International Trade and Manufacturing Employment in the South: Four Country Case Studies

Abstract

This paper investigates the impact of international trade on manufacturing employment in developing countries, by undertaking a comparative study of four countries—Bangladesh, Kenya, South Africa and Vietnam. It does so by employing a variety of methodological approaches: factor content; growth accounting; and econometric modelling. The main empirical finding is that international trade seems to be associated with the net creation of jobs in Bangladesh and Vietnam, with female workers being the key beneficiaries. In contrast, international trade has been associated with adverse employment outcomes in Kenya, and possibly in South Africa. This suggests that there may be crucial differences between Asia and Africa in terms of the impact of globalization on employment opportunities in manufacturing. Some alternative explanations for such differences are offered in the paper.

Notes

¹ For reviews of this literature, see Cline (1997) and Lawrence (1996).

² See Rama (2003) for a recent review of the limited research in this area. In addition to Ghose (2003), two other studies that examine the trade-manufacturing employment relationship in the developing country context are Revenga (1997) for Mexico and Moreira & Najberg (2000) for Brazil.

³ This is the assumption made by the individual country studies in the NBER project on trade and employment led by Krueger, and is a fairly plausible assumption for most low-income developing countries.

⁴ Strictly speaking, intra-industry trade would not lead to any change in the composition of output because imports and exports are in the same industry. In practice, however, if intra-industry is defined in terms of relatively broad industrial categories while a more detailed classification of production is used, then it is possible for intra-industry trade to have some impact on the composition of output and hence on employment.

⁵ In their model the factors of production are unskilled and skilled labour and they show trade increasing the demand for skilled labour in the less developed country (Feenstra & Hanson, 1996).

⁶ That trade reforms often lead to productivity gains in the manufacturing sector is well documented in the literature—see Levinsohn (1993) and Harrison (1994).

⁷ For reviews of such studies, see Wood (1994, chapter 3) and Lawrence (1996, chapter 2).

⁸ The intuition here is that a high ratio of domestic absorption to domestic supply will tend to magnify the effects of changes in import penetration (which is measured in relation to domestic absorption) on employment that is related to domestic production (supply).

⁹ There have also been firm-level econometric studies of trade–employment linkages, but these are not discussed here.

¹⁰ We define the import penetration ratio for a particular industry as its imports as a ratio of domestic demand (i.e. imports + output − exports); while the export–orientation ratio is exports as a ratio of output.

¹¹ The main emphasis of the paper is on the impact of trade rather than trade liberalization on employment. This is a broader question that focuses on changes in a country's integration with the global economy rather than just on its trade policy, although of course trade policy is a factor that affects trade flows. A similar point is made with reference to the literature on trade and poverty in UNCTAD (2004, part II, chapter 1).

¹² Rodrik (2000) classifies Kenya as among the 10 countries with the largest proportionate cuts in tariff since the early 1980s.

¹³ It should be noted that the paper uses the International Standard Industrial Classificiation (ISIC) definition of manufacturing, which is broader than the Standard International Trade Classification (SITC) and includes processing of many primary products.

¹⁴ The trade data for the four countries came from the International Economic Database of the Australian National University and have been reclassified from COMTRADE data according to the International Standard Industrial Classification (ISIC Rev. 2). Since the trade data are only available at the four-digit level and in a small number of cases Krause used a five-digit classification, we have had to modify his groupings slightly.

¹⁵ A more conventional trade theory approach could regard both these categories as capital-intensive.

¹⁶ The years were chosen such that these were the most recent for which data on female and male employment were available for most industries. Unfortunately, data on employment were not available for South Africa by gender.

¹⁷ To compute the employment coefficients, we use employment per dollar of output rather than per dollar of value added as the export and import figures are in gross terms.

¹⁸ One limitation of the data used for this exercise is that they cover only employment in the formal manufacturing sector. In the case of South Africa where the industrial census covers firms of all sizes and the informal sector is relatively small, the data cover 80% of all manufacturing employment as measured in the Labour Force Survey, which includes informal employment, in 2001. In Vietnam too, most manufacturing employment is in the formal sector and the data cover 86% of the broad figure for manufacturing derived from the 1999 population census, which includes informal manufacturing. In Bangladesh, where the size of the informal manufacturing sector is considerable, employment in the formal manufacturing sector comprises 42% of all manufacturing employment according to the 1995–96 Labour Force Survey. In Kenya, comparable figures are not available as the data on employment do not distinguish between the rural and urban informal sectors—in 1998, 50% of all employment was in the formal sector.

¹⁹ Gross output and employment data came from GSO (2000) and export and import data for 1999 from GSO (2001). The trade data for 1995 were calculated by Mekong Economics by reclassifying data from the Vietnam trade database from the HS system to the ISIC.

²⁰ The term “competitiveness”, while widely used in the business literature, remains controversial, particularly amongst trade economists. For a discussion of these issues and a defence of the concept of competitiveness, see Lall (2001).

²¹ This is somewhat surprising given the difference in labour intensity between exports and import-competing products in Vietnam, as indicated in Table 2. The data here cover only a relatively short period in the late 1990s and, as Table 1 indicates, the sharp rise in the share of unskilled labour-intensive products in Vietnam's exports took place in the early 1990s, so that the picture for the 1990s as a whole might indicate a more significant impact of exports on employment through the composition effect. However, in another paper one of the authors has shown that composition effects have had a relatively limited impact on manufacturing employment in Vietnam compared to scale and process effects for the 1990s as a whole (Jenkins, 2004).

²² The wage and output data in UNIDO are in nominal values—in the absence of price data at the ISIC three-digit level, we deflated nominal output data by the GDP manufacturing deflator of the country in question.

²³ For example, positive shocks to productivity leading to a fall in employment may lead to a fall in import penetration or an increase in export orientation (via a fall in the price of the industry's product).

²⁴ See Dollar & Zeufack (1999) quoted in Kingdon et al. (2005) for a comparison between firms in Kenya and Thailand, and Edwards & Golub (2004) for a study of South Africa.

Additional information

Notes on contributors

Kunal Sen

This research was funded by the Department for International Development of the UK Government as part of a programme of research on globalization and poverty. We would like to thank the anonymous reviewer for detailed and incisive comments, which have considerably improved the paper. The usual disclaimer applies.