Here’s Why the Fed Seemingly Ignores the Consumer Price Index 

Inflation grabbed headlines again recently when the Consumer Price Index (CPI) increased at an annual rate of 3.5% in March, up from 3.2% in February. But the Federal Reserve’s preferred measure of inflation—the Personal Consumption Expenditures (PCE) price index—most recently ran at a significantly lower rate of 2.5%. As a matter of fact, over the past 20 years, the annual PCE inflation was on average almost 0.4 percentage points lower than the CPI’s. 

Why are these two measures so different? And which one better reflects the cost of living and is more useful for the policymakers? 

Substitution between goods – The CPI is calculated using a fixed basket of goods and services. As a result, it doesn’t accommodate changes in consumers’ purchases—like switching from apples to oranges if oranges become cheaper. The PCE incorporates these changes and arguably better reflects changes in the cost of living. However, it requires detailed data on consumer spending for the month, which is why PCE inflation is usually published about two weeks after the CPI, a relatively long delay when dealing with monthly data. 

Similar, but not the same – The CPI generally focuses on prices consumers pay out of pocket, while the PCE looks at prices of all goods consumed by households. Although it sounds like a technicality, it means that the PCE includes prices of items paid for households by someone else; for example, the prices of medical services paid by health-insurance companies or by government programs. As a result, it focuses less on consumers’ wallets, and more on the prices in the broader economy. 

Different consumption baskets – As the two indexes are constructed using different surveys on consumer spending, the weights applied to items also differ a bit. The most relevant difference relates to the cost of shelter, which accounts for around 35% of the CPI, but only around 16% of the PCE. Importantly, rent payments paid directly by tenants are only a small part of shelter inflation. Most of it reflects the costs of owner-occupied housing, estimated by the potential rent that owners would have to pay for the house were they to rent it at a market price. This methodology is supposed to reflect the value of the stream of services provided by the owner-occupied housing and make inflation measurements more robust to changes in the ownership structure of the housing stock. It also avoids focusing on housing prices or interest payments which tend to behave more like financial assets. Given the large increase in the cost of shelter over recent years, this difference in weights accounts for most of the discrepancy between the CPI and PCE inflation rates.  

Because the PCE better reflects the changes in households’ consumption patterns, it likely better reflects changes in the costs of living and price trends in the economy. That’s why the Federal Reserve prefers it. But the CPI remains useful and popular because it’s timelier, focuses more on out-of-pocket costs and is comparatively similar to other inflation indexes used internationally.