Expanding the Nation's Productive Capacity

Chapter Three, Economic Report of the President, 1995

David S. Dahl | Regional Economist

Published April 1, 1995  | April 1995 issue

As referenced in "Productivity primer: How does economy grow?"

HOW FAST CAN THE ECONOMY grow on a sustainable basis? Most mainstream analysts currently believe that aggregate output can grow about 2 1/2 percent per year. Recently, however, some analysts— perhaps inspired by the outstanding performance of the economy in 1994—have asserted that much more rapid growth, possibly as fast as 5 percent per year, may be sustainable.

The answer to this question has profound implications for the future well-being of the American people. If the mainstream view is correct, aggregate output will double only every 28 years or so, and per capita output only about every 56 years (assuming population growth of 1 percent per year). But if the alternative view is correct, aggregate output could double every 14 years, and per capita output every 18 years.

The answer also has important implications for the conduct of government policy. Sensible Federal budget planning can proceed only in the context of a realistic assessment of the long-term outlook for the economy. If the outlook is robust, then a more expansionary fiscal policy may well be consistent with a responsible outcome on the deficit. If, on the other hand, the outlook is more subdued, a greater degree of fiscal restraint may be required.

Chart 3-1 illustrates one simple method for assessing the sustainable rate of growth of gross domestic product (GDP). (The estimates of GDP used in this chapter are based on so-called chain- type annual weighted data, which are discussed in Box 3-1.) The chart focuses on the growth of real GDP between the first quarter of 1988 and the fourth quarter of 1994. The reason for focusing on these two quarters is that the unemployment rate was very similar in both: 5.7 percent and 5.6 percent, respectively. This suggests that a similar fraction of the economy's overall productive capacity was being utilized in both quarters. Thus the average rate of growth of output in the interval between them should give a good indication of the average rate of growth of the economy's productive capacity during that period.

As the chart shows, real GDP increased at an average annual rate of 2.1 percent between the first quarter of 1988 and the fourth quarter of 1994. This suggests that the economy's productive capacity—potential GDP—also grew at about that rate. Over the same period, real GDP measured on the more conventional basis (1987 dollars) increased at an average annual rate of 2.3 percent. Therefore, this simple method suggests that the consensus view that the sustainable rate of growth is about 2 1/2 percent per year is slightly more optimistic than a purely mechanical reading of recent experience would warrant.

But does the simple graphical method, based only on historical experience, provide an accurate signal about the future growth of the economy's capacity? Historical experience does not yield certain knowledge of future trends. In particular, it does not take into account the influence of policies adopted by this Administration with the goal of enhancing the productive capacity of the economy. This chapter undertakes a systematic analysis of the factors contributing to the growth of the economy's potential, mainly for the purpose of assessing future growth prospects. The chapter begins by reviewing trends in the growth of GDP since the early l960s. Next it analyzes improvements in the productivity of American workers and increases in their hours of work—the two major sources of growth in the economy's productive capacity. This discussion also examines the shortcomings of existing measures of productivity growth and concludes that the economy's actual performance may be stronger than current estimates indicate. The chapter then turns to an examination of the appropriate role of government policy in enhancing the economy's sustainable long- run growth rate. The chapter concludes with a brief assessment of the outlook for trend productivity growth and for the growth of the economy's potential.

Box 3.1.
Chain-Weighted Measures of Output and Productivity Growth

Any index of aggregate output is constructed as the weighted sum of the output of the myriad types of goods and services produced in the economy. But what weights does one use? From an economic standpoint, it makes sense to use relative prices as weights. In the United States, government statisticians traditionally have used fixed weights, namely, the relative prices that prevailed in a particular recent year (currently 1987). The resulting index is appropriate for assessing economic performance in years when the relative price structure was similar to that in the base year.

Over time, however, relative prices can change greatly, making a fixed-weight index less useful for gauging long-term trends in output. Computers serve as a good example. The rapid increase in the quantity of computers produced over the past 30 years has been accompanied by a sharp decline in their relative price. Because the price of a computer in 1987 was far lower than it was in, say, 1963, the fixed- weight index understates the sector's share in total output in 1963, and hence understates total output growth between 1963 and 1987. After 1987, the effects are reversed: the price of computers has continued to decline, so use of 1987 weights for 1994 computer output causes an overstatement of the contribution of computers to 1994 output. Because the output of the computer sector has continued to grow faster than the economy as a whole, this overweighing causes the fixed-weight index to overstate the growth in output between 1987 and 1994.

Fortunately, the Department of Commerce, which prepares the traditional fixed-weight measures of GDP, also now publishes alternative GDP measures that eliminate this bias. One such alternative is the so-called chain-type annual weighted measure. The Department of Labor uses a similar chain-weighted measure (for the private nonfarm business sector) to construct the productivity measures cited in this chapter. According to the chain-type output measure, between 1963 and 1987 real GDP increased by an average of 3.3 percent per year, or 0.3 percentage point faster than the fixed-weight measure. Between 1987 and 1993, output as measured by the alternative index grew an average of 1.9 percent annually, or about 0.2 percentage point less than the official fixed-weight figures. Thus, correcting for fixed-weight bias makes the post-1987 performance of output (and therefore also of productivity) look somewhat less encouraging relative to its pre-1987 performance.

Factors Generating Growth of Potential GDP

Between 1963 and 1994 real U.S. GDP increased at an average annual rate of 3.1 percent per year. Because the economy appears to have been operating about at its potential in both those years, the average rate of growth of actual output between those dates should provide a relatively accurate estimate of the average rate of growth of potential output during the same period.

Growth of real GDP can be decomposed into two main components: growth of output per hour worked (or productivity) and growth of hours worked. As Chart 3-2 illustrates, these two components each contributed 1.7 percentage points to the growth of GDP between 1963 and 1994. (Strictly speaking, the data on productivity and hours worked pertain only to the private nonfarm business sector, whereas the data on output pertain to the total economy. As a result, and because the output of the private nonfarm business sector was increasing slightly more rapidly than the output of the total economy, the growth of output per hour and the growth of hours worked add up to slightly more than the growth of GDP).

Chart 3-2 also shows that the average experience since 1963 subsumes two very different episodes. Between 1963 and 1972 real GDP increased at an average annual rate of 4.2 percent. By contrast, since 1972 real GDP has increased only about 2.6 percent per year. (The economy appears to have been operating at about its potential in 1972; as a result, that year should also serve as a useful benchmark for purposes of estimating potential GDP growth rates.) The slower rate of growth of GDP since 1972 can be attributed to a slowdown in the rate of growth of productivity, since the growth of hours worked was about as rapid after 1972 as before.

