GEOGRAPHY, INTERNATIONAL TRADE AND TECHNOLOGICAL DIFFUSION

Sami REZGUI (University of Tunis El Manar)

We try in this paper to investigate the impact of geography on technological diffusion via international trade. On one hand, as is well developed in some theoretical literature since Marshall (Krugman ,1991) and in many empirical works inspired by the Jaffe's approach based on patenting (Eaton et Kortum, 1996, Keller, 2001), it is considered that technological diffusion could be influenced by geographical proximity. On the other hand, studies based on gravitational models show that bilateral trade between countries is significantly dependent upon geographical factors (Frankel & al,1996 ; Frankel & Romer, 1999). In this literature, physical distance, in particular, is showed to be negatively correlated to imports. Accordingly, it is supposed that this negative correlation could influence developing countries' growth (Gallups & ali, 1998).

On this basis, the Coe and Helpman (1995) study of technological spillovers between North and South and linked to international trade deserves more investigation with regard to the arguments of the economic geography. In fact, technological diffusion does not seem to be so international in scope as claimed by Grossman and Helpman (1990). It is also interesting to note that in his last NBER working paper, W.Keller (2001) showed that distance influence much more the intensity of technological diffusion between the frontier's technology countries than trade does.

Empirical investigation

Inspired by the Frankel and Romer work, we first examine the explaining power of geographical factors such as distance, population and country area for observed values of unilateral imports. Data considered concerns 42 countries that trade with the G6 countries (USA, Japan, France, Germany, United Kingdom and Italy). For this task, we use a unilateral trade model defined by equation (1):

Log M_ij = a₀ + a₁LogD_ij + a₂Log P_i + a₃ log S_i + e_ij (1)

M_ij = Imports of country i from country j (country j belongs to G6)

D_ij = Physical distance separating country i from country j

P_i = Population of country i

S_i = Area of country i .

e_ij = residual term

We then confirm previous results obtained by the authors especially the negative correlation between distance and unilateral imports for the countries considered. We find also that the fitted values of unilateral trade explain 30% of the observed values of unilateral trade for the years 1982 and 1995.

In the second part of our empirical investigations, we use the Coe, Helpman and Hoffmaister (1997) model to investigate the effect of geographical proximity on technological diffusion considering panel data for the Mediterranean countries, MENA countries, Central and South America countries and South East Asia countries. The data used covers the period 1982 to 1995. Estimations are done following equation 2:

Log TFP_it = l₀ + l₁ log M_it + l₂ logM_it*log(åS_ijt) + m_ijt (2)

TFP_it = Total Factor Productivity of country i for year t.

M_it = Imports of country i from G6

S_ijt = Foreign stock of knowledge benefiting country i proportionately to its imports from country j at year t.

m_ijt = residual term

Foreign stock of knowledge is measured according to the keller's approach which integrates a depreciation ratio of knowledge capital. The Data on R&D spending for the G6 countries considered are obtained from Main Science and Technology Indicators Data base (1999). Estimations are made for each region according to its proximity to G6 countries. For example, we consider imports from European G6 countries for Mediterranean and MENA countries and imports from USA for Central and South America countries. We then compare the intensity of technological diffusion by measuring elasticity of real GDP with respect to imports for each region.

Implications and Conclusions

Two main results are obtained :

- The geographical proximity of Mediterranean countries to the European G6 countries contribute to more technological diffusion in favour of these countries by comparison to MENA and Central and South America countries.

- Mediterranean countries observe a decreasing technological diffusion effect linked to their imports from European G6 countries. We show that a 1 percent increase in imports generates, via technological diffusion, a mean increase of TFP by 1.3 percent during the period 1982-1988, whereas technological diffusion effect on TFP are no more significant for the period 1989-1995.

The results we obtain are quite close to the conclusions of Keller's paper assuming less localized technological diffusion effects which materialize by a decrease of technological diffusion over time regardless to geographic location.