Long-Run Growth

Growth rates in the industrial economies over long periods are modest by comparison with the recent Asian experience. Angus Maddison (1982) estimates growth rates of per capita GDP since 1700 of 1 percent a year or less for France, the Netherlands (the world leader in 1700), and the United Kingdom. Performance for ten industrial economies since 1870 is summarized in Table 2. Japan’s 2.7 percent is the most rapid rate of growth of per capita GDP in that period; no country shows aggregate growth in excess of 4 percent.

Kuznets (1959), summarizing his fundamental research by describing “modern economic growth as the adoption of the industrial system” (p. 110), develops the implications—that modern growth requires minimum levels of skill and literacy on the part of the population, a shift from personal or family organizations to larger scale units, the creation of transportation and communication infrastructure before the division of labor within the country can become a reality, and that growth is accompanied by urbanization as the labor force shifts away from agriculture, changes in demographics, and changes in social values. He emphasizes too the disparities among national growth processes: “the observable cases of modern economic growth are best viewed as combinations of the industrial system (including its minimum social concomitants) with the distinctive initial structure of each country” (p. 113). Theory has the task, though, of developing principles or classifications that make it possible to describe the growth process more generally than to treat each national experience as an historical accident.

^{Table 2.}

Growth of Per Capita GDP, 1870-1985

(In annual percent)

Source: Maddison (1982), updated from International Monetary Fund, International Financial Statistics, various issues.

^{Table 2.}

Growth of Per Capita GDP, 1870-1985

(In annual percent)

Country	GDP Per Capita	Aggregate GDP
Australia	1.1	3.1
Canada	1.9	3.6
France	1.9	2.2
Germany, Fed, Rep. of	2.0	2.4
Italy	1.6	2.3
Japan	2.7	3.8
Netherlands	1.5	2.7
Sweden	2.3	2.9
United Kingdom	1.3	1.8
United States	1.8	3.5

Source: Maddison (1982), updated from International Monetary Fund, International Financial Statistics, various issues.

View Table

^{Table 2.}

Growth of Per Capita GDP, 1870-1985

(In annual percent)

Country	GDP Per Capita	Aggregate GDP
Australia	1.1	3.1
Canada	1.9	3.6
France	1.9	2.2
Germany, Fed, Rep. of	2.0	2.4
Italy	1.6	2.3
Japan	2.7	3.8
Netherlands	1.5	2.7
Sweden	2.3	2.9
United Kingdom	1.3	1.8
United States	1.8	3.5

Source: Maddison (1982), updated from International Monetary Fund, International Financial Statistics, various issues.

Descriptions of the growth process in the industrial economies stress industrialization, or in Maddison’s (1982, p. 17) more encompassing term, the scientific-technical revolution. Growth proceeds as labor shifts from low-productivity agriculture to high-productivity manufacturing, and to services, but with productivity growth in agriculture often exceeding that in the rest of the economy. Chenery and Syrquin (1975) characterize the transition to development not only by a shift to manufacturing from primary production, but also at a later stage by an increase in the share of manufactured exports in GNP.

Modern analyses of growth are supply oriented, starting from the production function. In the neoclassical framework (Solow (1956, 1957), Denison (1974)) growth is accounted for by the accumulation of capital and labor, and by technical progress.⁴ Two important empirical results derived in this framework are the major role of technical progress in explaining the growth of per capita output, and the relatively small share of growth attributed to capital accumulation.

Denison’s calculations for the United States, presented in Table 3, show advances in knowledge accounting for more than half the growth of output per employee over the period 1929-82.⁵ Despite the importance of technical change, its sources and the role of policy in promoting technical progress are difficult to quantify. Maddison (1982, Chapter 3) cites the “institutionalization of innovation” as a cause of both a more even pace of technical progress in modern economies and more rapid technical advance. He notes encouragingly that a productivity slowdown of the type seen since 1973 has occurred before, and that it is not necessarily permanent.⁶

^{Table 3.}

Sources of Growth of U.S. National Income, 1929-82

(In percent)

Source:Denison (1985), p.30.

