Why was U.S. economic growth in the 1980s not stronger? Appraisal of an optimistic projection for 1981 to 1990.

My 1983 presidential address to the Southern Economic Association was published in the April 1984 issue of the Journal under the title " Long-Term Economic Projection: Stronger U.S. Growth Ahead". After a general discussion of methodology, I presented my projection of real gross product, factor inputs, total factor productivity (TFP), unit costs and the implicit price deflator index for the U.S. business economy for the periods 1981-90 and 1990-2000. Publication by the Bureau of Labor Statistics (1991) of the relevant time series through 1990 makes possible an evaluation of the accuracy of the 1981-90 projection and an analysis of why it turned out to be on the high side.

Analysis of why forecasts went wrong may help to improve future efforts. Of more general interest, I believe that my analysis helps explain why U.S. economic growth in the 1980s was not stronger.

The acceleration in growth of real gross product from a 2.0 percent annual average rate between 1973 and 1981 to 3.2 percent 1981-90 did justify the subtitle of the article "Stronger U.S. Economic Growth Ahead." But actual growth was significantly below the projected rate of 4.0 percent. Actually, the range within which I expected the growth rate to fall was set from 2.5 to 4.5 percent [6,950]. Obviously, I would have done better to take the mid-point of the range, particularly since it was closer to the consensus of long-range projections available at that time. It is usually safer to occupy a position close to the consensus of forecasts rather than one further out on either side.

I. Comparison of Actual with Projected Rates of Change

As shown in Table 1, my projection of the growth rate of total factor input was on target. The over-shooting of the output growth rate was entirely due to overestimating the strength of the come-back in TFP at 1.8 instead of 1.0 percent average annual growth. Further, whereas my projection of the deceleration in unit labor costs from 8.3 to 3.2 percent a year was on the mark, I underestimated the deceleration in the implicit price deflator by one percentage point.

[TABULAR DATA OMITTED]

Even the precision of the input and unit cost projections must be credited to that guardian angel of forecasters - offsetting errors in components. In the case of inputs, I underestimated the growth of labor and overestimated the growth of capital. In the case of unit labor costs, an overestimate of the rate of increase in average hourly labor compensation was exactly offset by the overestimate of labor productivity shown in the table. The explanations of these errors are an important part of the explanation of why the projected growth of productivity and real gross product was too high.

II. Reason for Slower than Expected Growth in the Capital/Labor Ratio

It is clear that a major reason for the slower growth in the capital/labor ratio than projected was the Tax Reform Act (TRA) of 1986. For the projection, I had assumed ". . . that the provisions of the Tax Acts of 1981 and 1982 remain in effect, particularly the accelerated cost recovery system and . . . the 25 percent incremental R&D tax credit . . . If further tax changes are enacted this decade, I assume they will not be permitted to detract from the investment incentives now in place which are expected to result in a significant increase in the rate of growth of the real capital stock .. . as well as in the rate of technological innovation"[6, 951].

The political consensus that had led to pro-investment policies fell apart in the second Reagan administration. The 1986 TRA eliminated the ACRS and the investment tax credit, and capital gains were to be treated as income, raising the tax rate on gains. Despite the cuts in corporate as well as personal income tax rates that base-broadening made possible, TRA increased the mean marginal tax rate on capital from 34.5 percent to 38.4 percent[3, 91]. The marginal rate on labor income was estimated to have fallen from 41.6 percent to 38.9 percent. According to the Report: "These changes will increase labor effort and depress saving . . . In any case, the ratio of capital to labor is decreased"[3, 91].

The positive effect on labor supply has been documented by Bosworth and Burtless[2]. Poor households were unaffected; "But the labor supply gains in the rest of the population appear connected to tax rate changes. The largest increases occurred among earners in the most affluent families, who enjoyed the largest tax cuts, and especially among married women in those families, who were predicted to be the most responsive to tax cuts. The success of tax reform in raising labor supply was at least partly offset by its failure to raise or even maintain capital investment" [2, 36].

The negative effect of the TRA of 1986 on capital formation came on top of macro-economic policies that had perverse effects on saving and investment; namely, high budget deficits and relatively tight monetary policy. Although interest rates declined during the decade as inflation was subdued, the real cost of capital was very high. The real yield on Aaa corporate bonds averaged over 6 percent contrasted with around one percent in the 1970s. Yet restrictive monetary policy did not permit average prices to rise more than unit costs during the decade, as shown in the table. Profits of nonfinancial corporations averaged 8.9 percent of gross product in the 1980s, well below the 15.4 percent of the 1960s and 10.7 percent of the '70s. This had a dampening effect on investment demand, as well as on retained earnings as a source of funds. Capital formation would have been even lower had it not been for the net inflow of capital from abroad, attracted by the relatively high U.S. interest rates.

III. Reasons for Slower Productivity Growth than Projected

The slower than expected increase in the capital/labor ratio directly contributed 0.5 of the 1.4 percentage point shortfall in labor productivity (as indicated by the lower rate of substitution of capital for labor). But it also contributed indirectly to the 0.9 point discrepancy between actual and projected TFP, since capital goods are carriers of technological progress.

