The Influence of Garment Exports on Male-Female Wage Inequality in Sri Lanka

Third Draft May 2002 The Influence of Garment Exports on Male-Female Wage Inequality in Sri Lanka Jeevika Weerahewa Department of Agricultural Economics Faculty of Agriculture University of Peradeniya Peradeniya, Sri Lanka (94)-8-38777 jeevika@agecon.pdn.ac.lk Abstract: The objective of this paper is to assess the impact of garment exports on male-female wage inequality in Sri Lanka. A general equilibrium model is used treating female and male labor as specific factors of production in garment sector and rest of the economy respectively. Returns to specific factors are assessed under two scenarios: lower garment price and higher tariffs on textiles. Model is calibrated to Sri Lankan economy in 2000. Results show that wage gap is wider if a lower price for garments and a higher tariff for textiles are prevailed. Key words: General equilibrium model, Male-female wage gap, Sri Lanka

on technical Summary Globalization and its impact on wage inequality between skilled-unskilled and malefemale labour have become a crucial research issue over the recent past. Initial expectations are that with increased trade due to globalization developing countries will specialize in producing and exporting unskilled and/or female labour-intensive products and hence the wage rates of unskilled and/or female labor will rise. Production of skilled and/or male labor intensive products will decline with a consequent decline in the wage rates of skilled and/or male labor. The ultimate result will be a decline in male-female wage inequality, and in skilled-unskilled wage inequality. At the time of economic liberalization in 977, the garment industry played a minor role in the Sri Lankan economy. Since then, industrial exports have been dominated by the garment exports and currently the garment industry contributes 66 percent of total industrial exports. There is a high demand for semi-skilled female workers by this industry; more than 230,000 are employed in garment factories. The objective of this paper is to assess the impact of the expansion in garment exports on the male-female wage gap in Sri Lanka. The results of the study show that an increase in price of garments and a reduction in tariffs on textiles has helped to reduce male-female wage gap and hence inequality. A general equilibrium model is used to carry out simulations on the Sri Lankan economy, characterised as a small open price-taking economy. The analytical structure embraces a specific-factors model of trade. Two sectors are distinguished: garments and the rest of the economy. Female and male labour are treated as the specific factors in the garment and rest of the economy sectors respectively, and capital is treated as the mobile factor. Textile is the tradable input which is used in garment sector. Households are endowed with female labor, male labor and capital and they also receive transfer payments from the government funded through tariff revenue. The model is calibrated to a benchmark equilibrium data set for in Sri Lanka for the year 2000. Two policy experiments are conducted. The first experiment is to assess the impacts of a lower (export) price for garments. The second experiment is to assess the impacts if a higher import tariff for textiles prevailed. The results of the first simulation experiment suggests that if the world market price of garments falls then the gap between male-female wage rates will rise. The result of the second simulation experiment shows that if a higher tariff is charged on textiles the gap between male and female wage rates would have been higher. These results together suggest that the expansion in the garment industry may have helped to reduce the wage gap between males and females in Sri Lanka. 2

. Introduction Globalization and its impact on wage gap between skilled and unskilled laborers have become a crucial research issue over the recent past (Albergo and Whalley, 2000; Leamer, 995). In many countries the process of globalization involves expansion in trade due to reduction in import restrictions and due to promotion of exports. Generally, developed countries have the advantage of producing skilled labor intensive goods and developing countries have the comparative advantage in producing unskilled labor intensive goods. As a result of trade expansion, developed countries may specialize in skilled labor intensive products and export them and import unskilled labor intensive goods. Developing countries may specialize in unskilled labor intensive products and export them and import skilled labor intensive goods. According to neo-classical trade theory, this phenomenon could lead to rise in wage for skilled laborers and a decline in wage for unskilled laborers in developed countries and vise versa. Consequently, a rise in wage for unskilled laborers and a decline in wage for skilled laborers can be expected in developing countries. As a result, developed countries may expect a rise in the wage gap and developing countries may expect a drop in the wage gap with globalization. The same set of issues are arisen when female workers as unskilled workers and male workers as skilled workers are treated (DFID, 2000). Latter is evident in Sri Lanka, with the new era of globalization, which has started in 977 with the introduction of open economic policies. Liberalization of imports and promotion of exports have been the two key aspects of this process. Export promotion involved establishment of export development fund, free trade zone and export development board. Export incentives in the form of export rebates and tax holidays mainly for non-traditional exports were enhanced. Import liberalization also facilitated the importation of intermediate and investment goods for manufacturing exports. As a result of these policies, a significant growth is shown in the exports of unskilled labor intensive products, i.e., garments. Textiles and garments, which became Sri Lanka's largest single item of exports in 986, continues to maintain its dominant position, increasing its share of total exports from 2% in 977 to 28% in 986 and further to 54% in 2000. According to the Central Bank, 2000, garment sector alone contributed to Rs. 206,359 million. of total export earnings, 6.42% to the GDP and 63.3% of total industrial exports (table ). Garment industry absorbs a large number of female unskilled laborers and currently 230,000 work in the garment factories (Erzan et al., 99; Edwards, 996; Edwards, 997). These workers earn a relatively a higher wage than they earned earlier (Table 2). It should be noted that expansion in garment industry has increased the importation of textile, the raw material in the production of garment, and increased the exportation of garments, with a heavy dependence on the world market prices of textile and garments. As a large part of the gross input (65-70%) of the garment industry is imported, the value added and employment generated by the textile and garment industry is about 30% of the total manufacturing sector (Edwards, 996). 3

