zyxwvu zyxw zyx Journal of International Development: Vol. 5, No. 2, 135-143 (1993) zyxwv LABOUR MARKETS, POVERTY AND ADJUSTMENT TONY ADDISON Department of Economics, University of Warwick, U K LIONEL DEMERY Africa Technical Department, World Bank, USA Abstract: The labour market is one of the most important channels through which adjustment programmes affect the poor. Adjustment’s poverty impact will differ between competitive and imperfect labour markets. Two effects on poverty are highlighted: changes in real wages arising from changes in labour demand and priccs, and the reallocation of labour between sectors. These effects can work in the same or opposite directions, depending on the economy’s structure. Since some workers may gain, and some may lose, workers can have conflicting interests during adjustment. 1 INTRODUCTION The impact of adjustment policies on poverty remains the subject of heated debate. In part this is due to the numerous channels through which policy changes can affect the welfare of the poor. These include alterations in the incidence of government expenditures and taxes, together with adjustments in the markets for products, credit, and labour. Moreover, since poverty groups typically have a wide range of livelihoods, policy changes of benefit to some of the poor may be adverse for others. Finally, the distribution of welfare outcomes over time is important; adverse effects may be concentrated in the short run while gains only accrue o v e r the longer term.’ Given this complexity of effects, disentangling the processes through which policy changes affect the poor is an inevitably difficult business. This has led many to leap to immediate judgements about the merits (or otherwise) of current adjustment programmes for the poor. Such a reaction is understandable given the urgency of policy debate, and an informed weighing-up of the evidence will always be necessary. Nevertheless, it remains essential to simultaneously develop our analytical understanding of the relationship between policy change and poverty in order to build firmer foundations for policy decisions. In this spirit, the present paper concentrates on one aspect of the problem, zyx zyxwvut Recent overviews of the debate are provided by Jolly (1991) and Zuckerman (1991) 0954-1748/93/020135-0Y$O9.50 8 1993 by John Wiley & Sons, Ltd. zyxwvutsr zyxwvuts zyxwv 136 T. Addison and L . Demery namely the labour market effects of adjustment. Since the sale of labour services is often vital to the livelihoods of the poor, a fuller understanding of adjustment’s impact on their wages and employment is called for. To achieve this, account must be taken of the labour market ‘imperfections’ which commonly exist in developing countries. Much of the debate on adjustment and poverty either assumes competitive labour markets or invokes wage and employment rigidities leaving the analytical framework undefined. T o try to improve on the present situation, this paper introduces imperfections in a relatively simple manner through a two-sector model with search unemployment. Section 2 sets the scene with a brief account of the competitive model and its predictions about the poverty outcomes of adjustment. It is shown that the model is strictly agnostic about these outcomes. Moreover, some of this ambiguity still remains when the assumption of competitive labour markets is relaxed in Section 3. In Section 4 some implications of the analysis are highlighted. 2 ADJUSTMENT WITH COMPETITIVE LABOUR MARKETS In order to understand the labour market effects of policy adjustments we must first understand their product market effects. Adjustment programmes generally contain a multiplicity of policy changes depending on the country’s specific problems. Common to most programmes, however, is currency devaluation combined with fiscal and monetary restraint. To avoid over-complexity we focus on these measures, and abstract from any trade liberalization which may also be necessary.’ On the assumption that the relative price of importables to exportables remains unchanged, both goods can be aggregated into a composite commodity ‘tradables’, whose price is given in world markets (for a ‘small’ economy). The prices of nontradables are determined by domestic supply and demand. Both tradables and nontradables are final consumer goods, and we assume there are no intermediate goods. A useful model for analysing adjustment’s effects is the ‘dependenteconomy‘ model which contains a tradable and a non-tradable sector (see for example Knight, 1976, and Lal, 1988). For the short run (the period of most policy interest), it is usual to assume that capital is sector-specific. whilst homogeneous labour is mobile across the two sectors (capital becomes mobile in the long run). This implies that the rate of return to capital can differ between the two sectors, but full labour mobility results in the equalization of real wages. If the money wage is fully flexible, then full employment is assured, and the economy is on its production frontier. In the context of this model, adjustment will have the following effect. For a small open economy, fiscal and monetary restraint will reduce domestic demand, and therefore the price of non-tradables. As a result