The effect of international trade on labour-demand elasticities: empirical evidence from Tunisia

Abstract

This paper investigates the effects of trade liberalization on labour demand elasticities. The employment demand equation is estimated by using data for manufacturing industries in Tunisia covering the period from 1971 to 1996. The empirical results suggest a weak support for the idea assuming that trade liberalization will lead to an increase in labour demand elasticities: in the vast majority of industries considered, we cannot reject the hypothesis of no relationship between trade openness and labour-demand elasticities. This weakness of labour demand elasticity in practice is perhaps explained by the tight labour market regulations in place during the 1987–1996. However, these results are robust to the type of labour considered (contract labour and permanent labour). This supports the conclusion that under liberalization labour markets have become more flexible, and that employers prefer recruiting contract workers.

Notes

¹Important contributions to the literature on the impact of globalization on wages and wages inequality have been made by Feenstra and Hanson (1996, 1997) and Hanson and Harrison (1999).

²There have, of course, been other studies of linkages between trade policy and labour markets: Revenga (1992), Harrison and Revenga (1995), Kambhampati et al. (1997) and Curie and Harrison (1997) are recent examples. Each of these papers focuses on the relationship between trade policy and employment and wage levels rather than on elasticities. Slaughter (1997), on the other hand, uses industry level US data to investigate the link between variations in openness in the recent decades and labour demand elasticities.

³In this regard, our study is set in a better context than the one examined by Slaughter (1997) whose focus instead is on US trade policy changes over recent decades, which were rather less pronounced and more difficult to measure.

⁴For evaluation of the effects of trade liberalization on employment and wages in other developing countries, see Brecher (1974), Beyer et al. (1999), and Levinsohn (1999).

⁵GATT (1994), TPRM, for unweighted average tariff rates and ranges.

⁶Of course issues of counterfactuals and causality arise. The expansion of exports may have been positively influenced (in part at least) by other external and non-policy factors.

⁷According to Lakhoua (1998) these reforms have been accompanied by important social achievements; these include intensification of job creation, income seems to be more evenly distributed than in most comparable countries and the incidence of poverty to be correspondingly low.

⁸This approximates a situation in which there are a large number of varieties and each firm is an infinitesimal player but has some power over the pricing of its product.

⁹The theory clearly applies to all input factors. However, we focus only on labour demand since any effort estimating demand equations for other factors would be frustrated by the fact that firm-level data on the use of the other factors and/or factor prices is not available.

¹⁰Using the annual survey on firms made by the INS derived permanent and contract workers.

¹¹Capital price is assumed equal to the price of acquiring new capital or PFCFB (‘prix de formation de capital fixe par branche d’activité').

¹²We find that the results are mostly invariant to the assumptions regarding interest rates. This is discussed in more detail in the next section.

¹³To the extend that, say, aggregate demand or productivity shocks increase product demand and raise labour demand and increase wages (or any other factor prices for that matter) at the same time, the elasticity estimates delivered from Equation 6 would be biased due to the correlation between the error term and the right-hand side variable.

¹⁴This parameter correspond to the capital price variable interacted with the liberalization dummy.

¹⁵Indeed, we indicate that Tunisian labour regulations allow firms to recruit by fixed term contracts and seasonal personnel, and to dismiss them without any indemnity at the end of their contract.

¹⁶While instrumental variables estimation is perhaps a more satisfactory approach, it proves a little less feasible in this context: other than lagged endogenous variables, we have no variable in the data set that we may regard as being exogenous. Estimating (10) using lagged variables as IVs resulted in highly insignificant (and sometimes meaningless) estimates of the parameters of interest. One option that presents itself then is the pooling of data across industries to use lagged variables as instruments. These results are presented in Table 5 and are discussed later along with other results from the pooled sample.

¹⁷We are should to note some caveats associated with pooling data across industries.

¹⁸For a given level of value added, chemical could be better of by reducing employment by 63.8% and textile by 59.8%.

¹⁹Employment adjustment is 23% in chemical industry and 6.1% in textile industry.