interest on excess reserves. But once lending starts looking profitable, the Fed might have to raise its interest rate on reserves by a large amount to encourage banks to retain those excess reserves. And paying out such high interest on excess reserves would be impractical, since it would contribute to the Fed's short-term accounting losses.

The only reliable way to prevent an acceleration of inflation is to raise cash reserve requirements on bank deposits by, say, half a trillion dollars. The Fed should pay interest on the additional cash reserves to avoid further burdening banks. This would reassure markets today that the Fed is committed to a viable "exit strategy" from crisis-oriented monetary policy, and it would ensure stability of inflation expectations and interest rates as the economy recovers.

This will have virtually no immediate impact on credit availability in the economy because the excess reserves banks voluntarily hold above the minimum requirement are much greater than half a trillion dollars. Thus, the higher requirement will not pinch until loans and deposits get much higher.

In 1936-37 the Fed doubled reserve requirements as an insurance policy against inflation, and some monetary economists have argued that this contributed to the recession of 1937-38. But recent microeconomic analysis of bank reserve holdings by Joseph Mason, David Wheelock and me in a 2011 working paper available from the National Bureau of Economic Research showed that the higher reserve requirements were small relative to the banks' pre-existing excess reserves and had no effect on interest rates or the availability of credit.

Higher cash reserve requirements also make sense from a regulatory standpoint. A bank can reduce its risk of default either by increasing its capital ratio by selling more stock or by holding more cash assets. In practice, however, cash held at the central bank is real, while capital is an unreliable accounting fiction. Banks that fail to recognize loan losses often show higher book capital than they really have.

Moreover, the presence of cash increases the value of the bank during hard times, which reduces the incentives for banks to raise the riskiness of their loan portfolios during downturns (so-called risk shifting). In contrast, book capital requirements alone encourage banks to disguise losses and raise risks during downturns, which can leave taxpayers holding the bag.

Mr. Calomiris is professor of financial institutions at Columbia Business School.

overlay image