2012年-IMF国际货币组织全球_Tax_Composition_and_Growth_A_Broad_Cross_36页_1mb

Summary of "Tax Composition and Growth: A Broad Cross-Country Perspective"

Core Content

This working paper by Santiago Acosta-Ormaechea and Jiae Yoo explores the relationship between tax composition and long-run economic growth across a wide range of countries. The study uses a comprehensive and updated dataset covering 69 countries (21 high-income, 23 middle-income, and 25 low-income) over the period 1970–2009, with each country having at least 20 years of tax revenue data. The paper investigates how changes in the structure of taxation—specifically the shift between income taxes, consumption and property taxes, and value added tax (VAT) and sales taxes—affect growth rates.

Main Findings

1. Tax Composition and Growth: Full Sample

• A reduction in consumption and property taxes while increasing income taxes is negatively associated with long-run growth.
• Social security contributions (SSC) and personal income taxes (PIT) have a stronger negative impact on growth than corporate income taxes (CIT).
• A shift from income taxes to property taxes is positively associated with growth.
• A reduction in income taxes while increasing VAT and sales taxes is also positively associated with growth.

2. Income-Level Group Analysis

• High-income countries (HICs) and middle-income countries (MICs) show consistent and significant negative effects of income taxes on growth, and positive effects of shifting to property and indirect taxes.
• Low-income countries (LICs) show less robust results, possibly due to poor tax administration and enforcement, which may have affected the accuracy of tax policy data.

3. Tax Structure and Development

• As countries become more developed, overall tax levels increase (in line with Wagner's Law).
• Tax structures also evolve, with HICs relying more on direct taxes (income taxes) than LICs.
• Direct taxes (PIT, SSC, CIT) are more prevalent in HICs and MICs, while indirect taxes (VAT, sales, trade, and property taxes) are more significant in LICs and MICs, though property taxes remain modest even in high-income countries.

4. Empirical Strategy

• The paper uses a Pooled Mean Group (PMG) estimation method to analyze the long-run equilibrium and short-run adjustment dynamics of tax composition and growth.
• The PMG model is preferred over the Mean Group (MG) and Dynamic Fixed Effect (DFE) models, as it avoids the homogeneity assumption of DFE and provides more efficient estimates than MG.
• The error correction model is used to separate the long-run effects from short-run fluctuations.

Key Variables and Data Sources

• Tax composition variables:
  • Aggregate income taxes
  • Aggregate consumption and property taxes
  • Sub-components: PIT, CIT, SSC, VAT, sales taxes, trade taxes, and property taxes
• Control variables:
  • Investment ratio
  • Average years of schooling
  • Population growth
• Data sources:
  • IMF Government Finance Statistics (GFS)
  • OECD Revenue Statistics
  • UN public finance statistics

Methodology

• The paper uses revenue-neutral tax shifts to isolate the growth effects of tax composition changes.
• It omits one tax component at a time in regression analysis to interpret the effect of shifts from one type to another.
• Hausman tests are conducted to determine whether the long-run coefficients are homogeneous across countries, which supports the use of the PMG estimator.

Conclusion

The study highlights the importance of tax structure in determining long-term economic growth. It suggests that revenue-neutral shifts from income taxes to property or indirect taxes can have positive growth effects, while increasing income taxes (particularly PIT and SSC) tends to slow growth. The results are more robust in HICs and MICs than in LICs, possibly due to differences in tax administration and policy effectiveness across income levels. The paper contributes to the fiscal policy and growth literature by using a broader and more comprehensive dataset and by focusing on the composition of taxes rather than the overall tax burden.