This dissertation seeks to investigate the relationship between public sector employment and fiscal decentralization. We develop a theoretical model that helps us understand the interaction of the central executive's and subnational governor's decisions on the level of public employees at the central and subnational levels. Our empirical work shows that fiscal decentralization policy shifts central government employees to the subnational government level and that the increase in public employees at the subnational government level overwhelms the decrease in public employees at the central level. As a result, the level of total public sector employees increases with the degree of fiscal decentralization of a country. We also find that the levels of total public sector employees as a percentage of population are higher in unitary country systems than those in federal countries. The level of public employment also increases with the degree of urbanization and with the exposure to risk of a country.
This is a somewhat surprising result. Typically, more public employment is associated with an excessive number of public sector employees, and, therefore, with unproductive spending. On the other hand, fiscal decentralization policy has been generally thought to result in an increase in allocative efficiency, since a decision on public expenditures made by a level of government that is closer and more responsive to a local constituency is more likely to reflect the demand for local services than a decision made by a remote central government. In addition, decentralization has been thought of as having the potential of improving competition among governments and of facilitating technical innovations. Therefore, one might expect that fiscal decentralization should help to retrench the public sector employment. However, from our empirical result, we find that subnational governors without taking full responsibility for subnational public finance tend to bloat the levels of subnational government employees and ask the central government to pay the bill. As a result, the level of total public sector employees increases with fiscal decentralization policy. These findings are much in line with Oates' and Wallis' anticipated results, but they are based on different explanations. Employing the two most commonly used spatial dependency tests, Moran's I and Getis and Ord's G statistics, we also find evidence of spatial dependency in terms of the level of public sector employees as a percentage of population among the countries in our dataset. This finding suggests that while using a country's own domestic variables to explain the level of public sector employment, we should not ignore that the neighboring countries' policies also play an important role in determining it.
