Core Inflation and Price Stability

I’ll begin with a hypothetical question: If half the prices went down by 10%, and the other half went up by 10%, would that be price stability? If you have faith in the General Price Level, you may answer yes, but since the General Price Level has never revealed itself to me, nor did my parents teach me to believe in it, I must ask, “How can it be price stability if none of the prices are stable?” As a price level agnostic, I have to think that “reasonable price stability” (a phrase from the Fed’s mandate in the 1978 Humphrey-Hawkins Act) requires that at least some prices (perhaps as many prices as possible) be stable.

Now, food and energy prices are nearly impossible to stabilize, because they are so volatile, and because it’s extremely difficult, except over very long time horizons, to distinguish temporary fluctuations from longer-term trends. (One could, I suppose, choose a particular food or energy price and stabilize it by intervening directly in the market, but I think most people can agree that would not be a reasonable policy.) Given that food and energy prices cannot reasonably be stabilized, does it make sense to make a half-hearted attempt to stabilize them at the expense of destabilizing most other prices?

Many prices, on the other hand, can be reasonably stabilized, primarily because the people who actually set those prices prefer them to be stable and will be willing partners (up to a point) in any attempt to stabilize them. I would argue that pursuing “reasonable price stability” does not mean sacrificing Isaac or Iphigenia on the altar of the General Price Level. Rather, it means using monetary policy to discourage the aforementioned naturally stable prices from exiting the realm of stability.

You can see (I hope) why I think that the core inflation rate is more relevant to the Fed’s mandate than the overall inflation rate. You may say that I am just playing semantic games here, but I believe the semantics have substantive importance. For one thing, my reference to human sacrifice is not entirely metaphorical: the human cost of attempts to stabilize the general price level can be quite high when volatile commodities face upward price pressures.

More to the point, perhaps, the semantic problems with attempting to stabilize the general price level are indicative of the danger that such attempts will backfire even with respect to their semantically questionable objective. Food and energy commodities are typically traded in speculative markets, and speculative markets have been known to exhibit both excessively persistent trends and very dramatic reversals. I don’t think I need to remind anyone of what NASDAQ stocks did between 1995 and 2003. Suppose the same thing were to happen to oil. (It very well may be happening, though I don’t mean to suggest that I expect a reversal.) Suppose oil prices were to rise persistently for 5 years and then reverse dramatically. A central bank bent on achieving “general price level stability” would be forced to keep the economy weak for those first 5 years so as to drive down other prices and compensate for the rising price of oil. When oil prices began to fall rapidly, the economy would be weak, and the non-energy part of the economy would already be experiencing falling prices. Before the central bank’s reaction to the reversal had a chance to affect most prices, the “general price level” would be dropping dramatically. (The same logic also applies in reverse, if you imagine oil prices falling for 5 years and then suddenly jackknifing upward and consider what the general inflation rate would look like.)

It’s interesting to consider alternative core measures, such as the Cleveland Fed’s median CPI or the Dallas Fed’s trimmed mean measures, in the light of my agnostic semantics of price stability. A literal application of those semantics would suggest that means and medians are the wrong kinds of statistics to use. Rather, it is the mode of the price distribution that should be of concern if the objective is to stabilize as many prices as possible. My last example, though, suggests that the purely statistical month-by-month trimming of a price index is not an adequate approach. The prices that need to be stabilized are not the ones that have recently been most stable but the ones that are most likely to be stable in general. While there may still be a good case for ignoring outliers in the price distribution, there is an even better case for ignoring prices that are generally unreliable.