Inflation

What's the inflation rate? Depends who you ask.

Inflation doesn't hit all Americans the same way, a new paper finds. What does that mean for the economy? 

By Danny Vinik

How much did prices increase last year?

Economists and financial analysts will say 0.7 percent, as the Bureau of Labor Statistics reported in the Consumer Price Index. That's the indicator the government traditionally refers to as the official inflation rate, but many Americans believe it dramatically understates “real” inflation. ShadowStats, a website frequented by inflation skeptics, puts inflation closer to 10 percent in 2015.

That's a huge difference. In surveys, Americans also offer a wide range of estimates for inflation, anywhere from 15 percentage points or higher to actually decreasing.

Inflation—both the real number, and people's perception of it—is crucially important for economic policy. In Washington it helps drive major moves like interest rate-setting; at home, consumers may make spending decisions based on their expectations of price increases, so this level of uncertainty is a problem for understanding the economy. What’s going on?

Now a new study offers a clue. Economists have historically argued that Americans' estimates of inflation are so wrong because they either don’t understand inflation or misunderstand how much prices are actually increasing. But a new working paper, released by the National Bureau of Economic Research this month, suggests a potential new explanation: American households experience inflation very differently, and their estimates of inflation vary accordingly. In other words, inflation isn't just one thing: it varies widely from household to household. So their responses aren’t wrong; they’re just based on personal experience. The findings could have important implications for monetary policy.

The new paper offers the most comprehensive look at how prices vary for individuals at the household level. Economists have long known that the specific prices each American pays for goods and services will vary depending on geographic location, time and other factors. Chicken breasts in New York City may cost more than an identical product in Omaha, Nebraska. This paper wasn't looking at prices for a given time and place, but was instead looking at the annual change in prices—inflation—for specific households over time.

“When we measure inflation typically, we measure inflation for a single bundle of goods, a basket like in the Consumer Price Index,” said Greg Kaplan, an economist at the University of Chicago and co-author of the paper. “Effectively that assumes that every household is facing the same prices of goods. We all understand that there's a lot of differences in the population in terms of what prices we actually face for different goods and we wanted to get a measure of how big those differences were.”

Kaplan and his co-author, economist Sam Schulhofer-Wohl from the Federal Reserve Bank of Minneapolis, used a data set of 500 million transactions from 50,000 U.S. households from 2004 to 2013 to track the yearly changes in prices of goods purchased by individual consumers. The data is largely limited to goods with bar codes, and excludes costly durable goods like housing and cars, as well as many services, including education. In other words, the paper largely focuses on goods that people buy at the grocery store and retailers. This isn’t necessarily a bad thing. Given the frequency of these expenditures, they are likely to form the basis of people’s individual inflation estimates, especially compared to far more expensive items like a house or car, which are not purchased as often.

Taken in aggregate, the new paper’s data track fairly closely with the Consumer Price Index. (The authors reweighted the CPI for these calculations so the distribution of spending across different categories of goods and services matched up with their data.) But the average annual price changes masked a wide range of household-level inflation rates. The 25th percentile and 75th percentile of households experienced inflation rates that differed widely over the 10 year period. For the 10th and 90th percentile, the annual difference in inflation hit almost 20 percent.

Why do household-level inflation rates vary so much compared to the CPI? First, the CPI calculates inflation measures using broad expenditure categories that encompass many different products. Households will purchase different products within each category and in turn, their individual inflation rates will vary. Second, the cost of identical products will vary depending on many factors, like whether they are on sale at a given store or whether different stores charge different prices for the same item. Stores could even change different prices on different days of the week.

Previous research showed that households experienced different inflation rates because they purchased different goods. But Kaplan and Schulhofer-Wohl’s new research found that the more important reason for household-to-household variation is that prices for identical goods vary.

This doesn’t mean ShadowStats has been correct all along in its extremely high inflation estimate, which it generates from methodological changes to the traditional Consumer Price Index. In fact, this paper offers yet another confirmation that ShadowStats is wrong, as ShadowStats is looking at aggregate inflation numbers—just as the CPI does. But this does offer a potential explanation for why Americans have such varying beliefs about inflation: they come from personal experience.

“It’s possible some people think inflation is really high," said Kaplan, "because for them, inflation actually is high.”

When Federal Reserve officials meet to set monetary policy, as they are this week, they take into account a wide range of economic measures, most of which look at the U.S. economy on an aggregate level. But those aggregate statistics can often mask important changes within the country. For instance, while wage growth has barely budged since the Great Recession, more and more economists believe that workers are actually receiving raises—but in the aggregate statistics, those wage hikes are being masked by low-income people returning to the labor market and finding jobs. That’s why the top-line wage level has been stuck around 2-2.5 percent while the Atlanta Federal Reserve’s wage tracker, which looks at wage changes for continuously employed workers, puts wage growth at a much stronger 3.5 percent.

Kaplan suggests that the new paper has important implications for monetary policy. The Fed typically increases or decreases the Fed funds rate—the rate at which banks can borrow from one another overnight—to meet its dual mandate of full employment and 2 percent inflation. Higher aggregate inflation, all else being equal, should cause consumers to spend more as they know prices will increase in the future. But based on Kaplan and Schulhofer-Wohl’s new findings, households don’t face the “aggregate” inflation rate. They face a household level one, which will wary widely.

The Fed, of course, isn’t going to start targeting inflation at the household level. But it does mean that the central bank should be more aware that their monetary policy decisions will affect households in very different ways. It also offers a lesson for economic analysts who scoff at Americans who say inflation is 10 percent. For them, it might actually be that high.