That's what folks must have been speculating about when they read Neil Irwin's account of the Fed's decision to put off a hike in interest rates this week. Near the end of the piece Irwin tells readers:

"As Stanley Fischer, the Fed vice chairman, said in a television interview last month, if the Fed waits until it is absolutely certain it is time to raise rates, it will probably be too late.

"In other words, Fed officials inevitably have to make a decision based on what their models predict, not on cold hard evidence."

Huh? What exactly is the bad thing that happens if it's "too late" when the Fed acts? In the models I know, we start to see some acceleration of inflation. Given that the inflation rate has been well below the Fed's target for most of the last six years, the Fed should want the inflation rate to accelerate, at least if it is following its stated policy of targeting a 2.0 percent average rate of inflation. (This rate is too low, according to folks like I.M.F. chief economist Olivier Blanchard.) The inflation rate could average 3.0 percent over the next four years and still be consistent with the Fed's stated target.

None of the standard models shows a rapid acceleration of inflation as a result of the Fed being "too late." They show the inflation rate increasing very gradually. According to the most recent projections from the Congressional Budget Office being a full percentage point below full employment for a full year would lead to just a 0.3 percentage point rise in the rate of inflation. This would appear to be the cost of being too late in the standard models.

Perhaps Irwin could tell readers what Mr. Fischer was thinking about in giving his warning.



The headline writer for this piece deserves some grief for writing that "Yellen blinked" in reference to her decision not to support an interest rate hike. The implication is that this decision was due to a lack of will as opposed to good judgement. This is not the job of the headline writer to determine, nor the implication of the piece. Thanks Jeff for pointing this out.