Artisan Canvas

There’s A New Question About Inflation

The Consumer Price Index (CPI), which has declined for three straight months and is up just 0.1% over the past year, paints a picture of stagnant inflation. But ask any pit master about meat prices this summer, and they’ll point to some noticeable price surges. Meanwhile, gasoline has seldom been cheaper. Of course, any summer travel plans were probably canceled due to the pandemic. And that’s not necessarily a trivial point—because, while headline inflation numbers always mask some important facts about what’s happening with prices in the components, the pandemic’s rapid effects on consumption patterns may be altering some of these nuances in more fundamental ways.

Recall: The CPI measures price changes for a hypothetical basket of consumer goods. The Bureau of Labor Statistics painstakingly measures these price changes over time. Perhaps more importantly, it collects consumer spending data from the census and on a biennial basis uses that information to determine what weights to assign the sundry goods and services in the basket. In other words, the CPI fixes the relative weights in the basket and then tracks price changes. Crucially, there is a two-year time lag between the census survey data’s reflecting consumer habits and the contemporary CPI number. In most cases, the average consumption basket isn’t going to change materially over this lag period. We might have pandemic-induced meat supply chain disruptions for a few months, and prices may rise sharply, but overall this shock is unlikely to change the amount of meat consumed. But if the pandemic quickly and fundamentally changes a consumption pattern—how frequently we fly for work and leisure, for example—then it might take a couple years for the CPI basket weights to reflect this change (Exhibit 1). In the meantime, there would be a disconnect between the headline CPI being reported and measured inflation people would be experiencing.

Exhibit 1: Airline Fares Inflation Vs. Meats Inflation

Source: Federal Reserve of St. Louis as of 10 Jul 2020.

 

To explore this issue further, Harvard economist Alberto Cavallo recently reweighted the CPI based on  COVID-19 consumer spending patterns, finding a slightly higher growth rate than the Bureau of Labor Statistics (BLS) publishes. Cavallo’s analysis, he admits, doesn’t capture all spending, due to collection errors. He estimates the BLS’s measurements failed to capture 14% of spending in April 2019—and now, thanks to a growing data gap amid pandemic-inspired panic buying, it’s missing some 34%.

These questions aren’t just matters for eggheads at the BLS, either. The US Treasury Inflation Protected Securities market is indexed to CPI. Cost-of-living adjustments in Social Security and many defined-benefit public pensions use the CPI. A couple years of understated inflation might be particularly difficult for investors, retirees and senior citizens dependent on inflation-adjusted fixed payments. The US Government Accountability Office (GAO) concluded as much in a June 2020 report, noting “more timely expenditure weight updates could make the CPIs more accurate and relevant.” And what about the implications for monetary policy?

A number of Fed policymakers have begun suggesting the Fed return to its policy of outcome-based forward guidance. In this framework, the Fed would commit to an accommodative policy stance unless and until the economy hits certain unemployment and inflation benchmarks—most important to the Fed would likely be its 2% inflation target. An accurate inflation measure is essential.

The CPI’s basket issues are well known. In fact, they’re an expected and common feature of the Laspeyres-type calculations the CPI is based on. Keeping quantities the same over time simplifies the data collection and calculation. But the need for more accurate cost-of-living calculations led to more sophisticated techniques, aided in no small part by increasing computing power.

To avoid these issues, the Fed uses the Personal Consumption Expenditure (PCE) index—a chain-type index—as its official inflation measure. This index is constructed in a way that better captures how consumers might substitute goods within the basket as relative prices change. It also uses higher-frequency data, which helps raise the accuracy.

Exhibit 2: PCE Vs. CPI

Source: Federal Reserve Bank of St. Louis as of 10 Jul 2020.

 

As Exhibit 2 illustrates, CPI tends run hotter than PCE in high and rising inflationary periods and runs colder in low and disinflation/deflationary periods. And while there is a chain-weighted CPI series developed by the BLS, it’s not yet been widely adopted. With such high stakes surrounding this single data point, getting the inflation measurement right is crucial.