The Growth Puzzle

Productivity growth is the single most important determinant of our future standard of living.

People can expect their real wages and their living standards to double every 28 years, or roughly once a generation, if capital investment and technological change allow workers to increase their output by 2.5 percent per year. But if output per worker grows at only 0.5 percent, children can expect living standards only slightly higher than those of their parents.

In this regard, the numbers in recent decades look bad: Labor productivity growth in private businesses, excluding farms, declined from an average annual rate of 2.5 percent over 1948-69 to 2.0 percent over 1969-73, and then to 0.5 percent from 1973 to 1979. The recent numbers are somewhat better. Labor productivity growth has averaged 1.1 percent annually since 1979, but that rate is still well below the heights of the post-World War H period.

If output increases only because more people or inputs are used in production rather than because of higher productivity the welfare of the majority of society will not improve. Instead, all the increase in output is needed to pay additional workers and other suppliers of inputs at their old rates of compensation.

This is precisely what has occurred in the 1980s. Real output since the 1982 recession has grown at an average annual rate of 4.7 percent, but most of this growth stemmed from a dramatic increase in employment from the trough of the recession. Since the growth was due to more people working rather than to greater output per worker, real compensation per hour is roughly the same as it was in 1982. It is a mistake, therefore, to be sanguine about the health of the American economy based solely on steady growth in output. Growth based on continued increases in employment is bound to be limited because new entrants into the labor force come mainly from the "baby bust" generation. As employment growth has slowed, so has growth in output.

The important issue for the future, therefore, is not whether Gross National Product (GNP) growth was high or low in the 1980s, but rather why productivity growth has dropped from its previous levels. Economists have written extensively on the decline in productivity growth and have gone to great lengths to try to identify the reasons for the slowdown. No one has discovered a "silver bullet," and almost all observers end up concluding that a variety of factors are at work. The usual suspects include:

• the changing composition of the labor force due to the influx of teenagers and other less experienced workers;
• a slowing in the rate of growth of the capital-labor ratio as investment in equipment and structures failed to keep pace with the unprecedented increase in the employed labor force;
• a leveling-off in research and development expenditures;
• the diversion of investment funds to pollution abatement;
• the maturation of some industries, with little new technology; and
• changes in attitude towards work.

  In other words, no one has really been fully able to explain the slowdown, least of all through a single factor such as savings rates.

  Some recent work, however, has helped to give us a much better understanding of the forces shaping productivity growth today In their "long view" of Productivity and American Leadership, William J. Baumol, Sue Anne Batey Blackman, and Edward N. Wolff have amassed historical, international, and other new information to put into perspective the current productivity slowdown. Their approach is complex, and their conclusions not at all what some would have expected. "Our opinions did not determine our choice of facts -- rather, quite the reverse is true," they write in the preface. Indeed, the authors' conclusions contradict not only much conventional wisdom but also some of their own earlier writings.

  Lessons of the Long View
  Careful analysis of long-run data has forced Baumol and his co-authors to abandon their gloomy assessment of U.S. productivity performance. They conclude that although retention of American preeminence will not be automatic or easy, it is certainly not out of the question.

  The first part of Productivity and American Leadership addresses the major concerns about U.S. economic performance: the slowdown in our productivity growth; the decline of American preeminence as Japan, Germany and other industrialized countries with rapid productivity growth pull ahead; and fear of "deindustrialization" as increasing numbers of workers are employed in the service sector. The second part tries to disentangle the relative importance of the three factors that explain productivity changes -- labor, natural resources, and capital.

  Initially, Baumol and his co-authors feared that the United States had experienced a sharp and permanent decline in productivity growth. They are now convinced, however, that the short-term slowdown is merely a return to normal growth rates after an extraordinary wartime and postwar boom. Productivity grew rapidly during World War II as output remained high and labor in short supply After the war, the United States shared with other countries in a productivity surge as companies took advantage of an enormous backlog of technological innovations and savings available for productive investment.

  But the rates of productivity growth during the immediate postwar period were unsustainable. In part, they compensated for the poor performance of the 1930s, when consumer spending was depressed and technical innovations were slow to be commercialized. In part, postwar productivity reflected the burst of wartime investment. When the current slump in productivity performance is placed in this longer-term perspective, it does not appear all that unusual. In fact, the previous rapid growth in productivity constitutes the historical (although easily explicable) aberration. A slowdown from unrealistically high levels, which also occurred in virtually every industrialized country in the 1970s, reflects a return to normality rather than a long-term fall in the rate of productivity growth.