Table : Economic Significance of Garment Industry in Sri Lanka Year % contribution to GDP % contribution to export earnings Wage rate (index numbers) 980 3.08 0.27 38.8 985 6.40 2.90 203.9 990 9.60 30.50 379.5 995 9.27 43.46 65.6 2000 7.22 49.0 889.3 Table 2: Gender-Based Wage Rates in Manufacturing Year Average Males Females Difference (%) USD/month USD/month USD/month 985 33.0 38.2 27.9 26.96 990 4.7 43. 39. 9.28 993 48.85 5.4 46.3 9.92 996 64.06 66.89 6.23 5.66 Source: Gunatilake (999) Until the recent past an increase in garment price has been observed as shown in Table 3, which is another reason for expansion in the garment industry. This was mainly due to the negotiations with the exporting countries as suggested by Multi-Fiber Agreement. Furthermore, it was found that textile price has increased marginally over the last decade as shown in table 3. Though there is an increase in the nominal price of textile, reduction in import tariffs due to open economic policies helped to offset some of these increments. This change has also helped to expand the garment industry. Table 3: Price indices for textile and garments (990=00) Year Garments Textile 990 00 00 992 48.3 23.3 993 63.2 34.4 994 67.9 43.8 995 88.5 52.2 996 203. 203. 997 232.9 67.9 998 267.85 75.64 999 269.0 76.64 2000 304.65 98.47 The objective of this study is to assess the impact of increase in garment price and decrease in textile price due to liberalization of tariff on male-female wage gap in Sri Lanka. The changes in the economy such as production, consumption and trade will also be documented. A general equilibrium model is used for the analysis. 4

The paper is organized as follows. The next section summarizes the general equilibrium modeling done for Sri Lanka. The third section presents the model. The fourth section presents the data and calibration. Results are presented next. The paper ends with a conclusion. 2. General Equilibrium Modeling for Sri Lankan Economy A number of studies has been conducted for Sri Lankan economy to assess the impacts of economy wide policies on the wage rates in different sectors in Sri Lanka using general equilibrium models. These studies however did not specifically assess the impacts of different policies on the male female wage gap. Bandara (990) reviews the recent experience of general equilibrium modeling in Sri Lanka. The first major attempt at constructing a CGE model for Sri Lanka is by de Melo (978). His model is based on the SAM of 970 and belongs to the tradition of world Bank CGE models (Adelman and Robinson; 978). The model consists of six categories of households. Various categories of unskilled labor were considered as variable factors and other factors including skilled labor, land and capital were considered as sectorspecific fixed factors. This model has been used to examine the effects of selected agricultural policies (de Melo, 979), effects of alternative approaches to basic human needs (de Melo, 98), and the effects of alternative development strategies (de Melo, 982). Blitzer and Eckaus (986) developed a model for Sri Lankan economy, which departs from de Melo (978) mainly in terms of policy focus. Two consumer groups, urban and rural are identified and the base year for the model is 983. Jayawardena et al., (987) developed a structuralist model, considering structural rigidities in the supply side. The agricultural sector has a fixed supply and manufacturing and services sectors have demand determined output. There are four income classes: agriculture, estate wage earners, non-agriculture wage earners, and profit earners. Bandara (989) and Center for International Economics (99) have developed two models for Sri Lankan economy which uses Johansen (960) method for solving. The equation system of the model developed by Bandara (989) closely follows the ORAI model of Australian economy (Dixon, et al, 982). Location, i.e., urban, rural and estate is the basis for the categorization of labor and households into three types. It utilizes the 98 input-output data base which allows to include 24 industries each producing one of the 24 commodities. Model developed by Center for International Economics (99) is similar to Bandara (989), however consists of 35 industrial sectors and three primary factors. It uses a new input-output data base for 989 using the previous input-output table for 986. 5

