Trade reform and employment re-allocation in Kenya*
By
Jörgen Levin
Department of Business, Economics, Statistics andInformatics
Örebro University
Sweden
E-Mail:
Abstract
The Kenyan reform programme of the 1990s was disappointing with regard to both per-capita income and employment growth. When trade was liberalised in tandem with labour market rigidities and retrenchment of public employees, this had a significant negative impact on employment generation and households well being. In response to this, activities within the informal sector increased tremendously, and an increasing number of individuals were forced to live below the poverty line. When labour markets are functioning poorly facilitating operations of the union could generate a positive impact not only too members of the union but also to the non-union members.
Keywords: computable general equilibrium, trade liberalisation, labour market, Kenya
JEL classification: C68, F160, J500, O550
* The author is grateful to participants at a WIDER seminar.
1. Trade liberalisation and labour markets
Africa has been one of the last regions of the world to open up to the global economy, at least in the sense of putting together the policies and infrastructure to enable it to engage gainfully in world trade. But although there were scattered signs of economic recovery in the late 1990s, Africa enters the new millennium with one of the highest degree of income inequality in the world and the absolute number of poor people continues to increase (Ali et al. 2000).
Opening up to trade have raised concerns among policy makers and in particular on how to balance short-term cost versus long-term benefits. Labour markets are important transmission mechanisms, both for external shocks and in terms of possible economic integration. Traditionally, the study of labour markets in developing countries has focussed on medium and long-term issues, such as rural-urban migration, growth of the labour force and its implication for unemployment and poverty, and the effects of education on levels of earnings. In recent years, there has been a shift to understanding the role of labour markets in determining the wage- and employment-effects of trade reforms and other structural adjustment policies.[1] The role of the labour market in the transmission of short-run macroeconomic “policy” shocks has also attracted interest.[2] For example, Horton et al., (1994) argue that the labour market is central in determining the success of macroeconomic stabilisation and adjustment policies. It is also important in how macroeconomic policies affect poverty, since labour is very often the only asset owned by the poor (Demery and Addison 1994).
The market's flexibility determines the pace at which certain policy goals can be achieved: for instance, how quickly resources can be moved across sectors by shifting relative earnings, and how labour-market changes impact on the well-being of households and their individual members. However, complexities arise because labour is not homogenous: There are a huge variety of different skills. Moreover, differences in location, gender, and unionisation result in a large number of separate labour markets, each having its own characteristics. They are all linked to each other, and to other markets in the domestic economy. But when such imperfections in the labour market are considered, one thing is clear: The effects on poverty of a policy of expenditure switching are likely to be quite different from those in the orthodox model (Demery and Addison 1993).
In this paper we will analyse the Kenyan experience of trade liberalisation, labour-market rigidities and retrenchment of government employees. The following section briefly describes the adjustment process and labour markets in Kenya. The third section then gives a brief description of the modelling framework, and discusses policy experiments and their results. The final section summarises the results.
2. Labour markets and trade liberalisation in the Kenyan context
Kenya is one of sub-Saharan Africa's important economies in which political difficulties have slowed down the pace of economic reform in recent years. While the government succeeded in liberalising the economy in some areas, the country's adjustment experience has thus not been smooth; besides policy reversals, terms-of-trade shocks and unfavourable movements in aid flows have contributed to the uneven performance.[3] An expected boost in private-sector-led growth and employment-generation did not materialise. Two factors explain the dismal performance in job creation: First, because of government policy-uncertainties, private investors adopted a wait-and-see strategy, which slowed down investment. Second, with the liberalisation of trade and foreign exchange transactions, Kenyan firms became continuously exposed to competition from imports, and for many firms in the formal sector, this became a serious problem.[4]
In addition to private-sector layoffs and retrenchment of public employees, a half million new workers entering the labour market each year have added to the pool of job-searchers. In response, employment in the informal sector in Kenya has grown tremendously (Table 1). Informal-sector growth can be looked at as a residual absorber of labour, which the formal sector cannot absorb. Some of those in the informal sector are also actively looking for jobs in the formal sector, so besides acting as a cushion for the unemployed, the informal sector also acts as a queue for those waiting for formal-sector employment. Despite the employment generated in the informal-sector, there are some drawbacks. On average, the wage level is lower in the informal than in the formal sector, and the trend towards an increasing share of informal-sector workers might therefore negatively affect the urban poor (Manda 2002). Indeed, recent poverty estimates show that urban poverty has been increasing in the 1990s (Republic of Kenya 1998).