Chart 3-3 examines the slowdown in the growth of productivity in more detail. The chart illustrates one of the most significant economic developments of the postwar period. Whereas productivity in the private nonfarm business sector increased at an average annual rate of 2.8 percent between 1963 and 1972, it increased only 1.7 percent per year between 1972 and 1978, and only 1.0 percent after 1978 (yet another year in which the economy was operating close to potential).

By contrast, productivity growth in the manufacturing sector seems to have slowed much less during the past four decades. As Chart 3-4 shows, output per hour in the manufacturing sector is estimated to have increased on average about 3.3 percent per year between 1963 and 1972, 2.6 percent between 1972 and 1978, and 2.6 percent again between 1978 and 1987. (The chain-weighted data used in Chart 3-4 were only available through 1991. Growth in manufacturing productivity between 1987 and 1991 was quite weak, but this is not surprising given that the economy was still in recession in early 1991. Assessment of the more recent trend in manufacturing productivity will have to await publication of data for subsequent years, when the economy was once again operating closer to potential.)

Taken together, Charts 3-3 and 3-4 suggest that the slowdown in the growth of productivity after 1972 was concentrated outside the manufacturing sector. It has been argued that these and similar data exaggerate that concentration, because they do not control for the fact that the manufacturing sector may have increasingly "outsourced" some low-productivity activities. For example, if factories contract with security firms to do work formerly done by their own security guards, that activity will be counted in the services rather than the manufacturing sector, and if security guards' productivity is less than that of the factories' assembly-line workers, official statistics may report an increase in overall manufacturing productivity that does not reflect an increase in the productivity of any individual worker. What this argument ignores; however, is that high-productivity jobs may also have been outsourced, in which case the direction of bias in the official estimates would be ambiguous. On balance, the evidence suggests that the apparent strength of productivity growth in manufacturing is not a figment of job migration.

Much of the discussion in this chapter focuses on the slow rate of growth of productivity in the United States since the early 1970s, relative to earlier US experience and the experience of other countries. But it is worth noting that US workers remain among the most productive in the world. This suggests that the productivity "problem" in the United States has much more to do with the rate of growth of productivity than with its level. Box 3-2 discusses one possible explanation for the coincidence of a high-level and slow growth of productivity in the United States compared with other countries.

Factors Generating Growth of Productivity

Productivity can be raised by improving the quality of the work force (adding human capital per worker in the form of education or training); by increasing the quantity of capital (investing in new private equipment and structures and in public infrastructure); and by improving the efficiency with which these factors of production are used. Improvements in efficiency can come from advances in technology (due to basic research or applied research and development, or R&D), but they can also come from other sources, such as process innovation, that are not conventionally thought of as technology. Chart 3-5 summarizes the behavior of the main factors contributing to the growth of productivity since 1963. (Box 3-3 discusses whether an increase in productivity comes at the expense of a reduction in jobs.)

The Quality of the Work Force

One important determinant of worker productivity is the workers themselves and the skills and abilities they bring to the workplace. Increases in the hourly output of the average worker can reflect an improvement in the characteristics that allow workers to accomplish the same tasks in less time, to adapt to changing situations with greater flexibility, and to become the engineers of change themselves.

Box 3-2.
Technological Catch-up and International Differences in Productivity Growth

How could it be that the United States, with one of the highest levels of productivity in the world, is not also among the countries where productivity is growing most rapidly? Some economists have suggested that, far from being a paradox, this circumstance is to be expected. The slow-growing leader, fast-growing follower pattern may simply reflect the dynamics of technological "catch-up."

Standard models of economic growth assume that richer and poorer countries have the same production technologies at their disposal (even if they choose to implement them with different mixes of capital and labor). Recently, however, growth economists have begun to question the realism of this assumption. In practice, technological diffusion—the spread of ideas—from leader to follower is far from automatic. Firms in follower countries may lack the skilled workers (engineers, managers) needed to exploit technologies used in leader countries efficiently. In addition, firms in leader countries may attempt to guard their core technologies to prevent or delay their spread to potential competitors abroad. Technological diffusion may be particularly slow in the case of "soft" technologies (process technologies and work organization), which cannot be imported and reverse-engineered as new products can.

For follower countries a gap in technology creates an opportunity. Leader countries (such as the United States) will find their productivity growth limited by the rate of creation of new knowledge. But followers can grow more quickly by closing a portion of the technology gap. It appears that success in closing this gap helped spur the postwar growth of Japan and the East Asian newly industrializing countries, which invested heavily in technology acquisition and human resources and created business environments conducive to technological growth. Not every country succeeds, however, in closing the technology gap. Indeed, some followers have fallen farther behind, and follower countries as a group have not become richer faster than leader countries. Nevertheless, the evidence suggests strongly that, for followers, the upper limit on growth in per capita income and productivity exceeds that for technological leaders.

Two rough indicators of work force quality are average educational attainment (average years of schooling per worker) and average experience. Since 1963 the average educational attainment of the work force has increased by about 2 years. The Bureau of Labor Statistics (BLS) of the Department of Labor estimates that investment in education boosted productivity about 0.3 percentage point per year, on average, between 1963 and 1992. In contrast, the average experience level declined slightly between 1963 and 1992, knocking about 0.1 percentage point off productivity growth each year. On net, therefore, measured changes in worker quality have added an estimated 0.2 percentage point per year to productivity growth since 1963. Interestingly, worker quality appears to bear none of the responsibility for the post-1972 slowdown in productivity growth. In fact, the estimated contribution of improvements in worker quality to productivity growth increased, from essentially nothing before 1972 to about 0.3 percentage point per year between 1972 and 1992 (Chart 3-5).

One caveat is in order here. Although the BLS education measure captures changes in the average number of years of schooling, it does not capture changes in its quality. Clearly, quality matters: a worker who spent 12 years marking time in poorly taught classes is likely to be less productive than one who spent the same number of years actively learning from skilled teachers. Unfortunately, the evidence on whether any such decline in the quality of schooling could help explain the productivity slowdown is too scanty to support any firm conclusions.

