How Ecommerce Is Tamping Down Inflation: New Research

The concern is that traditional measures like the Consumer Price Index (CPI) no longer accurately reflect the impact of online purchasing. The data suggest that ecommerce has been deflationary to a statistically significant degree.

The implication is that unless updated to account for these factors, government policy and corporate forecasting decisions based on older models may be suboptimal. JL


Patricia Cohen and Jim Tankersley report in the New York Times:

Online spending now accounts for 10% of retail spending, and in some categories 50%. 44% of the products sold online did not exist in the previous year. A quarter disappeared from one year to the next. Federal Reserve officials see the internet’s increasing influence as why inflation routinely falls short of 2%.  The latest findings suggest that the official statistics are overstating inflation. “There is massively more deflation online — by as much as 2.5% points."

Unemployment is sinking and businesses are churning out more goods and services. Yet even with the economy standing on tippy toes, prices and wages are climbing a lot more slowly than anyone has expected.

Now a growing body of research is putting the blame more pointedly on e-commerce. The spectacular growth in online shopping, it turns out, is not only tamping down inflation more than previously thought, but also distorting the way it is measured.

The studies reinforce the views of Federal Reserve officials who see the internet’s increasing influence as a leading suspect in a continuing mystery: why inflation routinely falls short of their 2 percent target, considered the sweet spot for keeping the economy humming without overheating.

The caution is unlikely to prevent the Fed from raising benchmark interest rates at its meeting in Washington. Central bankers are engaged in a methodical effort to return rates to levels that prevailed before they were pushed near zero to fuel growth during and after the Great Recession. Some officials have raised concerns that the Fed may need to speed up that process, to prevent inflation from accelerating rapidly as the economy heats up even more.

But some of the latest findings suggest that those fears could be unfounded — and that the official statistics are overstating inflation.

“We don’t have as good measures of the economy as we should, in part because there are many new digital goods and new online channels,” said Erik Brynjolfsson, director of the M.I.T. Initiative on the Digital Economy. “At the same time, digitization of more and more of commerce creates a huge opportunity for much better measurement.”

One opportunity has come from the Digital Price Index, a compilation of data scraped from the web that Adobe Systems recently started collecting.

An analysis of that information by Austan Goolsbee of the University of Chicago and Peter Klenow of Stanford found that prices for goods sold on the internet rose much more slowly from 2014 to 2017 than indicated by barometers like the Consumer Price Index, or C.P.I.

Online prices of personal computers fell by 12.3 percent, for example, but the C.P.I. showed just a 6.9 percent drop. Toy prices online slumped 12 percent, while the C.P.I. put the drop at just 7.8 percent. Online prices for photographic equipment and supplies fell 9.2 percent compared with the 0.6 percent decline registered by the official measure.

The government said on Tuesday that the C.P.I. increased 2.8 percent in May from 12 months earlier, with much of that jump coming from higher prices for gasoline and other energy items. Not including food, energy and housing, which tend to change a lot from month to month, the index was up a much more modest 1.3 percent. The monthly figures are based on a sampling of 140,000 products sold both in stores and online. Adobe’s digital index solely includes internet sales, but the range of products covered is much bigger: 2.1 million transactions every month.

“A lot of what’s in the C.P.I. is not bought on the internet, like health care and housing,” said Mr. Goolsbee, who was also an adviser to former President Barack Obama. But if you compare the same set of goods, he said, “there is massively more deflation” online — by as much as 2.5 percentage points. A deflationary cycle, where prices keep dropping, as has happened in Japan, can mire an economy in decline.

“Prices are going down a whole lot faster than the C.P.I. shows for the same things,” he said.

After all, as bargain hunters know, comparison shopping is a cinch online. The result is that merchants are leery of raising prices. The ease of entering the marketplace, regardless of location, further ratchets up the competition. At the same time, internet retailers can often operate at a lower cost than their brick-and-mortar competitors.

Jerome H. Powell, the Federal Reserve chairman, has called it “the Amazon effect story.”

Research into online commerce in Europe also suggests that comparison shopping — even across national borders — keeps prices low. This work helps explain “why firms might be less able to change prices,” said Brent Neiman of the University of Chicago, one of the authors of the research.

New online research offers more evidence that “the more efficient method of selling through e-commerce is holding inflation down compared to what it would have been otherwise,” said Robert Gordon, an economist at Northwestern University.

Figuring out just how much of an impact online shopping has had on inflation, however, remains a challenge.

There are several measures of inflation — like the C.P.I., used to calculate cost-of-living increases in Social Security, and personal consumption expenditures (P.C.E.), an index favored by the Fed — and they all try to take account of buying and selling on the web.

The federal Bureau of Labor Statistics and the Bureau of Economic Analysis, which calculate those two measures, look at a typical basket of goods and sample prices, at times adding or eliminating items, like food and energy, that can vary widely from one month to the next.

Those agencies are constantly refining those calculations, but they inevitably have shortcomings. And as e-commerce grows and new products proliferate, some economists argue, the traditional measures have a harder time capturing the full scale of price changes online.

Even with thousands of price inspectors, the government can sample only a tiny portion of goods and services sold across the country. The traditional measures are also not well suited to take account of the constant introduction of new and upgraded products — like Instapots and athleisure yoga pants.

Cellphones, for example, were not included in the C.P.I. until 1996, when about 40 million Americans, or 15 percent of the population at the time, were already using them. In many cases, prices for those new goods — like computers — plunge over time, but price indexes fail to capture the change.

Mr. Goolsbee and Mr. Klenow discovered that new and updated products — from the iPhone X and digital assistant devices to electric bikes and toy monkey Fingerlings — were flooding online marketplaces much faster than estimated.

Putting aside apparel, where designers offer new fashions every season, the research found that 44 percent of the products sold online did not exist in the previous year. And nearly a quarter disappeared from one year to the next.

“What we have never had before is the quantity of stuff online,” Mr. Goolsbee said. Online spending now accounts for 10 percent of all retail spending, and in some categories 50 percent.

This isn’t the first time innovations in shopping have upended pricing structures. Over time, the introduction of the Sears catalog similarly held down prices compared with the small, local stores where people had always shopped, said Mr. Gordon, the Northwestern economist.

And more than two decades ago, when Amazon was just beginning to make its name as an online bookseller, a blue-ribbon commission appointed by the Senate Finance Committee — on which Mr. Gordon served — concluded that flaws in how the government measured prices had caused the C.P.I. to overstate inflation.

Recognition of measuring problems may be one reason that Fed officials said at their last meeting that they would be willing to let inflation at least temporarily rise above the 2 percent target.

Stephen Reed, an economist at the Bureau of Labor Statistics, said that “we constantly re-evaluate our methods and consider new research.”