There is a lot of concern about how the Federal Reserve plans to reign in the deluge of money it's dumped on the economy to stimulate lending and spending. When will the Fed sell the securities it took off the banks' balance sheets? When will it raise interest rates to stave off serious inflation? These are crucial questions, and as the Economist's Buttonwood blog explains, it's incredibly hard to know when to start tinkering with monetary policy because the information we get about the economy's health -- quarterly GDP estimates -- can be wrong. Sometimes, it can be very wrong:

Tim Bond of Barclays Capital has looked back through the American data since the mid-1960s. he points out that the difference between GDP as first reported (and therefore known to policymakers) and GDP (as subsequently revised) was around 1.4%. That is greater than the average level of the estimated output gap over this period.

As an example, policymakers thought that American GDP dipped 7.7% in the 1974-1975 recession; in retrospect, the decline was only 2.5%. No surprise then that the Fed eased monetary policy by too much and the result was high inflation.

Bond's conclusion is that, given the shock they have suffered, policymakers may end up overestimating the output gap, keeping policy too loose for too long, and tipping us more towards inflation than deflation.

Update: Ezra Klein writes:

At some point, however, the Fed going to need to raise rates. It's going to need to decide when reflation has given way to inflation. And that's what the argument over inflation is actually about. Not whether we're there now, but whether we can trust the Fed to know when we've gotten there, or whether it's going to be listening to the wrong people along the way.

Yup. I'm with Ezra on joblessness being a graver concern than inflation right now. But the interesting thing about Bond's finding, if it's accurate, is that it doesn't matter who you're listening to if the data is wrong. And our best data about output can be misleading, or even wrong.

We want to hear what you think about this article. Submit a letter to the editor or write to