Herath (994) developed a general equilibrium model to assess the economic implications of implementing a major irrigation project, Mahaweli, using SAM for 98. Three household types, urban, rural and estate are considered in this model. Bandara et al. (997) use a general equilibrium model, which incorporate on-site, and offsite impacts of soil erosion to the model of Bandara (989) to show that trade liberalization and combination of some tax and subsidy policies would reduce the environmental costs of soil erosion. 3. A General Equilibrium Model to Assess Male-Female Wage Gap In this study too, a general equilibrium model is used as the analytical approach. The model developed includes 2 sectors; i.e., garments and rest of the economy. Female labor and male labor are the specific factors of production for garments and rest of the economy respectively. Capital is considered as a mobile factor and textile is considered as a tradable input in the garment sector. Consumers obtain income from wages and government transfers made using tariff revenue and from capital rents. Following section describes the general structure of the model. Ricardo-Viner model with specific factors is an appropriate structure to analyze the structure in Sri Lanka as it can capture the asset specificity in the short run as described above. The model consists of 5 blocks, i.e., production, consumption, equilibrium conditions, price linkages, and government revenue. 3. Production Production functions are given below denoting Y for production, L for female labor, M for male labor, K for capital and for textile. Subscript i indicate garment and rest of the economy. ote that production of garments requires female labor, capital and textiles where as production of other commodities only require male labor and capital. Production technology is assumed to be Constant Elasticity of Substitution. Y Y ρ ρ ρ / ρ = A ( α L L + α K K + α ) () ρ 2 2 L K 2 ) 2 ρ2 / ρ2 = A ( β M + β K (2) Demand for female labor and textile can be derived from above production functions assuming that the objective of producers as profit maximization. The first order conditions of the profit maximization problem need to be obtained only with respect to variable inputs. The first order conditions with respect to capital and textile in sector and capital in sector 2 can be used to obtain the following expressions. ρ WK K α K = (3) ρ ρ i ρ W L + W K + W L K 6

α W K K W = ρ ρ i + W ρ i (6) ρ2 WK K 2 β K = (5) ρ2 ρ2 WM M + WK K Where, P i denotes output price and W i denotes input prices. 3.2 Consumption Consumers derive utility by consuming garments and the products produced by the rest of the economy. Demand functions derived by assuming utility maximization as the behavioral assumption of the consumers is given by the following system. Max U = D D λ λ 2. s. t I = P D + P D 2 2 (6) where U is the utility, D is demand and I is income. The solution to above problem are given by (7) and (8), I D = λ (7) P ( λ) I D2 = (8) P 2 Where D is the demand and I is the income by the households. Households derive income from the wages earned from labor, interest on capital and tariff revenue. ote that textile is a tradable input and expenditure on textiles go to the textile sellers in the rest of the world. Income of the households is given by the following equations. I = W L + W K + W M TR (9) L K M + 3.3 Equilibrium conditions Capital market clearing conditions are given by the following equations. K = K + K 2 (0) Good market clearing conditions are given by the following equations. Y i = Di + EXPi () Assuming perfectly competitive markets, zero profit conditions determine the return to each factor as follows. P Y = WL L + WK K + W (2) P2 Y2 = WM M + WK K (3) Trade balance condition is given by, P Y = P2 Y2 + W (4) 7