^{Table 3.}

Sources of Growth of U.S. National Income, 1929-82

(In percent)

	Total	Per Person Employed
Growth rate	2.9	1.5
Percent of growth attributed to:
Labor input except education	32	12
Education per worker	14	27
Capital	19	20
Advances in knowledge	28	55
Improved resource allocation	8	16
Economies of scale	9	18
Other	-11	-24

Source:Denison (1985), p.30.

View Table

^{Table 3.}

Sources of Growth of U.S. National Income, 1929-82

(In percent)

	Total	Per Person Employed
Growth rate	2.9	1.5
Percent of growth attributed to:
Labor input except education	32	12
Education per worker	14	27
Capital	19	20
Advances in knowledge	28	55
Improved resource allocation	8	16
Economies of scale	9	18
Other	-11	-24

Source:Denison (1985), p.30.

In international comparisons, productivity growth (roughly corresponding to Denison’s advances in knowledge) accounts for more than half of Southeast Asian growth, and relatively little of Latin American growth, for periods from 1955 to 1973. De Melo (1985) reports the results of studies of productivity growth in five Southeast Asian countries, five industrial market economies, and five Latin American countries. Productivity growth accounts for 46-64 percent of total growth in Southeast Asia, for 30-55 percent in the industrial market economies, and for 12-34 percent in Latin America. The capital stock increased more rapidly in Southeast Asia than in Latin America.

Growth of capital accounts in Table 2 for 20 percent of the U.S. growth rate, or 0.3 percent a year of growth, over the 1929-82 period. The resolution of the puzzle of the relatively small contribution of capital starts from the relationship between the rate of growth of capital per worker and the rate of growth of output per worker, which depends on the share of capital in total output. The relationship is that a 1 percent annual rate of growth of capital per worker adds a fraction of 1 percent, equal to the share of capital in output, to the growth of output per worker. In the United States over the period 1929-82 capital per worker increased at an annual rate of about 1.2 percent. The share of capital in output is about a quarter. One quarter of 1.2 percent is the 0.3 percent a year contribution of capital accumulation per worker to the growth of output per worker in Table 2.⁷

A slightly different perspective comes from consideration of the rate of return to capital, usually estimated in the range of 10-15 percent a year. Using for convenience an estimate of 12.5 percent, an increase in the share of investment in GNP by 1 percent raises the level of GNP by 0.125 percent, temporarily adding 0.125 percent to the growth rate. It would thus take a massive increase in net investment of 8 percent of GNP to raise the growth rate of GNP from the supply side by 1 percent within a period of three to four years. If productive projects can be found, investment programs of the pre-World War II Soviet type can contribute significantly to increasing the growth rate of potential output. But, within the constant returns neoclassical framework, an increase in the share of investment in GNP raises the level of GNP permanently but the growth rate only temporarily.

Kendrick (1980), Maddison (1982), and others have argued that the growth accounting framework of Denison significantly underestimates the role of capital in growth. Kendrick develops a more inclusive concept of capital that includes government capital,⁸ and intangible capital, including research and development, education and training, and health and safety. About half of GNP is investment by this definition. By Kendrick’s calculations, capital accounts for 70 percent of U.S. growth over the 1929-69 period.

Returning to the narrower definition, the role of capital in growth would be understated by Denison if it is the vehicle for the embodiment of new technology in production. Further, to the extent that technical progress is embodied in capital, an increase in the investment rate raises the level of output more than indicated in the preceding discussion⁹. Despite the plausibility of the embodiment hypothesis, it has been difficult to find decisive measures of the extent to which increases in productivity have been the result of new technologies embodied in new investment goods.¹⁰.

^{Table 4.}

Investment Share and the Growth Rate of GDP, 1965-84¹

Source: International Monetary Fund, International Financial Statistics, various issues.

