Right now, we are trying to understand the dynamics of inflation The headline personal consumption expenditures (PCE) price index fell 0.2 percent in December. Its 12-month increase was 0.75 percent, down from 1.6 percent in June. Should this low, and still falling, rate of price inflation retard the date of the liftoff from the zero-interest-rate policy we have been operating for more than six years?
I think not. We all know that headline inflation is being held down by the big decline in energy prices that began in the second half of 2014. We know that once energy prices stabilize, headline inflation is likely to bounce right back up. Policy needs to take past inflation into account, but it needs to take future inflation into account, too. Thats just another way of saying that, for policy purposes, its inflations medium-term trend that matterswhich is why analysts and policymakers pay so much attention to core inflation measures. The widely heralded FRB/US model that has been used by the Board of Governors staff since 1996 is an example: It is built around PCE inflation excluding food and energywhich is the traditional measure of core inflation. Ex-food-and-energy PCE inflation was essentially zero in December, month over month, while the 12-month rate slipped to 1.3 percent from 1.5 percent in June.
...
A good core inflation measure strips the noise out of headline inflation and leaves the signal. By that standard, recent analysis shows that the ex-food-and-energy PCE inflation rate that drives the FRB/US inflation forecast is a second-rate core inflation measure, at best. An alternative measure developed at the Dallas Fedthe Trimmed Mean PCEis superior in three respects.
First, trimmed mean inflation is better insulated from transitory energy-price swings. Since 1994 (the start of the current 2 percent-inflation era), conventional core inflations correlation with changes in the real price of oil is 0.26, while trimmed mean inflations correlation is just 0.05.
Second, as judged by root-mean-square error, it is more closely aligned with intuitive, direct measures of trend headline inflationlike the 36-month centered average, or headline inflations average over the coming 24-month periodthat we are only able to observe after the fact.
Third, trimmed mean inflation has shown substantially less systematic bias. Over the past 10 years, looking only at data that would have been available to policymakers in real time, conventional core PCE inflation has averaged 1.65 percentnearly 30 basis points below headline inflations 1.94 percent average. Meanwhile, trimmed mean inflation has come in at 1.83 percentjust 10 basis points below headline. Setting policy using conventional core as your guide is like navigating using a compass: It has a systematic bias and is influenced by local anomalies in the Earths magnetic field. Using the trimmed mean to set policy is more akin to navigating by GPS.