Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do lower population densities lead to lower per capita incomes via a reduced stimulus to technological innovation and reduced exploitation of economies of scale in production and infiastructure? Ihe proposition that rapid population growth and greater population density lead to higher per capita incomes has been advanced for both ancient and modem economies and both developing and developed countries. It can also be argued that the possibilities of realizing any economies of scale Trough international trade and of adapting technology developed by developed countries would make this proposition false for contemporary developing counties. This discussion distinguishes between manufacturing and agriculture and between changes that make it possible to use factors in different proportions and changes that yield more output for factors used in given proportions. It should be noted that Here is an important difference between Be proposition that greater density is beneficial and the proposition Tat rapid population growth is beneficial. The beneficial influences of greater density, if Hey exist, may accrue slowly over time; in the short and medium run, they may be overwhelmed by the costs imposed by rapid population growth. And it may sometimes be possible to capture the benefits of density by concentrating the existing population in a smaller area. In Denison's (19743 calculations for the United States, changes in all measured economic inputs account for only about 50 percent of total economic grown and 20 percent of growth in per capita income. The remainder Denison classifies as due to advances in knowledge, economies of scale, changes in the efficiency of the allocation of labor, and some additional minor categories. Ihus, in order to understand the consequences of population growth, one must understand the influence of those factors on He grown of per capita income. 47
48 MANUFACTURING POP ULA~ION GROWTH AND ECONOMIC DEVELOPMENT We begin with the question of whether larger populations may lead to economies of scale in manufacturing. Empirical research has shown that economies of scale in the provision of infras~ctural services do exist at the city level and that they are weakly associated with the overall size of the city. They are much more strongly associated with the local size of the particular industries in which the city specializes; these economies of scale are called "localization economies" (Henderson, 1985~. Localization economies, which are exhausted above moderate city size, arise from several sources. The easy and rapid exchange of information within an industry facilitates the adoption of new technology, and possibilities of specialization of tasks within He industry emerge. There is also the possibility of drawing on a larger experienced work force. The nature of these economies suggests that national size and population density should be of little relevance, except insofar as they provide a large enough market to allow the industry to reach a sufficient size within at least one city (Henderson, 1985~. It is also thought that these localization economies occur primarily for goods that are produced using technology (imported from developed countries) with relatively high capitalllabor ratios and that such goods are primarily demanded by He wealthiest segment of Be populations in He developing countries. Modern consumer durables are examples of such goods. Thus, the existence of manufacturing economies of scale in developing countries may occur principally when the income distribution is quite unequal. Labor-intensive manufacturing for a low-income mass market might well not experience such economies of scale (James, 1985~. Economies of scale at the national level may also occur. Denison (1974) concluded that for the United States over the period 1929-1969, economies of scale contributed a little more than 10 percent to the increase in income per worker and that their contribution was substantially greater Han the contribution of increased capital per worker and nearly as large as He contribution of education. However, the empirical and methodological basis for He calculated scale returns was very weak. Aside from economies of scale for a given technology, might greater population size or density lead to more rapid technological progress in manufacturing? Without a local capital goods industry, there may be less demand for locally produced technological progress. Larger economies are far more likely to support a local capital goods industry and are therefore more likely to generate indigenous technological progress (James, 1985~. Likewise, it has been argued that the rate of technological progress will be positively affected by the number of researchers, which will increase with the size of the total population (Phelps, 1980; Simon and Steinman, 1981~.
