Taxes Alone Won’t Power African Growth: Institutions Hold the Key, Landmark Study Finds

A sweeping new analysis of 35 African economies has delivered a finding that challenges decades of conventional fiscal wisdom: the relationship between taxation and economic growth is not fixed, but bends dramatically depending on the quality of a country’s regulatory institutions. The study, published in Discover Global Society by Solberg Horve Mishiwo of the University of Education, Winneba, Ghana, analysed a harmonised panel of 1,190 country-year observations spanning 1991 to 2024, and its results suggest that raising tax revenue in a weak regulatory environment may actually coincide with slower growth, while the same fiscal effort in stronger institutional settings aligns with faster expansion.

The research tackles one of the most persistent puzzles in development economics. Governments across Africa face mounting pressure to strengthen domestic revenue mobilisation to finance infrastructure, education, healthcare and private-sector development, yet many continue to grapple with narrow tax bases, weak compliance, rising public debt and institutional constraints. Traditional fiscal exchange theory holds that taxpayers comply more willingly when governments use revenues productively, and that stronger revenues should therefore support growth. But the empirical evidence has long been mixed, and the new study offers a systematic explanation for why: the fiscal-growth relationship appears to be institutionally conditioned rather than uniform.

Methodologically, the study is unusually ambitious. The core analysis employed a Two-Way Fixed Effects estimator with Driscoll-Kraay standard errors, which are robust to heteroskedasticity, serial correlation and cross-sectional dependence, the latter being a common feature of macro panels where countries experience shared global shocks. This baseline was complemented by Panel Threshold Regression to detect regime shifts, Panel Quantile Regression to examine whether associations vary across the conditional growth distribution, and Feasible Generalised Least Squares and Panel-Corrected Standard Errors as alternative robustness estimators. Crucially, the author also deployed two machine-learning algorithms, XGBoost and Random Forest, interpreted through SHAP feature-importance analysis, to assess the predictive contribution of each variable, while carefully distinguishing predictive importance from causal or econometric significance.

The baseline results were striking. Tax revenue mobilisation, measured as tax revenue as a percentage of GDP, showed a negative and statistically significant association with annual GDP per capita growth. Domestic credit to the private sector was also negatively associated with growth, while regulatory quality emerged with a strongly positive and significant coefficient. Most tellingly, the interaction between tax revenue and regulatory quality was negative and highly significant, indicating that the estimated tax-growth association varies systematically with the institutional environment. New business density, public debt, inflation, population growth and foreign direct investment did not reach statistical significance in the baseline specification.

The negative headline coefficients, however, tell only part of the story. The threshold analysis, using regulatory quality as the threshold variable, estimated a statistically significant breakpoint at a regulatory quality value of minus 0.412. Below that threshold, tax revenue mobilisation was negatively associated with growth; above it, the association turned positive. A similar pattern emerged for domestic credit to the private sector, which shifted from a negative association in weaker regulatory regimes to a weakly positive one in stronger regimes, and for new business density, which became positively associated with growth only in the higher-quality regulatory environment. Foreign direct investment likewise showed a positive association with growth only above the threshold.

The quantile regressions reinforced this picture of heterogeneity. Tax revenue mobilisation was negative at the 25th and median conditional quantiles of growth but turned positive and weakly significant at the 75th quantile. Domestic credit followed the same trajectory, moving from negative at lower quantiles to positive at the upper quantile, while new business density became positive at the 75th quantile. Regulatory quality, by contrast, remained positive and significant across the entire distribution, with its coefficient strengthening from 4.128 at the 25th quantile to 9.371 at the 75th. The interaction between tax revenue and regulatory quality stayed negative throughout, though its magnitude declined at higher quantiles.

The machine-learning component added a distinct predictive perspective. XGBoost outperformed Random Forest, achieving an R-squared of 0.781 against 0.724, with lower root mean squared error and mean absolute error, under a country-blocked validation design designed to prevent information leakage across a country’s own observations. SHAP analysis ranked regulatory quality as the leading predictor of economic growth, followed by tax revenue mobilisation, domestic credit to the private sector and public debt, with broadly consistent rankings across both algorithms. The author is explicit that these rankings indicate predictive contribution within the fitted models, not causal importance, coefficient direction or economic effect size.

The study is equally candid about its limitations, a transparency that strengthens rather than weakens its contribution. The balanced panel is a harmonised analytical dataset rather than a fully observed one: 64 percent of new business density observations, nearly 43 percent of tax revenue observations, 30 percent of public debt observations and 24 percent of regulatory quality observations were completed through interpolation, backward or forward filling, or cross-sectional median imputation. A restricted sensitivity analysis using only directly observed data, comprising 252 observations across 28 countries, preserved the direction of several central coefficients but with reduced precision, and some coefficients changed sign. The author also stresses that tax revenue as a share of GDP captures the scale of mobilisation but not tax composition, administrative efficiency, compliance burdens or the productivity of the expenditure it finances, and that the observational design cannot eliminate reverse causality or omitted confounding. All estimates are therefore conditional associations, not causal effects.

The theoretical framing helps explain why the results look the way they do. The study integrates four complementary perspectives: fiscal exchange theory, which links revenue to accountable public spending; endogenous growth theory, which emphasises entrepreneurship, innovation and financial development; debt overhang theory, which identifies constraints from excessive indebtedness; and institutional theory, which explains how governance shapes incentives and resource allocation. Within this framework, new business density and domestic credit are treated as selected observable dimensions of entrepreneurial ecosystem development rather than exhaustive measures, and public debt as an indicator of fiscal-sustainability pressure rather than a complete measure of fiscal health. The unexpected negative baseline coefficients for taxation and credit are interpreted not as evidence that these factors are inherently growth-reducing, but as reflections of conditions, such as credit allocation efficiency and expenditure productivity, that the aggregate measures cannot separately identify.

The policy implications are suggestive rather than prescriptive, but they point in a clear direction. Rather than focusing exclusively on raising the tax-to-GDP ratio, the findings are consistent with reforms that pair revenue mobilisation with regulatory effectiveness, accountable resource management and the productive use of public funds. Similarly, expanding business registration or private-sector lending alone may be insufficient where surrounding institutional conditions do not support viable enterprises and productive allocation of credit. As African economies confront rising debt burdens and intensifying demands for domestic revenue, the study’s central message resonates: the same tax shilling can be associated with very different growth outcomes depending on the institutional machinery that collects and deploys it. Future research, the author suggests, should pursue richer measures of tax design, entrepreneurial ecosystems and fiscal sustainability, alongside stronger identification strategies capable of distinguishing genuine causal mechanisms from the heterogeneous conditional relationships documented here.

Subject of Research: The relationship between tax revenue mobilisation, entrepreneurial and financial conditions, institutional quality, and economic growth across African economies

Article Title: Tax revenue mobilisation and economic growth across African economies under entrepreneurial fiscal and institutional conditions

Article References: Tax revenue mobilisation and economic growth across African economies under entrepreneurial fiscal and institutional conditions. (n.d.). https://doi.org/10.1007/s44282-026-00632-y

Image Credits: AI Generated

DOI: 10.1007/s44282-026-00632-y

Keywords: tax revenue mobilisation, economic growth, Africa, regulatory quality, institutional quality, fiscal policy, entrepreneurship, domestic credit, panel threshold regression, quantile regression, XGBoost, SHAP