Table 1: Employment by sector (thousands), 1990-2000
Private sector / Public sector / Informal sector / TotalNumber of persons / % / Number of persons / % / Number of persons / % / Number of
Persons
1990 / 709,5 / 30,2 / 699,8 / 29,8 / 937,4 / 38,9 / 2346,7
1991 / 726,6 / 29,0 / 715,1 / 28,5 / 1063,2 / 41,4 / 2504,9
1992 / 763,2 / 28,3 / 698,7 / 25,9 / 1237,5 / 44,8 / 2699,4
1993 / 789,5 / 26,8 / 686,0 / 23,3 / 1466,5 / 48,8 / 2942,0
1994 / 817,2 / 24,8 / 688,3 / 20,9 / 1792,4 / 53,3 / 3297,9
1995 / 867,0 / 22,8 / 690,0 / 18,2 / 2240,5 / 58,0 / 3797,5
1996 / 917,9 / 21,5 / 700,9 / 16,4 / 2643,8 / 61,0 / 4262,6
1997 / 946,8 / 20,4 / 700,6 / 15,1 / 2986,9 / 63,5 / 4634,3
1998 / 968,0 / 19,3 / 696,6 / 13,9 / 3353,5 / 65,8 / 5018,1
1999 / 990,3 / 18,3 / 683,3 / 12,6 / 3738,8 / 68,1 / 5412,4
2000 / 1002,9 / 17,2 / 673,9 / 11,6 / 4150,9 / 70,2 / 5827,7
Source: Republic of Kenya (2001)
Note: The informal sector covers all semi-organised and unregulated small-scale activities largely undertaken by self-employed persons or such persons assisted by a few employees. It excludes all farming and pastoralist activities.
Even among those still employed in the formal sector, the situation has deteriorated. On average, real consumption-wages are lower (public sector) or have changed very little (private sector) compared to the early 1970s (Table 2).[5] Although real wages have generally fallen over the period, employer costs have not been reduced, as indicated by the real product-wages, which increased (private sector) or stayed relatively constant (public sector).[6] This, together with the other impediments in the business environment mentioned earlier might have contributed to the sluggish growth in formal-sector employment during the period.
Table 2: Real average product, consumption and minimum wages 1972-2000 (thousands of 1982 Kenya Pounds)
Real average product wage, private sector / Real average product wage, public sector / Real average consumption wage, private sector / Real average consumption wage, public sector / Minimum real wagesRural areas / Urban areas
1972 / 0,70 / 0,98 / 0,95 / 1,33
1973 / 0,65 / 1,00 / 0,83 / 1,27
1974 / 0,64 / 0,90 / 0,84 / 1,19
1975 / 0,62 / 0,88 / 0,82 / 1,15
1976 / 0,61 / 0,91 / 0,84 / 1,26
1977 / 0,56 / 0,82 / 0,76 / 1,12
1978 / 0,62 / 0,89 / 0,75 / 1,07
1979 / 0,66 / 0,90 / 0,77 / 1,04
1980 / 0,74 / 0,87 / 0,83 / 0,97
1981 / 0,77 / 0,96 / 0,78 / 0,98
1982 / 0,73 / 0,92 / 0,73 / 0,92
1983 / 0,71 / 0,88 / 0,71 / 0,88
1984 / 0,72 / 0,85 / 0,72 / 0,86
1985 / 0,70 / 0,84 / 0,69 / 0,83
1986 / 0,70 / 0,86 / 0,73 / 0,90 / 0,13 / 0,24
1987 / 0,73 / 0,85 / 0,76 / 0,88 / 0,15 / 0,26
1988 / 0,73 / 0,90 / 0,75 / 0,93 / 0,15 / 0,26
1989 / 0,81 / 0,88 / 0,83 / 0,91 / 0,16 / 0,30
1990 / 0,79 / 0,85 / 0,77 / 0,83 / 0,18 / 0,34
1991 / 0,80 / 0,85 / 0,76 / 0,81 / 0,21 / 0,39
1992 / 0,80 / 0,81 / 0,64 / 0,66 / 0,24 / 0,44
1993 / 0,76 / 0,74 / 0,49 / 0,48 / 0,31 / 0,58
1994 / 0,77 / 0,75 / 0,54 / 0,53 / 0,36 / 0,70
1995 / 0,83 / 0,76 / 0,63 / 0,59 / 0,38 / 0,78
1996 / 0,92 / 0,84 / 0,70 / 0,64 / 0,41 / 0,86
1997 / 0,95 / 0,95 / 0,78 / 0,79 / 0,45 / 0,96
1998 / 1,04 / 1,12 / 0,92 / 1,00 / 0,52 / 1,11
1999 / 1,15 / 1,16 / 1,00 / 1,01 / 0,55 / 1,18
2000 / 1,25 / 1,17 / 1,08 / 1,00 / 0,59 / 1,25
Source: Republic of Kenya (various issues)
The Government has made several attempts to liberalise the labour market, but still minimum wages are adjusted annually in an attempt to keep real wages from falling. Table 2 shows the trend of rural and urban real minimum wages during 1986-2000.[7] Minimum wages have been increasing in both areas, particularly the urban.[8] The changes in minimum wages between rural and urban areas illustrate the political-economy aspects of the economic-reform process. While the outcome of the reform programme has been disappointing with regard to employment growth and per-capita income, the government has tried to compensate workers through wage adjustments. The urban bias in minimum wages could be seen as an attempt to satisfy political interests in urban areas.