  Even if the United States is not suffering a slowdown relative to its own past performance, it could still be falling behind other countries. Here again the authors say no, and build a convincing case that national incomes should converge over time, given the easy movement of technical advances and capital across national boundaries. The evidence of convergence suggests that the United States, though no longer preeminent, is not about to lose its economic standing.

  A purported corollary of poor productivity performance is that America will become a nation of people flipping hamburgers and doing each other's laundry. Supporters of this "deindustrialization" thesis point to the ever-growing shares of the labor force employed in the service sector and of services in gross national product. In response, the authors make two important points.

  First, the share of real output constituted by services has not changed at all. And, second, the United States is not shifting into services any more rapidly than other industrialized nations. They point to the consequences when one economic sector (manufacturing) is marked by rapid productivity growth, while another sector (services) is relatively stagnant. In that situation, the products of the leading sector will require less and less labor to produce, their relative price will fall, and they will account for an ever smaller fraction of the labor force and nominal GNP. For just the opposite reasons, the lagging sector, requiring an increasing share of the labor force, will see rising relative prices and will constitute an increasing share of GNP.

  The increasing share of the lagging sector, however, is solely the result of differing rates of price increase between the two sectors. Once output in each sector is "deflated," adjusting it for changes in its own price index, the relative shares of real output in manufacturing and services can be shown to be unchanged. This is what has happened in the United States over the last three or four decades.

  Nonetheless, while the apparent increase in the importance of services can be explained away, the authors acknowledge that the shift may not be costless. To the extent that the shift to services produces poorly paid jobs with little opportunity for advancement or impedes overall productivity growth in the country, the effects can be harmful.

  Turning to the factors underlying productivity growth, Baumol and his coauthors first address education and natural resources. They compare the historical records of developed and less developed countries and conclude that education is important to economic growth. They also present a comforting and convincing case that resource exhaustion is not a serious threat. They reiterate an argument, made by Simon Kuznets decades ago, that resources are not fixed but are a function of technology. Improved extraction procedures can partially offset the exhaustion of a particular resource, and technological advances can create substitutes for resources in short supply.

  Why Savings May Not Be the Key
  A key part of Productivity and American Leadership concerns capital. In the authors' view, the decline in the national saving rate and the relatively low level of investment need not condemn the United States to a future of mediocrity. They claim the apparent poor performance is partly a statistical illusion, reflecting the lower price of capital goods in this country. Moreover, the discrepancy between the U.S. saving rate and those of other developed countries may have less to do with basic attitudes toward thrift than with the stage of development. Japan's high saving rate may be transitory. Its pre-war levels of consumption have persisted even as income has risen due to the postwar productivity boom If other countries' historical experience is any guide, savings rates will decline as consumer aspirations catch up with rising incomes.

  In support of their sanguine view of the current low saving rate, the authors reiterate the point that technological change accounts for nearly two-thirds of productivity growth. That leaves a relatively small role for capital accumulation. Moreover, the causal relation between capital and productivity growth is unclear; technological advances may stimulate a demand for capital as well as capital contributing to productivity growth.

  In view of their optimistic reading of historical trends, comparative development, and the relation of savings to growth, the authors' relatively mild recommendations should not come as a surprise. They favor increased funding for basic research at the nation's universities and for long-run projects that may not pay off quickly. They want more government effort to facilitate the transfer of foreign technological advances to the United States. They would like to see entrepreneurial talent redirected towards more productive undertakings. And they make a case for investing in education, especially for deprived minorities, who will constitute an ever-increasing portion of the labor force. If these modest proposals prove insufficient, they suggest offering companies tax rebates based on their rate of productivity growth. This last option may sound somewhat difficult to implement, and the authors do not spend much time fleshing it out.

  But the proposals as a whole highlight a key contribution of this volume -- its de-emphasis on the importance of saving. There are no exhortations for Americans to increase their savings. None of the policy proposals involve saving incentives. The authors make clear that capital investment per se can play only a limited role; most of the action occurs in acquiring and adopting new technologies. Hence, they place their emphasis, I think appropriately, on doing business in a more clever fashion.

