Document Type : Research Paper
Authors
1 Associate Professor, Faculty of Economics, Allameh Tabataba’i University, Tehran, Iran
2 Ph.D. Candidate, Faculty of Economics, Allameh Tabataba’i University, Tehran, Iran
Abstract
The present study addressed the question of why countries exhibit substantial disparities in tax revenue performance, using a comprehensive multilevel meta-analysis of 48 empirical studies (799 effect sizes). After accounting for publication bias and relevant moderator variables, the analysis identified the key determinants of tax revenue performance. The findings indicated that gross domestic product (GDP), international trade, inflation, industrial sector value added, and the lagged value of tax revenue exerted significant positive effects on tax revenues. In contrast, agricultural sector value added and corruption had significant negative effects. Foreign direct investment (FDI), however, did not exhibit a statistically significant relationship with tax revenues. Moreover, how tax revenue is measured—whether including or excluding social security contributions—critically shapes the estimated relationships between these determinants and tax revenue. The analysis also demonstrated that methodological choices (e.g., model specification and estimation techniques), the study period, and control variables (e.g., population size and institutional quality) significantly contributed to the heterogeneity observed across prior empirical findings.
Introduction
The persistent disparities in tax-to-GDP ratios across countries pose a critical challenge for policymakers and economists. Numerous empirical studies have examined the determinants of tax revenue, highlighting factors such as economic size, trade openness, sectoral composition, inflation, and institutional quality. However, their findings remain fragmented and, at times, contradictory. These inconsistencies largely arise from differences in methodological approaches, data sources, model specifications, and contextual moderators. To address this gap, the present study aimed to conduct a comprehensive multilevel meta-analysis to synthesize the existing evidence, identify the core determinants of tax revenue, and explain the sources of heterogeneity in prior empirical findings.
Materials and Methods
Adopting a meta-analysis method, the present study systematically identified and synthesized quantitative evidence from 48 empirical studies, yielding 799 effect sizes. The analysis was centered on a meta-regression framework that models reported effect sizes as a function of their standard errors and a vector of moderator variables. This approach enabled the correction for publication bias and the systematic consideration of methodological and contextual heterogeneity across studies.
The empirical model was specified as follows:
where is the reported effect on tax revenue, is its standard error, represents moderators, and is the error term. A multi-level framework was also employed to address dependencies arising from multiple effects per study.
Results and Discussion
The meta-analysis yielded robust, synthesized findings on the key drivers of tax revenue, with the moderator analysis providing critical contextual insights. Among the positive and significant determinants, GDP (Effect Size = 0.20) emerged as a primary driver, confirming that economic scale plays a central role in revenue generation. This effect was the strongest in panel, static, and fixed/random effects models. International trade (Effect Size = 0.068) also exerted a positive and significant influence, indicating that trade openness enhances revenue performance. The effect was particularly pronounced when tax revenue was measured excluding social security (TRISSC), and it remained consistent in static and fixed/random effects models. Industrial value added (Effect Size = 0.079) demonstrated a stable positive impact, with its influence reinforced in model specifications that controlled for GDP, trade, and corruption. The strongest predictor was lagged tax revenue (Effect Size = 0.528), highlighting strong persistence in revenue collection over time. This effect was consistently observed across nearly all model specifications and definitions.
In contrast, several determinants exhibited negative and significant relationships with tax revenue. Agricultural value added (Effect Size = -0.185) significantly constrained revenue mobilization, suggesting that a larger agricultural share in the economy hampers tax collection capacity. This negative relationship became even stronger in models that controlled for inflation, population, and corruption. Corruption (Effect Size = -0.156) also consistently undermined tax revenue performance, with the effect most pronounced in panel and static models. Foreign direct investment (FDI), by contrast, did not display a statistically significant relationship with tax revenue. This finding suggests that its overall impact may be neutral or highly context-dependent, making it difficult to detect a systematic average effect across studies. A key insight from the moderator analysis is that the definition of the tax base matters profoundly. The relationship between several determinants (e.g., trade and agriculture) and tax revenue varies significantly depending on whether social security contributions are included (TRESSC) or excluded (TRISSC) from the revenue measure.
*Table 1. Meta-Analysis Results With Key Moderators*
Variable
Overall effect
TRISSC (Tax Excl. SSC)
TRESSC (Tax Incl. SSC)
Panel models
Static models
With corruption control
GDP
0.20***
0.002
0.194***
0.20***
0.239***
0.09***
Trade
0.068***
Insignificant
0.062***
0.053***
0.066**
0.055***
Agriculture
-0.185**
-0.361
-0.18***
-0.147***
-0.186***
-0.089***
Industry
0.079***
0.024
0.091***
0.071***
0.117***
0.09***
Inflation
0.03**
0.133
0.003
0.005
-0.013
0.023***
Corruption
-0.156**
0.038
-0.154***
-0.16***
-0.355***
-
Tax (t-1)
0.528**
0.53*
0.675*
0.566*
0.491*
0.381*
FDI
-0.019
-
-
-
-
-
*Note: *p<0.1, ** p<0.05, *** p<0.01. SSC = Social Security Contributions.*
Source: Results Research
Conclusion
This study offered a comprehensive synthesis of the determinants of tax revenue, demonstrating that economic structure (i.e., sectoral composition), the macroeconomic environment (e.g., trade and inflation), and institutional quality (especially corruption), play pivotal roles in shaping revenue outcomes. Importantly, the impact of these factors is not fixed or absolute; rather, it is significantly moderated by methodological choices and by how the tax base is defined. The findings can be translated into several clear policy implications. Governments should promote trade openness through well-designed, trade-liberalizing policies, as it can enhance revenue both directly and indirectly by expanding and formalizing economic channels. At the same time, agricultural taxation requires rationalization. Efforts should focus on formalizing the agricultural sector and reassessing blanket tax exemptions that may unintentionally create loopholes and narrow the tax base. In addition, a strategic emphasis on industrial development is also essential. Policies that support industrialization, particularly those enabling small and medium enterprises to scale up, can help create a more easily taxable economic base. Finally, combating corruption and strengthening institutions must remain a central priority. Improving governance and curbing corruption are indispensable for improving tax compliance and for increasing the overall efficiency of revenue collection.
Keywords
Main Subjects
)