Training workers on the job is another way of increasing their human capital and contributing to aggregate productivity growth. Solid quantitative estimates have not been made of the contribution of training to aggregate productivity growth because there are no reliable data on the aggregate amount of training taking place. Nevertheless, available microeconomic evidence suggests that training matters. Studies of the wages of individual workers indicate that the payoff to formal training (including apprenticeships)can be quite substantial: a year of training typically provides returns of a similar magnitude to those offered by a year of formal schooling (an increase in wages of about 6 to 10 percent on average). Other research has found that companies offering more training enjoy higher rates of productivity growth. (Chapter 5 discusses the importance of worker training in greater detail.)

Box 3-3.
Productivity and the Growth of Jobs

A persistent concern, voiced by many workers and business owners as well as some economic analysts, is that rapid growth of productivity may cause job losses. This concern seemed validated early in the current expansion, when strong growth of productivity seemed to be standing in the way of a vigorous pickup in the pace of hiring. Does this concern have any analytical basis?

At the macroeconomic level, a pickup in the rate of productivity growth need not be associated with any reduction in the aggregate number of jobs available in the economy—at least not once fiscal and monetary policy have been adjusted to reflect the favorable change in productivity growth. An increase in productivity growth allows GDP to grow more rapidly without generating inflationary pressures. Over the long term, macroeconomic policies can bring the growth of aggregate demand in line with the improved rate of expansion of the economy's productive capacity, and thus sustain the growth of employment.

At the microeconomic level, productivity growth may change the composition of available jobs, and thus may be associated with significant dislocation as workers are forced into new jobs, possibly requiring different skills and perhaps even relocation. In this context, the role of government is to facilitate the transition of workers and capital to their most productive uses, while setting fiscal and monetary policies to keep the economy on a sustainable trajectory of high employment with low inflation.

The Size of the Private Capital Stock

Increasing capital intensity—roughly speaking, the amount of capital per worker—has been a key source of productivity improvement over the postwar period. When new investment has been undertaken to support an improved technology, the gains have sometimes been especially impressive. For example, output per hour in the telecommunications industry increased an average of 5.5 percent per year between 1969 and 1989, as the industry invested heavily in new satellite, cellular, and fiber optic technologies.

Productivity increases through capital investment have often involved exploiting economies of large-scale production. Industries such as food processing, beverages, and electricity generation are cases in point. In the beverage industry, for example, high-speed canning lines have raised productivity, but their contribution has been made possible in part by the development of large markets. To operate efficiently, these lines must produce nearly 500 million cans per year!

Data from the BLS indicate that increases in capital intensity—also known as capital deepening—added about 0.9 percentage point per year to the growth of US productivity between 1963 and 1992. As Chart 3-5 shows, a reduction in the pace of capital deepening explains only a small portion of the post-1972 slowdown in productivity growth.


Historically, investment in public capital such as roads, bridges, airports, and utilities has made a significant contribution to the Nation's productivity growth. Yet the net public capital stock in the United States has declined relative to GDP, from 50 percent of GDP in 1970 to only a bit more than 40 percent recently. The net public capital stock has also declined relative to the net private nonresidential capital stock. These declining trends in public capital suggest that infrastructure investment has been a net drag on the growth of productivity since 1970, but there is no consensus as to the quantitative importance of this effect.

Research and Development

Total Federal and private spending for research and development has averaged about 2 1/2 percent of GDP since 1960 (Chart 3-6). In dollar terms, American investment in R&D in 1992 was greater than the R&D investment of Japan, Germany, and France combined. Even relative to national income, the United States was roughly tied with Japan for first place among major industrialized countries.

As Chart 3-6 shows, a much larger share of total R&D spending in the United States is privately financed now than used to be the case. Relative to GDP, Federal spending for R&D was at a high level in the early 1960s, after the Sputnik launch provoked a wave of concern that the United States was lagging behind the Soviet Union technologically. But that ratio trended down during most of the 1960s and 1970s and has been more or less flat since the late 1970s. In contrast, industry- funded R&D investment has been noticeably greater relative to GDP during the 1980s and early 1990s than during the 1960s and 1970s. Indeed, since 1980 the private sector has sponsored more R&D than has the Federal Government.

According to BLS estimates, investment in R&D contributed about 0.2 percentage point to the growth of productivity between 1963 and 1992, with essentially no difference before and after 1972 (Chart 3-5). In all likelihood, however, R&D has played a more important role than these estimates would indicate, for a number of reasons. First, given the difficulties involved in measuring the return to investment in R&D, part of it probably shows up in the unexplained residual (see below). Second, because it is very difficult for anyone investing in R&D to capture all of the benefits of that investment, part of the return to American investment in R&D probably is captured by foreign producers. (Similarly, American producers probably capture some of the benefits of R&D investment undertaken by foreign firms.) Finally, some investment in R&D has had important benefits in addition to whatever improvement in the measured growth of productivity it may have yielded. For example, medical research (which claims 18 percent of total US R&D) has substantial payoffs, but it is highly unlikely that these payoffs are fully reflected in the statistics on output per hour.

The Residual

Over the postwar period, increases in human and physical capital and investment in R&D fail to account for all of the measured growth in productivity. The remainder generally is presumed to reflect unmeasured improvements in the quality of the capital stock and the work force, as well as more efficient utilization of capital and labor in the production process. Available data suggest that this unexplained residual contributed about 0.5 percentage point per year to the growth of productivity between 1963 and 1992.

The nature of this residual has puzzled economists for 40 years and has stimulated a vast literature seeking to explain it and to understand the dramatic difference in its behavior before and after 1972. Between 1963 and 1972 the residual contributed about 1.5 percentage points per year to the growth of productivity. Between 1972 and 1992, however, the residual made no contribution at all (Chart 3-5).

Two possible explanations as to the source of the residual follow from the previous discussion. The data from the BLS do not quantify the effect of either on-the-job training or investment in infrastructure, so any contributions of those two factors end up in the residual. In addition, industries evolve in ways that increase productivity, and the contributions of these evolutions are not captured in existing measures of capital, labor, or R&D investment. For example, the shift from small food stores to supermarkets gave a substantial boost to productivity in food retailing in the United States in the 1950s and 1960s. Similarly, many American companies have improved their business systems, and the contributions of these improvements are likewise not captured in the official statistics except, by default, in the residual. For example, the redesign of production processes within the manufacturing sector (such as lean manufacturing of automobiles) and the redesign of products to make them easier to assemble have been sources of productivity growth.

Some observers have argued that an increasing burden of government regulation may account for part of the reduction in the contribution of the residual during the 1970s. Since the late 1970s, however, a number of important industries—including trucking, airlines, and rail—have been deregulated. In addition, competition has been introduced into the market for long-distance telephone services. These factors suggest that any role of regulatory burden in the post-1972 productivity slowdown probably has not been large.

