The empirical analysis investigates whether more financial market accessibility, especially to bonds, encourages PPP investments after controlling for their known determinants. This analysis goes further than the existing studies that confine their focus to whether economic and institutional developments are major determinants for attracting private investment in infrastructure projects.
While the literature emphasizes the need for developing countries to foster investment in infrastructure projects, there have been very few investigations into the determinants of PPP investment in terms of financial and capital market development, and economic development. This chapter tests whether the level of economic and finance sector development is a key driver of private investment in developing countries' infrastructure projects, while controlling for other features of a country's economy.
We estimate a set of model specifications with the level of total investment and private investment in PPP projects as dependent variables to evaluate whether economic and finance sector development is a key determinant of aggregated PPP project investments and private participation in PPP investments in developing countries. In addition to the independent variables of main interest, proxy variables standing for the levels of economic and finance sector development, and the government stability variable from the International Country Risk Guide dataset, are included.
As well as the major variables of interest, the first lagged dependent variable is included as an independent variable to capture potential dynamics because PPP arrangements are more likely in countries with previous PPP experience. After all, PPPs are complex arrangements between two parties. Previous PPP experience reassures private investors about the quality of their PPP counterparts. Arellano and Bover (1995) and Blundell and Bond (1998) show that the system generalized method of moments (GMM) approach allows lagged first differences to be used as instruments for dynamic panel models to correct any bias that might result from the standard GMM estimator. The system GMM estimator proposed by Blundell and Bond (1998), in particular, has become a common tool in applied economic research using panel data because it provides asymptotically efficient inference that assumes a minimal set of statistical assumptions. Blundell and Bond (1998) show that the standard GMM estimator has been found to have poor finite sample properties in cases in which the series are highly persistent. Here, the lagged levels of the series are only weakly correlated with subsequent first differences, consequently leading to weak instruments for the first-differenced equations.
To this end, we specify econometric models, including a benchmark panel regression and the difference GMM developed by Arellano and Bover (1995) and Blundell and Bond (1998), for dynamic panel models.4 All regressions include time dummies to capture time-specific global shocks or systemic risks, and country dummies to capture potential country-specific characteristics.