the price of tradables relative to t h e price of non-tradables ( P J P , ) rises.3 This change in the real exchange rate causes expenditure switching to occur, and capital and labour are encouraged to switch from producing non-tradables to producing tradables. When P , is inflexible downward, then fiscal and monetary restraint are insufficient to restore both ’ Trade liberalization’s poverty impact is especially complex. See Addison and Demery (1989) for a review. ’ The ratio PJP,, is one delinition of the real exchange rate (see Edwards, 1089) z zyxw zy z z zyxwvu zyxwvu Labour Markets, Povery and Adjustment 137 external and internal balance, and devaluation is required to raise P, relative to P , (Demery and Addison, 1987). The equilibrium wage that is established after labour has moved from nontradables to tradables is a linear function of output prices (see Dornbusch, 1980: 98): iv = g P , o<g<1 + (1 - g) P , where the dot refers to a percentage change and where 5 is a technology parameter. Rearrangement gives: iv - P , = - 5 (P,*- P?) iv - P , = (1 - 5) ( P t - P , ) Given the rise in P,the real product wage in tradables will fall, thereby encouraging more employment. In contrast employment in non-tradables will be discouraged by the rise in the real product wage in that sector (following the fall in P,). Since P , has risen, but P , has fallen, the final change in the real consumption wage is ambiguous when workers consume both goods. A fall in the real consumption wage is more likely, the more tradables relative to non-tradables which workers consume. In determining the poverty impact of adjustment, it is now common to use the class of indices developed by Foster et al. (1984). These have the virtue of being subgroup decomposable; in our case into poverty in tradables and non-tradables respectively. Kanbur (1987) formally derives the effects of expenditure-switching policy o n poverty, using the assumptions that only wage-earners are poor (enabling profit income to be ignored), and that within-sector income distributions are unchanged by policy. While these are limiting assumptions, they do have the virtue that they make the problem analytically tractable and permit us to focus on the key processes. In two separate papers we demonstrate how the labour-market outcomes of expenditure switching are likely to affect poverty as defined by the Foster et al. measure (Addison and Demery, 1993; Demery and Addison, 1993). If the real consumption wage is expected to fall-an effect which on its own would tend to be poverty-increasing-poverty may still not increase if poverty is significantly greater in non-tradables relative to tradables and the amount of labour reallocated from non-tradables to tradables is large. If these latter two conditions are met, then the effect of transferring labour to a sector (tradables) where the incidence of poverty is significantly lower than in non-tradables will tend to compensate for the povertyincreasing effect of the fall in the real consumption wage, and the poverty index will tend to decline. There is therefore an additional source of ambiguity as to the The index is given by: 1 pv,=-n c(Y ,y,- - Y i) q ;=I a zyxwvu where n is the population size, q is the number below the poverty line, Y , and Y , (i = 1,. . . q ) are the incomes of the poor; a is a policy parameter chosen by users to reflect their aversion to poverty. zyxwvutsr zyxwvut zyxw 138 T. Addison and L. Demery poverty effect of expenditure-switching in addition to that arising out of the composition of workers’ consumption baskets. 3 ADJUSTMENT WITH LABOUR-MARKET IMPERFECTIONS While the simplifying assumption of competitive labour markets helps to highlight their role in determining adjustment’s poverty effect, it is a rather extreme assumption given the institutional context of labour markets in many developing countries. Government intervention in labour markets has been common throughout the developing world, especially in Africa and Latin America (Fields and Wan, 1989). Therefore, we now introduce labour-market imperfections by dividing the labour market into a ‘formal’ sector, where the wage is set above a market-clearing level by institutional processes, and an ‘informal’ sector, which functions competitively. The formal sector is assumed to be characterized by downward rigidity of nominaf wages, the result (for example) of a legislated minimum-wage fixed in nominal units. An assumption of nominal rather than real wage rigidity is appropriate when the government is reluctant to index minimum wages, which is usually the case during periods of adjustment (see Section 4 below). Further it is assumed that unemployment takes the form of voluntary job-search activity, since jobs are always available at the going rate in the informal sector. In the context of our two-sector (tradablehon-tradable) world, the labour market can be segmented in several different ways. The most straightforward assumption is that one sector’s labour market is entirely informal (the wage is flexible), while the other sector’s labour market is entirely formal (the wage is rigid). This gives two possibilities which are discussed below. In case 1 nontradables production is characterized by formal employment while tradables production is characterized by informal employment. In case 2 the reverse is assumed: non-tradables is the informal employment sector. A more complex case, in which both the formal and informal sectors produce both tradables and non-tradables, is considered elsewhere (see Addison and Demery , 1993). zyxw zyxw Case 1: The Non-tradables Sector is Characterized by Formal Employment Consider an economy in which non-tradables production is characterized by formal employment contracts (and therefore inflexible nominal wages), while the tradables sector is characterized by informal employment. Such a stylized picture would be representative of an economy in which public sector employment dominated non-tradables, while tradables (both exportables and importables) were produced by ‘casual’ labour in agriculture and manufacturing. T o simplify matters, assume that P , remains fixed throughout the following discussion, which allows non-tradables to act as the numeraire. Downward rigidity in the price of non-tradables requires that adjustment must involve a devaluation to secure both internal and external balance simultaneously. For this reason in what follows adjustment will be modelled as an increase in P,IP, through devaluation raising P,. In Figure 1 the labour supply (OTON)is allocated to the two sectors, measuring Labour Markets, Povery and Adjustment zy 139 zyxwvu zy zyxw zyxwvutsr zyx zyxwv zyxwv Figure 1. Case 1: formal employment in non-tradables. along the horizontal axis employment in tradables from the left, and non-tradables from the right. Labour demand functions are denoted by VMPi (i = t ~ , t ) . ~ In non-tradables, employment is restricted to O , L , , by the nominal minimumwage W,*. Those labour-force participants unable to obtain formal contracts either take up employment (O&) in the tradable (informal) sector at the marketclearing wage (Wtl), or elect to search for non-tradable (formal) employment. The latter can be modelled by making the assumption that every worker has the same probability of obtaining a former sector job, and that a turnover rate of unity applies in the sector. Employment in the tradable (informal) sector is given where a rectangular hyperbola (labelled H and drawn through x ) intersects the tradables ~ all points on the H locus, the labour demand function (VMP,)-at point Y . Along following equation is satisfied: w,= pw, = Ln L , + u w, (3) where p is the probability of obtaining a high-wage job in the non-tradables (formal) sector. With a wage-differential of (W,* - Wtl) a portion ( L 1 L 2 of ) the labour force remains unemployed in order to job search. An adjustment programme aimed at expenditure-switching raises P,, which in turn causes labour demand in tradables to rise (VMP, shifts out to VMP,,). Given the assumption that P, is fixed, the position of VMP, remains unchanged, and non- This apparatus is now popular in the trade literature (see for example, Mussa, 1974, and Neary, 1978). It is used by Edwards (1988) to analyse the labour market effects of import liberalization. 'This is the Harris-Todaro (1970) locus, which has been extensively used in the literature (see Corden and Findlay, 1975 and Edwards, 1988). zyxwvutsr zyxwv zyxwvu 140 T. Addison and L . Demery tradable (formal) employment is unchanged. Employment in the tradable (informal) sector rises to O T L 3 and the wage in this sector rises to W,z. Given the reduced wage-differential between the two sectors, a reduced number of labourforce participants ( L ,L 3 ) opt for search employment. Since the price of tradables has risen, while the price of non-tradables and the nominal wage in non-tradables are unchanged, the real consumption wage in the non-tradables (formal) sector falls. However, the change in the real consumption wage in the tradables (informal) sector is ambiguous, since the real wage in terms of non-tradables has risen, but it has fallen in terms of tradables.’ The outcome for the real wage of workers in tradables therefore depends on their consumption propensities. Maintaining the assumptions made earlier, that only those selling labour services are poor, and that within-sector income distributions remain unchanged, the poverty impacts can be derived. This is done formally elsewhere (Addison and Demery, 1993; Demery and Addison, 1993), and the results are summarized here. First, the real-wage effect on poverty differs between sectors, in contrast to the case of no labour-market segmentation. Although the real consumption wage in non-tradables is certain to fall, and therefore likely to raise poverty, the real consumption-wage change in the tradables sector is subject to the ambiguity that we have identified. The net real-wage effect on overall poverty is therefore strictly uncertain. There is a clear-cut effect tending to reduce poverty if the incidence of poverty among the unemployed exceeds that in the tradables sector, since unemployment falls, as some of the unemployed enter the tradables sector. The overall poverty effect is therefore uncertain depending upon the relative strengths (and directions) of the real-wage and reallocative effects. zyxwv Case 2: The Non-Tradables Sector is Characterized by Informal Employment The assumption made above, that t