  The authors also enrich the debate by suggesting that saving is a residual, resulting from sticky consumption patterns and rapidly changing incomes. This notion not only is refreshing, but also can help explain what has happened in the United States. The usual story is that America's low saving led to low productivity growth and slow growth in income. This tale, however, does not square with the facts; productivity growth dropped in the late 1960s or early 1970s and the saving rate did not fall off until the 1980s.

  The more likely story is that America's low saving rates are the result of inflexible consumption patterns and a rapid drop-off in income growth due to the decline in productivity increases. Low saving did not cause the decline in U.S. productivity, and greater saving alone is unlikely to be the solution.

  In addition to putting the relative roles of capital and technological change in perspective, the authors also make an important contribution by placing the current productivity growth in historical perspective. For a supposedly intelligent group of people, economists have the bad habit of extrapolating last year's or the last decade's experience forever into the future. The Baumol, Blackman, and Wolff book will make this simplistic exercise much harder to justify.

  It would be a shame, however, if the book were read simply as documentation that all is really well with the American economy. Baumol and his co-authors should be read as a plea for complexity, not for complacency.

  One could quibble with specific omissions and assumptions. Most of the historical data end in 1980; it would have been useful to have them extended another ten years. The authors fail to address the common assumption that growth of the service sector has resulted in less desirable jobs. They dismiss too quickly the critics who argue that the way the numbers are constructed tends to overstate productivity growth in the manufacturing sector and understate the performance of the service sector. And it would have been useful to see more discussion of the role of public investment in stimulating productivity growth. But these specifics are offered only as mild reservations. This is a wonderful book that provides a fresh and enlightened look at the most important determinant of our future welfare. By refocusing the debate about the reasons for our current low level of productivity growth, Baumol, Blackman, and Wolff have steered the search for solutions in a far more useful direction than the simple call to increase saving.

  Lindsey's Story
  Lawrence Lindsey's The Growth Experiment has some superficial similarities to Productivity and American Leadership. Both books are concerned with the performance of the economy. Both are optimistic. In both books, the authors contend that they began their work with no preconceptions. Both end with tax proposals. There, however, the similarity ends. Baumol and his co-authors look over a long historical horizon, appreciate the complexity of economic growth, weigh statistics carefully, and present their conclusions with appropriate modesty. Lindsey explains everything through supply-side incentives, makes selective use of the data, and tells a simplistic and polemical story.

  Lindsey's claim that he began without preconceptions is disingenuous since it does not square with his analysis. Lindsey starts out by arguing that the 1981 tax cuts -- the Economic Recovery Tax Act of 1981, or ERTA -- brought down inflation, restored an eroding tax base, increased business investment, and led to nearly a decade of expansion, all without having any serious impact on the deficit. In his attempt to prove that low tax rates are the single most important factor in the economy, he is forced to disregard everything else, from the impact of the 1981-82 recession to interest rates to demographics to global economics.

  The second part of the book is devoted to building a case for a tax cut and the introduction of a new 19 percent flat rate tax. Lindsey's view of the future is as selective as his assessment of the past. He projects a rosy future of looming federal budget surpluses -- on the assumption that the emerging accumulations in the Social Security trust funds are available for general expenditures. He ignores a decade of debate about the purpose of the Social Security buildup and the reasons for keeping these funds separate from other revenues.

  Nevertheless, Lindsey's conclusions have been embraced by the business press. "It's a logical blueprint for Washington economics in this decade and on into the 21st century" say Forbes's editors. Fortune's reviewer claims, "Larry Lindsey has given a rigorous analytical book that refutes ... objections to lower taxes." A review in The Wall Street Journal endorses Lindsey's book as "the most credible and coherent supply-side account of the revolution in U.S. tax policy that triumphed in the 1980s."

  A Closer Look at the Evidence
  Because Lindsey's arguments have apparently been taken seriously in some quarters, it is not enough simply to list all the institutional and other factors that he omitted from his analysis. It is also necessary to consider the key estimates that form the basis for his conclusions.

  The magnitude of the supply-side response from the 1981 tax cuts is central to Lindsey's thesis and to his proposals for the future. Supply-siders argue that the revenue losses from tax cuts are partially or even totally offset by the increases in savings, investment, and growth that result from responses to the tax cuts. Lindsey contends that the tax cuts so changed behavior that the actual revenue loss was much less than the legislated tax cut.