Another source for wage adjustments is the trade union. Manda et al. (2001) found that it is primarily the less advantaged workers that make use of the union and that the members typically have a relatively larger fraction of their earnings in the form of various allowances. Thus it seems that union prefers to negotiate for higher allowances rather than higher wages as most allowances are exempted from taxation. In addition, the employers would also gain in terms of tax exemption if they can prove that they actually paid the workers. Wage adjustments in the Kenyan context can thus be seen as a two-stage process. First, the government is adjusting minimum wage guidelines annually and then the union negotiate primarily on the non-wage component.
3. Trade liberalisation, growth and labour-market regimes: methodology, experiments and results
Establishing whether trade liberalisation has any impact on growth and employment is not straightforward for three reasons (Greenaway et al. 2002). First, we need to frame an appropriate counterfactual. Second, we need to disentangle the effects of trade reform from other effects. Third, we need to consider how long to wait before conducting an assessment of the reforms. Different methodological approaches, such as cross-country and time series analysis, have been suggested to evaluate the outcome of trade liberalisation. A third approach, used in this paper, is computable general equilibrium modelling, which has the advantage of simulating different scenarios.
Three questions raised in this paper are: First, does labour market specification matter when trade is liberalised. Second, what is the impact when some sectors and labour categories are unionised and some are not? Third, what is the impact of trade liberalisation combined with government retrenchment on the structure of employment? The issue here is to explore the combined effects of expenditure-reducing and expenditure-switching policy measures.
The general equilibrium model used in this paper consists of nine sectors.[9] Due to the importance of agriculture in Kenya, it is divided into two sectors: coffee/tea, and other agricultural products. There are three manufacturing sectors: food, consumer goods and a capital and intermediate goods sector. There are two service sectors, private and public. The latter two along with building and construction and an “other rural” sector are considered nontradable. The “other rural” includes non-agricultural activities in the rural areas not covered by the formal sectors.
In each sector the production function is specified as a multilevel nest of different functional forms (equations 8-20). At the top level intermediate demand and value added are combined in fixed proportions (Leontief). Value added is a Constant Elasticity of Substitution (CES) function of aggregate labour and capital. However, it is possible to substitute domestically produced and imported inputs. In the two agricultural sectors capital is defined as a CES function of capital and land. Aggregate labour is a set of nested CES functions for five different labour categories: highly skilled, skilled, semi-skilled, unskilled and casual labour. In this nested production structure producers choose the optimal mix of primary factors at each stage in the production process. At any level, primary factors are demanded up to the point where the factor price, inclusive of sector-specific differentials, equals the marginal value product of the specific factor.[10]
Specification of export and import functions is different from the standard neo-classical trade model, which tends to yield specialisation in production and a one-to-one correspondence between changes in domestic relative prices and those in trade policy or world prices (equations 1-6).[11] Elements of imperfect substitution, and thus incomplete specialisation, are allowed for in our model. At the highest level of aggregation, the Armington (1969) specification is used to define a composite commodity for each sector as a constant elasticity of substitution (CES) function of commodities produced domestically and imported from abroad. Similarly a composite intermediate output is produced using domestically produced goods and imported inputs.
Output in each sector is either sold to the domestic market or exported. A constant elasticity of transformation (CET) framework allocates domestic output between exports and domestic sales. This implies that there is product differentiation or market penetration costs involved in reallocation of output between domestic markets and foreign markets.
