Patterns of internalization and third country effects : empirical evidence from the Czech Republic

Abstract

This master thesis aims at determining the patterns of internalization of multinationals and at evaluating the potential effects of third countries on inward FDI in the Czech Republic. To meet these objectives, I employed the knowledge-capital model of Markusen et al. (1996) and the extended spatial lag model of Blonigen et al. (2005) on a firm-level dataset of foreign affiliates located in the Czech Republic for the period from 2003 to 2008. Empirical analyses were conducted on data at different levels of aggregation to demonstrate the relevance of data disaggregation and heterogeneity. The results provided an empirical evidence for horizontal and vertical motives of internalization, with the prevalence of either motive varying across the sectors and the levels of aggregation. Effects of third countries on inward FDI were found to exert impact on a number of sectors of activity. Nevertheless, statistical significance of the results appeared to be highly sensitive to the geographical composition of data.

This master thesis aims at determining the patterns of internalization of multinationals and at evaluating the potential effects of third countries on inward FDI in the Czech Republic. To meet these objectives, I employed the knowledge-capital model of Markusen et al. (1996) and the extended spatial lag model of Blonigen et al. (2005) on a firm-level dataset of foreign affiliates located in the Czech Republic for the period from 2003 to 2008. Empirical analyses were conducted on data at different levels of aggregation to demonstrate the relevance of data disaggregation and heterogeneity. The results provided an empirical evidence for horizontal and vertical motives of internalization, with the prevalence of either motive varying across the sectors and the levels of aggregation. Effects of third countries on inward FDI were found to exert impact on a number of sectors of activity. Nevertheless, statistical significance of the results appeared to be highly sensitive to the geographical composition of data.