  Lindsey's estimates of the impact of reduced tax rates are based on a sample of 34,000 tax returns. His approach is to subtract from the direct revenue loss the offsetting revenue gains from the demand-side and supply-side responses. That approach is correct; the problems arise in its implementation. Since none of the reviews of Lindsey's work has questioned these estimates -- understandably, since they emerge from something of a black box -- a review of his calculations for a single year is worthwhile.

  Lindsey's first step is to calculate the base revenue loss from the 1981 tax cut. To take 1984 as an example, he starts by assuming that the economy in 1984 would have been exactly as it turned out to be without the tax cuts. Using his sample of actual returns from 1979, Lindsey extrapolates the income and deduction items on each return to reflect the size of the economy in 1984. In other words, he instructs the computer to fill out each return based on 1984 incomes and apply the pre-1981 personal income tax law. He then runs the sample through the computer again, this time using the new law. The difference between these two numbers is the direct revenue loss, which for 1984 turns out to be $98 billion. This number is very close to the figure published in the federal budget. So far, so good.

  Lindsey's first adjustment is designed to reflect the fact that the economy was stronger in 1984 because of the 1981 tax cuts -- a "demand-side" response to the economic stimulus of lower tax rates and higher disposable income. Using a model from Data Resources, Inc., he estimates that income would have been 3.2 percent lower without the tax cut. Since taxes increase more than proportionately with income, tax receipts would have been 5.1 percent (1.6 x 3.2 percent) lower. The arithmetic is straightforward: reducing baseline receipts by 5.1 percent yields an offset of $19 billion. That is, although the tax cut reduced government revenues by $98 billion, the tax cut made back $19 billion of that through economic stimulus. Lindsey, however, reports a $35 billion demand-side offset. This discrepancy appears to be due to an error in computation.

  Lindsey's second adjustment is aimed at capturing the supply-side increase in taxable income from additional labor, primarily by second earners and the self-employed, and from an increase in compensation in the form of taxable wages rather than tax-free fringe benefits. Since these changes were stimulated by the tax cut, income and tax receipts under the old law would have been lower. Lindsey argues that the dollar reduction in the baseline for 1984 to reflect the supply-side effects should be $15 billion. My suspicion is that labor economists m light find this high, but for now assume it is correct; the next adjustment raises more than enough questions.

  The final adjustment is what Lindsey terms the "pecuniary" effect. Lindsey defines this response as people reshuffling their assets in the face of lower tax rates; he offers as examples individuals switching from tax-exempt bonds to taxable bonds or choosing to give less to charity. He estimates the dollar amount of the pecuniary effect as the difference between the amount his model says should have been collected in 1984 under the new law and the amount of personal income tax that actually was collected. This discrepancy amounts to $20 billion in 1984.

  Indeed, revenues in 1984 were considerably higher than any forecaster had projected. Careful examination, however, indicates that the unexpected receipts arose from an unprecedented amount of one-time capital gains realizations. The receipts did not come from a shift from tax-exempt to taxable bonds or a similar long-lasting asset reshuffling. Most observers attribute the increase in realizations to the boom in the stock market, which rose sharply between 1981 and 1984 as it does usually in the early stages of recovery from a severe recession, particularly when accompanied by a disinflation that makes equity holdings more attractive to investors. Lindsey does not attempt to link the stock market boom to the personal income tax provisions of ERTA, which are the subject of his analysis. Hence, the unexpected revenues cannot be considered in any way an offset to the direct revenue loss from the lower personal income tax rates. This last adjustment is not supported by the evidence and should be disregarded.

  Putting together the corrected demand-side response and accepting the credible portion of Lindsey's substantial supply-side estimate suggests that instead of the direct revenue loss of $98 billion published in the budget, the actual revenue loss could possibly have been as low as $64 billion ($98 billion less $19 billion less $15 billion). There is no realistic basis for a lower figure, and no basis at all for Lindsey's estimate that the revenue loss was only $28 billion. In other words, any reasonable interpretation of the available data provides no support for a dramatic supply-side response and supports one's initial intuition that the tax cuts did indeed cost the government a lot of money and thereby contributed significantly to the deficit.