^1.This is the result of a regression of the growth rate of GNP or GDP over the period 1965-84 on the average investment share during that period, run separately for different groups of countries. The coefficient is that on the investment share, its standard error in parentheses.

^{Table 4.}

Investment Share and the Growth Rate of GDP, 1965-84¹

	Industrial market
	Economics	Africa	Asia	Latin America
Coefficient	0.15	0.22	0.03	0.08
	(0.04)	(0.12)	(0.29)	(0.09)

Source: International Monetary Fund, International Financial Statistics, various issues.

^1.This is the result of a regression of the growth rate of GNP or GDP over the period 1965-84 on the average investment share during that period, run separately for different groups of countries. The coefficient is that on the investment share, its standard error in parentheses.

View Table

^{Table 4.}

Investment Share and the Growth Rate of GDP, 1965-84¹

	Industrial market
	Economics	Africa	Asia	Latin America
Coefficient	0.15	0.22	0.03	0.08
	(0.04)	(0.12)	(0.29)	(0.09)

Source: International Monetary Fund, International Financial Statistics, various issues.

^1.This is the result of a regression of the growth rate of GNP or GDP over the period 1965-84 on the average investment share during that period, run separately for different groups of countries. The coefficient is that on the investment share, its standard error in parentheses.

The embodiment hypothesis would explain the positive cross-sectional relationship between the growth rate of GNP and the share of investment in GNP that is shown in Table 4. The relationship is particularly strong for the developed economies. Probably coincidentally, the coefficient on the investment share in the regression for the industrial market economies is close to standard estimates of the return to capital.

Denison attributes a slightly larger share of the growth of U.S. output per worker to increased education than to capital accumulation. The Denison calculations start from a labor input series in which weights were assigned to workers with differing levels of education.¹¹. The weights are based on earnings by educational level, adjusting as far as possible for differences in ability correlated with education.

Estimates of the returns to education in the United States are similar to those for physical capital, with the possibility that social returns may be higher.¹²

Despite a long tradition emphasizing increasing returns to scale in the growth process, their role has been difficult to pin down. Denison emphasizes the potential unreliability of his own calculations. Paul Romer (1986) argues that certain features of the growth process, such as apparently secularly increasing rates of growth, are more consistent with increasing than constant returns. Romer uses a one-sector production function.

Another approach focuses on what are sometimes called economies of agglomeration, or synergy. The notion is that coordinated expansion of many industries is more likely to be self-sustaining than more gradual processes. Suggestive descriptions of the early industrialization process point in the same direction:

. . .change begat change. . . A cheap supply of coal proved a godsend to the iron industry, which was stifling for lack of fuel. In the meantime, the invention and diffusion of machinery in the textile manufacture and other industries created a new demand for energy, hence for coal and steam engines; and these engines, and the machines themselves, had a voracious appetite for iron, which called for further coal and power . . . (Landes (1969, pp. 2-3)).

Rostow’s (1960) takeoff into economic growth has a similar emphasis. It is quite likely that economies of scale of this type are more important at early stages of development than in a large developed economy. Economies of scale may thus be more significant in the growth process in smaller less-developed economies than in the United States. It is in this context that access to export markets can become a significant source of increased growth.¹³ The expansion of international trade is given an important role in promoting growth in the periods 1820-1913 and 1950-73 by Maddison (1982, Chapter 3), though the contribution cannot be quantified.

The neoclassical analysis of growth places little emphasis on the role of government policy. Within that supply-side framework, investment incentives and educational spending could each contribute to raising the level of output; so too can government spending on infrastructure.

The political economy hypothesis that small governments are best for growth receives at best mixed support from the data. Dervis and Petri (1987) show that the share of government spending in GNP is smaller than average for the most rapidly growing middle income countries. Ram (1986) in a cross-section of 110 countries finds a positive partial correlation between the growth rate of GNP and the growth rate of government spending. Less formally, important counterexamples to the view that small governments are best for growth abound: although Japan has relatively low government spending, government plays an active role in coordinating economic strategies and in protecting domestic markets. Government spending in Germany is high. Government intervention in another of Europe’s growth successes, France, has been pervasive. The raw correlation between the share of government in GNP and the growth rate for the industrial market economies is in fact slightly negative; in the developing countries it is weakly negative in Asia and Latin America and positive in Africa.

