One of the tasks delegated to central banks like the U.S. Federal Reserve is to control the rate at which consumer prices rise—in other words, inflation. Yet prices seem to rise and fall every week. A quick trip down any grocery store aisle will confirm that hunch. Cheerios may be on sale one Sunday but not the next. Eggs may be buy-one-get-one-free for a few weeks, only to return to their original price. Is inflation stable when there is so much fluctuation over short time spans?
A long-held assumption among central banks is that prices are relatively constant over periods of about a year or so. This “price stickiness” is one of the reasons why central banks can influence the economy through monetary policy. But in the early 2000s, studies using microdata seemed to confirm what consumers saw on a daily basis, showing that prices actually flexed frequently, changing about once every three months. This research, by Mark Bils and Peter Klenow, fractured traditional consensus about how a central bank influences the economy through monetary policy. In the wake of their dissenting research, many macroeconomists wondered whether they had been barking up the wrong tree for years.
Fortunately for central banks, the old view of price stickiness has been vindicated in a recent study by Sergio Rebelo, a professor of finance at the Kellogg School of Management, Martin Eichenbaum, a professor of economics at Northwestern University, and Nir Jaimovich, a professor at Duke University. Their work confirms that the traditional view was right all along, but with a twist: prices are sticky, but it is important to distinguish between reference prices and sales prices. It turns out that reference price—the price that consumers pay when a good is not on sale—is quite sticky. It changes only about once a year. This price is important because it applies to half of the purchases that consumers make.
Sticky Reference Prices and Irrelevant Sales
The researchers used weekly observations of scanner data from individual consumer purchases of more than 60,000 products from a national supermarket retailer with 1,000 locations. Across the two years of data in their sample, the researchers discovered an intriguing pattern. The supermarkets changed their prices, sometimes as frequently as once a week, dropping them drastically for temporary sales geared to drive foot traffic to their stores. For example, snack chips that normally sold for $1 were temporarily knocked down to 50 cents.
Previous research predicted that these sales were important because rational consumers would ignore the regular price and delay their purchases until goods went on sale. But according to the scanner data, consumers bought products at the sale price and the regular price at similar rates. Rebelo says that in previous research, like Bils and Klenow’s, economists interpreted this oscillation between sales and reference prices as price flexibility. Yet this oscillation “is actually not what economists mean by price flexibility,” Rebelo says. “Conventionally, flexibility in prices means adjustments that respond to market shortages or surpluses, not temporary sales at grocery stores and the like. Clearly these sales have something to do with attracting customers to the store; they have nothing to do with macroeconomics.”
Because the vast majority of price changes were movements back to the reference price, the researchers argue that prices are actually quite rigid after all.
The data Rebelo and his coauthors gleaned from supermarket scanners showed that even though prices change often, most of these changes occur after a temporary sale—with prices generally going back to the reference price. In other words, this is “not the kind of price flexibility that would make monetary policy unable to influence the economy,” he says. As a result, Rebelo and his coauthors argue that the sales simply create noise in the data, and that when thinking about monetary policy the noise should be tuned out.
They found that, on average, reference prices endured for about one year, and that temporary sales simply gave the illusion that prices were flexible. Because the vast majority of price changes were movements back to the reference price, the researchers argue that prices are actually quite rigid after all: prices consumers pay on a day-to-day basis may change, Rebelo says, “but the reference price is quite sticky.”
This mechanism remained hidden for so long because to uncover it, researchers needed to access linked data about individual purchasing patterns over time and determine how reference and sales prices fluctuate. Until recently, such detailed microdata has not been available, Rebelo says.
Another finding of the paper confirms an old theory in macroeconomics about price rigidity, called menu costs. Since firms pay a fixed cost every time they change their prices, menu costs were thought to contribute to price stickiness by dissuading frequent small adjustments. Rebelo and his colleagues show that this prediction is largely true—supermarkets changed their prices only when they were very far away from their target markup. The researchers were able to tease out this result because they could use cost data to calculate target markups. This combination of data enabled them to see the connection between target markups, reference prices, and sales prices. Other researchers did not uncover this finding because they were not able to access this type of data, Rebelo says.
“What our paper shows is that even though we see these sales, the price schedule is basically fixed throughout the year, and that is why monetary policy can influence the economy throughout the year,” Rebelo says.
Though the supermarket data has its limitations, Rebelo says, the study reveals important insights about how retailers set prices, which in turn have implications about how monetary policy influences the economy. Everyone agrees that service goods, like doctor fees and tuition, are considered very sticky; some other products, like gasoline, change prices daily. Yet these prices are only part of the equation. “The real action is in apparel and food,” Rebelo says. “These prices are important because they account for a fair amount of people’s purchases.”
Since Rebelo, Eichenbaum, and Jaimovich published their paper, other researchers have found similar results by analyzing the data used to calculate the consumer price index data. These findings reinforce the strength of the original results.
“What this research does is re-establish the old consensus,” Rebelo says. “We show that the apparent price flexibility is fake because prices fluctuate equally between these values, a reference price and a sale price.”
Related reading on Kellogg Insight