Welfare and poverty impacts of trade liberalization: A dynamic CGE microsimulation analysis

Trade liberalization is one of the major policy reform issues in Bangladesh. Stretching back to the 1980s, when successive governments pursued reforms to make the economy more outward-oriented there has been a remarkable simplification in import procedures, a substantial decline in quantitative restrictions and sufficient opening up of trade in many restricted items. The rationalization of the tariff structure and floating of the exchange rates may have facilitated this process of trade liberalization. Trade policy reform also introduced generous promotional measures for exports including lower rate of interests on bank loans, duty drawbacks and exemption from value-added and other taxes. All this has helped reduce policy-induced anti-export biases. Between 1991–92 and 2004–05 the un-weighted average rate of tariff fell from 70 percent to 13.5 percent. A drastic reduction in unweighted tariff rates during the 1990s also resulted in the fall in import-weighted tariff rates which declined from 42.1 percent in 1990–91 to 11.5 percent in 2003–04. In 1991–92, average industrial and agricultural tariff rates were 73 percent and 76 percent respectively. However, there has been greater reduction in tariff rates for the industrial sector during that period which resulted in average industrial tariff to be 26.7 percent as against 39.7 percent average agricultural tariff rate in 2003–04 (Raihan, 2008).

This paper examines the macroeconomic, poverty and welfare impacts of complete and unilateral domestic trade liberalization in Bangladesh.

A dynamic top-down CGE microsimulation model for the Bangladesh economy has been applied in the present study. This top-down microsimulation model is developed in line with Savard (2003) and Robilliard et al. (2008), and the dynamic CGE model for Bangladesh is an extension of the model developed by Annabi et al. (2005). The CGE model is calibrated with a Social Accounting Matrix (SAM) of Bangladesh for the year 2005. The 2005 SAM identifies the economic relations through four types of accounts: (i) production activity and commodity accounts for 26 sectors; (ii) four factors of productions: skilled and unskilled labour, agricultural and non-agricultural capital; (iii) four institutional agents: households, firms, government and the rest of the world; and (iv) two consolidated capital accounts to capture the flows of savings and investment by private and public institutions.

This model is a sequentially dynamic CGE model where the economic agents do not have any intertemporal optimization behaviour. Rather, these agents are myopic. This amounts to a series of static CGE models that are linked between periods with an updating procedure for dynamic exogenous variables (e.g. labour force, capital stock, etc).

The loss in government revenue due to any tariff cut is compensated by indirect or direct tax mechanism, which is inbuilt in the model. In each period the nominal exchange rate acts as the numéraire.

The model has an investment demand function which determines the pattern of reallocation of new investment among sectors after any shock according to the ratio of the rate of return to capital and its user cost. Total labour supply increases at an exogenous rate, which is equal to the population growth rate and the labour force growth rate. Other nominal variables, such as transfers and the minimal level of consumption in the LES function, and government savings, current account balance also increase at the same rate. An adjustment variable, which is introduced in the investment demand function, helps in bringing the equality between total savings and total investment in each period. The model allows all variables in the baseline to increase at the same rate in level, and the prices remain constant. This method is useful for the welfare and poverty analysis since all prices remain constant along the business as usual (BAU) path.

The poverty and welfare effects of different policy shocks are estimated using the Bangladesh Household Income and Expenditure Survey (HIES) 2005. A total of 10047 households were covered in this survey. Changes in poverty indices are determined by changes in the poverty line and changes in nominal consumption (or income). The variations in consumption for each household group from the dynamic model are applied to generate new consumption vectors for individual households from the Bangladeshi household survey. Two different poverty lines for rural and urban households are used, which are endogenously determined by the model taking into account the rural and urban CPIs. Changes in poverty indexes are determined by changes in the poverty line and changes in nominal consumption (or income). The poverty line represents the cost of a basic-needs basket of goods. If the change in poverty line is greater (smaller) than the change in nominal consumption, then poverty is likely to decrease (increase).

According to SAM 2005, the tariff rates vary across the sectors and range from as low as 0 percent to as high as 40.19 percent (food). Other Textile has the highest sectoral import penetration ratio (42.66 percent), followed by Other Industry (39.94 percent). The highest share in total imports is for Other Industry (65.89 percent), followed by Other Textile (17.55 percent). The sectoral export orientation ratio is the highest for Knit RMG (99.32 percent) followed by Woven RMG (80.26 percent). Together Woven and Knit RMG exports account for 76.2 percent of total exports. In the case of value addition, all the service and construction sectors together account for 61.69 percent of total value added in the economy. The aggregated agricultural and the manufacturing sectors constitute 20.4 percent and 17.88 percent of the total value added respectively.

A scenario incorporating full trade liberalization, where tariffs on all imports are reduced to zero, is simulated. The base values of all other parameters are retained.

The macroeconomic impacts are reported in Table 1. The impacts on GDP and welfare illustrate the importance of analyzing trade liberalization in a dynamic framework; both measures decline in the short run (2008) and then strongly increase in the long run (2020) compared to the business-as-usual (BaU) simulation. The short-run negative impact is explained by the fact that trade liberalization contracts the import-competing and highly protected sectors, and capital cannot be quickly reallocated to the expanding export-oriented sectors. Positive growth is observed in the domestic terms of trade (TOT, the ratio of export to import prices on domestic markets) in both the short run and the long run, given the decline in domestic import prices. Imports and exports register strong positive growth, particularly in the long run. Reduced domestic import prices lead to a fall in consumer prices for both rural and, slightly more, urban households. Skilled and unskilled wage rates decline, although less so in the long run when capital is reallocated toward the expanding sectors. The reduction in unskilled wage rates is somewhat smaller, given the expansion of unskilled labour–intensive textile and garment sectors. The user cost of capital also declines in both the short run and the long run.

