Productivity Data: It's Official We're Missing By A Mile
The Bureau of Labor Statistics just made an important confession. Based on results from studies released on Mar. 30, the government has all but admitted for the first time something that many economists have argued for years: U.S. productivity growth is understated, especially in hard-to-measure service industries. The results offer the latest ammunition for those who believe that the U.S. economy is in the middle of a technology-driven productivity revolution that could have significant long-term impacts on growth, inflation, and economic policy.
The heart of the research is a study by the bureau's William Gullickson and Michael J. Harper, in which they break down productivity growth by various industries using a concept called multifactor productivity. This measure is a more comprehensive gauge of productivity than what has been traditionally used because it measures the productivity of both labor and capital. It also tries to capture the impact of technological change--so central to the New Economy debate.
The results show what even the government admits is implausible, if not impossible: Based on existing data, the contribution of nonmanufacturing industries--mainly services--to productivity growth was zero from 1979 to 1996. The BLS identified five service industries--banking, insurance, construction, utilities, and health care--that showed negative growth. Given the massive investment by many of these industries in efficiency-improving technology, especially by financial-services companies, that result strains credibility.
So how off are the results? Associate Commissioner Edwin R. Dean will only say, "We have not shown that the degree of underestimate is large." But if you arbitrarily reset the growth rate in these industries--just to zero--overall multifactor productivity growth would be raised by 0.4 percentage points. If you assign them a more likely 1%, the understatement for the economy would be a significant 0.8 to 0.9 points.
OPEN QUESTIONS. Those recalculations would more than double the overall growth rate of multifactor productivity since 1973. More important, they imply that the 1% annual growth figure that the BLS has used for labor productivity since then also severely misrepresents the efficiency of American workers.
To be sure, the BLS research leaves many questions unanswered. First of all, as Dean points out, productivity in some sectors may well be overstated. Manufacturing, for instance, is increasingly using temporary labor that is counted in the service sector, which could understate the number of hours worked in manufacturing, thus biasing productivity growth upward. Also, the studies do not settle the current raging debate: Does the U.S. economy's ability to grow at a rapid 4% clip over the past two years without generating inflation reflect a permanent upshift in productivity growth? Or is the good showing merely a temporary result of weak global conditions outside the U.S. and other transient factors?
The best thing to come out of this research is likely to be better productivity data, since it identifies the key industries where the BLS will now redouble its efforts. In particular, Dean says that work on more accurate banking-sector data already underway will result in better measurement for that industry by yearend. However, he admits that progress will be slow, not because of a lack of productivity at BLS, but a lack of money.