Measuring Inflation: CPI vs PCE

As inflation increases, there seems to be a division emerging between those who rely on the Consumer Price Index and those who prefer the Personal Consumption Expenditure measurement.

Normally, the difference between the two measurements is not worth debating beyond the eclectic circles of statistical index nerds. However, with inflation as high as it is now, the difference has come to play a role in politics; if you don’t want President Biden and the Democrat majority in Congress to look bad in view of high inflation, you might be inclined to go for the index that reports the lower inflation rate. This is not acceptable from a moral viewpoint – we do not manipulate or pick-and-choose statistics based on our political preferences – but the temptation is understanding: in August, the annual CPI figure stood at 5.25 percent while the annual PCE inflation was 4.26 percent.

In other words, de facto one percentage point’s difference. At that level, it is no longer peanuts which measurement you rely on. It gets worse, of course, the higher inflation we have: in 1980 the CPI reached 14-15 percent with the PCE stayed below 11 percent.

At the same time, the CPI also reports more pronounced drops in inflation when such episodes happen. We saw this at the trough of the Great Recession, but also as we were coming out of the stagflation era in the early 1980s:

Figure 1

Sources of raw data: Bureau of Labor Statistics (CPI), Bureau of Economic Analysis (PCE)

So which of the two indexes should we rely on? As always when you ask an economist anything (try “What time is it?” or “Did you remember to buy eggs?”), the answer is a solid “it depends”. That is frustrating, but it is better than lawyers who start a stopwatch, talk for 45 minutes and then send you a bill. But it actually does depend on what you want to do with the index numbers. There are three things to keep in mind when choosing between the CPI and the PCE.

To begin with, both are indexes, which is not only a good thing – the price index is a statistical invention worthy of a Nobel Prize – but uncontroversial in itself. The index is a tool for organizing and processing large amounts of data for a variety of purposes, in our case policy making. However, there is no natural law that prescribes what index to use when you measure such a phenomenon as inflation. There is a vast body of economic research behind the CPI and PCE indexes, making them valid and (despite what uninformed fools in social media will tell you) highly accurate.

At the same time, as with everything in applied, empirical science (hello, vaccine debaters?) there is no absolute truth as to how the index itself should be built. There is room for methodological differences, which leads to the small but visible gap between CPI and PCE numbers visible in Figure 1. Therefore, if we really want to choose one, we should familiarize ourselves with those differences.

Secondly, the indexes differ primarily on two points, the first being the basket of goods and services they measure. While the CPI focuses on out-of-pocket expenditures, the PCE also includes items we buy through indirect arrangements. Health insurance is a good example. The difference here is not an exhibit of CPI weakness, but instead the result of different focus. The PCE is part of our national accounts system, which means that it must cover “everything” in order to pull its load in reporting the size, composition and evolution of our economy. The CPI, on the other hand, is aimed at capturing the impact of prices themselves on consumers; long-term expenditures are not affected overnight by inflation, and sometimes not at all, as with a car loan.

In other words, if we want to study price changes for the purposes of anticipating short-term swings in consumer spending, the CPI is preferable. If, on the other hand, our purpose is to track longer-term trends in the economy, the PCE serves us better.

Third, and perhaps most important: there are differences in the demographics they base their surveys on. The CPI is based on a survey of consumers in urban areas. It is said to represent consumer expenditures and prices among 85-90 percent of the U.S. population, but it still leaves a small part out of its target demographic. The PCE, on the other hand, is universal in its application, making it the preferable index. At the same time, the CPI is based on data collected from consumers, while the PCE is based on surveys of sellers. This gives the CPI an accuracy edge over the PCE edge, especially insofar as consumer-experienced inflation is concerned.

So, back to the question – and please without the two-word standard economist answer: which of the two indexes is preferable?

If our concern is fiscal and monetary policy and how they can affect consumer behavior, the CPI is preferable. If our purpose is more long term and more academic in nature, we go with the PCE. Despite the shortages of the CPI in terms of capturing expenses on certain types of items, it gives us a better idea of how consumers perceive inflation and therefore a better idea of how they may respond in the near future.

For more reading:

A nice analysis from the Federal Reserve of Cleveland:

See chapter 17 of the BLS index handbook:

And chapter 5 of the corresponding publication from the BEA: