l 4440 M. -_E_C_4 t 14qS ROUNDTABLE DISCUSSION Employment and Development T he panelists for this roundtable discussion were Nancy Birdsall, executive vice president, Inter-American Development Bank; Paul Collier, professor of economics at the Centre for the Study of African Economies, Oxford Uni- versity; Richard B. Freeman, professor of economics at Harvard University and asso- ciated with the National Bureau of Economic Research and the Centre for Economic Performance at the London School of Economics; and Christopher A. Pissarides, professor of economics at the London School of Economics and Political Science- The moderator was Michael Bruno, chief economist and vice president, Development Economics, at the World Bank. Nancy Birdsall on Growth, Inequality, and the Labor Market Labor economics is at the heart of development because it is where the economics of growth and the politics of inequality meet. I hope that World Development Report 1995 on labor will advance the way we think about the relationship between growth and inequality. With this in mind, I want to state and explain three preju- dices of mine about public policy and labor market institutions in developing coun- tries. I wish I could call them propositions, or at least testable hypotheses, but they are based largely on crude generalizations. They are meant to provoke thinking and help shape new research questions-especially about the political economy of labor markets and inequality in developing countries. Perhaps new work, including the World Development Report, will vindicate them in some form. Prejudice 1: In the medium run (five years or more) labor market institutions are largely irrelevant to the welfare of the working dass. What matters is nonlabor poli- cies; what matters is the pattern of growth. Labor did well in East Asia between 1960 and 1990. Wages increased as fast or faster than GDP, nearly doubling or more in absolute terms during the 1980s (Fields 1993), while unemployment declined and the mix of jobs improved. This improve- ment occurred in Hong Kong, where unions rere tolerated but remained weak; in Japan, where unions were strong and recognized by government; and in Singapore, where the government deliberately and successfully repressed labor and wages in the Procedigs of the World Bank Annal Conferece on Devdopment Economics 1994 01995 The International Bank for Recnstruction and Development ! THE VORwD BAKN 439 440 Roundtable Discussion on Employment and Dreelopment 1970s but abandoned wage repression in the 1980s as labor shortages emerged. Wages grew rapidly in the Republic of Korea, too, where the labor movement was repressed throughout most of the 1980s (for political not economic reasons) but not wages-wages rose largely in line with productivity and GDP growth. What was common across East Asia was not the approach to labor institutions (which, of course, are not confined to unions but include minimum wages, tripar- tite mechanisms, severance arrangements, unemployment insurance, and the like) but the approach to growth itself and to the idea of Lshared growth" (World Bank 1993). Beyond macroeconomic stability, a necessary condition for growth was an emphasis on nonlabor policies and programs for ensuring that the working dass, the nonelite, would benefit from growth. East Asian governments used many different mechanisms to encourage the sharing of growth: land reform in Korea and Taiwan (China), generous subsidies for housing in Hong Kong and Singapore, credit tar- geted to small business and massive investment in rural infrastructure in Indonesia, Malaysia, and Thailand. But two policies can be singled out as favoring labor: the emphasis on exports, which ensured high demand for labor and drove up employment and wages, and heavy public investment in universal basic education, which supported increasing labor productivity of the initialy unskilled and, thus, productivity-based wage growth that was broadly based and fally shared. Widespread basic education also encouraged employer-financed training by reducing the costs and raising the retums to such training. In one of many of the virtuous circles of growth, reduced inequal- ity, faster accumulation of human capital, and productivity-based wage growth fed back to demand for education, ensuring the future supply of more educated and trainable workers and thus continuing gains in labor productivity and wages (Birdsall and Sabot 1994). These three fundamentals-macroeconomic stability, exports, education-matter much more for labor's welfare than unions or their repression, minimum wages or their absence, or the degree of nonwage costs The converse also seems to be painfully true: labor institutions cannot insulate workers from wage declines when the fundamentals are all wrong. Declines in real wages in Africa and Latin America, even in countries where labor institutions have been relatively strong, illustrate this vulnerabiity. Preudice 2: Labor market policies and institutions are a perverse vehide for pro- gressive social policy, especially in developing countries and most especially in developing countries (like most in Latin America) with unequal distributions of income, political power, and privilege. In Latin America and elsewhere there has been a conflation of labor policy with social policy. The political left, in a tradition rooted in European conditions described by Marx, has defined social justice as fundamentally an issue of workeds rights or entitlements. This tradition is alive and well in much of Latin America. An example: the social rights section of Brazil's 1988 Consitution devotes one four-line paragraph to education, health, and social security, and four pages with thirty-four numbered items to such (constitutionally guaranteed) entidements as protection against layoffs Nancy Birdsail on Growtb, Inequality. and the Labor Markit 441 without just cause (which becomes a matter for the courts), the right to severance pay, the right to a maximum forty-four-hour work week and eight-hoalr day, and the right to job security in the face of automation. An entirely different and equally detailed article specifies the rights to union representation-also in the section enti- tled social rights. Ideas do make a difference. The idea that social justice can be attained by guar- anteeing certain entitlements to the wage worker seems to be a bad one, and one worth challenging. The resulting labor institutions have probably had efficiency costs in many Latin American economies. Cox-Edwards (1993) points out several such costs-for example, the chilling effect on the entry and growth of high-tech- nology industry of severance pay obligations that increase with the tenure and expe- rience profile of a finm's labor force. The efficiency costs have probably been higher in the public sector since these entitlements cannot be effectively enforced in the private sector. In some countries job security entitlements in the public sector make civil service downsizing legally impossible and have created costly resistance to the privatization of public enterprises. But worse than any efficiency costs are the probable costs in reinforcing and exacerbating inequality in Latin America's societies. There are two main reasons for this. First, the truly underprivileged are outside the formal sector-women working as domestics, women and children self-employed in low-productivity street busi- nesses, and the landless and otherwise disenfranchised in rural areas. Higher labor costs in the protected sector reduce these workers' chances of obtaining higher-pro- ductivity jobs. Second, emphasis on worker entitlements has distracted government and the well-meaning left from giving attention to the fundamentals of growth- including a labor-demanding growth strategy-that are key to overcoming poverty and inequality. Happily, these ideas are changing. In Chile and Mexico, for example, market-ori- ented governments are trying to reshape labor institutions to be market enhancing while developing complementary social policies that do not rely on labor institu- tions alone to address the structural problems of poverty and inequality. Prejudice 3: A laissez-faire, let-the-market-work approach is not good public policy. Why should government get involved at all? One reason is that societies may want to collectively agree on enforcing certain standards, such as prohibiting child labor and ensuring minimum safety in the workplace. This is true even though there is a tradeoff between mandated standards and the efficient working of incentives. Laws in some Latin American countries that guarantee a thirty-day vacation seem to me to reflect the wrong tradeoff but so does the absence of any minimal safety stan- dards. As the NAFTA debate in the United States showed, the issue of standards is now central to intemational trade, as well as to domestic labor policy. Another reason for government involvement is that ensuring a competitive mar- ket, in labor as in other areas, often requires some government intervention. The East Asians were not laissez-faire in their approach; they managed the labor market to minimize monopoly rents and to ensure that the market was competitive. In Latin America government is not reducing monopoly rents but rather is an instrument of 442 Roundtable Discussiun on Employment and Development such rents. Government institutions are also often the victim of strong public sector labor unions-as in Venezuela, where 80 percent of unionized workers are in the public sector. A third, and most important, reason for government to get involved is that healthy labor institutions, including democratic unions, can ease the difficult process of structural adjustment and the often necessary sharing of the costs of adjustment. The costs of adjustment have not bcen shared in Latin America-the elite has too often managed to insulate itself This brings me full circle to the politics and economics of inequality-where I hope new work on labor will start. Chile's recent reforms of labor institutions pro- vide an interesting model, combining a tripartite discussion on the guidelines for wage increases in an agreed framework that says the minimum wage should be increased in line with economnywide productivity gains and projected inflation. The government refuses, however, to engage itself in any bilateral collective bargaining. This kind of limited government intervention, which enhances rather than under- mines the labor market, seems to be the right direction to take. I look forward to continued discussion and experimentation at the country level to further our under- stnding in this important area. Paul Collier on African Labor Markets African economic performance is problematic. Indeed, on plausible scenarios, Africa has more person-years of future poverty than any other region. To what extent might failures in the labor market account for this? Posed more bluntly: if Africa had East Asian-stle labor markets, with an emphasis on education and flexibility, would its performance start to converge with that of East Asia? I first discuss two models. One analyzes African lablr market failure, the other is ostensibly orthogonal but, I suggest, potentially more pertinent I then turn to the political economy of African labor markets, arguing that an underlying feature is aversion to the high risks currently endemic in African economies. The Harris-Todaro Model Only one really well-known model has come out of the analysis of Africa, and it hap- pens to be about labor market failure. The Harris-Todaro (1970) model postulates that wages were institutionally rigid at a high level and that this rigidity induced high unemployment through a jobs lottery. This model has proved to be a false start. Wages are not institutionally rigid, and labor markets are not job lotteries. Timne-series data on real wages in Africa have shown ar astonishing degree of down- ward real wage flexibility as inflation has eroded real wages (Jamal and Weeks 1993). The exception is the franc zone, where price stability impeded labor market adjustment, something that the recent devaluation will correct But even in the franc zone real wages have been flexible. One test of wage flexibility is the cross-section relation between local wages and local unemployment rates. In North America this relation is Paul Collier on Afican Labor Markes 443 negative: high unemployment depresses wages. The Harris-Todaro model predicts that this relation should be positive in Africa: high wages attract job seekers In fact, the rela- tion is negative -ven more negative than in North America (Hoddinott 1993). African labor markets are highly flexible; unfortunately, they have needed to be so. Nor are African labor markets a lottery. The context that inspired the Harris- Todaro model was Nairobi in the late 1960s, where unemployment was indeed high. But this labor market should be subjected to closer scrutiny. Unemployment was con- centrated among secondary school leavers, a pattern reflecting the abrupt change in secondary education policy from famine to feast at the time of independence. The stock of secondary school-educated labor was expanding by 20 percent a year. Wage adjustnents could not fully equilibrate this massive supply shock, and firms became more selective, discriminating in terms of such characteristics as examination results. In 1968, for example, the group that had failed the examination had the highest unemployment rate. But they learned: four years later those who had failed the exam- ination had the lowest unemployment rate because their expectations had adjusted to the new reality (Collier and Lal 1986). Note that firm screening in terms of exami- nation performance and school subjects does not induce an inefficient signaling response: students simply try harder and study more useful subjects. This result can be contrasted with the bumping model (Fields 1974), in which firms screen only by years of education, which induces a socially cosdy and explosive growth in average years of schooling. In Kenya the urban unemployment rate dropped to less than 6 per- cent, typical of rates elsewhere in Africa with a few exceptions. To summarize, whatever might be the problems of African labor markets, high unemployment is not among them: markets are equilibrated by both iages and selection criteria. Yet this flexibility does not seem to have alleviated the problem of poor economic performance. Employment in a High-Risk Environment I now turn to a model that is neither inspired by an African context nor is about the labor market, but that I think has more to offer. The model (Dixit 1989) concerns the relation between fixed investment and risk. It shows why even agents who are not risk averse have an incentive to defer investment in a high-risk environment. By retaining liquidity, asset holders gain by keeping their options open. This model is directly pertinent to why Africa has failed to attract investment despite reform of incentive structures. Risk is an important deterrent to investment in Africa not only because there is a high incidence of policy shocks and trade shocks, but also because these shocks occur against a background of slow growth. Large shocks and slow growth imply that investments can easily get beached, whereas in high-growth economies investments hit by negative shocks are soon likely to be bailed out by generalized expansion. Over the past decade foreign direct investment to develop- ing countries has tripled, yet Africa's share has fallen to negligible levels. This neglect has not been offset by domestic investment. The investment failure is in part accounted for by the high-risk environment. 444 Roundtabke Discussion on Emptoyment and Development This relation has a direct corollary for African labor markets, because labor is the predominant mechanism for resource mobility. When macroeconomic reform changes relative prices and newly profitable firms are reluctant to invest, changing employment becomes the only way to achieve an output response. However, I think that the main importance of the investment-under-risk model for African labor markets may be in the form of a simple extension. Many employ- ment decisions have an investment component, and so a high-risk environment dis- courages such employment just as it discourages fixed capital formation. I offer three examples of the investment content of employment. First, there is evidence for industrial countries that hiring and firing decisions are lumpy because they have fixed costs (Caballero and Engel 1993). These fixed costs represent an investment. Second, a standard result of labor economics is that firms must finance the accu- mulation of firm-specific skilis. Third, in agriculture the long lags between employ- ment and output and the high risks in that cutput give labor hiring the risk-taking characteristics of investment. If indeed the high-risk, slow-growth environment discourages investment-inten- sive employment, this relation affects the character of labor processes. In the formal labor market it induces slow growth of employment. This conclusion is consistent with African evidence: formal employment accounts for a small and diminishing share of the labor force: 12 percent in 1980 and 9 percent in 1990 (Mazumdar 1994). Within the formal sector the discouragement of investment-intensive em- ployment induces a greater reliance on casual employment and a slow pace of skill accumulation Evidence from Zimbabwe shows firms employing a relatively small core of long-term workers in whom they invest and a large group of casual workers (Velenchik 1994). There is evidence that rates of return to education are falling despite the fact that the African labor force has little education. This may be hap- pening because the casualization of the labor force does not provide a context in which education can be well harnessed. Enrollment rates in schools are dropping in many parts of Africa, probably due to falling returns. In agriculture the problem manifests itself as too little labor hiring (Collier 1989). In most of eastern and southern Africa less than 10 percent of the labor force on smallholdings is hired. Yet land-labor endowment ratios on holdings are diverging. Land has ceased to be abundant and is overwhelmingly acquired through inneri- tance rather than purchase or tenancy. Labor transactions are insufficient to equal- ize factor returns across African farms of different sies. Hence, within rural Africa tbn same analytic question arises as in the debate on globalization and wages: to what extent can specialization in the goods market substitute for a labor market? Not only does the high-risk agricultural environmient reduce labor hiring, but there is also a feedback from the resulting thinness of the rural labor market to increased risk. Agricultural households facing adversity do not have reliable access to wage labor as a safety net. Nor does the other obvious rural risk-reducing strategy, diversification, appear to work very well. In rural Asia there is a dear negative relation between income and diversification: households choose to buy security at the price of income. Yet in Africa, where exposure to risk is probably greater, there is ^ positive relation PIul Collier on African Labor Markets 445 between income and diversification (Krishnan 1994). Poor households at, beast able to diversify because there are entry barriers to nontraditional activitims in thc form of capital and information (Bevan, Collier, and Gunning 1989). The Political Economy of a Risk-Averse Society I now want to suggest that this high-risk environment with limited coping strategies may help account for two political-economy phenomena that are of major impor- tance in African formal labor markets. The common theme is that we would expect households to be highly risk averse. First, risk aversion might help account for the intense popular pressure to pro- vide public sector employment. The share of the public sector in wage employment became markedly larger during the 1970s and 1980s in Africa than in other conti- nents. Under the pressure of this employment expansion, public sector wage rates collapsed. The public sector is no longer lucrative: indeed, it is usually not possible to live on a public sector salary. Yet such employment is safe. In effect, the public sector labor market has become the domain of social security policy, offering a low but secure income supplement to many households. A by-product is that the public sector, despite being very large, is in a poor position to deliver services: the public sector is both too large and too small Second, risk aversion may help account for at least part of the opposition of the formal sector labor force to macroeconomic reform. The centerpiece of such reform has been devaluation. It is conventional wisdom that labor loses from devaluation. Real wages need to fall as part of the reduction in expenditures involved in balance of payments improvement, and this drop is achieved by a failure of nominal wages to keep up with the inflation that devaluation triggers. This conventional wisdom is cor- rect for a devaluation such as that of the CFA franc, an unusual case in the African context More commonly, African governments have already contained their pay- ments deficits through trade restrictions by the time they devalue, so that the deval- uation is trade liberalizing. Relative prices change, but there need be no reduction in real expenditure. Real wages need not fall during trade liberalization; the losers are those who received quota rents and capital owners in the import-substitution sector. Yet the opposition to reform is typically more widespread, including more or less the whole of the formal wage labor force-a potent force. Part of the reason is that beneficiaries of reform are typically not in the labor market but in the smallholder export sector, which is politically weak in Africa (Bates 1981). But the active oppo- sition of the formal labor force goes beyond this. The import-substituting manufacturing sector naturally tends to lose from trade Eberalization. This loss is exacerbated if the reform is accompanied by foreign assis- tance, since the aid will appreciate the exchange rate (Collier and Gunning 1992). If there was a fully integrated national labor market, workers would not mind this effect: the Stolper-Samuelson mechanism would ensure that all workers gained and that capital owners bore the losses. However, studies of trade policy lobbying in industrial countries have found that workers and firns tend to take the same side: 446 Roundtable Discussion on Employment and Deuelopment rents are sufficiently shared that interests are common. In Africa this seerns to be less the case. For example, in the Zimbabwean trade liberalization, manufacturing firms were instrumental in lobbying for the liberalization, but employee organizations were vocal opponents. The manufacturing sector in Zimbabwe is largely import substituting, but some firms anticipate becoming substantial exporters. In the labor market this pattern implies both net labor shedding and a reallocation of workers as import-competing firms contract and exporting firms expand. Freeman (1993b) has modeled gainers and losers in such a process, but in his model there is no risk aversion: at the announcement stage all workers support reform because positive income changes are expected. More commonly there is acute opposition at the announcement stage because workers correcdy anticipate a greater risk of job loss. The hope is that as the reform proceeds workers in the expanding export sector constitute a lobby of gainers to counter this perception of enhanced risk. Whether this group of reform promoters emerges depends on whether workers in the firms that become more profitable share in the benefits. Recent results for Ghanaian manufacturing (Jones 1994) show that, as in industrial country labor mar- kets, wages are higher in firms that are more profitable: there is rent-sharing. This relation is true only in the upper echelons of the labor market. For production workers there is ilo relation between profitability and earnings. Production workers in export manufacturing have no reason to counter the opposition to reform that will come from production workers in the import-substituting sectors. The con- junction of an efficient labor mariker for production workers within the manufac- turing sector and a failure of the labor market to encompass the bulk of the labor force produces a bias toward opposition to liberalization. Manufacturing workers either lose or are indifferent, whereas the major export sector of most African economies has few formal wage employees. The other major group of workers opposing reform is in the public sector. Ordinary public employees are unlike;y to lose direcdy from a trade-liberalizing devaluation, since they are not usually the beneficiaries of quota rents. However, public sector retrenchment is generaUly induded as part of the adjustment package. A political by-product of this packaging may be that groups that lose from logically distinct polircy changes can find common cause in opposition. Public sector retrench- ment is a dassic example of a reform in which the gains are so diffused that nobody realizes that they have benefited, whereas the risk of job loss creates a well-defied group of losers. The high degree of opposition that reforms have encountered in Africa has in turn provided a feedback into the high-risk environment. The public does not agree on the reforms, and so there is a significant risk of policy reversal. Conclusion I started with a question of whether Africa, if it had more flexible labor markets and a more educated labor force, would start to perform more like East Asia. I have sug- Richard & Freeman on the Global Labor Market 447 gested that neither rigidities of the labor market nor shortages of educated labor are central to African problems. African labor mr-cets are remarkably flexible. But a high-risk, low-growth environment is not conducive to high returns to human cap- ital. Rather, it is conducive to the politics of risk aversion: large, badly paid public sectors and opposition to reform. Africa cannot become East Asia by directly emulating its labor markets. Rather, its labor markets will start to look like East Asia's only when African societies suc- ceed in producing a convincing consensus committed to the maintenance of reform. Richard B. Freeman on the Global Labor Market The following developments are, arguably, bringing labor markets closer together, leading to a global labor market with consequences for workers in industrial and developing countries: * The rapid growth of employment in industry and manufacturing in develop- ing countries-from 136 million workers or 43 percent of the world total in 1970 to 219 million workers or 53 percent of the world total in 1990. * Inaeased education in many developing countries that in part underlies the growth of employment in industry and that makes an increasing share of the world's work force able to do similar tasks. Mean years of schooling in devel- oping countries rose from 3.4 years in 1970 to 5.3 years in 1990. With 75 percent of the world labor force in developing countries, the increase in mean schooling implies millions more workers with relatively high levels of education. • The growth of trade in manufacturing between developing and industrial countries. Through trade, manufacturing workers in many industries in developing countries have become substitutes for manufacturing workers in industrial countries. Perhaps the most striking indicator of this change is the rising share of manufactures in developing country exports to industrial countries, from 16 percent in 1970 to 53 percent in 1989 (Wood 1994). * The growth of promarket and protrade institutions in many developing economies. The successes of Hong Kong, the Republic of Korea, Singapore, Taiwan (China), and, more recently, Chile and the collapse of communist autarkic regimes have shown that the path to sustainable development rests on market institutions and membership in the world trading community. * Increased immigration from developing countries or economies in transition (whose GDPs per capita would put them in the developing countries cate- gory) to industrial countries. Between 1970 and 1990 the imnmigrant share of the U.S. population nearly doubled, with the bulk of new immigrants com- ing from developing countries. The immigrant share of the population also rose in Australia, Canada, and Western Europe. * The transmission of modern technology through the expanded operations of multinational firms around the wvorld and the education in industrial coun- tries of high-level students from developing countries. 448 Roundtable Discussion on Employment and Development These changes are related in various ways that are susceptible to detailed analy- sis. Here I simply note them as indicators of the growing interconnection of labor in the emerging global labor market. If the globalization of the labor market is important, we would expect to see it show up in changes in the economic well-being of workers among countries. Do have we evidence of such changes? There are two facts that some regard as evidence of the growth of a global labor market. The first is the decline in labor demand for less-skilled workers in industrial countries-a decline that shows up in falling real wages in the United States and falling employment and high unemployment in Western Europe. The second fact is the increased demand for modern sector workers in developing countries that have joined the world economy-an increase that shows up in growing modern sector employment and higher real wages. These developments have the flavor of a world in which factor price equalization has begun to operate in a substantive way. If workers around the world are in the same labor market, workers with similar skills will ultimately be paid comparable wages. Wood (1994) has argued that these two facts are linked not only conceptually but also empirically: falling demand for less-skilled workers in the West is the flip side of the improved position of comparable workers in newly industrializing countries. My reading of Wood's empirical evidence and that of others (Freeman 1993a; Sachs and Schatz 1994) does not, however, support such a sweeping generalization, appealing though it is as a unicausal explanation of ongoing changes. It is no easy task to quantify the effects of labor market globalization on work- ers in industrial or developing countries, but evidence suggests that the magnitude of the effects of trade with developing countries does not come close to explaining the deteriorated position of less-skilled workers in industrial countries. At the same time the evidence does not support the view that trade and immigration have had no effect on wages around the world. Just as trade has benefited workers in some developing countries, it has created problems for some workers in the West, though it is far from the main cause of their economic problems. Most important, perhaps, as modern technology spreads and more developing countries with huge popula- tions (such as China, India, and Indonesia) join the world economy as major exporters of manufactured items, it is hard to imagine that this will not have even greater effects on labor markets than globalization did in the 1980s. To dismiss the potential effects of globalization on the job markets in diverse countries may be to miss the change in the world economy that will most affect those who earn their living from work in the next several decades. Globalization of the labor market gives rise to several policy issues. The first and currently most visible issue relates to the demands by some industrial coun- tries for higher labor standards in developing countries. How will governments and the international trading community respond to the demand for labor stan- dards-minimal modes of treatment of workers-that arise when countries pur- chase products from countries with different levels of output per capita and labor codes? Richard B. Freeman on the Global Labor Market 449 In industrial countries the demand for global labor standards has two sources. One is the desire of consumers that the products they consume be made under decent conditions (Lhis section summarizes Freeman 1994). For goods sold at the same price most consumers, when given a choice between goods produced under decent conditions and those made under slavery, by prison labor, or by ill-treated workers, would choose the products produced under decent conditions. Many would even be willing to pay more for those products, up to a point. This yields a demand curve for standards: the extra money consumers will pay for goods they know are produced under better working conditions. The ideal free market way to meet this demand is through information-labeling products with the labor condi- tions under which they are produced. If a labeling strategy is impractical or unreli- able (say, because producers or retailers will stick "Made under good conditions" labels on products made by forced labor in prison camps), other methods for regu- lating the market on behalf of consumers may be needed, ranging from government pressures to muck-raking campaigns and organized boycotts. The net effect of a consumer-driven demand for standards should be a rise in labor standards, paid for by consumers willing to pay for an "extended product" that includes better condil-ons for those who make the good. Legislation in China in 1994 on child labor, motivated by concern over the attitudes of Western con- sumers, provides a good example of the development of consumer-driven stan- dards. While economists have trouble arguing against demands based on consumer preferences, it would be remiss to ignore the danger that consumers in industrial countries may demand standards that are too high for developing countries to afford, with consequences that run counter to the underlying desire of consumers. For instance, an unwillingness to buy any products made by child labor could push child labor from the formal sector into the streets, rather than into schooling or other desirable activities. Information about what is economically plausible as well as about the conditions of work in developing countries is needed for the push for standards to be a boon rather than a detriment to the economic well-being of workers. The second reason behind the demand for global labor standards is protection of existing workers or firms. Many trade specialists believe that the demand for global labor standards is just old-fashioned protectionism disguised as altruism. That some groups that argue for labor standards in international settings-labor unions in industrial countries in sectors that potentially compete with woucers in developing countries-have a dear self-interest in limiting competition from lower-paid com- petitors suggests reason for such concern. In the debate over NAFTA in the United States, it was the AFL-CIO labor union, historically a supporter of free trade, that opposed NAFTA on the grounds that increased integration of the Mexican and U.S. economies would harm low-skill U.S. labor, while the manufacturing community, proponents of cariffs in decades past, supported the agreement, in part because it gave greater surety to investments in Mexico. In this case a higher social safety net in the economically less advanced country might be the appropriate means of defus- ing protectionist sentiments and enhancing the move toward an open economy. 450 Roirndtable Discussion on Employment and Deuelopment Globalization makes workers in developing countries potential beneficiaries of pro- tectionist sentiments in industrial countries. The second policy issue is how quickly the wages of workers in developing coun- tries that enter the world trading system approach wages in industrial countries. If we are entering an era of increased pressure toward factor price equalization, what we want are huge increases in wages and working conditions in developing coun- tries, not huge decreases in wages and working conditions or long-term unemploy- ment in industrial countries. In the newly industrialized economies of Asia that are widely heralded as devel- opment successes, wages have in fact increased at historically unprecedented rates. Wages in manufacturing have zoomed upward in Hong Kong, Korea, Singapore, and Taiwan (China) in the past twenty years, from about 5 percent of U.S. wages to roughly 30 percent in 1993 (at current exchange rates). The living standards of workers in these economies are even closer to those in the United States and other countries-on the order of 60 percent of U.S. living standards. These differences are potentially largely skill-related, considering the lower education of workers. Korea has not yet reached the standing of a high-income economy according to World Bank classifications, but Hong Kong and Singapore have-and Taiwan (China) would be close if it were included in the Bank's data. If other developing economies that enter the world economy in a big way experience equally rapid increases in real wages, increased pressure toward factor price equalization will take the form largely of improved earnings for traditionally low-paid developing-country labor, rather than of a threat to the jobs of less-skilled workers in industrial countries. The best outcome for workers in a global economy is a rapid increase in the pay of labor in developing countries. What policies and institutions might best bring the wages of workers in develop- ing countries toward the levels in industrial countries? In the West improvements in labor conditions were accompanied by the rise of trade unions and various labor mar- ket regulations. For the most part the newly industrializing Asian economies have not followed this route, at least not until recently. But unions, once suppressed in Korea, have struggled toward a legitimate place in society. Korea and Taiwan (China) have enacted social legislation protecting labor. There appears to be a natural pattern in which economies "buy" greater labnr standards and protection for workers as they develop and in which unionization and other forms of collective activity by workers similarly grow with democratization and economic advancement. Some economists regard these patterns as inimical to economic advancement. I do not. They are part and parcel of successful market economies. The danger is not that economically successful countries will choose too much regulation and inter- ference with markets (though some may do that) but that those that have not (yet) succeeded may try to "buy" a social product that they cannot afford. Uruguay is often cited in this respect as having a social security system that exceeds its produc- tive capacity. Many of the transition economies in Eastern and Central Europe have similar problems. Just as consumers want to know that the products they buy are produced under decent conditions, the citizenry in all countries wants certain regu- Cbristopber A. Pissarides on Lessons of European and U.S. Labor MaTrkets 451 lations of job market outcomes and processes. But the demand for these 'goods' must be conditioned on what the economy can afford, not on best practices in more advanced economies. The third policy issue is the extent to which many developing countries can fol- low the same path of development in a global economy at the same time, or whether some countries will be forced to the wayside in the new international division of labor. Will China's increased movement into the world economy mean that the Philippines will lose its opportunity to advance through exporting, say, children's toys? Will Colombia, Korea, and Malaysia be able to move up to the next rung in development? Will the transmission of advanced technologies reduce demand for certain types of labor in developing countries, as has occurred in industrial coun- tries? Or will some countries simply fail to develop the institutional infrastructure needed to compete in a global economy? One surprising result emerging from studies of tariffs and quotas in developing countries is that some countries have protected products made with low-skill labor, such as textiles or footwear, where these countries would be expected to have a comparative advantage relative to industrial countries. In Mexico the reduction of tariffs and quotas appears to have harmed less-skilled labor because trade barriers were highest in unskilled, labor-intensive industries (Feliciano 1994). Whether such trade policies reflected fears of competition from even lower-wage countries or from higher-wage but technologically more advanced countries is an interesting question. Globalization does not necessarily hiean shifts in demand toward the products of any particular country, regardless of its wage level. In sum, the ongoing globalization of the world's labor fo.ce is a major develop- ment that has the potential of affeeiing workers throughout the world. We may never reach the point where whrt happens in the labor market in Calcutta deter- mines wages in Amsterdam, Manila, or New York. But what happens in Calcutta (or Shanghai or Dacca) has the potential for increasinly affecting wages throughout the world. Indeed, I can envisage a map of the world based on the skills of workers and their place in the global labor market-a map based not on political boundaries but on economic competency. Some segments of developing countries would be in the global market and other segments would not, just as some groups in industrial coun- tries would be part of the market and others not In the ideal global labor market education and skills will ultimately matter more than residence in a particular polit- ical entity. Christopher A. Pissarides on Lessons of European and U.S. Labor Markets I structure my discussion around a comparison of the experience of the United States with that of the major European countries and then draw some lessons about the role of labor market po qy for developing country labor markets. In making this comparison i do not argue that one labor market is better t'han the other. One can- not point to a model to emulate but only to some lessons of experience and, occa- 452 Roundible Discusion on Employment ad Detelopment sionally, to solutions. The views expressed by the leaders of the G-7 nations at the jobs summit in Detroit in March 1994 are instructive. Each leader went to the sum- mit hoping to learn from the experience of the others, with the Europeans envying the U.S. job creation record and the Americans envying the Europeans' earnings record. I concentrate here on unemployment and earnings. Since the 1960s unemployment in the United States has fluctuated within a S to 10 percent band. There have been cycles, but since the 1973 oil crisis unemploy- ment has been on a flat trend. By contrast, unemployment in the European Union (EU) has been on an upward trend. Again, there have been cycles, but each new trough has been higher than the previous one. As a result unemployment, which was only about 2.5 percent in the early 1970s, has fluctuated between 8 and 10 percent in the 1990s. Higher unemployment is associated with a longer duration of unemployment in all countries, though the relation is closer in countries where the cause of higher u-nemployment is lack of job creation (rather than more job destruction). In the United States in 1990, of the 5.5 percent unemployed, 5.6 percent were unem- ployed for more than one year. In the European Union the incidence of long-term unemployment is much higher. In Germany, 46.3 percent of the 4.9 percent unem- ployed had been unemployed for more than a year; in the United Kingdom it was 36.1 percent of 5.9 percenL In the European Union as a whole long-term unem- ployment averaged 51 percent of an 8 percent unemployment rate (OECD 1993). Every economy needs some unemployment to function properly. Just how much is a matter of debate. 'What is certain, however, is that long-term unemployment is wasteful in terms of the loss of skill and the disenfranchisement it entails from labor market institutions and processes. There is evidence that long-term unemployment ceases to act as an inflation deterrent on unions and that the long-term unemployed are less likely to be offered jobs by employers, other things equaL They are also more likely to be discouraged (Layard, Nickell, and Jacknan 1991; Blanchard and Diamond 1994). The high incdence of long-term unemployment must be regarded as a failure of European labor markets. An associated failure is the rising trend of unemployment. Even if there is disagreement about the right level of unemployment, it is difficult to find factors to justfy an increase in the European equilibrium unemployment rate from 2.5 percent in 1973 to more than 8 percent in 1993. The unemployment experience of Europe and the United States contrasts with their earnings experience. Real average earnings in the European Union have been on a steep upward trend since the 1960s. In the United States they have increased hardly at all. More striking is the behavior of earnings at the low end of the earn- ings distribution. Unskilled earnings in the United States have been on a declining trend for several years, decreasing by 10 percent since 1972 (at the top end, real earnings increased by 4.4 percent over the same period). European workers in the bottom 10 percent of the wage distribution earn 44 percent more in real terms than their U.S. counterparts, despite the higher average standard of living in the United States. ChristopherA. Pisarides on Lessons of European and U.S. Labor Markes 453 Europeans often envy the U.S. record on job creation. A comparison of unem- ployme,t records seems to justify that envy. But closer examuination shows that look- ing only at the number of jobs created is misleading. In the long run job creation is driven largely by the supply of labor. The supply of labor has risen faster in the United States than in Europe, which accounts for much of the higher job creation in the United States. In the short run job creation is also driven by job destruction. A country that destroys a lot of jobs is also likely to create more, because those that lose their jobs will be looking for new ones. Job destruction rates are on the whole higher in the United States than in Europe. Fmally, the quality of the new jobs is also relevant. A lot of the jobs created in the United States require low skills, are of short duration, and are mainly in services; although they keep young and untrained workers off the unemployment books, they are dead-end jobs that are not likely to facilitate the transition to more permanent employment Why the disparity in the unemployment and earnings experience of Europe and the United States? To find the answer we have to look at the low end of the labor market, since unemployment in Europe affects mainly the unskilled. It is clear from the record that in the past twenty years die industialied world has expienced a negative shock at the unskilled end of the labor market. The task that confronted the labor markets of the industlized world was how to shift employment from low-skdll manufacres to other sectors. In this regard the U.S. labor market has been more successful than have European markets, but at the cost of a large class of "working poor.' Although a higher mobility of labor in the United States than in Europe has something to do with this record, the main reason, I believe, has to do with labor m=rket policy. Labor MArket Policy At the risk of oversimplification, I can put the main lesson of the diversity of expe- rience just described as follows: in the face of negative shocks a welfare policy that does not pay attention to labor market incentives can be much more costly, in terms of forgone output and financial cost, than a policy that does. The absence of a labor market policy encourages the creation of many unstable, low-income jobs that do not offer productive training opportunities, while a passive labor market policy that protects low incomes gives rise to inflexibility, espedaly at the low end of the mnar- ket, and to long-duration unemployment. Labor marke: policy is of two general kinds: policy that regulates the employ- ment relationship without money transfers (restrictions on the dismissal of employ- ees, legally enforced consultation with workers) and polcy that transfers money from one part of the labor market to another. Regulation is a form of tax on employers and the less there is, the more job cre- ation there is likely to be. Although some basic worker rights need to be protected by law (for example, the grating of rest periods), restrictions on the firm's ability to quickly recruit and dismiss workers tend to reduce flexibility at small obvious 454 Round&abk Discussion on Employmen and Deveomen benefit to workers. There is a clear relation between the degree of employment- protection legislation and job turnover (Nickeli 1982) and, consequently, between employment-protection legislation and long-term unemployment. Countries that impose more restrictions on the dismissal of labor tend to have longer durations of unemployment (figure 1), presumably because the restrictions discourage both job destructon and job creation. Transfers from the govemment to the private sector take the form of spending on passive" policy measures and spending on "active" measures. Passive labor market measures usually indude income support for the unemployed and other low-income groups without preconditions, and active measures mean programs that encourage faster transition from unemployment to employment. Passive measures are good for reducing poverty and for achieving a more equitable distribution of income, but they reduce the incentives for work at the low end of the market. Passive measures are much more common in European countries than in the United States, one reason for the difference in behavior of their labor markets. Generous income support in Europe acts as a floor on wages and so discourages low-income job creation. As a result, unemployment tends to be of longer duration. When income support is absent, as in the United States, unemployment turnover is higher and job creation and destruction are more frequent. Experience in the OECD shows that the most important policy influence on the duration of unemployment is not the level of income support but its duration (Jackman, Pissarides, and Savouri 1990). Most unemployed workers leave unem- ployment quicldy. Even generous unemployment benefits, provided they do not last long, have little effect on the rate at which people leave unemployment. But when benefits are available for long periods of time, more of the unemployed enter long- term