dc.description.abstract | Policing encompasses a wide range of services, which can be assigned to different levels of governments. Because there are potential advantages and disadvantages related to each option, the decision to adopt a more centralized or decentralized system is an important policy issue for countries around the world. Since the 1990s, there have been calls for empirical studies of effects of the structural arrangements on police performance, but the literature lacks generalizable studies of the effects of police systems. The lack of standardized classification makes it difficult to examine empirically the effects of police systems. The objectives of this study are threefold. First, to develop a typology of police systems by integrating theories of new institutionalism, decentralization, and fiscal federalism. Second, to empirically examine the effect of centralized and decentralized police systems on police performance and demand for police. Third, to provide an exante analysis of the potential effects of Thailand's decentralization of police services and to derive policy implications. This study constructs a new typology of police systems -- i.e., the police decentralization index -- that measures the varying degrees of police decentralization. The index is employed to examine the effects of police decentralization on citizen trust, demand for police, and crime rates by utilizing an unbalanced panel dataset from 2001 to 2012 for 72 countries. Findings indicate that the structure of police systems is not significantly related to citizen trust in the police. This finding is opposite of expectations given that new institutionalism argues that structure affects conduct and performance (North, 1990, 1991) and that decentralization is argued to move the government closer to the citizens and enhance relations between them (Oates, 1972, 1977, 1999; Pollitt, 2005). The structure of police systems is inversely associated with demand for police: countries with a more decentralized police system tend to employ fewer police officers. Fiscal federalism argues that decentralized government is more responsive to citizen preferences and, thus, more efficient (Oates, 1972, 1977, 1999; Tiebout, 1956, 1961). This finding suggests preferences for police are lower in decentralized systems and is consistent with prior studies by Ostrom (1976), Ostrom and Parks (1973), and Ostrom and Smith (1976). The effects of the structure of police systems on crime rates are mixed: decentralized police systems tend to have more homicides but fewer thefts and have no significant effects on robberies. These findings suggest that decentralized police systems may be more effective in preventing property crimes but not violent crimes. While the finding about homicides is opposite prior research in the United States by Ostrom and Smith (1976), the finding about thefts are consistent with that prior research. Based on these findings, if Thailand were to adopt a more decentralized police system, there would be no changes in the level of citizen trust in the police and the robbery rate. The homicide rate would increase by 5.32 per 100,000 inhabitants, and the theft rate would decrease by 110 per 100,000 inhabitants. The demand for police would decrease by 29.83 officers per 100,000 inhabitants. | eng |