Another commonly mentioned explanation for the reduction in the contribution of the residual to productivity growth is the rise in energy prices during the 1970s. According to this logic, efforts to reduce energy consumption reduced measured productivity growth. This explanation is not very convincing, however, because energy costs do not bulk large in total costs, and because productivity growth has not revived despite the reversal of most of the 1970s runup in real oil prices.

Finally, it is possible that part of the slowdown in measured productivity growth is not real but reflects measurement error. This could be the case if, for example, measurement error has caused the official statistics to understate productivity growth by more since 1972 than before. Even if measurement error does not help explain why productivity growth has been slower since 1972 than before, it may help explain why it has been so slow in absolute terms. A later section of this chapter examines the extent to which the productivity problem might reflect faulty measurement.

Has the Trend in Productivity Growth Improved Recently?

Since 1987, according to current estimates, productivity growth in the private nonfarm business sector has averaged 1.2 percent per year, somewhat faster than the average during the previous decade. And since 1991, productivity growth has averaged about 2.0 percent per year—more than twice the 1978-87 average. Are recent claims of a pickup in trend productivity growth justified? (Provided there has been no offsetting reduction in the growth of hours, such a pickup would translate into an increase in the economy's potential growth rate.) This question is not easily resolved because the recent behavior of productivity has been heavily influenced (for the better) by the faster pace of economic activity during the last two years. A proper assessment of the trend in productivity growth can be made only by abstracting from cyclical influences.

Chart 3-7 focuses on the behavior of productivity since 1976. Between 1978 and 1982—a period that included the deepest recession of the postwar period—productivity actually declined slightly according to official estimates. Then, as recovery took hold, productivity rebounded. By 1987 the economy once again was operating in the neighborhood of its full potential. Between 1978 and 1987 the growth of productivity averaged about 0.9 percent per year.

Since 1987 this chain of events has essentially repeated itself: a period of slow growth in productivity as the economy endured a recession, followed by a period of rebound as the recovery gathered strength. Today, well into the expansion, the economy once again appears to be operating in the neighborhood of its potential. Between 1987 and 1994—as was noted above—productivity growth averaged about 1.2 percent per year. Thus, currently available data do seem to hint that the trend in productivity growth has picked up in the last few years. However, the magnitude of that pickup pales in comparison to the decline that occurred earlier in the post-war period. Moreover, the evidence in support of a pickup is still inconclusive. For example, if trends are computed for the periods 1978-86 and 1986-94 rather than 1978-87 and 1987-94, the suggestion of a pickup is much weaker: productivity growth averaged 1.0 percent per year in the earlier alternative subperiod and 1.1 percent in the later one. On the other hand, if the breakpoint chosen is 1988 or, especially, 1989, the evidence in favor of a pickup appears stronger. However, the averages over these later periods, especially the one since 1989, are dominated by the cyclical recovery and so may create a false impression of an improvement in the trend.

Furthermore, the Labor Department released data in 1994 suggesting that the growth of hours worked between 1993 and 1994 may be revised upward by enough to shave 0.1 percentage point off the average rate of productivity increase for the period 1987-94. Thus, while the evidence in favor of a slight improvement in the productivity growth trend is encouraging, it is not yet decisive. The experience of the next few years will be quite telling for this issue.

Issues Related to the Measurement of Productivity

To many in the business community, the idea that there has been a slowdown in the rate of improvement of business efficiency would simply be implausible. International comparisons based on detailed case studies suggest that the level of productivity is higher in the United States than in Germany or Japan and that many important innovations—especially in the services sector—have originated in the United States.

Examples of such innovations abound. Retailers have invested heavily in information technology to improve efficiency and the quality of service. New specialty formats provide customers with a wider array of choices. Financial institutions have simultaneously improved their efficiency and expanded their product lines dramatically. Mortgages are now processed much more quickly and in much greater volume. Customer service has been enhanced by the widespread introduction of automatic teller machines as well as automatic deposit and withdrawal services. The mutual fund industry now provides individual investors with diversification possibilities that would have been barely conceivable 30 years ago. In the field of medicine, with the introduction of microsurgical techniques, a cataract operation performed today is faster and safer than one performed even a decade ago. And with the advent of arthroscopic surgery, repair of a torn knee ligament involves a shorter stay in the hospital, less chance of collateral damage during surgery, and a faster recovery time. Telecommunications companies have introduced many new services, including high-speed data transfer and mobile cellular telephone service.

To some extent, these dramatic changes in service industries are not reflected in the productivity data presented in this chapter. Either they do not enter the standard productivity calculations at all, or their contribution to growth is understated. For example, within the financial services area, productivity growth in the banking industry has averaged more than two percent per year in recent years, according to BLS estimates. However, these estimates are not used in the construction of aggregate measures of output and productivity. Instead, for these measures, growth of real output in banking and other financial services is assumed equal to the increase in hours worked in the industry, so that growth in labor productivity is roughly zero by assumption.

Measurement issues are particularly important in the area of health care, both because that sector now accounts for 14 percent of GDP and because the conceptual difficulties there are so great. For example, current productivity measures would not reflect the influence of a technological advance that allowed a gallbladder patient to be treated and to recover in a much shorter time than before. As for telecommunications, productivity data understate the benefit to consumers of newly available services.

These examples reflect underlying problems in productivity measurement associated with the changing character of the economy. But there are also other general problems in measuring productivity. Roughly speaking, official measures of average labor productivity are calculated by dividing the nominal output of a given sector (e.g., the private nonfarm business sector or the manufacturing sector) by an estimated price index and a measure of hours worked. The trends in all three of these variables are subject to measurement error.

In concept, the task of measuring nominal output is straight- forward: one need only calculate the current dollar value of total production of "final" goods and services—that is, goods and services that are used for either consumption or investment at home or abroad, by either individuals, businesses, or governments. In practice, however, the task is challenging. One important set of difficulties involves the definition of investment goods. Traditionally, investment goods have been defined as tangible assets, such as factories or drill presses, that have a useful lifetime of more than one year. As a result, intangibles such as computer software and research and development have for the most part been treated as intermediate goods and services—that is, as inputs into the production process—and therefore not as part of final demand.

