Determinants of foreign direct investment in Spain

Abstract

This study utilizes panel data as a means of examining the determinants of foreign direct investment (FDI) in Spain. Data that are taken in the period 1993–2002 are used in order to estimate the determinants of FDI, at the sectoral level, by differentiating the manufacturing sectors, and at the regional level. The analysis investigates the sectoral, regional and macroeconomic variables that have successfully attracted FDI inflows from those that have not. Empirical results suggest that the differential between labour productivity and the cost of labour has been an important determinant of FDI in Spain during the period 1993–2002. Factors related to demand, the evolution of human capital, the export potential of the sectors and certain macroeconomic determinants that measure the differential between Spain and the European Union average, also play a very important role in attracting flows of FDI. Certain policy issues that are relevant to the results are also discussed.

Notes

¹ Porter (1986) and Esteban and Vives (1994) underline the fact that neither production nor productivity depend solely upon the level of physical capital and work but also upon other variables, factors that are external to the firms, such as the level of infrastructure, human capital and technology.

² This generalization involves carrying out an extrapolation from the micro (or firm-level) to the macro (or country-level) and includes a number of determining factors which constitute a series of economic variables that might influence firms’ costs and incomes and, in consequence, the decision to invest in the foreign country.

³ These techniques provide certain statistical and econometric advantages, advantages that are highlighted in the works of Hsiao (1986), Dielman (1989), Baltagi (1995), Matyas and Sevestre (1996) or Greene (1998) among others.

⁴ In this article, the properties of the time series that were used were not studied because the number of annual observations was relatively small. As Shiller–Perron (1985) and Davison–MacKinnon (1993) point out, the power of the unit root test basically depends upon the chronological span of the data. In order to guarantee that the resultant regressions are not spurious, the relationships established by existing economic theory and other empirical analyses within the same field are taken into consideration.

⁵ The informatics system Econometric Views 5.0 has also been used in an attempt to include additional specific temporal effects. This option does not improve the results, rather, on combining the two types of effects (group and temporal), given the characteristics of the sample, tends to provoke situations in which there is perfect multicolinearity.

⁶ The Bartlett, Levene and Brown–Forsyhe tests indicated the presence of heteroskedasticity.

⁷ The result of the individual inclusion of SECPLP and SECCOST in the model was not significant.

⁸ To a certain extent, this indicates that, when foreign investment by companies holding foreign stocks is not included, as is the case of the manufacturing sectors, this variable ceases to be significant. This is because the main factor underlying the rise of this type of investment, which has an important relative weight as a proportion of total foreign investment, has been the fiscal incentives that Spain has offered with respect to the incomes obtained through foreign business investments (Ley 43/1995 del Impueste de Sociedades Boletin oficial del Estado (BOE) 1995).