Printer Friendly

The long-run economic outlook.

I BELIEVE it is my job to give an upbeat analysis of the longer-term prospects for the economy. I'll do my best, although it would be difficult to beat Herbert Hoover, who in June 1930 said, "Gentlemen, you have come sixty days too late. The depression is over."

Our current projection is for relatively slow growth, averaging only 2.6 percent a year between 1992 and 1997. This reflects both the expansion phase of the business cycle, which we assume will take place sometime between now and 1997, and a projection of the potential growth of the economy, which we place at around 2.1 percent. That rate of potential growth is projected to continue through 2002.
Table 1
Sources of Potential Growth (in percent)
 1960-70 1970-79 1981-89 1992-97
Gross Domestic Product 3.6 3.2 2.3 2.1
Business Sector
 Output 3.6 3.5 2.5 2.4
 Labor 1.2 2.2 1.5 1.4
 Capital 4.5 4.4 3.5 3.7
 Total Factor Productivity 1.3 0.5 0.3 0.3
Contributions to Growth
 Business Sector 2.7 2.6 1.9 1.8
 Other Sectors 0.9 0.6 0.4 0.3
Source: Congressional Budget Office.

This projection is really quite slow compared with past performance: between 1960 and 1989, for example, we estimate that potential output was growing at a 3 percent rate. A 3 percent growth rate will double real output GDP after about twenty-three years, while a 2.1 percent growth rate takes thirty-three years.

What I would like to do today is to look at the main factors -- labor force growth, productivity growth, and the growth of capital -- that underlie the projection of real growth. Then, remembering President Hoover, I want to spend a few minutes talking about how much confidence we can have in these projections, I think that's a very important issue that often gets lost when we present either short-run or long-run projections to our clients or our boards.

Having a careful economic projection is certainly important for planning, but it doesn't mean that we can pretend to know what the economic environment will be five or ten years down the road.


Before getting into the three factors underlying growth, I want to say a few words about how the Congressional Budget Office's projections have evolved over time. We have over the years used several different methods, reflecting in part how ambitious we wanted to be. First, we focused very much on various statistical methods of trend extraction. I don't want to dismiss that approach: it's not at all clear that we can make a much more accurate projection than such methods would produce. But those trend-extraction methods don't meet our needs in other ways. In particular, they don't allow us to show how the economic outlook would be affected by different fiscal policies, and particularly by different federal deficits. At the time of the 1990 budget agreement, for example, we had a major change in fiscal policy, i.e., the actions taken would greatly reduce the federal deficit below where it was projected to go without the budget agreement. The conferees went to a huge amount of trouble to cut the deficit, in part because economists told them and the country that there would be a serious long-term economic cost if they didn't take action. So when the agreement was concluded, we had to have a method of incorporating the policy actions into our projections.

The method we chose was to use a version of a Solow growth model, whose projections would be sensitive to the growth of capital, of labor, and of all the other factors -- technical change, changes in regulation, etc. -- that are summed up in total factor productivity. The model we use is a modification of one developed at Brookings, and it separates out housing and government production from the core growth model. This growth model lets us reflect different projections of federal borrowing, as well as assumptions about private saving and changes in patterns of investment.

Some of the inputs to this model come from a parallel set of more fully specified macroeconomic projections, but information moves in both directions between these models. The growth model gets from the parallel model its assumptions for borrowing from abroad, for the relative prices of capital goods, and for personal and corporate saving. In turn, the growth model passes to our broader projections its numbers for capital accumulation and potential GDP. I don't want to spend too much time on those issues; although they are important, we can still analyze the outlook perfectly well in terms of the three main factors in the growth model: labor force growth, growth of total factor productivity, and projections of public and private saving. So now let me turn to those three factors, beginning with labor.


The labor force projections are largely driven by two factors: the end of the period of fast labor force growth that occurred while the baby boomers entered the labor market, and the slowing of the upward trend in labor force participation by women. Both these projections are perfectly conventional: like most of you, we use the Bureau of Labor Statistics' midpoint estimates of labor force growth. The labor force grew at an average rate of 2.0 percent between 1960 and 1989, after adjusting for cyclical effects, and in our projection the growth slows to only 1.3 percent, cyclically adjusted. The growth in the native labor force in the next ten years is so slow that immigration accounts for nearly half of the projected labor force increase.

