TECHNOLOGICAL SPILLOVER FROM FDI: EVIDENCE FROM TUNISIAN MANUFACTURING INDUSTRIES

Sami REZGUI (University of Tunis El Manar)

There has been a lot of interesting contributions on technological spillovers generated by FDI and benefiting to countries and also to firms (R.Caves, 1974, Wang and Blomstrom 1992, Harrisson and Haddad, 1993, Blomstrom and Sjohlm 1998, Aitken and Harrisson, 1999, V.Kathuria 2000…). The main purpose of this paper is to examine to what extent the presence of FDI within a firm and within an industry could explain a positive evolution of the firm efficiency. Based on computed indices of total factor productivity (TFP) from Tunisian panel data, and following Farell's stochastic production frontier's approach (1957) we could determine the evolution of efficiency at the firm level. We use the Schmidt and Sickles (1984) approach assuming time-invarying technical efficiency. This approach could be justified by the short period (1996-1998) considered in this study.

Using microeconomic data, Harrisson and Haddad (1993) concluded for Morocco that there was no spillover from foreign investment. Firms' TFP is not affected by the presence of foreign ownership. The authors used the Schmidt and Sickles (1984) approach of measuring technical efficiency on panel data. Aitken and Harrisson (1999) consider quite small the net effect of foreign ownership on the Venezuelian economy. A positive effect of FDI on TFP is only found in firms with less than 50 employees and is assumed to be largely internalized by joint ventures. The authors conclude at the absence of spillovers from FDI benefiting to domestic firms. For the case of India, V.Kathuria (2000) found that the presence of foreign owned firms in a sector leads to reduced dispersion in efficiency mostly for the firms investing in R&D activities. The author used the Cornwell & ali (1990) approach to estimate time varying technical efficiency levels for individual firms justified by the period considered (14 years).

Our own hypothesis is that technological spillover from FDI varies and depends not only on the characteristics of the firm (size of the firm, % of FDI, exports, Market share, skills), but also on the characteristics of the industry to which the firm belongs (concentration ratio, foreign capital penetration).

TFP' indices are derived from fixed effects panel data assuming Cobb-Douglas production function with constant returns to scale (Y_ijt = a_ijt F(K_ijt, L_ijt) The residual terms (TFP) are derived from the LS regression :

Log Y_ijt = log (a_ijt) + b log (X_ijt) + e_ijt, Y_ijt = output of the firm i belonging to the industry j at time t ; X_ijt = vector of inputs ( Labor and Capital).

According to the Schmidt and Sickles method, relative technical efficiency at the firm level is measured as follows:

We consider Log a_ijt as the estimated TFP of the firm i belonging to the industry j at time t computed in logarithm. Let log a_j be the highest efficiency level in the industry j for the three years of our study :

Log a_j = max [log a_ijt]

Let q_ijt = log a_ijt - Log a_j Ş (a_ijt / a_j) = exp (q_ijt)

Relative Technical Efficiency at the firm level that we note p_ijt, is:

p_ijt = exp (q_ijt) = (a_ijt / a_j)

p_ijt could also be considered as a measure of the dispersion in productivity. Our purpose is to explain the variation of p_ijt. A positive variation of p_ijt should be interpreted as a move of the firm toward the efficient frontier.

Let Dp_ij = p_ij98 - p_ij96

We then estimate the following model :

Dp_ij = F(skills, market share, export, %FDI, concentration ratio, FDI at the sector level, interactions, dummies).

We also allow for interaction effects in order to capture the role of the human factor measured by the skills variable and to measure the intensity of spillovers linked concomitantly to FDI and to the market structure. Dummies are also used to capture specific industry effects.

This model is estimated using two digit industry data for 165 firms belonging mainly to the food, textile and clothing, mechanical and electrical industries, chemicals, paper and plastic products and cement. We find that technological spillovers from FDI benefit essentially to small and medium size firms (with less than 100 employees).