The relationship between tax structures and income inequality

La relación entre las estructuras tributarias y la desigualdad de ingresos

Carlândia Brito Santos Fernandes

State University of Maringá (Brazil).
Email: cbsfernandes@uem.br

Marcos Roberto Vasconcelos

State University of Maringá (Brazil).
Email: mrvasconcelos@uem.br

Marina Silva da Cunha

State University of Maringá (Brazil).
Email: mscunha@uem.br

Thomas Obst

German Economic Institute, (Germany).
Email: obst@iwkoeln.de

Recibido: 31 de agosto de 2023

Evaluado: 30 de junio de 2024

Aceptado: 8 de febrero de 2025

DOI: 10.13043/DYS.100.4

Abstract

The global rise in income inequality has heightened interest in how fiscal policy—particularly the composition of tax revenues—affects income distribution. This study examines the relationship between government revenues from direct and indirect taxes and income inequality, contributing to the tax incidence literature through three key innovations: it focuses on a recent period (2000–2012), compares countries with different income levels and tax structures, and addresses endogeneity using the System Generalized Method of Moments (System GMM) estimator. The empirical analysis reveals that a greater reliance on indirect taxation is associated with higher income inequality, while a larger share of direct taxes correlates with reduced inequality. These patterns hold across both OECD and non-OECD countries. Given the higher Gini coefficients observed in non-OECD economies, the findings highlight the importance of enhancing tax progressivity as a policy strategy to reduce income disparities and promote more equitable economic development.

Keywords: Fiscal policy, tax structure, income distribution, economic inequality, developing countries.

JEL Classification: O15, O23, H23, E62.

Resumen

El aumento global de la desigualdad en los ingresos ha intensificado el interés por comprender la manera como la política fiscal, en particular, la composición de los ingresos tributarios, afecta la distribución del ingreso. Este artículo investiga la relación entre los ingresos gubernamentales provenientes de impuestos directos e indirectos y la desigualdad de ingresos, con lo cual se aporta a la literatura sobre incidencia tributaria, a través de tres innovaciones claves: el enfoque en un periodo reciente (2000-2012), la comparación entre países con diferentes niveles de ingreso y estructuras tributarias, así como el tratamiento de la endogeneidad, mediante el estimador de momentos generalizados del sistema (system GMM). El análisis empírico revela que una mayor dependencia de los impuestos indirectos se asocia con el aumento de la desigualdad, mientras que una mayor participación de impuestos directos se correlaciona, por el contrario, con su reducción. Estos patrones son consistentes en países de la OCDE y en países que no pertenecen a ella. No obstante, dados los mayores coeficientes de Gini observados en las economías no OCDE, los resultados destacan la importancia de fortalecer la progresividad tributaria como estrategia para mitigar las disparidades de ingresos y promover un desarrollo económico más equitativo.

Palabras clave: política fiscal, estructura de impuestos, distribución del ingreso, inequidad económica, países en desarrollo.

Clasificación JEL: O15, O23, H23, E62.

Introduction

Income inequality has steadily increased across many of the world’s major economies since the 1980s. In England, the Gini index for the distribution of market income rose significantly from 0.331 to 0.454 between 1975 and 2016, while Sweden saw an increase from 0.314 to 0.366 over the same period. For the United States, Yellen (2014) reports that between 1989 and 2013, the median household income of the wealthiest 5 percent increased by 38 percent, compared to just 10 percent for the remaining 95 percent. Even Japan, often seen as a country with strong social cohesion, has experienced a rise in income inequality in the 21st century, as noted by Yokoyama et al. (2019) and Yamada and Kawaguchi (2015). These examples suggest that income concentration has become a widespread pattern, affecting countries with very different economic and social system.

This development has drawn attention to the formulation and understanding of economic policies capable of reversing—or at least mitigating—the current trend (Stiglitz, 2017; Atkinson, 2015; Piketty, 2014; Gornick and Jäntti, 2014), with fiscal policy gaining prominence in most of these studies. However, fiscal policy has at least three primary functions. First, it provides the government’s main source of resources—taxes—which finance public services and goods. It also influences the allocation of the economy’s available resources and alters the distribution of income and wealth within society. Each of these core objectives can be pursued and realized in different ways. How, then, can policymakers reconcile the multiple goals of fiscal policy? Traditionally, fiscal policy has been framed in terms of the efficiency-equity trade-off (Stiglitz and Rosengard, 2015). Moreover, as Martinez-Vazquez et al. (2014) and Gobetti (2022) observe, in recent decades, the pendulum may have swung too far in favor of efficiency. Is it possible to escape this dichotomy, or must a balance always be maintained? While answering these questions lies beyond the scope of this paper, the focus here is on tax policy, which may contribute to finding an answer.