3.4 Price linkage equations Textile price is linked with the rest of the world price through tariff as follows. W W = W ( tariff ) (5) + 3.5 Tariff revenue Tariff revenue is given by, W TR = W tariff (6) The model consists of all the equations listed above. Endogenous variables of the model are: production, consumption, trade, returns to fixed factors and rental rate of capital. Exogenous variables of the model are prices of garment, rest of the economy and textiles, tariff on garment and endowment levels. 4. Data Data to represent equilibrium in year 2000 were used for the analysis. ational Data on Sri Lankan economy are obtained from the Central Bank Report, 2000. Gross Domestic Product (GDP) in year 2000 is taken as Rs.,255 billion at market prices and it is considered that it comes from 2 major sectors: garments and rest of the economy. Value of production of garments is Rs. 25.686 billion. To obtain the production in the rest of the economy garment production is deducted from GDP. Value of garment exports is Rs. 206.359 billion and the value of textile imports is Rs..367 billion. The value of exports was considered as equal to value of imports and the value of imports by the rest of the economy is obtained by deducting the value of garment exports by the value of textile imports. Using the value of exports, consumption of garments and consumption of the products produced by the rest of the economy are generated. Using the proportion of wage bill in garment sector, value of the female labor in garment sector is obtained. Value of textile input in the garment sector is considered as the value of textile imports. Assuming zero-profits the value of the capital input is obtained. Similarly, using the proportion of wage bill in agricultural sector, value of male labor in the rest of the economy is obtained and assuming zero-profits the value of the capital input is obtained. Output and input price levels are set at one to generate respective quantity levels. Table 4: Values of production, trade and consumption (Rs. Billion) Sector Production Female labor Capital Male labor Textile Exports Garment 25.65 25.882 78.405 0.367 206.324 Rest 039.86 0 623.923 45.940 0-94.957 8

5. Results and Discussion The model is used to assess the impact of a decrease in price of exportable in the world market and increase in tariff on textile. Two assumptions are made regarding the elasticity of substitution between two inputs. First, elasticity of substitution is treated as one. In order to assess the impacts under this assumption, Cobb-Douglas production functions and input demand functions derived from Cobb-Douglas are used. Second, in the CES production function, ρ value is treated as 0.5. In the following discussion, impacts under both assumptions are reported (Table 5-7). Columns with the heading Cobb-Douglas and CES show the results of Cobb-Douglas production functions and CES production functions respectively. The columns with the heading garment price and textile tariff refer to results after reducing garment price by 0% and after introducing textile tariffs by 35% respectively. Table 5: Impact of policy shocks on wage, income and production Variable Base Cobb-Douglas CES Garment price shock Textile tariff shock Garment price shock Textile shock Male wage rate.040.05.026.028 Female wage rate 0.449 0.303 0.533 0.485 Wage Gap 2.38 3.468.924 2.9 tariff Table 6: Impact of different policies on capital allocation Base Cobb-Douglas CES Garment price Textile tariff Garment price Textile tariff Rest 78.405 36.08 24.54 23.07 9.3 Garment 623.923 666.246 677.787 679.257 683.97 Table 7: Impact of different policies on Consumption and Trade Measure Base Cobb-Douglas CES Garment price Textile tariff Garment Price Textile tariff Demand for 9.327 0.22 9.238 0.246 9.52 garments Demand for rest 34.87 9.9 23.92 2.99 25.73 Imports of textiles.367 49.93 24.983 3.689 4.386 Exports of 206.324 97.2 56.057 65.5 4.57 garments Imports by the 94.957 37.56 3.092 27.27 27.3 rest 9

