President's Message: Should the Fed Target the CPI?

As Fed Chairman Alan Greenspan has been saying for some time, and the Boskin Commission's recent report confirms, the consumer price index appears to overstate the cost of living by about 1 percentage point. If this estimate is accurate, then annual cost of-living increases in government benefit programs and upward adjustments in tax brackets have been too high.

Predictably, the political controversy surrounding this subject centers on whether we should correct the bias in the CPI to help balance the budget. But lost in the debate is another important question: Should the Federal Reserve use the CPI—biased or not—to define its price-stability objective? The answer, in my opinion, is yes.

As the Boskin Commission reported, roughly half the bias in the CPI, called substitution bias, arises because current data-gathering methods fail to account for the public's ability to shift its purchases quickly, in response to changes in relative prices. This bias can be eliminated, but only with a shifting of priorities or an increase in resources at the Bureau of Labor Statistics. The other half of the bias arises because it is difficult to account for new products and improvements to existing products. These issues are thorny and unlikely to be resolved anytime soon.

Obviously, we should continue the debate about how to make the CPI as accurate as possible. That said, as a long-run objective for monetary policy, the CPI has several things already going for it. It is one of the most carefully constructed and timely of all economic measures. Moreover, it is widely recognized and often used in economic calculations and indexing arrangements.

If we're trying to achieve price stability in terms of cost of living, by definition we should be looking for CPI growth that is equal to its estimated bias. The CPI increased 3.3 percent in 1996. Subtract the 1 percentage-point bias in the measure, and you can see that we're still 2.3 percentage points away from having stable prices. Let's say, though, that the Fed were to set a 0 percent to 2 percent long-run objective for the CPI, and it ended up growing roughly 1 percent a year. We would then have effectively eliminated inflation.

My point is that uncertainty about the best way to measure the cost of living shouldn't keep us from the important task of choosing a price index to measure our progress toward price stability. As resources allow, I am certain that the BLS will continue to fine-tune the CPI. In the meantime, the Fed can define its price stability objective in a way that takes any known bias into account