The role of demand factors in growth has received less quantitative attention, though steady and rapid expansion of markets is often informally cited as an important stimulant to growth. The econometric evidence for the United States implies that expanding demand is a more important factor explaining investment than the cost of capital (Clark (1979)). Expanding international markets likewise serve as an investment-stimulating source of demand.¹⁴ The relation between macroeconomic policy and expanding markets is, however, complicated. In the short run, monetary and fiscal expansion serve to increase demand and stimulate demand-led growth, but once the economy nears full employment, the task of demand management becomes more difficult and supply factors become dominant. However difficult the task of demand management at full employment, there is no doubt that restrictive aggregate demand policies retard growth.

Strategies of Development

The neoclassical supply-side analysis of growth does not emphasize the role of economic policy. Nor does it provide signposts to rapid development, except perhaps in cases of Soviet-style industrialization through massive capital accumulation.

Two basic approaches to growth policy can be detected. One, coming down from Adam Smith, is laissez-faire, seeing growth policy as consisting mainly of reducing governmental impediments to growth. The other is more technocratic, looking to government policy to direct economic development in particular directions.

In the absence of guidance from theory, strategies for growth developed in the post-World War II period have seized on particular aspects of the structures of developing economies or of the growth process as the keys to rapid growth.²⁰ Aside from the laissez-faire strategy of relying on the market, development strategies have focused on the most obvious characteristic of growth spurts, industrialization. They differ according to the methods used to encourage the development of industry.

Import substitution was the first such strategy. Developed in the late 1940s and early 1950s, the analysis of the problem of the developing economies started from their international role as suppliers of raw materials and agricultural products. The terms of trade had and would turn against these goods. Industrialization would not only bring rapid productivity growth but also reduce vulnerability to the swings and trends of primary product prices. The strategy was to develop industry to replace manufactured imports and to reduce reliance on primary exports. It emphasized the development of the internal market rather than exports as the source of demand for the products of the new industries. Domestic industry could be protected by high tariffs, and domestic agriculture, where supply was thought to be inelastic, could be taxed.

Despite rapid growth in the 1950s and 1960s, particularly of industrial output, several Latin American countries ran into difficulties—in the form of external and sectoral imbalances and on the fiscal front, leading to inflation.

Fishlow (1985), reviewing Latin American experience, argues that import substitution paradoxically increased foreign exchange vulnerability. The anti-export bias of the policy reduced the growth rate of exports. Reductions in marginal imports made the remaining imports even more crucial; further, the increasing sophistication of domestic manufactures increased the demand for imported inputs. Increasing current account deficits in the late fifties were solved in part by direct foreign investment, which meant increased rather than reduced dependence on the outside world.²¹

Although the import substitution strategy was developed in the Latin American context, the development strategies followed in much of Africa and Asia were similar.²² The difficulties that resulted on all three continents are not an inevitable consequence of an industrialization strategy. Rather they result from the emphasis on import substitution at the expense of the development of the primary sector and of exports.

Balance of payments difficulties and inflation in Latin America produced several reactions pointing in different directions: that the import substitution strategy should be pursued on a larger scale through a Latin American free trade area; that increased official capital inflows could solve both the government budget and foreign exchange problems; and that structural reform reducing foreign dependence and increasing the role of the state was needed. The free trade area solution brings the hard-to-quantify benefit of expanding markets that growth analysts have long emphasized. Nevertheless, the proposal has repeatedly failed to leave the ground, despite recent stirrings between Brazil and Argentina. Increased aid appeared for a while in the late fifties and early sixties, but was not a permanent solution. The third approach, closing the economy off further, was not attempted. In Africa, where the problems appeared later, aid inflows did increase, but fell far short of amounts required to maintain living standards.