TECHNOLOGIES INNOV~ON ~ ECONOMIES OF SCOW 49 But these arguments are generally advanced at the global level, and one must ask why a developing county would generate its own technological progress rather than importing technology from developed countries. One answer might be that die technology from developed countries uses much more capital per worker than is appropriate in labor-abundant developing countries. However, it is possible that the imported technology is in fact more efficient in the sense Hat the productivity of both capital and labor is higher, in which case the argument is undermined. Countries with larger populations and therefore larger markets should, in principle, be better situated to develop technology appropriate to local factor proportions. The argument has been made in great detail that technological change in England- in contrast to the United States-was historically guided in the direction of saving capital rather Han labor, reflecting the relative scarcities of factors in these countries (Haba~uk, 1962; David, 1975~. In many contemporary developing countries, however, a variety of government policies distort local factor prices so severely that there is no incentive to develop techniques that exploit the relative abundance of labor. Such policies include overvalued exchange rates and credit policies that favor industry and artificially reduce the price of capital goods, particularly those imported from developed countries. It should be noted, however, that the markets for which final production is targeted may influence the choice of production technique. For example, goods for high-income consumers, for use in the modern economic sector of developing countries, or for export to He developing countries may require relatively capital-intensive production techniques to ensure high levels of standardization or quality (James, 1985~. For these reasons, most local technology appears to mirror the factor proportions of developed countries, with a few East Asian countries being He impressive exception (lames, 1985). Similarly, higher national population densities do not appear to offer manufacturing any advantages arising from reduced costs of infrastructura1 investment since what makers is not national density but rather urban concentration, which today does not depend on overall density, although it might have in He past (Boserup, 1981~. Based on existing research, there is little reason to expect technological progress in manufacturing to be favorably affected by greater population size or density in any individual country, particularly given government policies that are now common. James (1985), in a cross-national sample of developing counties, found no statistically significant association between the rate of industrial labor productivity growth (which reflects changes in capital, He scale of production, and technology) and He population grown rate from 1960 to 1970. Research in this area is in its infancy, and it would be premature to rule out altogether the possibility of positive effects.
so AGRICULTURE POPUl~ON GROWTH AND ECONOMIC DEVELOPMENT Because agriculture is dispersed and carried out in small units under varying agroclimatic conditions, there are no national economies of scale operating through the size of production units. However, it has been plausibly argued that greater population density on agricultural land should favorably influence infras~uctural investment in transportation, communications, iraga- tion, markets, agricultural extension stations, repair shops, and so on, since with denser populations each location can serve more people and therefore have lower per capita costs (Boserup, 1981; Simon, 19773. Limited empirical work supports these arguments (Boserup, 1981; Simon, 1975; Glover and Simon, 1975; but see Evenson, 1984a, who sometimes finds negative effects). Without such infrastructure, the transfer of modern agricultural technology is difficult. For example, irrigation becomes profitable only after a certain density has been reached (note, for example, the failure of many irrigation schemes in suW5aharan Africa), but in some counties, it is a prerequisite for adoption of many modem, high-yielding plant varieties. Consequently, lower density agricultural populations that have not in the past invested in irrigation may be less well situated to benefit from new agricultural technology (Hayami and Ruttan, 1985; Pingali and Binswanger, 1985~. It also appears to be even more true in agriculture than in manufacturing that each region needs its own technology, suited not only to the relative supplies of land, labor, and capital, but also to He local agroclimatic conditions. Therefore, local research and development are even more important for agriculture than for industry. More densely settled populations would appear to generate a greater demand for such local research and be better able to fund it (Pingali and Binswanger, 1985~. Very little empirical research has been done, but Evenson's (1984b) analysis of data from northern India does not support this argument. It is useful to distinguish three different aspects of technological change: first, change in the basic agrarian system in use, with each system employing factors in different proportions; second, the direction of technological change, in the sense that new knowledge tends to economize on either land or labor within any given agrarian system; and third, the pace of technological change. Examples of agrarian systems, running from least to most intensive use of labor, include forest fallow, bush fallow, grass fallow, annual cropping, and multicropping, each win its distinctive tools and over features. Agrarian systems using the plow are often viewed as more advanced than slash and burn, but they are not necessarily so: each system is most efficient at a particular population density, and farmers have been observed to switch from plow cultivation to slash and bum when density has declined. As He population in an area becomes denser, more labor-intensive systems are adopted if the population changes occur slowly enough so that the appropriate
