The Road to Good Governance: Via the Path Less Accountable? The Effectiveness of Fiscal Accountability in Liberia

Abstract

Accountability mechanisms are touted as a path to “good governance.” But are accountability mechanisms a sure route to achieving the objectives of “good governance”? Limited case studies have offered inconsistent evidence (Blair, 2000; Charlick, 2001; Devas & Grant, 2003). But empirical evidence of the relationships among principles of good governance—high citizen participation, low levels of corruption, high-quality service delivery—and accountability mechanisms is lacking. We examine the effectiveness of accountability mechanisms in Liberia and find relationships between measures of county level fiscal accountability and measures of good governance do not always produce expected results, making fiscal accountability mechanisms no guarantee for achieving goals of good governance.

Keywords:

• good governance
• fiscal accountability
• citizen participation
• corruption

Notes

Kelly Krawczyk now works in the Department of Political Science at Auburn University in Auburn, Alabama.

¹Although the rule of thumb for treating a categorical variable as continuous is that the variable must contain at least eight or more categories, due to limitations of available data we have treated the three indexed dependent variables in each of the three models as continuous even though they contain fewer than eight categories. All other variables in the model are either binary, continuous, or they have been statistically tested to see if they can be treated as interval-level variables. In cases where categorical independent variables did not meet requirements for treatment as interval-level variables, dummy variables for each possible response category were created and one less than the number of dummies was used in the regression equation.

²The Afrobarometer survey separates group membership into “religious” and “other voluntary and community group memberships.” Religious group membership was not found to have any statistical significance in any of the models.

³In model 2, both the political efficacy and local political efficacy variables did not meet the standard requirements for inclusion in the regression analysis as interval-level variables. Thus, they are included as separate dummy variables for each response.

⁴Both the electoral cleanliness and satisfaction with democracy did not meet the standard requirements for inclusion in the regression analysis as interval-level variables in this model. Thus, they are included as separate dummy variables for each response.