经济周刊：选择自己的通胀指标

2 July 2026 Richard de Chazal, CFArdechazal@williamblair.com+44 20 7868 4489 Economics WeeklyChoose Your Own Inflation Metric Louis Mukamalmukama@williamblair.com+1 312 364 8867 William Blair As we discussed in last week’sEconomics Weekly, theWarsh-led Fed appears to be looking to jettison thereaction function of more recent FOMCs, but Warsh isseemingly waiting to announce what its replacementwill be. Warsh also seems to be suggesting that he wouldlike investors to take on more risk themselves and relyless on any hand-holding and moral suasion from theFed. This likely means the strike price for the Fed put isnow lower, and investors may have to stomach a littlemore volatility in the future. In many ways, this feels likeNassim Nicholas Taleb’s antifragile approach to mon-etary policy. It also seems sensible in a world where theinflation regime has shifted and where inflation risksare now either symmetrically or even asymmetricallyskewed to the upside, as opposed to being solely to thedownside from 2000 to 2020. Under this new regime,where inflation is harder to predict and more prone torise, the Fed’s reaction function necessarily needs to betweaked, and Warsh is right to want to at least questionthe nature and measurement of the inflation metricsthat the Fed uses to gauge inflation’s proximity to its tar-get.In thisEconomics Weekly,we discuss some of thealternative measures of inflation and what might bethe best approach in this uncertain environment. the projects I would hope to undertake as one of thefirst reforms at the Fed is a data project where wewould go off and we would evaluate with the publicsector and the private sector, including the Bureauof Labor Statistics, a survey of a billion prices.WhatI’m really most interested [in], Senator, is what’sthe change of that 500 million and one price?Because that’s inflation. That’s a change in thegeneralized level. In a market economy, priceschange all the time. And I don’t want to beconfused by that. I want to know what inflationreally is, and I still think there’s some work todo.[Emphasis added] Once again, in a world where inflation is not permanentlyhigh, but seemingly more volatile and increasingly beingimpacted by recurring supply shocks in a way that wasnot the case in previous decades, central banks can stillafford to look through first-round supply shocks; however,they need to be exceptionally vigilant when it comes to thethreat from second-round effects. As a result, gauging alternative measures of inflation,including trimmed means and multivariate measures,makes a lot of sense and is useful in providing a broaderpicture of the inflation environment. This is particularlytrue when, due to budget cuts at the Bureau of Labor Sta-tistics, there has been a sharp increase in the share of theCPI basket that is no longer resulting from direct surveyevidence, but is being imputed (exhibit 1). Questioning Everything About Inflation In his first press conference, Fed Chair Warsh said that hewas commissioning a series of taskforces, one of whichwas on inflation: First on the inflation framework review, theirremit is, what are the drivers of inflation? What’sthe Fed’s responsibility for inflation? In part, howdo we measure inflation? As Warsh discussed at his April confirmation hearing, heprefers the trimmed mean measures of inflation: My broad sense is that these inflation risks and theinflation damage the last several years is improv-ing somewhat. It has improved somewhat in thelast year.The measures I prefer are looking atthings that are called trimmed averages, wherewe take out all of the tail risks, all of the one-off items, and we ask ourselves whether thegeneralized change in prices is having secondorder effects on the economy.Again, they’re notwhere they should be, but I think that the trend isquite favorable. Headline PCE inflation is a weighted average of ratesof change in the prices of every good and service in thebasket, with each item weighted by its share of totalspending.Warsh’s favored trimmed mean measure ranksthose price changes from high to low and removes (ortrims) a fixed share from the top and bottom of the rank-ings. The inflation rate is the (expenditure-weighted) Some in economics profession are increasinglylooking to these median type measures, but among William Blair average of what is left. An extreme form of the trimmed-mean approach is median inflation, which excludes allprice movements other than the median—Warsh’s refer-ence to the 500 million and 1 price. The Dallas Fed, which publishes the most closely watchedversion, asymmetrically eliminates the top 31% of thefastest-rising price categories, but only the bottom 24% ofthe slowest. The cutoffs are set to reflect the behavior ofseveral measures of headline inflation between 1977 and2009, a period when the distribution of price changes wasnegatively skewed (exhibit 2). Another alternative measure that we have followed closelyover the years, in part because it a