This paper describes the determinants of FDI location among Italian provinces, focusing on the role of market potential (in terms of market access to Mediterranean and European countries) and institutions (presence of Mafia-type crime, corruption of public officers, inefficiency of the labour judicial system). Provinces with better access to neighbouring countries are expected to receive larger amounts of investments. Better efficiency in settling disputes relative to labour subjects, fewer corruption episodes and lower presence of organized crime should also be associated with higher flows of FDIs. Data have been drawn from different sources (FDIMarkets Database, OECD, IMF, EuroStat Regio, and various Italian statistical, economic and political institutions) and were elaborated in order to create specific indices. All together, they constitute a detailed collection of information on this subject (the appendix contains maps to better visualize data on the Italian territory). The particular structure of the data ? characterised by high presence of zeros, large overdispersion and high volatility of the dependent variable ? required extensive econometric testing. Baseline analysis was carried out through Zero-Inflated Negative Binomial, Poisson Pseudo-Maximum-Likelihood and Logit estimations. Additional tests were conducted adopting fixed effects methodologies (including a Mundlak procedure on a Negative Binomial random effects estimation) and other procedures able to deal with excess zeros and overdispersion in the dependent variable (Zero-Inflate Poisson, Hurdle Poisson, Hurdle Negative Binomial). Conclusions partially confirm initial hypotheses, while giving new insights on this issue. Market potential has a positive impact on FDIs, especially in relation to Mediterranean countries. Provinces with better infrastructural networks, in particular port and railway facilities, are more able to grasp the opportunities offered by economic growth in the Mediterranean. Inefficiencies in the labour judicial system are associated with lower probability to receive any investment (binary process). The relation between FDIs and Mafia follows a more insightful pattern, as the presence of Mafia reduces the probability of receiving inflows of foreign capital, but it is associated with larger projects, implying the investment of higher volumes of money. Similarly, higher levels of corruption are associated with larger investment projects, although they do not have any significant relation with the probability of receiving any investment (binary process). Regressions were further estimated with year fixed effects and additional covariates. Results were also tested introducing 1- and 2-period lags of the dependent variable. Dummies for specific areas of the country (north, centre, south) were constructed and interacted with market potential indices.
