Income taxation across countries
What this paper finds — and why it matters
The paper provides the most comprehensive cross-country empirical characterisation of effective income tax functions to date, estimating the two-parameter log-linear tax function — pioneered by Feldstein (1969) and applied in structural macroeconomics by Heathcote, Storesletten, and Violante (2017) — for over thirty countries across approximately four decades using harmonized household microdata from the Luxembourg Income Study (LIS). The log-linear function fits income tax systems worldwide with median R² of 0.984 (mean 0.976), extending a finding previously known mainly for the United States to essentially all LIS countries. Five main facts emerge. First, income tax progressivity (τ) and average tax level (λ) are positively correlated across countries: Northern European countries with the highest average tax rates — Belgium, Netherlands, Germany, Finland — also have the highest progressivity; countries such as Brazil, Colombia, Peru, and the Republic of Korea exhibit effectively flat income taxes (τ near zero or negative) despite progressive statutory codes, because actual enforcement and effective coverage are limited. Second, progressivity increases with economic development: richer countries systematically operate more progressive income tax systems, consistent with greater institutional capacity to enforce income taxation. Third, progressivity differs significantly by family structure: married couples with children face the highest progressivity across countries, single households without children the lowest, reflecting child tax credits, joint filing rules, and other family-based provisions. Fourth, the United States ranks toward the lower end of progressivity among high-income countries, with τ ≈ 0.046 in 2010; Belgium, Finland, Germany, Iceland, Ireland, the Netherlands, and Spain are more than twice as progressive as the US. Fifth, transfers account for most redistribution: the combined tax-and-transfer system’s progressivity substantially exceeds that of income taxes alone, indicating that analyses focusing solely on income tax progressivity understate total redistributive effort.
Summary of a forthcoming paper, AI-assisted and human-reviewed. See the linked original for the authoritative claims and full conditions.
What is the log-linear tax function and why does the paper adopt it for cross-country comparison?
The log-linear tax function expresses post-tax income as T(y) = λy^(1−τ) + (1−λ)y, equivalent to log(y − T(y)) = α + (1−τ)log(y), where τ measures progressivity (τ > 0: marginal rates rise with income; τ = 0: flat tax) and λ captures the average tax level. The function is attractive because it (a) is used widely in structural macro models, enabling direct calibration from these estimates; (b) can be estimated consistently from microdata with just two parameters; (c) permits clean cross-country and over-time comparisons. A richer functional form would sacrifice the comparability across 30+ countries and 40 years of data.
How well does the log-linear function fit income tax systems across all countries in the sample?
Very well. Across all 200+ country-wave regressions, the median R² is 0.984 and the mean is 0.976. The fit is robust to different income definitions, imputation methods, and country-specific data sources. This extends the well-known finding for the United States (HSV 2017) to countries with very different income tax structures, suggesting the log-linear form is an adequate empirical approximation to real-world progressive tax schedules worldwide.
What is the cross-country pattern of progressivity in 2010?
Spain (τ ≈ 0.157), Belgium (τ ≈ 0.139), and the Netherlands (τ ≈ 0.127) have the most progressive income taxes in 2010. The Republic of Korea (τ ≈ −0.006) is slightly regressive in effective terms, along with Peru (τ ≈ 0.013) and other low-income countries where income tax coverage is limited. The United States has τ ≈ 0.046, placing it toward the lower end of progressivity among developed countries. In terms of the Progressivity Tax Wedge (PTW) — how much marginal tax rates rise between the average income earner and one at twice the average — Belgium, Finland, Germany, Iceland, Ireland, the Netherlands, and Spain are more than twice as progressive as the US.
How does income tax progressivity relate to economic development?
The paper documents a systematic positive relationship: richer countries (measured by median income, mean income, or GDP per capita) have more progressive income tax systems. Low-income countries like Peru and Guatemala collect most revenue through goods and services taxes and exhibit low income tax progressivity; high-income Northern European countries have both high tax capacity (the institutional ability to enforce income taxation) and high progressivity. This complements the tax capacity literature and suggests that the development-progressivity link operates through institutional channels, not solely through political demand for redistribution.
How does family structure affect income tax progressivity?
Estimated separately for four household types — single without children, single with children, married without children, married with children — progressivity is consistently highest for married couples with children and lowest for single households without children. This pattern holds across countries and over time, reflecting child tax credits, joint filing rules, and other family-based tax provisions that steepen the effective marginal tax schedule. The paper quantifies this heterogeneity by family type, filling a gap in cross-country comparisons that typically focus on single households without children.
What do transfers add to the redistributive picture, and what is the implication for welfare analysis?
When estimating a combined tax-and-transfer function (post-tax-and-transfer income regressed on pre-tax income), the progressivity of the combined system substantially exceeds that of income taxes alone. Countries with high income tax progressivity also tend to have high transfer system progressivity, but the transfer channel dominates. Analyses that focus solely on the income tax progressivity parameter τ therefore understate the total redistributive effort of high-income countries and overstate the tax-side role. This has direct implications for welfare analyses and cross-country comparisons using the log-linear framework.