Tax policy offers a possible solution to the growing problem of income inequality, as higher marginal tax rates and the resulting increase in progressivity imply that the wealthy pay a relatively larger share than the poor (Duncan and Sabirianova, 2016). In this way, Piketty (2014) argues for the strengthening of the progressive income tax system. This paper contributes to the tax incidence literature (as in Martinez-Vazquez, 2008) by shedding light on the relationship between tax policy and income inequality, a distinct approach in at least three aspects. Firstly, we focus on a more recent period 2000-2012 whereas most previous studies examine earlier timeframes (Duncan and Sabirianova, 2016; Martinez-Vazquez et al., 2011). Second, our analysis compares OECD and Non-OECD countries, while much of the literature has relied on a mixed country samples (see Woo et al., 2013).¹ To the best of our knowledge Martinez-Vazquez et al. (2014) are the only authors to include the first decade of the current century, but they analyze a combined sample of developed and developing countries. Third, we employ an econometric approach— system GMM—to address the issue of endogeneity.

Our sample hence includes countries with varying levels of income and tax structures, allowing us to examine whether the level of economic development influences the relationship between fiscal structure and income distribution. The main questions we address are: Does the more significant share of the revenue from indirect taxes, relative to direct taxes, worsen income distribution? Do these effects vary across the groups of countries analyzed? This study contributes to the ongoing debate by empirically assessing the relationship between fiscal structure and income distribution and by emphasizing the potential need for this recalibration. The main findings indicate a correlation whereby greater dependence on indirect taxes is associated with a more unequal income distribution in both country groups. However, this result has limitations, as we do not yet measure the degree of progressiveness or regressiveness within each direct and indirect tax category observed in our sample. After all, indirect taxes can, in some cases, be progressive, just as direct taxes can be regressive. However, if such variation is present in the sample, it would only strengthen the results. We intend to measure the degree of progressiveness of each tax category in future research.

This paper is organized as follows. Section 2 outlines the interdependent relationship between the tax system and income distribution. Section 3 describes the data and key variables. Section 4 summarizes the methodology and discusses the results. In Section 5, we present robustness checks. Section 6 concludes.

I. Trade-off Between Efficiency and Equity: The Role of the Tax Structures

At the microeconomic level, rising income inequality may create stronger incentives and conditions for a segment of society to commit fully to productive entrepreneurial activities. This includes the willingness to take on high-risk ventures—such as technological innovation endeavors or long-term productive projects—as well as to invest in education in pursuit of higher future earnings. Therefore, if monetary incentives are indeed a key driver of human effort, then economic agents are more likely to realize their economic potential when they can expect to retain most or all of the gains from their labor or business activities. The outcome would be increased productivity and, in the long run, a higher level of output that benefits all economic agents.

Inherent to this reasoning is the assumption of the existence of a perfect credit market. That is, all economic agents have equal access to borrow the resources necessary to invest in physical or human capital, regardless of their initial level of wealth (see Acemoglu, 2009). Under this assumption, there is no consideration that beyond a certain threshold, the negative effects of income inequality could outweigh its potential benefits for economic growth. In other words, relaxing the perfect credit market assumption is enough to reveal mechanisms through which income and wealth inequality may undermine economic growth. For example, if inequality reaches a point where a substantial portion of the population struggles to attain adequate health or invest in education, the economy’s overall productive potential declines. Without the capacity to accumulate human capital, some individuals with high entrepreneurial and innovative potential may be unable to realize their capacities, leading to a loss in aggregate productivity (Cingano, 2014).

Furthermore, the segment of society that benefits from income concentration is often both incentivized and well-positioned to influence economic policymaking in its favor (North, 1990; Acemoglu and Robinson, 2012; Stiglitz, 2017). For instance, this group may lobby for lower tax rates on high-income brackets or seek preferential tax treatment. All else being equal, such actions can undermine the government’s ability to raise revenue by increasing public debt or cutting public spending (Stiglitz, 2017). Beyond exerting pressure on the state, wealthier households may also adopt more defensive and rent-seeking behaviors, reducing their entrepreneurial initiatives and instead focusing on preserving their position through market and political influence (Atkinson, 2015). This dynamic creates a vicious cycle between income inequality and economic growth (North, 1990; Acemoglu and Robinson, 2012). Over time, it can foster political instability, as economically disadvantaged groups more receptive to populist rhetoric (Berg and Ostry, 2011; Cingano, 2014). Such populist movements often discourage productive investment by increasing uncertainty, further reinforcing this harmful cycle.