unemployment and stay there (figure 2). With the rise in European unemployment, and the impressive unemployment record of Sweden and the other Scandinavian countries that pursued active labor market policies, most European countries started devoting more attention to active policy measures in the 1980s. These have taken the form of spending on public employment services and administration, labor market training, measures to help young workers and members of disadvantaged groups find regular employment, and subsidies paid directly to firms for job creation. In the early 1990s OECD countries spent half as much on active policy measures as on passive (Jackman, Pissarides, and Savouri 1990; OECD 1993). Although the administraion of such measures is more cosdy than the administration of passive measures, they have proved popular, mainly because of their success in reducing long-term unemployment. The proportion of long-term unemployment is plotted against spending on active measures per unemployed worker (as a percentage of GDP) in figure 3. Sweden does not appear in the chart because its high spending on active measures places it too far out on the horzontal axis and distorts the picture (Sweden's place on the hori- zontal axis is 119). Its long-term unemployment rate is one of the lowest in Europe, at 4.8 percent in 1990. Germany is also missing from the chart because during 1990-92 its spending on active measures was distorted by unification. The chart reveals a negative correlation between spending on active measures and long-term Christopher A. Pissardes on Lessons of European and U.S. Lcbor Markets 455 Figure 1. Employment Protection Legislation and Long-Tenn Unemployment, 1990 Pcrcent so E 60 o2 a Spain E * Nhelands* ,,mm * panmgA 40 * France E *i~~~~~~~ ennwkc *United Kingdom 20 * J3 *Nway e New Zaland 01 0 Sweden 0 S 10 IS 20 Maximum lcngth of notice prior to dismissal plus covcrage of severance pay (months) Figure 2. Passiue Policy Measures and Long-Term Unenployment, 1990 Pcrcent 80 U *2rdasl E 60 o * Spai E *c NetheIams E: 40 * Gittmhgiy ~~~~ 40 U~~~~~~~~~~~~ France U United Kingdom Dcnmark 2 u 20 *J ENorway in C UthitedStes Canod- 0 S Sweden 0 0 so 100 150 200 Mximunm duration of benefit (weeks) Figure 3. Active Policy Measures and Long-Term Unemployment, 1990 Percent 30 >. 60 pni, a Greece E r 40 *F France E United Kingdom u 20 * *Ncw7eand Norway o United Stnes 0 0 0 5 10 is 20 25 Active policy spending per unemployed (percentage of GDP) Note Long-teem unemployment is the percentage of unemployed who have been unemployed for more than one year. Dam are for 1990, 1991, or 1992, depending on data availability. Fcw 1-Maxmmm length of nodee or severance pay is the sum of the nmamum length of norke that bas to be given before a dismisslt can bc brought into force and the number of months of pay tdat have to be given as severace payment at the fime of dismissaL Fgure 2-Mxdmum dura- tion of benefit is the maxmum number of weck that an el gble worker is entided t receive unemploymcnt compen- stion- Figwe 3-The horizontal axis measures expendimre on active policy measures, as defined by tbe OECD, ex- pressed firt as a percentage of GDP and then divided by the proportion of uncmployed workers. Soure OECD 1993. 456 Roundlable Discussion on Employment and Development unemployment, especially for EU countries. Multiple regression analysis confirms the claim that active measures reduce both overall unemployment and the incidence of long-term unemployment Uackman, Pissarides, and Savouri 1990; OECD 1993). Lessons for Developing Countries The labor market has to provide both a flexible framework to allow for the adop- tion of new technologies and growth and satisfactory incomes for people who com- mitted to their skills at some time in the past. Even mature ecornmies cannot cope with change without disadvantaging some sections of the labor torce. In the United States the disadvantaged are those whose only access to the market is through unsta- ble, low-income jobs; in most of Europe the disadvantaged are the long-term unem- ployed. What lessons can developing nations learn from this experience? I concentrate here on the lessons that have been learned from policy. First, mea- sures that support low-income groups can lead to rigidities at the unskilled end of the labor market and reduce the market's adaptability to new circumstances. A wel- fare policy has to pay attention to the incentives that firms have for job creation and those that unskilled wor:cers have for seeling employment and acquiring new skills. A second lesson is that an active labor market policy, such as the subsidization of training and job searches, helps remove the disincentives of a passive income sup- port policy. The challenge for emerging labor markets is how to combine income support policies to avoid poverty with active measures that can remove the disin- centives of the policy and provide the right incentives for disadvantaged groups to reenter regular employment. The experience of European countries in this respect, espeially of the Scandinavian countries that pioneered actve policy, can shed light on the design of such policies. References Bates, RH. 1981. Markets end States in aopical Afria. Bcrkeleyr University of California Press. Bevan, D.L., P. Colier, and J.W Gunning. 1989. Peasants and Government: An Economic Andysis. Oxford: Clarendon Press Birdsall, Nanq, and Richard SaboL 1994. "Virtuous Circles: Human Capital, Equity, and Growth in East Asia:' Blanchard, Olivier, and Peter Diamond. 1994. "Ranking, Uncmploymnent Duration and Wages.- Review of Economic Studies 61 (October): 417-34. Cabalero, RJ., and ESLRA. EngeL 1993. "Microeconomic Adjustment Hazards and Aggregate Dynamics." Quanrl Journal of Economsics 108: 3S9-84. Collier, P. 1989. "Contractual Constraints on Labor Exchange in Rural Kenya.' Intenational Labor Review 128 (6): 74S-68. Colier, P., and J.W Gunning. 1992. "Aid and Exchangc Rate Adjustment in African Trade Liberalisa- tions." EconomicJounal 102: 92S-39. Collier, E, and D. Lal. 1986. Labor and Poverty i Kenya. Oxford: CLarendon Press. Cox-Edwards, Alejandra. 1993. -Labor Market Legislation in Latin America and the Caribbean." Report 31. World-Bank, Regional Studies Program, Latin America and the Caribbean Technical Departmnent; Washington, D.C Dixit, A. 1989. "Intcr-Sectoral Capital Reallocation under Price Uncertaintyf Journal of Intnationa Economics 26: 309-25. Roundtable Discussion on Employment and Development 457 Feliciano, Zadia. 1994. "Workers and Trade Liberalization: The Impact of Trade Reforms in Mexico on Wages and Employment? Ph.D. diss, Harvard Univcrsit, Cambridge, Mass. Fields, Gary. 1974. "The Privatc Demand for Education in Rclation to Labor Marker Conditions in Less Developed Countrics." Economic Journal 84: 906-25. - . 1993. "Changing Labor Marker Conditions and Economic Development in Hong Kong, Republic of Korca, Singapore, and Taiwan, China." World Bank Economic Reiew 8 (3): 395-414. Freeman, Richard. 1993a. 'Is Globalization Impovcrishing Low-Skill American Workers?" Paper pre- sented at the Forum on Policy Responses to an Intcrnational Marker, The Urban Institute, November 17, Washington, D.C. . 1993b. 'Labor Markcr Institutions and Policies: Help or Hindrance to Economic Dcvelopment?" In Lawrencc IL Summers and Shekhar Shah, ds, Proceedings of the World Bank Annual Conference on Devlopment Economics 1992. Washington, D.C . 1994. "A Hard-Hcaded Look at Labor Standards." Paper presented at the U.S. Dcpartment of Labor conference on Standards and Global Economic Integration, April 2S, Washington, D.C Harris, J., and M.P. Todaro. 1970. lMigration, Unemployment and Development A Two-Sector Analysis.? American Economic Review 60: 126-42. Hoddinott, J. 1993. "Wages and Unemployment in Urban CBtc dt'voire. Working Paper 93/3. Centre for the Study of African Economics, Oxford. Jackman, Richard, Christopher Pissarides, and Sawas SavourL 1990. "Labour Market Policy and Unemployment in the OECD." Economic Policy 11 {October): 449-90. Jamal, V., and J. Weeks. 1993. Africa Misunderstood London: MacMllaL Jones, B 1994. "Education and Labour Productivity in Giana- An Applied Production Analysis." Working Paper 94/12. Centre for the Study of African Economies, Oxford. Krishnan, B 1994. "Getting By and Getting On? Centre for the Study of African Economies, Oxford. Layard, Richard, Stephen Nickell, and Richard Jaclknan. 