Factor incomes of capital and land are distributed to three institutions: government, enterprises, and proprietors of land (equations22-24). Enterprises keep a fixed proportion of capital income, net of depreciation and taxes, as retained earnings, while the remaining share is distributed to households. Sectoral factor incomes of five labour categories, as well as other sources of income such as remittances from abroad and government transfers, are distributed in fixed proportions to the various households (equations 28-33).[12] Other assumptions usually made in CGE models concern whether capital stocks are fixed within a period and/or whether capital is mobile or not. Here the model is run with the assumption of either fixed sector-specific capital or mobile capital capturing short/medium- and long-term aspects, respectively.[13]
The numeraire chosen is a bundle of domestic products, whose “aggregate” price becomes the numeraire price index. Thus, the exchange rate variable in the model is a real price level deflated (PLD) exchange rate.[14] Given an exogenous shock the real exchange rate variable will equilibrate the trade-balance constraint through changes in relative prices of traded and non-traded goods on the domestic market. Traded goods prices in the model are the domestic prices of exported and imported goods while the nontraded goods prices are defined as the prices of domestically produced goods sold on the domestic market. All of them are endogenously determined in the model.
In order to quantify the impact of trade liberalisation under different assumptions in the labour market four different types of labour market regimes are explored in the model.[15] The first is a neoclassical regime where the labour market clears through adjustments in the real wage (flex). The second is a combination of the neoclassical regime, under which the wage-rate adjusts to clear the labour market, and the Keynesian closure, with a fixed real wage-rate and unemployment (rigid). Sticky wages are assumed with resulting unemployment among skilled, semi-skilled, and unskilled labour categories. Unemployed workers spill over to a “casual” category, adding to the supply there. Since wages for the casuals are market-determined, there will be downward pressure on them (Mitra 1994). The third and the fourth regime introduce a union where the former is combined with flexible wages (uflex) and the latter is combined with rigid wages (urigid). Hence, in the latter regime there are, as in the second regime, unemployment and spill-over effects.
In the first and second regime it is assumed that intersectoral wage differences are constant. The wage differentials are exogenous, suggesting that factors acquire sector-specific skills upon entry into the sector and lose those skills upon exit. However, introducing the union in the model we explicitly model the behaviour that can generate the observed wage differentials.
There are many views on union behaviour, depending on the specification of the union’s utility function. Here the union takes the demand for labour as given (Lu) and chooses the wage differential (WDu,l) that maximises its utility (UNUTIL) according to equation 1 where WFl is the economy-wide average wage and (Lmin) is the minimum acceptable level of employment. This specification coincides with the behaviour observed in the Kenyan labour market as the wage differential can be approximated to a wage premium including allowances.
(1)
Given a CES production function substituting the optimal labour demand in the union sector (Lu) into the union’s utility function the optimal wage differential is:
(2)
where and p,u are exponents in the union’s utility function and the unionised sectors production function, respectively. This implies that when a sector contracts, perhaps as a result of lower protection, the decline in the wage differential (WDu,l) can dampen the reduction in employment.[16] This is the case when the economy-wide average wage is flexible. In the other case (unionrigid) when real wages are assumed fixed adjustment in the wage differential can dampen unemployment and spill-over effects.
3.1 Trade liberalisation and the labour-market regime
What is the quantitative impact of trade liberalisation under different assumptions in the labour market? More specifically would there be any dramatic changes in the labour market whether the union is included or not. Moreover, does it matter whether the labour market is flexible or not. The second column in Table 3 shows the results of a 50% tariff reduction while assuming a “standard” neoclassical labour market where the real wage adjusts to clear the labour market. The results show what one would expect, there are no dramatic welfare gains[17]. Most traded sectors would be favoured by this strategy; the capital/intermediate sector is the major exception. As expected labour demand generally increases in the expanding sectors and contracts in those where output is falling or constant. Casual workers experience the highest increase in their wage, rate followed by unskilled and semiskilled workers, while skilled workers are hurt. Thus, lower protection seems to benefit more those factors common in the rural areas, land and lower- skilled workers.
How would results change if some sectors and categories of labour are unionised and some are not? It is assumed, based on the Kenyan context discussed above, that skilled, semi-skilled and unskilled labourers in the food, consumer and capital /intermediate industry are unionised. If the labour market is assumed to clear through adjustments in the real wage introducing the union (flex scenario versus uflex scenario) add some additional premium to those unionised sectors which see an increase in labour demand. For example, increased demand of labour in the food sector induce a premium to the unionised workers while in those sectors where labour demand is reduced the wage differential see a decline. Thus the union are able to increase the wage differentials in sectors where output and labour demand is increasing. But the union is also adjusting the premium downwards in order to save jobs in those sectors, which see increased competition from imports.