Recently, however, a number of observers have suggested that the traditional definition of an investment good should be expanded to include business expenditures for computer software. A move in this direction would raise the measured level of GDP and hence would also raise the measured level of productivity. Moreover, to the extent that business expenditures for computer software have been growing more rapidly than the economy as a whole, such a redefinition would also raise the rate of growth of both output and productivity. Finally, such a redefinition would temper the apparent slowdown in productivity growth since 1972, assuming that, as seems likely, the growth of software production has been more rapid since 1972 than before. Box 3-4 discusses issues related to treatment of software as an investment good in the national income and product accounts (NIPAs).

Box 3-4
Business Expenditures for Computer Software in the National Income and Product Accounts

Much of computer software is treated as an intermediate good in the national income and product accounts rather than as an investment good. (Software that is sold with computer hardware as part of a package is, however, included in the current NIPA measure of investment if the machine itself is so treated.) In part, the current treatment of software reflects a presumption that much computer software has a useful lifetime of less than one year, and thus does not quality as an investment good under current definitions. In part, however, it also reflects a lack of information; many companies probably do not themselves know how much they spend on computer software, and the Department of Commerce certainly does not know, because none of its ongoing surveys requests this information.

If computer software were to be included in the national income accounts as an investment good, estimates would have to be developed not only of nominal outlays for computer software but also of a quality-adjusted price of software. To estimate such prices, analysts would have to determine, for example, how much more "word processing power" was provided in a new release of a word processing package than in the one it superseded.

It is difficult to know how much the treatment of computer software as an intermediate good affects the overall productivity pictured. But because the volume of software purchases is vastly greater today than it was three decades ago, it may help explain part of the productivity puzzle. The case of computer software also illustrates some of the serious conceptual difficulties involved in improving current measures of productivity.

Measurement of prices is the critical problem in the measurement of productivity. The output of the economy increasingly is shifting away from standardized commodities with easily definable characteristics that change little over time, toward goods and services for which issues of quality and even definition are of primary importance. And if the trend in prices is mismeasured, so will be the trend in output and hence productivity. As an illustration of the difficulties involved in measuring prices, consider the increased prominence of discount outlets in the retail sector. In constructing the consumer price index, government statisticians treat goods sold at discount retailers as distinct from similar or identical goods sold through traditional outlets. When a discount retailer adds to its product line an item already being sold by traditional retailers, but offers it at a lower price, the difference between the discounter's and the full-service merchant's price is treated as signaling a difference in the quality of a total package: item for sale, service provided, and possibly other consumer amenities. Hence, the lower price suddenly available at the discounter is considered not to imply a reduction in the cost of living, and it is not allowed to drive the index down. But while it may be true that discounters provide less attentive or complete service and a less enjoyable overall shopping experience than their full-price counterparts, it is also plausible that part of the difference in initial price reflects operating efficiencies and hence does represent a true reduction in the cost of living; if so, it would argue for taking at least partial account of the discounter's initial prices in computing the index.

Even measurement of hours worked is more difficult than one might imagine. Estimates based on surveys of employers and households show different trends. In part this divergence may indicate that employers have a relatively poor idea of how many unpaid overtime hours their employees are working at home. For their part, workers have been shown to overstate hours worked on average.

It is easy to point to deficiencies of existing elements of the measurement system—deficiencies that could be alleviated by a reallocation of resources for data collection and analysis—but it is much harder to pinpoint the quantitative significance of such deficiencies. The Bureau of Labor Statistics has been in the forefront of research into methodological improvements in both price and productivity data and, indeed, has implemented many improvements in both types of data in recent years.

What are the implications of possible measurement errors? First, they are likely to provide at least a partial explanation for why the measured growth of productivity has been slow in recent years. Second, as was noted earlier, they help explain the post-1972 slowdown in productivity growth to the extent that they have been more severe since 1972 than before. Although the magnitudes involved are not known with any precision, it is likely that error-contaminated data understate the economy's productivity growth rate and hence its capacity growth

Factors Generating Growth of Hours Worked

In addition to increases in output per hour worked, the other source of growth in the productive capacity of the economy is increases in the total number of hours worked. Of course, the implications of increases in work hours for the economic well-being of the American people are not the same as the implications of increases in productivity, because increases in hours worked impose some cost in terms of time no longer available for other activities.

Growth in hours worked can come from four main sources: growth in the number of hours worked each week by the average employed worker; growth in the fraction of the labor force that is employed; growth in the fraction of the working-age population that is in the labor force; and growth in the size of the working-age population. Chart 3-8 summarizes the behavior of each of these factors since 1963.

According to the Department of Labor, the number of hours worked per week on the average job in the nonfarm business sector declined from just over 38 hours per week in the mid-1960s to about 34 hours in the early 1980s. Since then it has been about flat (Chart 3-9). (The nonfarm business sector differs from the private nonfarm business sector in that it includes government enterprises such as the US Postal Service.) On net, the decline in the average workweek has taken about 0.4 percentage point off the growth of aggregate hours worked since 1963—a bit more between 1963 and 1972, and a bit less since 1972 (Chart 3-8).

Changes in the employment rate have contributed essentially nothing to the trend growth in hours over any of the periods shown in Chart 3-8. This outcome reflects two facts. First, the years 1963, 1972, and 1994 were chosen as endpoints precisely because the employment rate was near its so-called full-employment level in those years. Second, the full-employment level of the employment rate has not changed greatly over the periods examined here.

One of the most striking macroeconomic developments of the postwar period has been the convergence in the labor force participation rates of men and women (Chart 3-10). Thirty years ago fewer than 40 percent of working-age women were in the labor force; today that fraction stands at nearly 60 percent. The largest increases in labor force activity took place among younger women, but substantial gains were also registered by women in their forties and fifties. The trend among men has been in the opposite direction. In 1960 more than 83 percent of working-age men were in the labor force, but by the early 1990s that fraction had dropped below 76 percent. The reduction in the labor force participation of men was particularly pronounced among older workers.

On balance, the influx of women into the labor force was the more important of the two gender-related trends, and the aggregate participation rate displayed a marked upward drift over the last 35 years, contributing about 0.4 percentage point per year to the growth of hours. The contribution of the participation rate to the growth of hours has been a shade greater since 1972 than before.

Since 1989, however, the growth in labor force participation has been unusually slow. In fact, the average participation rate in 1993 was below the average rate in 1989. The average rate did move up noticeably in 1994, but it is still too early to know whether the upward trend in this variable has resumed. Moreover, the interpretation of the participation data for 1994 has been made more problematic by the introduction in January 1994 of the redesigned Current Population Survey (the Labor Department survey that is one of the key sources of monthly data on the labor market). Data collected over the next few years should help resolve whether the pause in the increase in the participation rate between 1989 and 1993 was a temporary aberration or a signal of a new, permanent state of affairs.

