e theoretical discussion above, the hypotheses H1 to H8 are derived to empirically investigate the determinants of FDI in Turkey:
Hypothesis 1 (H1) An increase in the sum of world income leads firms to switch from exporting from home to host country to local production in host country.
Due to high transport costs, firms may serve the foreign country through local production rather than exports to meet the demand in the foreign location as a result of increasing world income. Exports are an option with high marginal cost, while local production involves the fixed cost of building a plant (Markusen, 1997).
I measure the national income size of the countries by gross domestic product (GDP) and an increase in the sum of incomes lead firms to switch from export to FDI because of transport costs (Markusen, 1998 and Markusen and Venables, 1998). I use the logarithm of total income (ln Sum) and the expected sign is positive.
Hypothesis 2 (H2) Similarity in size of two countries increases FDI between them.
The logic behind this hypothesis is that when countries differ in size significantly, production will be located in the one with the larger market, and the smaller market will be served through exports. Similarity of countries in terms of market size (GDPs) signifies the convergence in country size. The expected sign is positive since convergence in country size should motivate horizontal FDI (Markusen, 1998).
Similarity (lnSim) is calculated as the logarithm of similarity index (Simindex). Simindex takes values between 0 and 0.5, values close to 0 representing high difference in country size and values close to 0.5 indicate similarity in country size. Following Helpman (1987), similarity index is calculated by the formula;
(1)
where i subscript stands for home country i and h is Turkey and t denotes time.
Hypothesis 3 (H3) Difference in relative endowments (skilled labour or capital) will motivate vertical direct investment to exploit factor-price differences between the countries.
Knowledge generating activities are skilled-labour or capital intensive in contrast to production. Firms can locate knowledge generating activities in skilled or capital abundant countries and production in unskilled labour abundant or capital scarce countries. Skilled labour ratio difference is aimed to capture the endowment difference with respect to population with high education. High endowment difference in skilled labour should encourage vertical FDI given the ability of multinationals to locate production in countries abundant with unskilled labour. In order to measure skill difference, Markusen and Maskus (1998) and Carr et al., (2001) use the ratio of professional, technical workers, administrative and managerial workers to total employment. The authors use the Yearbook of Labour Statistics. However, the Yearbook of Labour Statistics has many gaps in time dimension; therefore I turn to other proxies. Gast and Hermann (2008) use the difference in share of population in agriculture as a proxy. This proxy does not control for endowment difference in skilled labour. Carstensen and Toubal (2004) use the number of students in education as a proxy. According to Carstensen and Toubal (2004), I employ the difference between home countries and Turkey in the ratio of gross education enrolment in tertiary education to the total enrolment covering primary,
