On Trade Policies and Wage Disparity: Evidence from Egyptian Microeconomic Data

Abstract

This paper proposes an empirical investigation of the effect of different trade barriers on wages in Egypt. The effect of trade barriers on wage disparity has been widely discussed at both empirical and public policy levels. This debate has mainly dealt with traditional tariff barriers. Less attention has been attributed to other barriers, such as non-tariff measures and red tape costs. However, these barriers – and in particular red tape costs –impede more than tariffs in developing countries. Thus, using a microeconomic dataset, this paper assesses to what extent different trade barriers affected wage disparities and employment in Egypt. These disparities are studied in three dimensions: on gender (males versus females), qualification (skilled versus unskilled), and regional (urban versus rural workers). The main findings show that red tape barriers have a higher impact than traditional tariffs on wage disparity. Female and blue-collar workers are more affected by such barriers. The effect of trade barriers on regional wage disparity seems to be less important then gender and qualification. Finally, when the effects of observable worker characteristics are filtered out, it turns out that wage premia are negatively affected by all trade barriers.

Keywords:

• Trade policies
• wage inequality
• tariffs
• non-tariff barriers
• gender
• Egypt

Notes

¹The definition of such a process splits trade facilitation aspects into four major parts: simplification of commercial procedures; harmonization of trade rules; transparent information and procedures and the recourse to new technologies allowing trade promotion (Zaki, 2008).

²For an extensive literature review, see Goldberg and Pavcnik (2004 and 2007b)

³The difference between the two papers lies in the liberalization of the agricultural sector that may have a significant effect on poverty in the short and medium run. While India experienced significant tariff reductions in the agricultural sector, agricultural trade liberalization in Colombia was limited. This is why Colombian poor in rural areas were not affected by the liberalization waves.

⁴According to Milanovic and Squire (2007), the link between policy reforms and wage inequality is likely to be stronger than the link between policy reforms and inequality in total income. The latter is affected by a number of other institutional factors such as the role of social transfers (pension spending or family benefits), demographics of the population and labor force participation. By contrast, wage inequality does not depend on such factors. Thus, the link between trade policy and wages must be stronger than that between trade policy and the distribution of total income.

⁵Egypt is divided into 27 governorates (sometimes called provinces), which include four city governorates: Alexandria, Cairo, Port Said and Suez. Also included are the nine governorates of Lower Egypt in the Nile Delta region, the nine governorates of Upper Egypt along the Nile River south from Cairo to Aswan, and the five frontier governorates covering Sinai and the deserts that lie west and east of the Nile.

⁶For the sake of robustness check, skilled and unskilled workers have been distinguished on the basis of their education levels. It turns out that figures do not change.

⁷For more details about the ERSAP effect, see Korayem (1997).

⁸All ad valorem tariff equivalent for the whole sample are available upon request

⁹Hourly real wages are calculated as the sum of wages earned in the reference month from primary jobs, adjusted for average number of work days per month and average hours per day.

¹⁰For more details about the wage premium estimation, see Gaston and Trefler (1994), Attanasio et al. (2004), Dutta (2007), and Said and El Azzawi (2009)

¹¹To remedy for the sensitivity of the estimated wage premia with respect to the omitted industry dummy, I follow Haisken-DeNew and Schmidt (1997) that have calculated the wage premia as deviations from an employment weighted mean as follows where is the normalized wage differentials, I is an identity matrix and W is a matrix of industry employment weights with each element , where n is the number of workers in industry s. Thus, equation 3 is estimated using the normalized wage differentials not the estimated ones.

¹²Those sectors are: food and beverages, tobacco, textiles, garments, leather goods, wood products, paper, publishing, coke and petroleum, chemical products, rubber products, non-metallic substances, basic metal, metallic products, machinery and equipment, electrical equipment, radio and television, medical equipment, transport equipment, and furniture

¹³Centre d'Etudes Prospectives et d'Informations Internationales. They are available on CEPII's website.

¹⁴According to the World Trade Organization, sanitary and phytosanitary measures are defined as any measures applied: to protect human or animal life from risks arising from additives, contaminants, toxins or disease-causing organisms in their food; to protect human life from plant- or animal-carried diseases; to protect animal or plant life from pests, diseases, or disease-causing organisms; and to prevent or limit other damage to a country from the entry, establishment or spread of pests. Measures for environmental protection (other than as defined above), to protect consumer interests, or for the welfare of animals are not covered by the SPS Agreement.

¹⁵To estimate the tariff equivalent of non-tariff and administrative barriers, the methodology of Kee et al. (2009) has been used.

¹⁶To formally test for the differences in the interaction terms, an F-test has been made based on a restricted model (in which the coefficients on the interaction terms are restricted to be equal) and an unrestricted one. The sum of squares from the restricted and the original unrestricted model are used to compute an F-statistic to test the hypothesis that the coefficients are the same. The F-test showed that the linear constraint (coefficient on the interaction terms) is rejected and therefore the coefficients are significantly different at 1% level of significance.

¹⁷I run the regressions on males only since the number of females in my dataset is quite small (around 150 women).

¹⁸Skilled workers are those who have higher than intermediate level of education and unskilled ones have a lower than intermediate education.