Would better state institutions increase tax collection, or would higher tax collection help improve state institutions? In the absence of conclusive guidance from theory, this paper searches for an empirical answer to this question, using a panel dataset covering 110 non-resource-rich countries from 1996 to 2017. Employing a panel vector error correction model, the paper finds that tax capacity and state institutions cause and reinforce each other for a wide range of country groups. The bi-directional causality results suggest that developing tax capacity and building state institutions need to go hand in hand for best results, particularly in developing countries. Based on the impulse response analyses, the paper also finds that the causal effects in advanced economies are generally low in both directions, while in developing countries, both tax capacity and institutions shocks have larger positive impacts on institutions and tax capacity, respectively.

• Cover
• Content
• ABSTRACT
• I. INTRODUCTION AND BACKGROUND
• II. POTENTIAL CHANNELS OF CAUSALITY
• III. STYLIZED FACTS
• IV. EMPIRICAL APPROACH
• V. EMPIRICAL RESULTS
  • A. Panel-Causality Tests
  • B. Impulse Response Function
• VI. CONCLUSION AND POLICY IMPLICATIONS
• REFERENCES
• BOX
  • 1. State Institutions and Revenue-Specific Institutions
• FIGURES
  • 1. Cross-Section Correlation Between Revenue and State Institutions
  • 2. Tax Capacity and Institution by Income Group
  • 3. Tax Capacity and Institution of Countries Above/Below Tax Tipping Point
  • 4. Impulse Response Analysis Results for All Sample Countries
  • 5. Impulse Response Analysis Results by Income Group
  • 6. Impulse Response Analysis Results Above and Below Tax Tipping Point
• TABLES
  • 1. Cross-Section Regression of the Impact of State Institution on Revenue Institution
  • 2. Average Cross-Section and Time Variations (Standard Deviation) in Tax Capacity and Institutions
  • 3. Long Run Panel Causality Test Results
• APPENDICES
  • I. Data Summary and Sources
  • II. Deriving Overall Institutional Index
  • III. Unit Root and Cointegration Tests
  • IV. Engle-Granger Procedure for Causality Testing
  • V. Long Run Bivariate Panel Causality Analysis
  • VI. Long Run Panel Causality Test (Controlling for Political Stability)