  Lindsey's "Great Surplus"
  Lindsey's second central calculation, which appears in a chapter entitled "The Great Surplus of '99," provides the justification for his call for a new tax cut. How, you may ask, does Lindsey claim surpluses when the administration and Congress are sweating over efforts to reduce deficits? The answer lies in part in Lindsey's underestimation of expenditure growth, which even under the assumption of no new legislation tends to outpace the increase in GNP. But the bulk of the problem is Lindsey's lack of understanding of the scheduled buildup of assets in the Social Security trust funds Subtracting the projected Social Security surplus from Lindsey's 1999 figure eliminates his surplus; correcting the expenditure growth assumption and other problems brings his numbers in line with the Congressional Budget Office estimate of a 1999 deficit of over $200 billion in the non-Social-Security portion of the budget.

  Is it legitimate to include Social Security surpluses with other revenue? Are they available to cover general expenditures? The answer is clearly no. The buildup of reserves in the Social Security trust fund is designed to ease the burden of supporting future retirees after the turn of the century, when retirees are numerous and workers are relatively few. These accumulations will ease the burden by increasing national saving and investment today so that total income from which benefits will be paid in, say, 2020 will be higher than it would have been otherwise. Whether or not additional saving actually occurs depends on how Congress treats the annual Social Security surpluses. If Congress uses the surpluses to pay for current outlays in the rest of the budget, no real saving will occur and income will be no higher in 2020 than it would have been in the absence of the buildup.

  The necessity of prefunding Social Security may be open to debate. Many would argue that a return to pay-as-you-go financing is an equally viable option. The result that all experts and commentators agree should be avoided, however, is one where the reserves amassed in the Social Security trust fund are spent on current consumption. This outcome would have the undesirable distributional consequence of financing general government activities with the relatively regressive payroll tax rather than with the more progressive personal income tax. Senator Moynihan's concern that Congress was heading in this direction prompted his controversial proposal to reduce payroll tax rates and return the system to pay-as-you-go financing. In short, for Lindsey blithely to include the Social Security surpluses in the revenue totals, declare surpluses in the unified budget, and call for tax cuts makes it appear as if he has missed five years of debate on the financing of the nation's largest social insurance program.

  Lindsey's proposals for tax reform include a personal income tax with a flat tax rate of 19 percent, substantially higher zero-tax thresholds, limitations on deductions for homeownership and state taxes, inclusion of fringe benefits and all transfer payments in the tax base, inflation indexation for interest income and capital gains, and the introduction of tax sheltered Individual Savings Accounts. On the corporate side, he advocates the use of a cash flow tax which he believes would give an enormous boost to investment by offering an immediate deduction for capital expenditures, offset in part by the elimination of the interest deduction.

  One could debate endlessly the specifics of his tax proposal, but the major criticism must be that these provisions are a very circuitous way of accomplishing the goal of increased national saving and investment, let alone increased productivity and economic growth. Incentives for individual saving embedded in the personal income tax have been ineffective and tend only to reward the wealthy, who would have saved in the absence of the incentive. The most direct approach is to move toward balance in the non-Social Security portion of the budget and to use the Social Security surpluses to increase national saving as they were initially intended.

  In short, the reader gains little from Lindsey's book. Can it really be that the 1981 tax cuts, not the Federal Reserve, were responsible for wringing inflation out of the economy in 1982? More fundamentally for most people the 1980s were not that great. As noted earlier, the rapid growth in output came from increases in employment and did not produce commensurate gains in wages or family incomes. The most recent data show that between 1979 and 1988 average family income for the nonelderly declined for the bottom 40 percent of the population, remained steady for the middle 20 percent, and rose for the top 40 percent, particularly for those with the highest incomes. The only way to improve family incomes and living standards for the entire population is to have rising real wages, and the only way to have rising real wages is to enhance productivity.

  Improving productivity growth, however, is a complicated and difficult task, requiring major changes in institutions and in how we organize the workplace; it is not a problem susceptible to a simple single-factor solution. Sadly, the business press seems either indifferent to the debate, or committed to preordained conclusions. While the editors of Forbes rushed to print prepublication excerpts from Lindsey's book, no business magazine has paid serious attention to Productivity and American Leadership. A simple thesis supported by selective facts unfortunately sells much better than a rich, even-handed, complex story -- particularly when the pleasant remedy is lower taxes.