Essays on investment climate in developing countries

Resumen

This thesis consists of three chapters on investment climate in developing countries. The first chapter examines the investment climate (IC) effects on the growth of developing countries. In contrast to past studies in the IC literature, which narrow their investigation to a subgroup of IC variables, I measure multiple dimensions of the investment climate in a single index. To construct an IC index, I use the methodology proposed by Escribano and Hacihasanoglu (2012 and 2013) to combine 87 firm-specific variables for 27,624 firms in 113 developing countries. This index overcomes multicollinearity and the dimensionality problems of the World Bank s enterprise survey database and makes it possible to compare different countries. I show both theoretically and empirically that IC as a whole does matter to explain cross-country income differences in developing countries. Once I control for the IC, trade or macroeconomic policy does not have any explanatory power on GDP per capita while geography has a weak effect on it. My results contribute to the institutions literature by showing IC as a specific type of institution to achieve higher levels of income. In chapter 2, joint with Alvaro Escribano, we develop an investment climate index (ICI) using enterprise surveys of the World Bank. We propose a simple methodology that allows us to combine a large set of continuous and binary IC variables into an index which proxy the good and the bad qualities of the investment climate. As a byproduct, we also construct IC sub-indices for five different blocks of IC variables: infrastructure; corruption and crime; nance; quality, innovation, and labor skills; and other control variables. When aggregating, we use two options, one with equal weights and the other with unequal weights of those IC variables. The unequal weights of ICI are obtained using principal component analysis (PCA), after transforming all IC variables to binary variables. We identify at least three important advantages of using our IC indices. First, they minimize the loss of information in regression analysis when compared with individual IC explanatory variables. We show, by using a probability of export equation with Turkish data, that our IC indices can proxy a large set of IC variables in regression analysis. Furthermore, we show that these IC indices make a regression analysis, with more than 100 explanatory IC variables, simpler and avoid a serious multicollinearity problem. Second, the ICI offers the possibility of making cross-country comparisons based on the description of the investment climate. For that purpose, we calculate the ICIs for 113 developing and transition countries, and then we show the cross correlation of the aggregate ICI with other aggregate indices like the World Economic Forum s Global Competitiveness Index (GCI), which is equal to 0.52, etc. Third, these IC indices allow us to incorporate the aggregate investment climate information of a country as an interesting determinant in macroeconomic models analyzing the IC impact on economic growth, cross-country convergence, etc. In chapter 3, I analyze the competitive restrictions of the exporters of Turkish manufacturing sector in terms of the IC variables. Although the Turkish economy experienced certain improvements after 2000, a poor investment climate of Turkey is one of the most important factors of the country's competitiveness with respect to its competitors. Improvements in IC may raise competitiveness by increasing firm-level performance, provide a sustainable growth perspective through higher productivity, and moderate the severe unemployment problem of Turkey by encouraging both domestic and foreign investment. The aim of this chapter is to investigate how IC constraints in Turkey affect exports by showing which components of IC have particular importance for the exporters. This analysis enables us to exactly determine the microeconomic structural reforms for the long-run prospect of Turkish manufacturing export. I estimate the model by using the Heckman model because of the nature of the data at hand.