Between 1963 and 1972, growth of the working-age population averaged nearly 1.8 percent per year. By contrast, since 1972 this growth has averaged 1.4 percent per year, and since 1982 only about 1.1 percent per year.

Since 1963, aggregate hours worked in the private nonfarm business sector have increased at an average pace of about 1 3/4 percent per year, with little difference in the growth rate before and after 1972. By happenstance, the slower rate of decline in the workweek after 1972 and the slight step up in the rate of change of the participation rate (both pluses for the growth of hours) were about offset by the slower growth in the working-age population.

What can the Government do to Improve the Economy's Long-Run Growth Potential?

Without a doubt, the future rate of increase in the economy's productive capacity will be largely determined by the decisions of the millions of individual businesses and households in the private economy. The role of the government is, and will continue to be, a limited one: to foster an open and competitive market environment, and to help the market work better when it would otherwise generate an inefficient result.

Government policies to advance these objectives generally fall into two broad categories. First, government must address the question of national saving. Historically, nations that have saved the most have also invested the most, and investment has been strongly correlated with productivity. Therefore, it is a matter of considerable concern that the national saving rate in the United States is low by international standards and has declined in the last 20 years. Second, government must address market failures. Depending on the context, pursuit of the second objective may require the government to strengthen market forces already in place (as, for example, when it subsidizes student loans or provides support for worker training and skill acquisition); to impose regulation (as, for example, when it takes actions to curb excessive market power or to protect the environment); to enhance competition (as, for example, when it reduces barriers to international trade); or to provide public goods (as, for example, when it funds R&D). The need for public goods arises especially in situations in which private market incentives on their own would result in less than the optimal amount of investment being undertaken because the returns from that investment are not fully appropriable by the private investor. Investment in basic research is a case in point. It should go without saying that government policies to address market failures should be designed to achieve their objective while imposing the lightest possible burden on the economy. (Chapter 4 discusses this point further.)

Boosting Productivity by Increasing Domestic Saving

During the 1960s and 1970s gross saving in the United States averaged about 17 percent of GDP. As Chart 3-11 shows, gross saving declined markedly thereafter, averaging roughly 15 1/2 percent during the 1980s and only about 12 1/2 percent between 1990 and 1993 (fiscal- year basis). In part this decline reflected the deteriorating fiscal position of the government sector (defined to include all levels of government—Federal, State, and local). Measured on a national income accounts basis and averaged over fiscal years, the deficit of the government sector was only 0.2 percent of GDP during the 1960s and about 1 percent during the 1970s. But during the 1980s the average deficit widened to 2 1/2 percent of GDP, owing entirely to a dramatic increase in the Federal deficit. And the average between 1990 and 1993 was even a bit worse because of a decline in the surplus of State and local governments.

Personal saving has also declined, from about 4 1/2 percent of GDP during the 1960s and 5 1/2 percent in the 1970s to only 3 1/2 percent during the early 1990s. Meanwhile, the trend in business saving—which accounts for the bulk of gross saving—has been remarkably flat since the 1960s.

In fiscal 1994, gross saving, private and public, reversed course and edged up to nearly 13 1/2 percent. The main cause of this development was a considerable reduction in the deficit of the consolidated government sector, almost exclusively the result of a sharp improvement at the Federal level: measured on a national income accounts basis, the Federal deficit in fiscal 1994 (the first year in which this Administration's budget plan was in effect) declined to 2.6 percent of GDP, a full 1.5-percentage-point reduction from the preceding year.

Gross saving serves as a good measure of the Nation's saving effort, but saving net of depreciation may be a more meaningful measure of the domestic resources available for increasing the capital stock. Unfortunately, the trend in net saving has been even more disturbing. As Chart 3-12 reveals, the decline in net saving—from an average of eight percent of GDP in the 1960s to an average of two percent of GDP between 1990 and 1993—has been even steeper than the decline in gross saving. Net saving increased in 1994, and it is in this light that the reduction in the Federal deficit is especially significant: the fiscal consolidation at the Federal level accounts for all of the improvement in the Nation's net saving rate in 1994 over the average for the early 1990s.

In theory, domestic investment need not be tightly linked to domestic saving, and a country that succeeds in boosting domestic saving may not be rewarded with an increase in domestic investment. In that event, however, it would be rewarded with a reduction in its current account deficit (roughly speaking, its balance of trade in goods and services with other countries). In the case of the United States, either outcome—an increase in investment or a reduction in the current account deficit—would be a desirable result of an increase in the domestic saving rate.

In this light it is relevant to ask what the government can do to stimulate the rate of gross saving. Fundamentally, two approaches are possible: one is to boost public saving (that is, cut the deficit of the government sector), and the other is to stimulate private saving.

Increasing Public Saving
As has been documented in Chapters 1 and 2, this Administration has made a very substantial contribution toward the reduction of the Federal deficit (Chart 2-9 in Chapter 2). Even so, the longer term outlook for the deficit remains troublesome, owing in part to the projected shift in demographics, as the baby-boom generation moves into retirement and begins collecting Social Security and medicare benefits. This aspect of the long-term outlook suggests that, despite the progress achieved under the Omnibus Budget Reconciliation Act of 1993 and the additional deficit reduction proposed in the Administration's 1996 budget package, more work remains to be done to put the budget on a secure footing for the long term and hence to ensure a healthy national saving rate.

Increasing Private Saving
The Federal Government has often sought to increase national saving by inducing the private sector to save more. The evidence on the effectiveness of such efforts is mixed.

Many of these attempts have focused on increasing the after-tax rate of return to the owner of a particular type of asset. For example, individual retirement accounts (IRAs) increase the rate of return on saving by allowing tax-free accumulation of funds held in qualified accounts, from which the funds cannot be withdrawn without penalty until the owner reaches the age of 59 1/2. The Administration has proposed an expansion of IRAs, to allow tax-deductible contributions by all couples with incomes below $100,000 (and individuals with incomes below $70,000), and to allow penalty-free withdrawals before age 59 1/2 for the purpose of purchasing a first home, paying for postsecondary education, defraying large medical expenses, and covering long-term unemployment expenses. Chapter 1 discusses this initiative in greater detail.

Boosting Productivity by Helping Markets Work Better

Aside from increasing domestic saving, a government can increase the productivity of its citizens by improving the quality of the labor force, increasing the quantity and improving the quality of the available capital stock, promoting the development of new technology, and fostering a free market characterized by vigorous competition.

