The strong performance of productivity growth in the second half of the 1990s was in fact attributable to accelerating technical change, not to poor measurement or to temporary factors.
Productivity is one of the most closely watched indicators of long-term economic prospects. Rising productivity is the key to making possible permanent increases in the standard of living. In Productivity Growth in the 1990s: Technology, Utilization, or Adjustment (NBER Working Paper No. 8359), authors Susanto Basu, John Fernald, and Matthew Shapiro present new estimates of the role of technological change in creating the unusual increases in measured productivity during the second half of the 1990s.
Changes in technology are the only source of permanent increases in productivity, but a number of transient factors can affect both true and "measured" productivity. For example, workers may work harder during periods of high demand and firms may use their capital assets more intensively by running factories for extra shifts; both factors can lead measured productivity to be too high relative to actual technological progress. Similarly, during periods of high demand, productivity can rise because firms take advantage of increasing returns to scale; the authors argue that this effect is not permanent and should be discounted when measuring long-run technical change. The strength of the latest economic expansion in the second half of 1990s has led many commentators to argue that the rapid increases in measured productivity during that period were attributable to bad measurement or to temporary factors of this type.
The expansion that began in the 1990s also is distinguished by a large and long-lasting increase in business investment. Although labor force employment, labor force participation, and rates of unemployment have been comparable to what occurred in earlier expansions, the share of investment in information technology rose from a baseline of roughly 3 percent of GDP in the late 1980s to almost 6 percent of GDP by 1999. The authors suggest that this unusually rapid rate of investment actually may lead measured productivity growth to understate the underlying rate of technical change -- because rapid capital investment disrupts firms' ability to produce output, for example because their workers often are diverted from their normal tasks to install new equipment and learn to use it effectively. These "adjustment costs" lower output growth, and thus lower measured productivity growth as well.
Controlling for this range of confounding effects, the authors find that the strong performance of productivity growth in the second half of the 1990s was in fact attributable to accelerating technical change, not to poor measurement or to temporary factors. They find that in the first half of the 1990s, true technology grew at an annual rate of 1.2 percent, but this rate rose to 3.1 percent for the 1995-9 period. In fact, the rate of technical change over 1995-9 exceeded even the measured growth rate of 2.5 percent, because of the temporary damping effect of higher investment on productivity growth noted above.
In the aggregate, the authors conclude, there is "evidence of a substantial increase in the pace of technological change in the latter half of the 1990s." More intensive use of capital and labor accounted for some of the increase in measured productivity in the first half of the 1990s, but that utilization was flat or declining over the second half. Adjustment costs masked a substantial fraction of the increases in true technology that occurred in the second half of the 1990s.
These results also suggest that productivity increases were distributed widely, if unevenly. Durable manufacturing experienced the fastest rate of technology growth and its largest acceleration, with increases of over 6 percent per year during the second half of the 1990s. Technological growth in the private non-manufacturing sector--which includes the large and important service sector--increased from 0.9 percent to 2.7 percent over the same period. In non-durable manufacturing, however, technology growth was "very slow," although the authors suggest that this result may arise from data problems at the end of the sample.
-- Linda Gorman