There’s no productivity miracle hiding in the data
New study finds productivity measurement problems no worse than 20 years ago
Why hasn’t the technology revolution lifted the U.S.’s miserable productivity growth rate? Silicon Valley evangelists have a ready answer: The growth is there, we’re just not measuring it.
Popular though the stealth productivity boom is, it is a myth. A new study finds that official statistics do tend to understate productivity, but by less than two decades ago. So even if all the benefits of social networks, online shopping and less invasive surgery were being measured, it probably wouldn’t change the overall picture: Productivity growth is historically slow.
The study is to be released Wednesday by the Hutchins Center on Fiscal and Monetary Policy at the Brookings Institution. Author Brent Moulton worked on federal statistics from 1985 until 2016 when he retired as associate director of the Commerce Department division that computes gross domestic product.
When the dollar value of GDP rises, part of it is because of inflation, and part of it is because workers and firms are pumping out more or better products. If statisticians overestimate inflation, they will underestimate how fast actual output is growing, and thus productivity.
In 1996 a commission led by the economist Michael Boskin concluded the consumer-price index overstated inflation by 1.1 percentage points. Mr. Moulton calculates the bias has since dropped by a quarter of a percentage point. For the personalconsumption expenditures price index, which is used to calculate productivity data, the bias has dropped half a point. He concludes that overall, productivity
growth was understated by 1.1 percentage points in 1996, but just 0.65 points last year.
Mr. Moulton’s estimates of bias are in line with what other economists have found, but unlike many of those others, he concludes the bias hasn’t grown; it has shrunk. He bases this in great part on the extensive changes statistical agencies have made to price data since 1996, which were retroactively incorporated into the productivity statistics. One of the most important substituted in a new formula for averaging price changes for an old formula with an inherent upward bias.
The stealth productivity story rests on the fact that price data doesn’t always capture changes in quality when a new item replaces
an older item, especially for computers whose prices dropped as their microprocessors, storage, screens and software got better. In 1998 federal agencies adopted “hedonic” adjustment to better capture improved quality in computers, then extended it to televisions in 1999, audio and video equipment in 2000, and hospitals and nursing homes in 2009.
Except for computers, the impact was pretty tiny, which Mr. Moulton sees as evidence that not a lot of quality is going unmeasured. In fact, quality is sometimes overstated: This happens when a company raises the price for a new item that isn’t much better than the old, yet the entire price increase is attributed to higher quality. This turned out to
be the case with audio equipment.
Mr. Moulton concedes the data still doesn’t fully capture the benefits of an entirely new product, i.e., one that isn’t simply replacing an older model. The first smartphones were better and cheaper than a combination of a cellphone, music player and camera, but the price data don’t capture that. But this problem isn’t new: the same was true of cars, home appliances and antibiotics.
“When it was discovered aspirin, an extremely inexpensive drug, worked well for heart attacks and strokes, that led to an improvement in health and quality of life, but not a change in consumer spending patterns,” Mr. Moulton says. In any case, the Bureau of Labor Statistics is updating its smartphone sample every six months to capture the evolving quality.
The data do count Google search and Facebook: They are treated as an input to companies that pay for the ads. True, they also have large benefits to consumers who get them for free, but, Mr. Moulton points out, so did television. Also, as their benefits have risen, some benefits of other “free” ad-supported media like newspapers and magazines have shrunk. Economic data simply can’t capture all the benefits of innovation, but those benefits aren’t necessarily larger now than when Americans got indoor plumbing and electricity.
Mr. Moulton estimates the shift from shopping in brickand-mortar stores to online has driven prices down—the socalled Amazon effect—by more than the official data capture. But he estimates this effect is smaller than when consumers shifted to big box and warehouse stores decades ago, which similarly wasn’t captured in the data.
Elsewhere, Mr. Moulton finds other sources of under-measured productivity growth, such as in the quality of business investment in information technology and software, and, going in the other direction, the fact that outsourcing has overstated domestic sources of productivity improvement.
This won’t be the last word on the subject; there are other sources of mismeasurement Mr. Moulton doesn’t tackle and new products on the way that will pose tricky problems for statisticians.
But for now, it is safe to say that if there is going to be a technology-driven economic boom, it is going to need more than just a different set of measuring sticks.