1991. Unemployment Miaceconomic Perfor- mance and the Labo r aw*e Oxford: Oxford University Press. Mazumdar, D. 1994. "Wages in Africa." World Bank, Africa Technical Dcpartment, Washington, D.C NickeDl, Stephen. 1982. "The Determinants of Equilibrium Unemployment in Britain? EconomicJournal 92: SSS-75. OECD (Organization for Economic Cooperation and Development). 1993. Employment Outlk. Paris. Sachs, Jeffrey, and Howard Schatz. 1994. "Trade and Jobs in U.S. Manufacturing." Broolings Pancl Papcr. Brookings Institution, Washington, D.C. Vclenchik, A. 1994. "Labor Markets? In J.W. Gunning, cd., "Enterprise Survey of Zimbabwe: Fmal Report" Free University, Amsterdam. Wood, Adrian. 1994. North-South Tda, Employmen, and Inequality. Oxford: Clarendon Press. World Bank 1993. The East Mian Mirace: Economic Growth and Pubic Poicy. A Policy Rcsearch Report Washington, D.C. FLOOR DISCUSSION OF THE ROUNDTABLE DIscussIoN ON EMPLoYMENT AND DEVELOPMET F reeman, said a participant from the World Bank, has stated that demand for labor standards comes from workers, consumers, and groups seeking in- creased protection. What happened recently in Marrakesh, however, illus- trates that Western countries also apply pressure for labor standards to protect their industries from labor-intensive exports from developing countries. High unemploy- ment rates in the European Union, the participant felt, would only exacerbate the tendency toward this form of protectionism, and the newly emerging markets were unlikely to benefit from the globalization of labor markets. Whether labor standards would hurt some countries' chances was a politically hot question, said Freeman, and in a sense it depended on where the demand for labor standards came from and how it operated. If it was workers demanding greater safety, the result might be lower wages, because the workers were effectively saying that they would rather take some of their compensation in the form of safer worldng conditions. If the demand for labor standards came from consumers, the consumers would be paying for the standard. Basically they would be saying, rm willing to pay a nickel more for a shirt made in a country with decent standards rather than buy one from a country without them. It would be silly, even disastrous, to insist on U-S.-level minimum wages in developing countries, or European mini- mum wages in the United States. But many rights that elicit concern-the right for women not to be harassed, for example, or the right to decent working conditions- are not expensive to guarantee. Many employers, faced with the prospect of not sell- ing their product, would instantly improve working conditions. The problem, then, is govemment demands that really reflect pressure from trade unions or textile lob- bying groups, not out of concern for workers' rights but to keep competitive prod- ucts from coming into their country. This problem can be phrased in a much better way if we grab control of the issue, said Freeman. Alan Gelb (speaker in another session) asked Pissarides to summarize the current state of thinking about the efficiency of active labor market policies. To what extent do such policies facilitate improvements in the labor market and to what extent do This session was chaired by Michael Bruno, chief economist and vicc presidcnt, DevelopmentEconomics, at the World Bank. Proceedings of the World Bank Anal Confeence on Deuelopmt Economics 1994 458 01995 The Intcrnational Bank for Reconstruction and Developmcnt / THE WORlD BANK Floor Disassion of the Rowndtable Discmssion on Employmet and Developmem 459 they simply keep officially registered unemployment low while subsidizing jobs? Pissarides replied that there had been some exploitation of labor market policy for political reasons, to reduce the number of unemployed. But what he meant by active policy measures was measures that get unemployed workers working. Sweden, for example, subsidizes training in the private sector: after a year of unemployment the government conve:s what would have been an unemployment benefit (a direct sub- sidy to the worker) into a subsidy to a firm for employing the unemployed worker. That gives the unemployed worker some skills and facilitates the transition to a job. Another example was the RESTART program in the United Kingdom, in which the Department of Employment interviewed the long-term unemployed and tried to find them jobs. The program has been successful because it turned out that there were jobs for the long-term unemployed but that they had become discouraged and had stopped going out and applying for thenL A participant from the University of Massachusetts asked Collier for more details about firm-specific training. How significant was it as a proportion of firm costs? In Africa, did it matter what type of firm it was (multinational, public, or private)? Was there a difference in firm costs within an industry or across national boundaries or regions? Collier responded that they had just gotten data on whether firm training differed by type of firm. They would soon be able to provide a robust answer to that question from surveys conducted in the manlfacturing sectors of half a dozen African countries. A participant from Congressional Research asked Birdsall to compare prospects in Latin America with prospects in Asia. If the question was, where should you put your money, replied Birdsall, she would say that the potential returns were greater in Latin America, but so were the risks. It was a question of how much of a risk- taker you are. (She was speaking on that point as Nancy Birdsall, she reminded the audience, not as executive vice president of the Inter-American Development Bank.) The biggest challenge in Latin America, both politically and economically, was inequaiity, which was probably reducing hoisehold savings and investment in chil- dren's health and education. Inequality also threatened the social and political sus- tainability of difficult but essential ongoing economic reforms. Another paricipant asked Birdsall how important labor market outputs and aggregate growth were to women's employment and investments in women's human capital. There is no question, responded Birdsall, that returns to labor and returns to education in the labor market affect parents' and people's choices about how much education to invest in and that in the long run labor market outcomes dearly affect women's welfare. Whether they affect women's employment in the formal sector, as Freeman implied in a different context, depends very much on institu- tional and other factors. A participant asked Collier to what extent were demographic trends a factor in Africa's wage decline? Collier responded that the demand for labor had failed more than the supply; a successful economy can generate enough growth in labor demand to accommodate growth in the labor supply. In some countries the African labor supply has been endogenous to economic failure. In Ghana in the 1970s, for exam- 460 Floor Dission of the Rowdtable Discsion n Employmnt and Development ple, the labor supply didn't grow because people migrated out of the country. But the main factor was demand failure, irn Collier's view, not demographics. The Harris-Todaro model might not work for places like New Delhi, said a par- ticipant from the George Washington University, but there was some evidence that it helped to explain Botswana and South Africa (in the work by Robert Lucas, for example). Yes, responded Collier, South Africa was a distinctive labor market, by African standards; in particular, there was virtually no smallholder sector, so alter- natives to wage employment in the formal sector were very different. Also, the polit- ical economy in South Africa was radically out of phase with the rest of Africa, so South Africa today looks somewhat like the rest of Africa did in the early 1960s. A participant from the George Washington University asked Freeman if there were any lessons for developing countries on works councils, gain-sharing, and the like, which under certain conditions might boost efficiency and even human capital accumulation. Freeman did not know of any good examples from developing coun- tries. The issue was, he said, what level of education and stage of development you must reach before you could move to that form of organization. In the United States Xerox and Texas Instruments do this kind of thing very well, but even in the United States most companies don't do it. It is a complicated question, and Freeman hoped someone would do more work on it.