Several empirical studies have sought to estimate this relationship between income inequality and economic growth. For example, Persson and Tabellini (1994) found evidence that a more equal income distribution is positively associated with both the growth rate and overall income levels. More recently, Berg et al. (2012) identified income inequality as a barrier to long-term, sustainable economic growth. This has contributed to a growing recognition that income inequality is not solely a matter of distributive justice—it also has broader macroeconomic implications (Berg and Ostry, 2011). As a result, there has been a renewed push for research aimed at identifying potential policy responses to address the issue (Cingano, 2014; Ostry et al.2014; Kennedy et al., 2017).

In this context, fiscal policy—more specifically, the tax policy—has attracted increasing attention. This raises the question of whether market-determined distributional outcomes can be reduced by a given tax system, thereby promoting a more equal distribution of disposable income among agents, that is, income after payment of taxes and receipt of transfers. In theory, the answer is yes: it is sufficient to tax higher-income individuals and redistribute the proceeds to those with lower incomes. However, high marginal tax rates may discourage economic agents with greater human capital and productivity from supplying labor (due to the substitution effect), engaging in entrepreneurial efforts, or accumulating capital. At the firm level, increased taxation on capital and income can disincentivize domestic capital formation and, in the context of international capital mobility, may encourage firms to relocate to jurisdictions with a less burdensome tax regimes. As a result, the pursuit of equity may come at the cost of economic efficiency. The challenge for modern societies is to avoid reaching this trade-off’s breaking point.

The belief in the market’s efficiency in allocating resources has fueled criticism of direct taxation. According to Martinez-Vazquez et al. (2014) and Gobetti (2022), since the 1980s, many developed countries have reduced both average and top marginal rates of direct taxes in an effort to minimize distortions in private agents’ allocative decisions. Consequently, direct taxes—such as those levied on labor income and capital gains—have been losing prominence within overall tax structure and have become less progressive, as highlighted by Duncan and Sabirianova (2016). Similarly, as Joumard et al. (2012) emphasize, taxes on inheritance, wealth, and capital income have been reduced in several countries, weakening the redistributive capacity of tax systems.

In turn, tax structures have increasingly shifted toward reliance on indirect taxes—such as consumption taxes, excises, and customs duties—which are presumed to be more neutral with respect to labor supply decisions and, in particular, entrepreneurial activity (Vartia, 2008; Djankov et al., 2009). For Stiglitz (2017), this shift may also reflect attempts by wealthier individuals to influence tax systems in their favor through political pressure. Notably, rising income inequality has been observed in many of these economies during the same period. This raises the question of whether these two trends—the move toward a greater share of revenue from indirect taxes and the increase in income disparity—are actually related².

Several authors have identified a relationship between a country’s tax structure and its level of income inequality—for example, Bastagli et al. (2012), Chu et al. (2004), and Woo et al. (2013).³ The first two studies conclude that both the level and progressiveness of taxation can partially explain disparities in income distribution. In general, they argue that direct taxes tend to favor a more equitable income distribution, while indirect taxes tend to exacerbate inequality. Woo et al. (2013) examined the effects of taxation on income inequality in a panel of advanced and emerging market economies over the last three decades. According to Bastagli et al. (2012) and Chu et al. (2004), progressive taxation and social benefits are consistently associated with lower disposable income inequality—a conclusion also supported by Joumard et al. (2012) in their analysis OECD countries. Decoster et al. (2010) applied microdata from five European countries to simulate the effects of reducing the weight of direct taxes and increasing indirect taxes. Their findings suggest that such a shift would reduce the overall progressiveness of the tax system.⁴

Thus, the tax system can play an essential role in reducing income inequality, as noted by Burman (2012) and Joumard et al. (2012). Furthermore, Woo et al. (2013) found that fiscal policy may favor long-term trends of both equality and growth by promoting education and training for low-and middle-income workers. These findings align with those of Persson and Tabellini (1994), Berg et al. (2012), and Lee and Son (2016) who report a negative relationship between economic growth and income inequality. In summary, indirect taxes—such as those on goods and services—tend to be more regressive, as lower income individuals spend a more significant share of their income on consumption and therefore end up paying a proportionally higher average tax rate than those with higher incomes.

Thus, the efficiency-equity trade-off inherent in fiscal policy must be reconsidered and recalibrated. This study contributes to this debate by empirically assessing the relationship between tax structure and income distribution, thereby highlighting the potential need for such recalibration. The following section outlines how this is accomplished.

II. Data and Variables

In this paper, we use the Government Revenue Dataset (GRD) compiled by the International Center for Tax and Development (ICTD) and United Nations World Institute for Development Economics Research (UNU-WIDER)⁵ (UNU-WIDER, 2018) to clarify and further the discussion on the relationship between direct and indirect taxation and inequality. Specifically, from this dataset we use: a) total direct taxes—excluding social contributions⁶ and resource taxes—including non-resource taxes on income, profits, and capitals gains, taxes on payroll and workforce, and taxes on the property; b) total indirect taxes—including resource revenues, comprising taxes on goods and services, international trade and other taxes; c) total government expenditure as a percentage of GDP.