The import substitution strategy is widely regarded as a failure, but manufacturing output in fact grew rapidly, on average 6 percent a year, in Latin America in the 1950s and 1960s.²³ Development thinking and strategy in the 1960s continued to emphasize industrialization, but with greater attention to constraints.

Economic Structures and Models

Broad guidelines are useful, but the effects of alternative adjustment and growth strategies have to be examined for each economy individually. In this section I first discuss common problems facing different groups of countries and then briefly advocate the use of economic models in economic policymaking, for both national governments and international institutions.

Both the heavily indebted and the primary producing countries face particular difficulties in the present world economy—and those which are both have an even harder task. Among the debtors it is convenient to separate countries into African and Latin American, equivalently those whose debts are primarily to official and private lenders, respectively, and to a significant extent, those which are low income and middle income, or less and more industrial.

The debt problem dogs economic policymaking in the debtor countries. The debtors face a choice among accepting and attempting to adjust to their current debt burdens and levels of net capital flows, treating the current difficulties as temporary and attempting, if necessary by arrears, to obtain more financing, and trying to reduce the debt burden through negotiation with creditors. Whether debtor governments have the internal political strength to enforce the radical reductions in living standards that the first strategy requires in several countries is not known. The miracle so far is that the debt crisis has seen a movement toward rather than away from democracy.

Unless the debtor countries themselves, the international institutions, and responsible governments bring pressure for a longer-term solution than the constant renegotiating that has been the pattern of the last five years, adjustment strategies will be difficult to map out. A similar strategic problem faces the creditor banks and the debtor countries: to the extent either makes a serious adjustment effort—in the banks’ case the creation of loss reserves, on the debtors’ side, macroeconomic adjustment—it weakens its case at the next bargaining session. One way or the other, it is important to resolve the uncertainties over the treatment of the existing international debts.

Whatever the treatment of the existing debts, one lesson of the current crisis will ease the task of policymaking. It is now clear that floating rate bank lending should be avoided for anything other than short-term balance of payments financing. Capital inflows can certainly contribute to development, but they should take the form of direct investment, or equity investment, or long-term official capital. Here the role of the international organizations comes to the fore: they will be more important sources of external development finance than they were in the 1970s.

The debtor countries cannot in the next few years look forward to any significant private inflows of funds and will have to plan their growth and adjustment strategies accordingly, with details of course differing from country to country. This is easily said, but the burden on low-income debtors, particularly in primary exporting sub-Saharan African countries with low-income levels, is severe. Their hope must be for a recovery of commodity prices; their strategies will have to be based on the assumption that those prices will rise little in the next few years. These are the countries where the removal of remaining domestic distortions, especially between agriculture and urban economies, may be extremely important, and where positive policies to increase agricultural output with the aim of attaining food security are possible.³⁴ They are also the countries where official aid and the policies of the international organizations can have a large effect and where improved weather has already helped.

The problem for most countries is to make needed short-run macroeconomic adjustments while both macroeconomic policy and longer-run supply-side policies encourage growth.³⁵ The short-run macro adjustments have to focus on the basics: the government budget, the balance of payments and the exchange rate, wages, and monetary policy. Both the balance of payments and the need for investment suggest tight budgets, with the exception that productive government investment spending should not usually be cut back. There is likely here to be a tradeoff between growth and the distribution of income, as governments may have to cut back on subsidies and transfer payments as a major part of a budget balancing strategy.