Improving the Skills of the Work Force
The Federal Government has an important role to play in improving the quality of labor. Individual workers have an incentive to acquire productive skills on their own, without government involvement, if for no other reason than that better skills usually mean higher earnings. As is discussed in Chapter 5, however, individuals and organizations left to themselves are likely to underinvest in skill acquisition. To help overcome this problem, the Administration has devised a comprehensive set of education policies centered on the theme of lifelong learning. Together these policies are aimed at ensuring that students enter school ready to learn (thanks to Head Start and other programs); that schools work as effectively as possible in helping students to live up to their potential (through the Goals 2000 program); that students make a smooth and well-planned transition from high school to a job or further training (through the School-to-Work program); and that workers are given an opportunity to upgrade their skills (for example, with the help of a tax deduction for postsecondary training or through a grant for retraining in the event of unemployment). Each of these initiatives is described in detail in Chapter 5.

Increasing Investment in Technology
Firms that invest in technology often are unable to capture all of the benefits of their investment. That is, there appear to be important spillovers or "positive externalities" from such investment, in the form of benefits captured by other firms without compensation to the firm making the investment. These externalities imply that the social return to investment in R&D is higher than the private return, and that a private market left to its own devices would invest too little. As a result, government has an important complementary role to play, either in sponsoring research itself or in subsidizing private-sector research, or both.

Increasing investment in research and development is one way to promote technological innovation and productivity growth, because well-directed R&D spending has a very high growth payoff per dollar. Indeed, estimated social rates of return to R&D average around 50 percent—much higher than the average estimated private rate of return of 20 to 30 percent. (Box 3-5 discusses empirical evidence on average rates of return on R&D investment.)

For this reason the Administration has supported extending the research and experimentation (R&E) tax credit. (Box 3-6 examines the R&E tax credit in more detail.) The Administration is also increasing funding for government-industry research partnerships and is working to restore a 50-50 balance between the military and civilian components of its technology investment. (The defense share of Federal R&D spending has already fallen from 69 percent in the government's fiscal year 1986 to a projected 55 percent in fiscal 1995.) In addition, the Administration is working to focus a larger portion of the Federal R&D effort on so-called dual-use technologies (those with both military and civilian applications). Other Administration research initiatives reflect a strong continuing commitment to basic science, to the creation of improved information and transportation infrastructure, and to the development of technology in pursuit of other national goals, such as environmental protection and world-class manufacturing. These initiatives and others are designed to speed the pace at which new technological ideas are discovered and disseminated in the private sector. Chapter 4 provides more details on the Administration's reorientation of Federal R&D policy in light of the end of the cold war.

Box 3-5
Research and Development Pays Off

Investment in R&D appears on average to have an impressive payoff. One recent study concluded that the private rate of return—that is, the return to the firm performing the R&D—averages perhaps 20 to 30 percent. For comparison, the average rate of return to investment in the business sector as a whole is thought to be in the neighborhood of 10 percent.

Estimated rates of return in R&D to society as a whole are even higher, thanks to the spillovers described in the text. For specific innovations, estimates of the returns have ranged as high as 423 percent in the admittedly atypical case of optical fiber. In a wide range of areas, however, case study evidence points to rates of return of between 30 percent and 80 percent.

By choosing particular technologies for study, case study research runs the risk of choosing only "winners" (that is, R&D investments that have paid off handsomely), thus biasing the results upward. But the case study evidence has been widely corroborated by industry-level studies. By estimating the industry-wide returns to R&D carried out within the industry itself and within related industries, these studies have provided additional evidence that social rates of return greatly exceed private returns. On the basis of such evidence, a recent survey concluded that, with spillovers taken into account, the returns to R&D average perhaps 50 percent.

Typically, we might expect such high returns to encourage firms to spend more on R&D, driving down the rate of return until it equals the return to other activities. Why have returns remained so high? In the case of private returns, one probable explanation is that investing in R&D is risky. For every idea that yields a high payoff there may be dozens of "losers" into which a firm sinks resources in vain. If the firm were unconcerned about risk—for example, if it were able to farm out its risk by selling shares of its R&D activities to mutual funds—the variability of returns would not matter. But in practice, because of the problems of communicating the quality of a potential innovation to investors, the firm is likely to have to shoulder some of the risk itself. As a result, unless it is large enough to withstand the resulting variability of returns without difficulty, the firm will probably require a higher return as compensation for the greater risk.

Box 3-6
The Research and Experimentation Tax Credit

The research and experimentation tax credit is a Federal tax subsidy available to firms engaging in certain research activities. To address concerns that the subsidy be focused as narrowly as possible on research that otherwise would not have taken place, the credit is made available only on the increment of domestic research expenditures over a threshold amount.

The incremental nature of the credit means that some tax-paying firms (those with total research spending below the threshold) will not receive a subsidy for their research activities, worthwhile though they may be. The Congress recognized this concern but believed that an incremental credit was a more efficient subsidy mechanism than one that subsidized all research spending—in other words, that an incremental credit could achieve most of the benefit provided by a flat (non-incremental) credit at a lower budgetary cost.

Empirical research on the effectiveness of the R&E credit has yielded mixed results. Many of the early studies found that the credit was not very effective: an additional dollar of Federal tax subsidy was estimated to generate less than a dollar of additional research. However, the credit was substantially restructured in 1989, and more recent studies have indicated that the R&E credit is more cost-effective than previously thought.

The spillovers from both basic research and more applications- oriented activities cross national boundaries. In recent decades such transnational spillovers have probably been magnified by the revolution in communications, which allows news about innovations to be transmitted instantaneously around the world. Importantly, the existence of these spillovers suggests that the global return on R&D investment exceeds the national return. As a result, even national governments, acting on their own, will tend to sponsor too little basic research and applied R&D. If this analysis is correct, there may be a role for international coordination in support of such research. By instituting a formal mechanism for sharing research costs, such coordination could reduce the incentive of each country to free-ride on innovations financed by others.

Working to Reduce Trade Barriers
Barriers to international trade inhibit the efficient allocation of production across industries and countries and lower the real purchasing power of consumers. Trade barriers at home permit inefficient industries to continue using labor and capital resources that could be used more productively in other sectors. And trade barriers abroad limit the access of our efficient industries to foreign markets. One of the most beneficial aspects of an open world trading environment is that it exposes businesses all over the globe to greater competition, and forces firms and industries either to improve their efficiency or to free up their productive resources (labor and capital) for use elsewhere in the economy. Box 3-7 describes a recently developed theory suggesting that traditional analyses have been far too conservative in their conclusions regarding the costs of protectionism.