I should say a bit about the past few years, which have really complicated labor force projections. As you know, labor force participation dropped in 1990 by much more than we could account for by the recession, and has recently picked up again. Should we treat the swing in participation as an unusual effect of the recession, or as the first evidence that labor force growth would be even slower than the demographics would indicate? We chose to assume that the drop in participation was related to the recession and would be reversed as the recovery progressed. That meant we could assume that potential GDP would not be much affected; it also means, of course, that the unemployment rate is likely to stay high for even longer than usual.


Now let me turn to the projections of total factor productivity, i.e., the residual of growth left unaccounted for by growth in labor or capital inputs. Out of the 2.1 percent potential growth in overall GDP that we project, about 0.3 percent is total factor productivity. It would be nice if this residual were either unimportant or at least constant; unfortunately, it is neither. The stylized facts are that the residual grew fast in the 1950s and 1960s and slowed dramatically in the early 1970s. With the revision of the accounts late last year, there has been no sign of an increase in the growth of the residual in the 1980s.(1) There are lots of possible explanations for these shifts in growth rates, but our assessment of the literature is that they still leave a lot unexplained. Oil price changes probably played a role in the 1970s and early 1980s, as did the growth of environmental regulation and changes in education and the accumulation of human capital. But these factors, as well as we can measure them, don't by any means account for the swings in productivity growth that have occurred.

So what do we do about the projection? We have to try to extract the trend statistically. In practice, we adjust the residual for the business cycle and then assume that its growth will be the same in the 1990s as it was from 1981 to 1991, or 0.3 percent. This could be either too high or too low: too high, because it doesn't account for the tightening of environmental regulations following the 1990 Clean Air Act, and because real oil prices fell in the 1980s, something we cannot expect to be repeated; or too low because of interactions with the labor force projections. There is some weak statistical evidence that slow labor force growth goes along with higher total factor productivity growth, and, as the rate of new entrance into the labor force slows, the average experience level of the labor force should rise, helping to raise the growth of productivity. Other factors are presumably at work as well. The most prudent assumption seems to be that total factor productivity in the 1990s will be similar to that of the 1980s. Obviously, the projection for total factor productivity is not very well tied down. I'll come back in a few minutes to what that means for the confidence we should attach to the projections.


The last major piece is the projection of saving and capital formation. Saving, like productivity, has undergone some major, partly unexplained swings in the past, which make it difficult to project. The stylized facts are these: gross private saving wobbled around 17 percent of GDP in the 1960s, rose to about 20 percent in the 1970s, and then dropped sharply during the 1980s to around 16 percent of GDP. I am including state and local government saving in those figures: it's largely saving in pension plans. Federal government borrowing used a fairly small amount of that saving up until about 1980, except in a few unusual years. Since 1983, however, government borrowing has never fallen below 2 percent of GDP (using NIPA concepts so as to avoid the distortions coming from the savings and loan cleanup). As a result, gross national saving, which was about 17 percent of GDP in the 1960s and 1970s, has recently been only 12 or 13 percent. And because we now devote a larger proportion of business investment to short-lived assets, depreciation has risen, and net saving has dropped from about 7 or 8 percent of GDP in the 1960s and 1970s to around 2 percent recently.

One reason for the decline in private saving is the poor income growth that resulted from sluggishness of total factor productivity growth in the 1980s, together with the winding down of the baby boom's entrance into the labor force. With slower income growth we should expect to see a slower rate of accumulation of wealth. This doesn't bode too well for the next ten years, when we are projecting about the same productivity growth and a slight slowing of labor force growth from the 1980s.

Some other factors, however, might help to raise saving a little in the next few years. First, the 1980s saw lots of capital gains from the stock market, which probably held down saving by boosting asset growth. Because that boom was a correction of many years of apparent market undervaluation, there is no reason to expect it to be repeated in the 1990s, so that household saving could rise a little. Second, changes in the age composition of the population could raise saving slightly, though the evidence is not at all clear. In any case, our projection does assume a slight increase in the gross saving rate in the 1990s.

The other part of the saving projection is the projection for federal borrowing. Our most recent projections are that federal borrowing -- excluding the costs of fixing the savings and loan mess -- will fall very slightly relative to GDP through about 1995, as long as the restraints of the 1990 Budget Enforcement Act are in place. From then on, under current policy, rising health care costs will mean the deficit will once again begin to increase relative to GDP, reaching 5.3 percent by 2002, excluding deposit insurance. To put that in perspective, the deficit on the same basis averaged 3.8 percent of GDP in the late 1980s.