Tariff elimination leads to an immediate reduction in the domestic price of imports that is proportional to the initial sectoral tariff rates. Domestic consumers respond by increasing import demand, once again in rough proportion to the fall in import prices, with the strongest increases in the grain mill, food, other textile and other industry. The sectors that had low initial tariff rates (grains, livestock, other fish, woven and knit ready-made garments) register negative import growth in the short run as consumers substitute toward goods for which prices drop more dramatically. In the long run, import volumes grow more (or contract less) in all sectors except leather.

The current account balance is fixed in the short run and subsequently increases at a fixed rate equal to the exogenous rate of population growth. Thus, the increase in imports leads to a real devaluation and an increase in exports. The export response is generally smaller in the long run, with the dramatic exception of woven and knit ready-made garments and other textiles. In the long run, the woven and knit ready-made garments sector flourishes, and their export volume increase by nearly 18 and 32 percent respectively compared to the BaU scenario (Table 2). With a negative sloping demand curve for exports, FOB export prices fall.

Output expands most in woven and knit garments and other textile sectors. Export-intensive ready-made garments benefit from export expansion, and all these sectors register input cost savings, as evidenced by the positive evolution in value added prices despite falling output prices. Greatly increased import competition for textiles is offset by increased input demand from the ready-made garments sector. In contrast, production contracts in the heavier manufacturing sectors for which export demand stagnates or declines. As a result, non-agricultural capital and labour migrate to the textile and garments sectors and away from the other manufacturing sectors, with relatively little movement in the agricultural sectors. In the long run, the non-agricultural capital stock response is much larger and tempers the reallocation of skilled and unskilled labour. There are also moderate capital stock increases in the agricultural and service sectors.

In the short run, nominal factor returns fall by roughly 10 percent as a result of declining domestic prices. Overall investment falls in response to the average reduction in capital returns relative to the user cost of capital. This makes the long-term reduction in wage rates somewhat smaller, especially for unskilled wages. The average returns to capital fall slightly more in the non-agricultural sector, although these rates converge after long-term adjustment in sectoral investment rates.

Under both the scenarios, a fall in nominal income for all households is observed in both the short run and the long run (Table 3). This reduction is smallest among the poorest households—urban households with illiterate or low-educated heads and rural landless or marginal households—given their reliance on unskilled wages. Medium- and high-educated urban households, as well as non-agricultural rural households, are the biggest losers as a result of their high endowments in non-agricultural capital and skilled labour. In the short run, real consumption decreases for all households as nominal income falls more than consumer prices. However, the opposite is true in the long run. The figures of Equivalent Variation (EV) are very much in line with real consumption growth, with the poorest household categories emerging as the biggest winners.

The poverty effects of two simulations are reported in Table 4. Foster-Greer-Thorbecke (FGT) poverty measures (Foster et al., 1984) are applied in this study. The FGT indices allow us to compare three measures of poverty: head count ratio, poverty gap index and squared poverty gap index. In the short run, the poverty head-count increases for all households, except those headed by highly educated heads, for which there is no change, and those headed by illiterate heads, for which poverty falls. Also the depth of poverty and the severity of poverty increase in the short run. However, in the long run, poverty indices fall dramatically for all households, especially among the poorer households. This suggests that accumulation effects captured by the model play a major role in alleviating poverty.

Use of a dynamic computable general equilibrium model coupled with household survey consumption data suggests that there are marked differences between the short run impacts and the long run impacts of tariff liberalization. In the short run there are possibilities of reduced welfare and increased poverty. However, in the long run, resources are reallocated towards the more efficient and expanding sectors, generating positive outcomes in terms of welfare gains and poverty reduction.

1. 1
   WTO agreements and domestic policy reforms – implications for poverty in Bangladesh: a study in a dynamic sequential CGE framework

   1. N Annabi
   2. B Khondker
   3. S Raihan

   et al. (2006)

   In: T Hertel, A Winters, editors. Poverty and the WTO: impacts of the Doha development agenda. Washington D.C.: The World Bank and Palgrave Macmillan. pp. 429–466.

   • Google Scholar
2. 2
   A class of decomposable poverty measures

   1. JE Foster
   2. J Greer
   3. E Thorbecke

   (1984)

   Econometrica 52:761–776.

   • Google Scholar
3. 3
   UNESCAP, Emerging trade issues for policymakers in developing countries in Asia and the Pacific

   1. S Raihan

   (2008)

   153–172, Trade liberalization and poverty in Bangladesh, UNESCAP, Emerging trade issues for policymakers in developing countries in Asia and the Pacific, Bangkok, UNESCAP.

   • Google Scholar
4. 4
   Crisis and income distribution: A micro-macro model for Indonesia

   1. A-S Robilliard
   2. F Bourguignon
   3. S Robinson

   (2008)

   In: F Bourguignon, L da Silva, M Bussolo, editors. The impact of macreconomic policies on poverty and income distribution: macro-micro evaluation techniques and tools. Houndmills: Palgrave-Macmillan. pp. 112–123.

   • Google Scholar
5. 5
   Poverty and income distribution in a CGE-household micro-simulation model: top-down/bottom-up approach. CIRPEE Working Paper 03-43

   1. L Savard

   (2003)

   Québec: Laval University.

   • Google Scholar

Author details

1. Selim Raihan

   Dept. of Economics, University of Dhaka, Bangladesh

   For correspondence

   selim.raihan@econdu.ac.bd