Box 3-7
A New Analysis of the Costs of Protectionism

Traditionally, in extolling the virtues of free trade and warning against excessive tariff protection, economists have focused on trade- induced efficiency gains of the type discussed in the text. But estimates of the costs of protectionism obtained from traditional economic models have typically turned out to be quite small. The inefficiencies caused by a 20-percent tariff, in one such analysis, turn out to cost the economy perhaps 4 percent of national income—hardly trivial, but far too little to explain why highly protected developing economies have often remained very poor. This finding has become more puzzling over the past decade or two, as mainstream opinion in development economics has swung firmly toward the view that integration in the world trading system has been critical to the success of the fastest growing developing nations.

Recent research has suggested one possible solution to this puzzle. If international trade barriers prevent new goods and technologies from being introduced into an economy, rather than simply raising the cost of goods that are currently available, then the cost of protection may be much higher. In one simple new-goods model, for example, a 20-percent tariff exacts costs equal to an astounding 39 percent of income—nearly 10 times as much as in the standard model. No highly abstract model is likely to give definitive estimates of the costs of protectionism, of course, and models with different assumptions yield very different results. Nevertheless, the new research does suggest a way to bring theory more closely into line with experience.

In light of the significant long-run benefits accruing to the economy from the pursuit of open markets, the Administration strongly supports the creation of a world trade and investment environment free of international barriers and has made historic progress toward that objective. After securing the ratification of the North American Free Trade Agreement (NAFTA) in 1993, the Administration scored several major achievements on the trade front in 1994. Most important was the signing of the Uruguay Round agreement of the General Agreement on Tariffs and Trade and its subsequent congressional approval. The Administration also made strides toward achieving freer trade and investment flows within Asia and Latin America. Chapter 6 describes at greater length the accomplishments of the Administration on the trade front.

Although removal of trade barriers leads in the long run to an improvement in the standard of living in all countries that participate, it can involve significant costs in the short run for some industries and some workers. For example, the transition to a new job from one lost because of trade liberalization can be difficult and may require significant retraining for the new job and even relocation. However, part of society's overall income gain from the move to freer trade can be used to reduce the cost of dislocation borne by individual workers. To ease the transition of workers affected by the implementation of NAFTA, as well as of other displaced workers, the Administration has introduced a number of innovative programs focusing on worker retraining. These programs are described in Chapters 5 and 6.

Improving the Efficiency of Regulation
Government regulation plays a central role in shaping the competitive environment in which firms operate. In many cases an improvement in regulation can simultaneously promote the more effective attainment of policy objectives and increase the efficiency of the economy. For example, a traditional approach to the problem of reducing emissions of sulfur dioxide (a major cause of acid rain) might have entailed mandatory investment in costly new pollution reduction equipment by all emitters. Instead, a market-oriented system, based on tradable emissions allowances, is achieving the same results while allowing the efficient allocation of the task of reducing pollution across emitters. Chapter 4 addresses in much greater detail the important contribution of efficient regulation to overall productivity.

Conclusion: Prospects for Growth

In sum, the preponderance of the available empirical evidence supports the conventional wisdom that the economy's productive capacity is expanding at roughly a 2 1/2-percent annual rate. Growth in the productivity of American workers appears to have picked up slightly in recent years, to about 1 1/4 percent per year, measured on a chain-weighted basis (this is roughly equivalent to 1 1/2 percent on the more usual fixed-weight basis). However, trend growth in the aggregate number of hours worked in the economy probably will be somewhat slower than it has been during the past decade or two, owing largely to a decline in the rate of growth of the working-age population. On balance, the sustainable rate of growth of the economy's potential appears to be nearly the same as it has been over the past two decades, with the increase in the trend growth of productivity offsetting part of the decline in the population growth rate.

The Administration's economic projection for the next five years reflects this analysis. Thus, among other factors, the projection reflects a cautious assessment of the beneficial effects of Administration policies to enhance the Nation's productive capacity and to foster more rapid growth of productivity. The projection also places the Administration squarely within a broader consensus about the longer term outlook for the economy. The Administration has attempted to adopt a balanced assessment of the outlook, grounded in rigorous analysis and consistent with recent experience. Although some observers maintain that the economy can grow much more rapidly on a sustained basis, currently there is no convincing empirical evidence to support such claims.

To illustrate the difficulty of improving the trend in the growth of the Nation's productive capacity, consider the following example. Suppose that a particular set of policies were to result in an immediate and permanent increase in the investment rate of one percentage point of GDP. Given that investment now constitutes about 14 percent of GDP, this would be an impressive accomplishment indeed. Under plausible assumptions, a standard approach to modeling the long- term growth of the economy suggests that such an increase in investment would boost the average annual rate of growth of potential GDP only by about 0.2 percentage point per year for the first 10 years. Thereafter the growth effects would diminish, fading eventually to nothing—but leaving the level of potential GDP an estimated 3 1/2 percent higher than it would have been without the investment push.

The analysis in this chapter also indicates that currently available official statistics probably understate the true rate of growth of productivity, and hence the rate of expansion of the Nation's productive capacity. Furthermore, to the extent that problems of measurement have become more acute during the last two decades (as might be suggested by the shift in the economy toward the services sector, where measurement is particularly difficult), the slowdown in the trend rate of productivity growth during the mid-1970s apparent in the official data is probably overstated.

Clearly, a full understanding of the scope and magnitude of measurement error is important for the proper design and conduct of economic policy. In particular, measurement error may cause official statistics to understate the performance of the American business sector, both relative to its international competitors and relative to its earlier performance. At the same time, measurement error does not provide a basis for adjusting one's view of the appropriate stance of monetary and fiscal policy. An upward revision in the estimated pace of innovation and growth in the economy would have similar implications for estimates of both actual and potential output, and thus would result in no revision in the estimated gap between the two.

The improvement in the trend rate of growth of productivity that is embedded in the Administration's economic forecast has important implications for the wealth and welfare of the Nation. If policies to boost the annual growth of productive capacity by 0.2 percentage point had been implemented a decade ago, the American economy would now have the capacity to generate an additional $150 billion in goods and services every year. Fortunately it is not too late to lay the foundations for comparable gains in productivity and incomes 10 years hence. The disappointing growth record of the last 20 years, and the anxieties that so many Americans have about their own and their children's economic prospects, demand that every effort be made today to expand the economy's capacity in the future.