The effect of this lower saving is of course to lower the potential growth of the economy and eventually to lower living standards below where they might otherwise be. The model we use for growth projections says that for every permanent 1 percentage point reduction in gross saving as a percent of GDP, if it persists, consumption possibilities fall by about 1 percent after thirty or forty years -- and forever thereafter. In other words, we can consume a little more now by reducing our saving rate, but doing so will permanently cut into future consumption. In our projection, the net national saving rate is around 3 percent during the 1990s, well below the 7 or 8 percent average of the 1960s and 1970s. If it persists, that will correspond to about a 4 percent permanent reduction in consumption in the future.


Now let me turn to the question of how confident we can be in any long-term projections. Is there a reasonable way that we can tell our clients and corporate boards how much uncertainty they should plan for? And how do we tell the difference between a reasonable and an unreasonable projection?

It would be nice if we could look to track records to assess long-term projections skill, but few people -- and certainly not CBO -- have a long enough track record to make any such analysis convincing. Alternatively, we could look at other people's projections to see if they perhaps have thought of something we haven't included in the projection. I've done that in the next figure, where I've marked CBO's and OMB's projections for real GDP in 1997 and added a few unlabelled projections from other forecasters. You can see that they cluster reasonably well around levels of real GDP of around $5.6 trillion to $5.76 trillion in 1987 prices, which according to our calculations would imply growth rates of potential GDP of between 2.0 percent and 2.6 percent.

This picture might induce some complacency on our part, at least with respect to the projections to 1997, but unfortunately I think it is seriously misleading. The private forecasters, together with CBO, all use basically the same method of projections, i.e., we use models that either are growth models or have growth models embedded in them. So it is not surprising that we all come to basically the same set of conclusions about the implications of slowing labor force growth and low national saving.

But the models just don't support that degree of confidence. They are capable of capturing only certain features of the historical record. As we have seen, for instance, the projections for total factor productivity are quite hard to pin down.

We can get a better estimate of the uncertainty of these projections with a statistical procedure that uses the record of how well this model fits past data. Technically, we used a bootstrap method. The solid lines in the figure show the results of that procedure, calculated for 1997 and 2002. The way to interpret the lines is that they enclose one standard error of the projection, which means that we estimate that real GDP has about a 15 percent chance of falling below the lower line, and a corresponding 15 percent chance of rising above the upper line, in 1997 and 2002. Alternatively, we think that the projection for real GDP in 1997 is $5.62 trillion plus or minus 3 percent, with about a 30 percent chance of being wrong -- either too high or too low.

We've also calculated the uncertainty of the estimate of the federal deficit that derives from this uncertainty of the GDP estimate. Accordingly, our projection of the deficit under current policy for 1997 is about $290 billion, plus or minus $90 billion because of the economic uncertainty. Again, the chance that we are wrong -- that the deficit will fall outside those limits, either high or low -- is about 30 percent.


We project a relatively slow growth of real GDP of about 2.6 percent over the next five years, based on a potential growth rate of about 2.1 percent. This projection is the result of assuming that total factor productivity growth will follow the trend of the 1980s, that labor force growth will remain slow, and that net national saving will remain depressed.

The uncertainty we attach to this projection is unsettling, and it certainly surprised me when we began to do these calculations. They say that there is about a 15 percent chance that actual GDP growth -- including the expansion phase of the business cycle -- will be lower than 2.0 percent between 1992 and 1997, and a corresponding 15 percent chance that it will exceed 3.2 percent. Perhaps that range seems too wide, but I am reminded of Mayor Jean Drapeau of Montreal, who said, "The 1976 Olympics could no more lose money than I could have a baby."

Robert A. Dennis is Assistant Director, Macroeconomic Analysis Division, Congressional Budget Office, Washington, DC. This paper was presented at the 34th Annual Meeting of the National Association of Business Economists, Dallas, TX, September 13-16, 1992.


1 Measurement issues are important here, but alternative measurements probably would not overturn the main conclusion, that productivity growth in the 1980s has not improved much from that of the 1970s. The main measurement problem is that measures based on 1987 prices are reasonable for years in the late 1980s, but probably distort earlier years' growth rates. Improved measures of total factor productivity that avoid this distortion are not yet available.
COPYRIGHT 1993 The National Association for Business Economists
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1993 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:growth projections for the 1990s
Author:Dennis, Robert A.
Publication:Business Economics
Date:Jan 1, 1993
Previous Article:The excesses of the eighties.
Next Article:International perspective: structural changes and the integration of Europe.

Terms of use | Copyright © 2017 Farlex, Inc. | Feedback | For webmasters