Essays on Public Economics and Spatial Econometrics

Abstract

The primary purpose of this study is to investigate the spatial interdependence of fiscal variables (budget balance, tax revenue, and public expenditure) and the spillover effect of conflict on the level of local jurisdictions. The first study provides an empirical investigation of fiscal interactions in the context of a developing country. I examined three fiscal components—budget balance, tax revenue, and public spending to measure spatial interactions between Colombian municipalities from 2000 to 2010. I used a quasi-experimental identification strategy exploiting exogenous variation from the global oil price shocks that affect the Colombian municipalities to different degrees depending on local oil endowments. I found that there is a significant spatial interaction in taxes, but no significant interaction concerning budget balance and total public spending. This suggests that even though there is local tax competition, municipalities do not mimic their neighbors in the decision whether to offset tax changes by changes in borrowing or spending.The second study examines how the conflict in Colombia responded to a fiscal decentralization reform in 2001 using municipal-level data for Colombia. While there is a rich literature on the economic drivers and impacts of conflicts, the literature that examines the relationship between fiscal decentralization and conflicts are relatively small. The study is contributing to the literature by examining this relationship at the local (municipal) level, which has not been explored widely or conclusively in other studies. This study also addresses spatial dependence among municipalities. This is important because it is likely that conflict and other economic variables lead to spillovers that impact municipalities in close proximity. We employ spatial econometric techniques, particularly Spatial Durbin model, to account for spatial dependence in the data. The preliminary results showed a decrease in conflict in response to the decentralization reform but the exact response varies by conflict region. Oil reserves in some municipalities seemed to have played a role in the conflict outcomes as well.The third study examines how cluster size affects the testing performance greatly. The standard asymptotic theory requires the identical size of the clusters. Carter, Schnepel, and Steigerwald (2017)[1] relaxed this assumption and analyzed the effects of the cluster heterogeneity on testing performance. This study extended the results to the panel data (i, t) and showed that the temporal serial correlation has a critical effect on the cluster-robust testing. The optimal clustering parameter can be obtained using a simulation-based approach and it depends on the cluster heterogeneity and the relative sizes of i and t.