clock menu more-arrow no yes

Filed under:

Robert Samuelson: Think there's been a productivity boom? Think again

In this July 27, 2011 file photo, assembly line worker Edward Houie moves a door into position for a 2012 Chevrolet Volt at the General Motors Hamtramck Assembly plant in Hamtramck, Mich.
In this July 27, 2011 file photo, assembly line worker Edward Houie moves a door into position for a 2012 Chevrolet Volt at the General Motors Hamtramck Assembly plant in Hamtramck, Mich.
Paul Sancya, AP

WASHINGTON — Let's travel back in time to 1995. Most Americans still remembered the calamitous inflation of the late 1970s (prices rose 13 percent in 1979). Many federal benefits, including Social Security, were (and are) tied to inflation. But was the inflation overstated, as many economists thought? If so, the economy might be doing better than reported. To answer that question, the Senate appointed a commission, headed by Stanford economist Michael Boskin, to weigh the evidence.

In late 1996, the Boskin Commission, as it was known, released its findings: Inflation — defined as increases in the consumer price index, or CPI — was overstated by 1.1 percentage points. If, for example, the year-to-year increase in the CPI was 4 percent, the actual rate was closer to 3 percent.

Since then, have we made any progress? Well, yes and no. But first let's recall what the Boskin Commission found.

The largest problem, accounting for about half the overstatement, involved new products or improvements in existing ones. Suppose you bought a tire that got 20 percent more mileage than its predecessor, but there was no price increase. In reality, you got a 20 percent price cut. But the index often missed these price cuts, because adjusting for quality is hard.

A similar problem afflicted new products — personal computers, internet services, microwave ovens and the like. When they were introduced into the CPI, their existing gains weren't counted as price cuts. Only after the product was inserted into the CPI were future price cuts recognized.

There were other omissions. One was shifting buying habits — what economists call "substitution." If the price of beef went up, some consumers would switch to chicken at lower prices. These price cuts were undercounted. So were consumers' tendency to shop at stores (Walmart, Target) with lower prices. These price cuts were also undercounted. (Economists' label for this miscounting is "outlet bias," meaning that the statistics don't accurately reflect consumers' actual behavior.)

Just how much of the economy's changes in output represent price increases and how much represent real shifts in production is a crucial question. If price increases are overestimated, then the economy's output, productivity and living standards are underestimated.

Two decades later, there's been some good news.

In a just-released paper by the Brookings Institution, Brent Moulton — a former top official of the Commerce Department's Bureau of Economic Analysis — argues that dozens of small, technical changes over the years by the BEA and the Labor Department's Bureau of Labor Statistics have reduced inflationary bias. Moulton now estimates the bias at 0.85 percentage points, down from the Boskin Commission's 1.1 percentage points. The improvement is almost a quarter.

The CPI is "updated much more frequently, enabling new goods and services to enter … more rapidly," Moulton wrote. The producer price index, which measures wholesale prices, has increased its coverage of service providers — lawyers, architects, insurance agents.

To Greg Ip, The Wall Street Journal's economics columnist, the data debunk the notion of a "stealth productivity boom" — a vast amount of production, enabled by new technologies, that is missed by conventional statistics. "Even if all the benefits of social networks, online shopping and less invasive surgery were being (properly) measured, it probably wouldn't change the overall picture," he wrote. Because the gap between measured and unmeasured inflation has narrowed, so has the likelihood of a huge, invisible productivity dividend.

Still, the problem of counting accurately will be with us forever. The reason: There's a constant inflow of new products and services that aren't easy to measure — health care and education being obvious examples. There's another point: Groups whose government benefits are tied to inflation — notably Social Security recipients — don't necessarily want better statistics. So even when improvements are within our grasp, they are not adopted.