The common problem in fiscal policy making: an exploratory experimental analysis

Abstract

This thesis explores implications of the common pool problem in fiscal policy making on fiscal performance. Three main research problems were investigated, using various laboratory experiments. First, we examine the effect of dynamic common pool problem on fiscal performance. Second, we test the delayed stabilization hypothesis as predicted in the dynamic common pool model. Third, we investigate the effect of multiyear budgeting on fiscal performance. Along the highlighted research problems, the thesis is structured into three core chapters that follow the much broader and general introductory chapter—chapter one. Chapter two examines the effect of the dynamic common pool within budget institution on fiscal performance. Dynamic common pool problem occurs when fragmentation within budgeting institution creates uncertainty in the future fiscal path, thereby increasing the proclivity for budget actors to strategically incur deficit and draw down on the public-sector wealth. Given the limitation of field data in exploring a dynamic and strategic setting, the experimental method is adopted. The experimental data generated are estimated using non-parametric techniques and system GMM technique. Overall, the findings indicate that dynamic common pool problem leads to poor fiscal performance, as measured by deficit level. In addition, we observe that budget actors react to the declining public-sector wealth with more aggressive appropriation behaviour, which further exacerbates the poor fiscal outcomes. Furthermore, the effect of dynamic common problem was found to be propagated through the strategic channel. In essence, despite the absence of the conventional fundamentals for incurring fiscal deficit—the divergence between discount rate and interest rate— poor fiscal performance persists due to the dynamic common pool problem. Chapter three empirically tests the delayed stabilization hypothesis, as predicted by the dynamic common pool model. The hypothesis suggests that there is stabilization threshold at which the presence of dynamic common pool problem no longer exerts a negative effect on fiscal outcome. In testing this hypothesis, we adopt an experimental design which builds on and extends the legislative bargaining game of “divide-the-dollar”. This is implemented by applying the random stopping rule after a pre-specified period. Also, tax is imposed on the groups that exhaust their initial public-sector wealth. However, to eliminate possible identification problem, the initial public-sector wealth is restored after some periods of paying tax, although participants are not V

aware of this replacement ab initio. Thus, it becomes possible to evaluate if the appropriation behaviours follow the pattern predicted by delayed stabilization hypothesis. Furthermore, the experimental data generated are analysed using both fixed-effect panel threshold regression model and piecewise linear regression model. The results do not support the prediction of delayed stabilization. In fact, the deficit level is highest in the periods after which the group’s initial public-sector wealth is restored; a clear contradiction of the delayed stabilization hypothesis. However, we observe a temporary stabilization during the periods when resources pool is generated by taxing the players. This evidence in part supports the crisis hypothesis, which suggests that probability of successful stabilization is enhanced when it coincides with economic crisis. Thus, this suggests a possible role for resources composition (whether taxation or natural endowment) and economic crisis in stabilization policies and plan. In chapter four, we examine the implication of the voting equilibrium model for the relationship between multiyear budgeting and fiscal performance. Voting equilibrium model recognizes that budget actors are rational individual, who are also seeking to maximize their utility in relation to the budget proposal. Thus, voting equilibrium model predicts that voting preference of the budget actor will influence the eventual budget outcome, thereby influencing any possible effect of multiyear budgeting on fiscal performance. This prediction is tested in a laboratory setting for a five-member budgeting institution, under two different treatment conditions. In the baseline treatment, participants make their budgeting decision using multiyear budgeting procedure, while annual budgeting procedure is adopted in the secondary treamnent. The budget outcomes under two treatments are compared using various non-parametric techniques and Selten’s predictive success measure. We find no signifcant different in budget sizes under the multiyear and annual budgeting proceses. As predicted by voting equilibruim, the multiyear budgeting generates lower budget size than annual budgeting and vice versa, according to the configuration of the voters’ preference. It is also observed that the predictive success of the voting equilinruim model reduces as the dimension of the planning horizon increases. In essence, increasing the dimension of planning horizon has the tendency to increase the uncertainty in the budgeting outcomes. Chapter five concludes with a summary of the key findings and discusses the important policy implications of our findings. An important policy addition from this study points to the significant improvement in fiscal performance that could be gained by addressing the dynamic common problem within budget institution. Similarly, the study demonstrates the importance of adopting active stabilization policy such as fiscal rules or centralization of the budget institution, as against relying on budget actors to act endogenously to correct deficit bias. Finally, in designing the possible fiscal interventions, our findings suggest that the voting preference of budget actors will be crucial.