radables are characterized by informal employment, and non-tradables by formal employment, can be reversed. In this case tradables are characterized by nominal wage inflexibility, while the wage is flexible in non-tradables. This stylized picture describes an economy in which tradabes are produced by ‘modern’ firms and non-tradables consist of small-scale service providers and manufacturers. In this case, shown in figure 2, the minimum wage of W,* restricts employment to OTLl in t h e tradables (formal) sector. The non-tradables (informal) sector wage settles at Wnl, with employment in that sector being O N L z .The remainder of the labour force searches for formal employment; search-unemployment is therefore LIL2. Again, adjustment through expenditure switching raises the price of tradables, and increases labour demand in tradables production. Provided that the minimum wage continues to bind in the tradables (formal) sector (despite the upward shift in the V M P curve), employment in tradables increases to OTL3.The market-clearing wage in non-tradables will rise to Wn2, and employment in that sector will fall to ONL4.Unemployment falls to L3L4since the expansion in tradables employment (which is given along the horizontal line at W,*) is certain to exceed the contraction zyxwvu ’ The rise in W J P , is less than the increase in P,IP,, so that W J P , falls Labour Markets, Povery and Adjustment zy 141 zyxwvu zyx Figure 2. Case 2: informal employment in non-tradables. in non-tradables employment (which is along the V M P , curve). The real consumption wage in tradables is certain to fall, but the real-wage change in nontradables is ambiguous and depends on workers’ consumption patterns. In this case the fall in the real wage in tradables increases poverty, while the influence of the real wage change for workers in non-tradables is uncertain. The effect of the reallocation of labour on poverty again depends on the incidence of poverty across the two sectors and the unemployed. If poverty is greatest among the unemployed, for example, and lowest in the tradables (formal) sector, then the reallocation of labour will tend to reduce overall poverty. So once again we have an ambiguous real-wage effect and an unambiguously beneficial reallocative effect. In both cases 1 and 2 the real consumption wage falls in the formal sector, and its direction of change in the informal sector is ambiguous, regardless of the tradability, or otherwise, of the sector. This occurs because in the formal sector the nominal wage is unchanged over the short run, and consequently any policy which raises the price level reduces the real wage. The situation of informal workers is likely to improve, especially if they consume mainly non-tradables. The unemployment effects of both cases are beneficial since in these circumstances adjustment will reduce the real-wage gap between the sectors. Given that search unemployment is a function of the wage gap, this reduces unemployment in the model. The sector in which the real consumption wage is certain to fall (i.e. the formal sector) is unlikely to influence the overall poverty index significantly. First, the weight of such workers in the labour force is likely to be small, since formal contracts typically have limited coverage. Second, poverty in the formal sector is likely to be negligible, since such employees typically benefit from wage support in the years preceding adjustment. The uncertain real consumption-wage change in the informal sector, therefore, is likely to dominate the overall poverty impact of adjustment. This is confirmed in the illustrative exercise for Cote d’Ivoire reported in Demery and Addison (1993). zyxwvut zyxwv zyxwvutsr zyxwvutsrq zyxw 142 T. Addison and L. Demery 4 SOME CONCLUDING OBSERVATIONS The two cases above illustrate some of the basic processes determining whether adjustment will tend to raise or reduce poverty through its labour market effects. They show that, for a stylized economy, the poverty effects of employment and wage changes under adjustment cannot be predicted without country-specific information on some of the key parameters, in particular the consumption basket of workers. Moreover, since the tradability of the output of the formal and informal sectors varies across countries, it is quite possible for adjustment’s poverty effects to differ significantly across countries. Finally. the mobility of labour between sectors varies between countries, especially when intersectoral mobility implies geographical mobility (between urban and rural areas, for instance), and the costs of mobility to workers differ between countries at different stages of development. The sources of wage inflexibility, and therefore labour-market segmentation, will differ across developing regions. Unionization has traditionally been strongest in Latin America, and organized labour has attempted (with partial success) to defend wage levels and to index wages to inflation. More recently, in those Latin American countries that have pursued adjustment intensively, the power of organized labour has been reduced, and in some cases wage-freezes have been agreed (most notably in Mexico). This has resulted in the de-indexation of wages from inflation, and labour