Also important is the fact that R&D outlays fell short of my projection that they would rise from 2.35 percent of GNP in 1981 to 3.0 percent in 1990, which was only slightly above the previous peak in 1964. In fact, the ratio crested at 2.83 percent in 1985 and '86, and then fell back to 2.69 percent in 1990 according to preliminary estimates by the National Science Foundation[7]. Not only was publicly funded R&D cut back, but private R&D was adversely affected by the 1986 TRA reduction in the incremental R&D tax credit and by falling corporate profits after 1988.

Based on our growth accounting methodology explained in the article[6], the R&D shortfall reduced the contribution of technological advance to growth of real product and productivity by several tenths of a percent. Slower economic growth itself reduced the contribution of scale economies by about 0.1 percentage point, using Denison's formula[4].

IV. Monetary Policy Effects

Acceleration of economic recovery to almost 7 percent in 1984 led the FRB to tighten monetary policy. This brought the average annual expansion rate of real GNP down to less than 3.5 percent over the next three years. Unemployment remained above six percent of the labor force through 1987, and inflation ratcheted down further to a 3 percent annual rate.

Although I had not anticipated as restrictive a stance by the Federal Reserve Board so early in the recovery when writing the SEJ article, I soon recognized that its actions would defer the eventual recession[5]. But when expansion accelerated in 1988, unemployment fell below 5.5 percent, inflation began to rise, and FRB tightened again. This led to a sharp deceleration in growth throughout 1989 and the first half of 1990, culminating in a recession beginning in July. The downturn was precipitated by the Gulf War but the stage had been set by FRB policy.

The slowdown and recession brought modest absolute declines in productivity as capacity utilization and employment rates fell, particularly in 1990. If growth rates are calculated for the period 1981-88, prior to the productivity declines, they are 0.6 and 0.5 percentage points higher for real gross product and TFP, respectively, and only 0.2 and 0.4 percentage points below the projected rates.

As a general proposition, it is desirable for long-term economic projections to be made explicitly to approximate the expected average rate of growth in productive capacity. Cyclical developments can be factored in to the extent they affect secular growth rates, but not the position of the economy in the tenninal year of the projection period.

V. Conclusions

What can be learned from this post-mortem? One clear lesson is that economic trends are importantly influenced by policy variables that are difficult, if not impossible, to anticipate a decade or more into the future. Given the bi-partisan support of the tax acts of 1981 and '82 designed to promote investment and growth, I did not anticipate the shift in policy represented by the 1986 TRA which slowed the growth of tangible investment and R&D outlays. The outcome of the November 1982 elections that confronted a Republican president with Democratic control of both Houses of Congress should probably have been a tip-off.

At the time the projection was made in the spring of 1983 the country was pulling out of recession and it was expected that the federal deficit would be brought down significantly over the following years. Certainly no one expected it would average almost $200 billion in fiscal years 1983 through 1990, keeping the cost of capital high and putting the major burden for fighting inflation on monetary policy. In that regard, the continuing restraint by the FRB was not generally anticipated, although Mr. Volcker's toughness in disinflating the economy during the years 1980-82 did provide an important clue.

Perhaps economic forecasters should consult political forecasters, depending on the batting average of the latter. More to the point, there should be a systematic evaluation of all the major long-range economic projections made over a period of years, such as Victor Zarnowitz has done for short-term forecasts. We would learn of more pitfalls and types of errors to avoid than those pointed up by this review. But my judgment is that unforeseeable changes in policies and exogenous future shocks will always introduce errors into the best of forecasts. Yet I also believe that a review of long-range projections would reveal that they do a better job, on average, than naive extrapolation of recent trends. At least that much can be said in favor of my 1981-90 projection. A decade from now we can see if my projection for the 1990s does as well, and if its inevitable shortcomings can be as readily explained.

References

[1.] Bureau of Labor Statistics, U.S. Department of Labor, "Multifactor Productivity Measures, 1990," USDL 91-412, August 29, 1991. [2.] Bosworth, Barry and Gary Burtless. "Effects of Tax Reform on Labor Supply, Investment, and Saving." Unpublished typescript of paper presented at 1990 meetings of the American Economic Association (available on request to the authors at The Brookings Institution, Washington, D.C.) [3.] Council of Economic Advisers. Economic Report of the President, transmitted to the Congress February 1987, together with The Annual Report of the Council of Economic Advisers. [4.] Denison, Edward F. Accounting for United States Economic Growth 1929-1969. Washington: The Brookings Institution, 1974. [5.] Kendrick, John W., "Cost Containment Prolongs the Expansion." The AEI Economist, May 1985. [6.] _____, "Long-Term Economic Projection: Stronger U.S. Growth Ahead." Southern Economic Journal, April 1984, 945-64. [7.] National Science Foundation. National Patterns of R&D Resources: 1990. Final report, NSF 90-316.