The links between openness and productivity in Mediterranean countries

Abstract

We examine the relation between the international trade, the foreign direct investment and the total factor productivity of the Mediterranean partner countries of Europe within the framework of a cointegrated panel model. The results, obtained from data on seven Mediterranean partner countries of Europe (Algeria, Egypt, Israel, Jordan, Morocco, Tunisia, Turkey), show that FDI and human capital are complementary in the acquisition of productivity gains. We identify the threshold level of human capital from which the received foreign investments generate beneficial effects. In a more general way, the improvement of the total factor productivity via the international openness results only from the indirect effects related to the transfer of technology.

Notes

¹ Maghreb (Algeria, Tunisia, Morocco), Mashrek (Egypt, Jordan, Lebanon, Syria, Palestine), Turkey, Israel, Cyprus and Malta.

² Romer (1990) and Grossman and Helpman (1995) proposed the first theoretical models.

³ Lucas (1988), among others, builds a model that clarifies the role of human capital accumulation for growth.

⁴ Their work can be seen as an extension of the research carried by Griliches (1988) and Coe and Helpman (1995) on the positive impact of domestic R&D accumulation on productivity.

⁵ The first indicator is a variable that captures interventions by the World Bank. The second one is inspired by Dean et al. (1994) and takes into account four elements: tariffs, quotas, exports subsidies and impediments and exchange-rate misalignments. The third indicator is the one derived by Sachs and Warner (1995), that takes five criteria into account: the scope of tariff-free products, the average tariff rate, the black market premium, the tact that the economy is socialist-ruled or not and the existence of a State monopoly on main exports. The conclusions of this article remain unchanged whatever the indicator in use.

⁶ Endogenous growth theory considers this possibility. Grossman and Helpman (1991, chaps. 18 and 19) set out several cases where free trade leads to lower productivity growth in technologically backward countries.

⁷ There are three sectors, and each of them has its own constant rate of technical progress (one sector has slow technical progress, another one has medium technical progress and the third one has fast technical progress). Least developed countries have access to the best technologies, but with a longer time lag. They have, therefore, a comparative disadvantage in those sectors where productivity growth is the strongest. The idea according to which international trade facilitates technological transfers is incorporated through the assumption that the technological gap of a given country in a given sector is inversely related to the openness rate in that sector. Long-run overall TFP growth rate is the production-weighted average of sector growth rates.

⁸ For example, Coe and Helpman (1995) find that investment in R&D from OECD countries creates some 30% additional income for other OECD countries (which is well above the income impact of domestic R&D investment), through an increase in their TFP. According to Bernstein and Mohnen (1998), R&D spillovers from the USA towards Japan is well above those from Japan towards the USA.

⁹ Their study covers Oman, Morocco, Saudi Arabia, Jordan, Tunisia and Egypt over the period 1978–1998. It is worth mentioning, though it is often the case, that the explanatory power of their equations is rather weak.

¹⁰ They find that a positive impact of FDI on growth only appears when the capital stock is above 0.52 years of secondary school for adults. They compute the average education level as follows: if 10% of the adults above 25 reached the secondary school level and if 75% completed the full 6-years cycle whereas the remaining completed the first 3 years, the secondary school level is: 0.1 (6 * 0.75 + 3 * 0.25)  = 0.52.

¹¹ See Sjoholm (1999) for the technology spread and Nelson and Pack (1999) about absorption capability.

¹² First constant returns to scale are assumed, so that one attributes to TFP what should perhaps be attributed to externalities. However, choosing a production function that displays increasing or decreasing returns to scale requires to give a correct value to the parameter driving returns to scale, which seems no less restrictive. Second, capital stock estimates are even less reliable for DCs that they are for developed countries. Finally, in connection with the previous argument, TFP measures are highly sensitive to the share of each factor in value added. Therefore, we relied on previous work about PM countries to get an estimate of labour and capital share.

¹³ Senhadji (2000) weights labour by human capital in order to get his labour share. He then obtains weaker  β coefficients. However, quality-adjusted labour series  à la Senhadji are only available for the period 1960–1994 and are very tricky to construct. Therefore, we prefer to work directly on the basis of employment. Qualification will be incorporated later in our regressions on TFP determinants.

¹⁴ In particular, the aforementioned papers, which make use of data sources partly different from ours.

¹⁵ Average yearly growth rates, which are close from the average of yearly growth rates, amount to 0.46% for Morocco and 0.36% for Israel.

¹⁶ Keller (1998) also strongly criticized the measure put forward by Coe and Helpman: using random weights for trade partners, he obtains more foreign technology diffusion effects and a better fit in terms of R ².

¹⁷ The authors obtained similar results when they replaced gross fixed capital formation of country j by its GDP.

¹⁸ These variables are government consumption, the black market premium, a measure of political instability, a measure of political freedom, a proxy for development level of the financial system, the inflation rate, and a measure of the quality of institutions.

¹⁹ The number of data available for Mediterranean countries is, anyway, very low. This makes the construction of domestic R&D stocks series particularly tricky and difficult.

²⁰ In the building of the interaction variable FDI × E, the schooling rate is not expressed as a percentage, in order to get a coefficient whose order of magnitude can be compared to the coefficient on FDI.

²¹ Again, one should be aware of both the advantages and drawbacks of the schooling rate as a proxy for human capital.

²² More precisely, this level was reached in 1984 by Egypt, in 1987 by Algeria and in 1994 by Tunisia and Turkey. Jordan had a higher rate in the early 1980s, which has deteriorated afterwards up until 1992, and then became higher again in 1995.