markets have become more flexible. Unionization is considerably weaker in Africa, being concentrated in traditionally capital-intensive sectors such as mining. In East Asia, limitations o n worker organization have ensured considerable labourmarket flexibility. For most countries, government interventions have been more important than unionization in explaining wage rigidities. Both minimum wage regulations and public-sector pay policies have in the past been important sources of rigidity. Most African governments have permitted the real value of minimum wages to be eroded by inflation, or have made only token upward adjustments in money wages, allowing these to be exceeded by inflation. In Latin America the traditional power of the protected public sector has been reduced. For instance in Costa Rica wages fell sharply in response to devaluation and other stabilization measures taken in the 1980s (Grindling and Berry, 1992). Across many developing countries the adjustment decade of the 1980s saw the start of major changes in the structures and institutions of labour markets. When labour market segmentation exists, conflicts of interest among employed workers are created. These conflicts of interest are intensified during adjustment, since, as we have seen, adjustment can potentially reduce incomes and increase poverty among one group, while improving the situation for others. Each group may then engage in the lobbying of government to express its support or opposition to the adjustment programme. This accounts for some of the complexity in workers’ attitudes to adjustment reported in Nelson (1991). Many countries are in the midst of fundamental labour-market changes which, through altering institutions and the behaviour of employers and employees, will change the parameters determining the future impact of macro-policy reforms on employment and wages. Further analytical work is needed to increase our understanding of how these institutional changes affect adjustment’s poverty impact, and the success of adjustment itself. zy zyxwvut Labour Markets, Povery and Adjustment REFERENCES 143 Addison, T. and Demery, L. (1989). ‘The impact of liberalisation on growth and equity’. In Renshaw, G. (ed.), Market Liberalisation: Equity and Development. Geneva: International Labour Office, pp. 1-88. Addison, T. and Demery, L. (1993). ‘The poverty effects of adjustment with labor market imperfections’. In Horton, S.. Kanbur, R. and Mazumdar, D. (eds), Labor Markets in an Era of Adjustment. Washington DC: World Bank, pp. 26-42. Corden, W. M. and Findlay, R. F. (1975). ‘Urban unemployment, intersectoral capital and development policy’, Economica, 43, pp. 59-78. Demery, L. and Addison, T. (1987). ‘Stabilization policy and income distribution in developing countries’, World Development, 15, pp. 1483-1498. Demery, L. and Addison, T. (1993). ‘The impact of macroeconomic adjustment on poverty in the presence of wage rigidities’, Journal of Development Economics (forthcoming). Dornbusch, R. (1980). Open Economy Macro-economics. New York: Basic Books. Edwards, S. (1988). ‘Terms of trade, tariffs, and labour market adjustment in developing countries’, World Bank Economic Review, 2, pp. 165-187. Edwards, S . (1989). Real Exchange Rates, Devaluation, and Adjustment: Exchange Rate Policy in Developing Countries. Cambridge, MA: MIT Press. Fields, G. S. and Wan, H. (1989). ‘Wage-setting institutions and economic growth’, World Development, 17, pp. 1471-1483. Foster, J., Greer, J . and Thorbecke, E. (1984). ‘A class of decomposable poverty measures’, Econometrica, 52, pp. 761-766. Grindling, T. H. and Berry, A. (1992). ‘The performance of the labor market during recession and structural adjustment: Costa Rica in the 1980s’, World Development, 29, 1599-1616. Harris, J. and Todaro, M. P. (1970). ‘Migration, unemployment and development: a twosector analysis’, American Economic Review, 60, pp. 126-142. Jolly, R. (1991). ‘Adjustment with a human face: A UNICEF record and perspective on the 1980s’, World Development, 19, pp. 1807-1822. Kanbur, R. (1987). ‘Structural adjustment, macroeconomic adjustment and poverty: a methodology for analysis’, World Development, 15, pp. 1515-1526. Knight, J. B. (1976). ‘Devaluation and income distribution in less-developed economics’, Oxford Economic Papers, 28, pp. 200-227. Lal, D. (1988). ‘A simple framework for analyzing various real aspects of stabilisation and structural adjustment policies’, Journal of Development Studies, 25, pp. 291-312. Mussa, M. L. (1974). ‘Tariffs and distribution of income: importance of factor specificity, substitutability and intensity in short and long run’, Journal of Political Economy. 82, pp. 1191-1203. Neary, P. (1978). ‘Short-run capital specificity and the pure theory of international trade’, Economic Journal, 88, pp. 448-510. Nelson, J . M. (1991). ‘Organized labor, politics, and labor market flexibility in developing countries’, World Bank Research Observer, 6 , pp. 37-56. Zuckerman, E. (1991). ‘The social costs of adjustment’. In Thomas, V . , Chhibber, A., Dailami, M. and de Melo, J. (eds), Restructuring Economies in Distress: Policy Reform and the World Bank. New York: Oxford University Press for the World Bank, pp. 247271. zyxwv zyxwv zyxwvut