5. Garment price shock: In order to simulate the equilibrium in the economy when lower prices were faced for the garment, policy analysis was conducted by setting garment price at.9. If lower prices were prevailed in the world market, wage rate of the male would be higher compare to the wage rates of females. The gap between two wage rates would have been higher. A higher impact is seen with the Cobb-Douglas formulation for all the variables suggesting that elasticity of substitution between two inputs play a key role here. 5.2 Textile tariff shock: In order to simulate the equilibrium in the economy when higher prices were faced for the tradable input, policy analysis was conducted by setting tariff rate on textiles at 35%. If higher tariffs were in place, wage rate of the male would have been higher and the wage rates of females would have been lower. The gap between two wage rates would have been higher. A higher impact is seen with the Cobb-Douglas formulation for all the variables suggesting that elasticity of substitution between two inputs play a key role here. 6. Concluding comments These results suggest that recent policy change implemented by the Sri Lankan government to liberalize textile market has improved the economic growth and reduced the male-female wage gap. The higher prices received for garments, as a result of the Multi-Fiber Agreement also show similar impacts. Sri Lankan economy may face adverse impacts once the export quotas received by Muti-Fiber Agreements over in year 2004. Furthermore, the results show that though a similar pattern can be observed in models with different degrees of substitutability among inputs, magnitude of the impacts are relatively higher when elasticity of substitution is lower. Researchers hence should pay special attention on the parameter, elasticity of substitution. References Adelman, I. and S. Robinson (978) Income Distribution Policy in Developing Countries: A Case Study of Corea, California: Stanford University Press. Albego, L. and J. Whalley (2000) Decomposing Wage Inequality Change Using General Equilibrium Models Unpublished manuscript, University of Warwick. Bandara, J.S. (989) A Multisectoral General Equilibrium Model of the Sri Lankan Economy with an Application to the Analysis of the Effects of External Shocks, Ph.D. Thesis, La Trobe University, Melbourne. Bandara, J.S. (990) Recent Experience of Computable General Equilibrium (CGE) Modeling in Sri Lanka: A Survey, The Journal of the Sri Lanka Association of Economists, Vol. 5, o. &2, pp.57-75. Bandara, J. S. (99) Computable General Equilibrium models for Development Policy Analysis in LDCs. Journal of Economic Surveys. Vol.5, o.. University of ew England. 0

Blitzer, C.R. and R.S. Eckaus (986) Modelling Energy- economy Interactions in Small Developing Countries: A Case Study Sri Lanka, Journal of Policy Modeling, Vol. 8, pp. 47-50. Annual Report, Central Bank. Sri Lanka, various years. de Melo, M.H. (978) A General Equilibrium Investigation of Agricultural Policies and Development Stratergies: A Case Study of Sri Lanka: Ph.D dissertation, University of Maryland, California. de Melo, M.H. (979) "Agricultural Policies and Development: A Socio-economic Investigation Applied to Sri Lanka", Journal of Policy Modeling, Vol., pp. 8-00. de Melo, M.H. (98) "Modeling the Effects of Alternative Approaches to Basic Human eeds: Case Study of Sri Lanka", in D. Leipziger (ed.) Basic needs and development (pp. 37-80). Cambridge, Mass: Oeleschlager, Gunn and Hain. de Melo, M.H. (982) "A Simulation of Development Strategies in an Economy- wide Model", Economic Development and Cultural Change, Vol. 30, pp. 339-349. DFID (2000) Trade Liberalization and Job Creation: Making Globalization work for Women. DFID White Paper. Dixon, P.B., Parmenter, B.R., Sutton, J. and D.P. Vincent (982) ORAI: A Multisectoral Model of the Australian Economy, Amsterdam: orth Holland. Edwards, C. (997) Export Processing Zones in Sri Lanka - Costs, Benefits, Profits and Policy Issues. Research Studies : Working paper series o. 6. Institute of Policy Studies. Sri Lanka. Edwards, S (996). The Uruguay Round and MFA Quotas; The textiles and Garment industries in Sri Lanka-The next 0 years. Institute of Policy Studies. Colombo. Erzan et al (99) In The Uruguay Round and MFA Quotas; The textiles and Garment industries in Sri Lanka-The next 0 years. Institute of Policy Studies. Colombo. Gunathilake, R. (999). Labour Legislation Female Employment in Sri Lanka s Manufacturing Sector. Research Studies. Labour Economics Series o. 4. Institute of Policy Studies. Colombo. Hearth, P.W.R.B.A.U. (994) An Investigation of the Impact of Some Recent External and Internal Shocks on the Economy of Sri Lanka, Central Bank of Sri Lanka. Jayawardana, L., Maasland, A. and P.. Radhakrishnan (987) Stabilization and Adjustment Policies and Programmes: Sri Lanka, WIDER Working Paper, WIDER, Helsinki. Leamer, E.E., (995) In Search of Stolper-Samuelson Linkages Between International Trade and Lower Wages Johanson, L. (960) A Multisectoral Study of Economic Growth, Amsterdam: orth- Holland. Sandarathne,. (999) Economic Growth, Structural Change and Economic Infrastructure in Sri Lanka in the Second Half of the Twentieth Century. Postgraduate Institute of Agriculture, University of Peradeniya, Sri Lanka