TECHNOLOGICAL INNOVATION ANZ) ECONOMIES OF SCALE 51 complementary investments in infrastructure can be made and institutional changes can occur (Rosenzweig et al., 1984~. Even if technology does not change, returns to labor appear to decline at most quite gently with increasing density, far more gently than they do when labor inputs are increased within any single system (Boserup, 1965, 1981; Pingali and Binswanger, 1985; Hayami and RuDan, 1985~. Declining returns to labor are often offset by increased hours of work. Within any agrarian system, it may be possible to substitute animal or mechanical power for labor, and irrigation, fertilizer, or new seed varieties for land. Ibis substitution is often effected by technological change. As noted in our discussion of renewable resources, research reveals a close association between population density and the labor intensity of technology, an association that is consistent with the view that He substitution has been historically quite responsive to differences in relative availabilities of land and labor as reflected in their prices (Hayami and Ruttan, 1985; Pingali and Binswanger, 1985~. Thus, although the evidence is not conclusive, population growth and density apparently play an important role in directing technological change in agriculture, in contrast to manufacturing, for which technologies and their labor-saving bias typically seem less appropriate to labor abundance (James, 1985~. Unfortunately, it is difficult to determine empirically the direction of causation leading to observed associations. And, as in manufacturing, government policies governing prices of inputs and outputs also exert an influence, but in agriculture these are less dominant. This review of the evidence suggests Hat population density strongly influences the choice of agrarian system and the direction of technological change. But there is nothing in these arguments to suggest that denser or more rapidly growing populations are better off; rather, they show that the choice of system and direction of technological change typically adjust to the negative effects of higher density and more rapid grown. These results raise the question of whether countries with denser populations generate a pace of technological advance that gives them a net advantage over those with less dense populations. Were are a number of reasons why technological progress in agriculture might be more rapid in denser populations. Some of these derive from the infrastructural advantages of denser populations, which facilitate the flow of infonnation about new technologies and, by increasing He possibilities for marketing output, also increase the gains to, and incentives for, raising productivity. Denser populations may also be better able to bear the fixed costs of agricultural research relevant to an area's particular conditions. If such positive influences on technology do occur over some range of densities, there might well be a point beyond which furler increases were disadvantageous. Analyzing a cross-sechon of 45 developing counties, James (1985) found
52 POPUl~lON GROWTH ANl) ECONOMIC DEVELOPMENT that the rate of grown of labor productivity in agriculture between 1960 and 1970 was significantly positively related to the national population density in 1965, while productivity gains in manufacturing were not. While this finding is based on a very rough analysis and should be regarded as vely.preliminary, it is consistent with the view that density has nonnegligible positive effects on technological advances in agriculture. However, EvensQn's (1984b3 study of northern India found the negative effects of density on production grown to be about twice as great as the positive ones. A number of other studies have examined the relationship between population density in developing countries and the rate of growth of output per capita or per worker, but very little can be concluded from them. They all report a positive association for at least some range of densities (Lefebvre, 1977; Simon and Gobin, 1980), but in some studies He association becomes negative after densities of about 100 persons per square kilometer are reached. CONCLUSIONS In manufacturing, economies of scale exist principally at He urban level and are exhausted at a moderate level of city size. Hence, there is no significant relation between national population density and economies of scale. Moreover, given a widespread dependence on imported technology and He existence of international markets for many manufactured goods, national population density offers little stimulus for technological progress in manufacturing. We therefore conclude that slower population growth would not have any negative effect on productivity in manufacturing. In contrast, we find that the choice of agricultural technique is responsive to population density. Economies of scale in agriculture are also likely to occur, especially by spreading fixed costs in infrastructure and research over a larger number of people. Although there is no conclusive evidence on this point, Were are probably more direct methods for stimulating research and development. It seems unlikely that the stimulative effects of increased population density on agricultural productivity could more than offset the effects of diminishing returns to labor (discussed under Question 2~. Thus, for most developing countries, slower population growth is unlikely to result in a net reduction in agricultural productivity and might well raise it.