New Estimates of the U.S. Economy's Potential Growth Rate
Kahn, George A., Contemporary Economic Policy
According to standard textbook aggregate supply and demand models, keeping inflation stable at full employment requires real GDP growth to match closely the long-run GDP growth potential. If the GDP grows faster than potential, capacity constraints will be overshot, and inflation will tend to rise. If the GDP grows slower than potential, an output gap will open up, and inflation will tend to fall. From an initial position of full employment, monetary policy might need tightening if growth were projected to exceed potential or easing if growth were projected to fall below potential. However, the GDP's long-run growth potential is unobservable. Therefore, policymakers need a reliable estimate of the potential growth rate in order to use it in practice.
Most analysts placed the GDP's potential growth rate at about 2.5% in the 1980s, but recent developments call into question this estimate for the 1990s. On the one hand, business downsizing and expanded investment in plant and equipment - especially computers - suggest an increase in long-run labor productivity growth (LPG) in the 1990s. To the extent that long-run LPG has increased, the GDP's potential growth rate currently may be above 2.5%. (Throughout this paper, the term productivity refers to average labor productivity as opposed to multifactor productivity.)
On the other hand, the dramatic computer price decline in the late 1980s and early 1990s has led to an increasingly large, upward, substitution bias in fixed-weight measures of real GDP and productivity growth. The Bureau of Economic Analysis (BEA) has addressed this problem by adopting more accurate chain-weighted indexes. The newly featured measures of real GDP and labor productivity show less growth in the 1990s than do the old measures that apply fixed weights based on 1987. To the extent that recent output and productivity growth - properly measured - are less than previously thought, the GDP's potential growth rate may have fallen below 2.5%. Analysts may not yet fully appreciate this measurement issue. For example, slightly over half of the respondents to a recent National Association of Business Economists survey (NABE, 1995) said the shift to chain-weighted GDP did not affect their potential growth estimates while only 32% thought their estimates would be lower.
Using an Okun's law framework, this paper estimates potential growth for the 1990s as measured by both fixed- and chain-weighted GDP. It then decomposes estimated potential growth rates into LPG and labor input growth (LIG) using a regression analysis to separate secular from cyclical changes. Finally, it compares estimates of potential GDP and trend productivity growth for the 1990s with estimates from earlier periods using both fixed and chain weights.
Eliminating the substitution bias associated with fixed-weight real GDP measures raises estimated potential GDP growth in the 1980s but lowers it in the 1990s. As a result, estimates indicate that the potential growth has slipped from roughly 2.5% per year in the 1980s to roughly 2.0% in the 1990s. Decomposing potential growth into LPG and LIG shows that this slowdown has occurred despite a modest increase in estimated trend productivity growth. Chain-weighted data indicate that trend productivity growth increased from 0.9% per year in the 1980s to 1.2% in the 1990s - perhaps boosted modestly (but statistically insignificantly) by business downsizing and investment in new plant and equipment. Finally, a secular decline in the growth rate of aggregate hours worked has prevented the LPG increase from translating into an increase in potential GDP growth.
II. OUTPUT, EMPLOYMENT, AND PRODUCTIVITY IN HISTORICAL PERSPECTIVE
The rapid LPG early in the current economic expansion led some observers to conclude that long-run LPG had increased. This conclusion, in turn, contributed to the view that the GDP potential growth rate also had increased. …