The wage problem, or more generally the problem of adapting the standard of living to reduced resource availability, is a key to both current account and inflation adjustment. Generally, the adjustments will imply an initial cut in real wages, followed by later increases as productivity rises. If there is indexing of wages, it will be necessary to adjust the base wage downwards. Incomes policy may take its place here as a means of reducing adjustment costs to the new structure of supply and demand.³⁶

The problem of adjusting without causing a recession is difficult, but a successful devaluation that begins to shift resources into exports, the use of incomes policy, the avoidance of excessively high interest rates and continued government investment spending can help. Fund adjustment programs that recognize the effects of inflation on the government budget, and that recognize government investment³⁷ as a source of growth will require less rigorous and more believable budget adjustments than was the norm in the past.³⁸

To the extent that countries are also attempting to adjust from high inflation, they can look to the Argentinian, Israeli, and Brazilian stabilizations of 1985-86 for lessons. The primary lessons are that it is possible to disinflate without causing a recession—and, at least as important, that monetary and fiscal policy and wages have to be consistent with the new non-inflationary equilibrium.

At a general level, the long-run growth policies are to encourage investment in both physical and human capital, to remove distortions by liberalizing, to secure the economy’s agricultural base, and to encourage the development of industry in part through export promotion. Details, which are all that matter, have to be left to a careful analysis of the structure of each economy.

Such analyses draw on explicit or implicit economic models. Both the Fund and the World Bank in their consultative and lending roles have developed models for individual countries and groups of countries.³⁹ The models need to be developed further for use in policy analysis, both by the international institutions and the countries themselves. The alternative to using models is to rely on general principles—for instance, that so long as policy is sound, market adjustment will produce satisfactory economic performance. The latter principle is right in the long run, but it is nonetheless advisable to use explicit models to try to anticipate difficulties that may arise in the adjustment process.⁴⁰

The likelihood that the economy will achieve targets derived using models is of course slight. Policy mistakes or other shocks will produce deviations of outcomes from target levels. But the models enforce discipline on the analysis of adjustment, ensure the consistency of policy measures, and provide a framework in which the prospects of meeting balance of payments and growth targets can be coherently discussed. The target paths derived using models provide a baseline from which to judge events and to justify deviations that will inevitably be needed.

Lessons for Policymakers

The previous sections have summarized the conclusions of studies of long-run economic growth, discussed alternative growth strategies, and touched on the problems of adjustment and growth that now confront primarily African and Latin American countries. In adjusting and mapping out strategies, countries need to use models that connect their targets with their policies.

The approach suggested by this paper is eclectic—the fundamental reliance is on markets, but government policies make a difference for good or ill. But, as has been emphasized, success or failure depends on the details as they are developed for each country. There policy economists, both internal and external, can play an important role.

Keynes hoped economists would eventually attain the humble status of the dentist. The development economist active in policymaking is more like a doctor, called on to diagnose diseases and prescribe cures (always aware that the medicine may not be taken, and if taken may have unexpected side effects) for illness and sometimes to produce good health. Although the modern doctor is an impressive technician, many prefer the old-fashioned GP, whose experience and intuition more than compensate for the latest in technique.

Harberger (1984) has summarized the pragmatic lessons professionals have learned from studying and advising on economic policy in developing countries—and most apply also to developed countries. Here are the most important of his lessons:

1. Keep budgets under adequate control. Absolute balance may be impossible, but there is not much room for deficits.

2. Take advantage of international trade.

3. If tariffs are used, keep effective tariff rates reasonably uniform. If tariffs become excessive, use export incentives.

4. Use price and wage controls sparingly if at all.

5. It is rare that quotas, licenses, and quantitative restrictions are justified. They are inefficient and breed corruption.

6. Adopt a technical view of public enterprises. The right criterion is the efficient operation of the enterprise.

The missing, most important, commandment is:

7. Don’t let the exchange rate become overvalued.

This list is evidence not only of what should be done, but also of what often is done. The domestic distortions and problems that new policies have to cure were usually introduced within the same political system that now has to be reformed. This is not the place to expand on the political economy of reform,⁴¹ except to note that the outcome of policy packages depends not only on the economics but also on the political equilibrium.

There are two reasons for hope. One is that deep-rooted distortions are most likely to be eliminated at times of deep crisis. The other is that well-designed conditionality of outside lending can provide incentives for the adoption of better policies.