To measure inequality, we use the Gini index based on equivalized household disposable income (post-tax, post-transfer), obtained from the Standardized World Income Inequality Database (SWIID), version 7.1. Our sample covers 26 OECD and 26 non-OECD countries from 2000 to 2012.

Table A1 in the appendix presents the descriptive statistics and the correlation matrix. The Gini index ranges from a maximum value of 60 and a minimum value of 22, indicating a high degree of inequality variation across the countries studied. On average, direct taxes account for 7 percent of GDP, while indirect taxes are higher, averaging 11 percent. In terms of correlations, both types of taxes show a negative relationship with income inequality. Similarly, government expenditure is also negatively correlated with inequality. This finding slightly diverges from the discussion in the previous section, where the literature suggests that only direct taxes tend to affect income inequality negatively.

Table 1 reports the average shares of direct and indirect tax revenues and the Gini index of disposable income, for both OECD and non-OECD countries. In both groups, there is a relatively higher dependence on indirect taxes than on direct taxes. However, the extent of this reliance differs: in non-OECD countries, indirect tax revenues are more than double those of direct taxes, whereas in OECD countries, the gap is narrower—only six percentage points. According to Pickering and Rajput (2018), this pattern reflects the limited capacity of low-income countries to raise income taxes. The authors observe that income taxes account for approximately 32 % of total revenues in OECD countries, compared to just 20 % in non-OECD countries. Likewise, Mahler and Jesuit (2018) argue that in lower-income countries, political compromises often lead governments to favor tax instruments that do not fall on capital owners.

As expected, the Gini index is significantly lower for the OECD group than for the non-OECD group. Inequality tends to be lower in high-income countries, which appears consistent with the Kuznets hypothesis (Kuznets, 1955). This raises an important question: to what extent can direct taxes effectively reduce income inequality? And does this relationship differ between the two groups?

III. Methodology and Results

We employ the Generalized Method of Moments (GMM) estimator developed by Arellano and Bover (1995) and Blundell and Bond (1998), which fully exploits the dynamic panel data structure to address endogeneity. This approach uses lagged values of the endogenous explanatory variable and/or the dependent variable as instruments⁷. The Arellano-Bover/Blundell-Bond estimator assumes that the first differences of instrument variables are uncorrelated with the fixed effects and applies orthogonal deviations—that is, instead of subtracting the previous observation from the current one, it subtracts the average of all future available observations of a variable (Roodaman, 2009). In our specification, all variables are treated as endogenous when selecting instrumental variables. It is important to note that the validity of the System GMM estimator relies on two key conditions: the absence of second-order autocorrelation in the error terms and the exogeneity of the instruments.

We begin with the following baseline equation:

Each observation is indexed by \( i \, (=1,\ldots,N) \) for cross-sectional groups (countries) and \( t \, (=1,\ldots,T) \) for time periods (annual observations). \( Y_{it} \) represents income inequality measured by the Gini index for the 53 OECD and non-OECD countries \( i \) in period \( t \). \( Y_{i,t-1} \) is the lagged dependent variable. \( DI_{it} \) captures the correlation of direct taxes (as a share of GDP), \( IND_{it} \) denotes the correlation of indirect taxes (as a share of GDP), \( \mu_i \) represents a country-specific random effect that controls for all unobservable effects on the dependent variable that are exclusive to the country and do not vary over time, and \( e_{it} \) is an error term that varies over both countries and time. The lag of the dependent variable from the previous period generates dynamics in the model, which may be crucial for recovering consistent estimates of other parameters (Bond, 2002).

Applying the Arellano-Bond test, we reject the null hypotheses of no first-order autocorrelation [AR(1)] in all specifications, while failing to reject the null for second-order autocorrelation [AR(2)]. This confirms the validity of using the lagged variables as instruments, that is, the instruments are internal. To assess the exogeneity of the instruments, we apply the Hansen tests of over-identification. The resulting J-statistic follows a χ² distribution under the null hypothesis that all instruments are valid⁸. A rejection of this null hypothesis would imply that some of the instruments are invalid⁹. However, a large number of instruments may generate a possible over-identification of the endogenous variables, not allowing adequate treatment of endogeneity and generating biased estimates. Excessive instrumentation may also reduce the power of the Hansen test itself. To mitigate this risk, we collapse the instrument matrix, creating one instrument per variable and lag distance, rather than generating separate instruments for each period, variable, and lag distance (Wintoki et al., 2012).

Table 2 presents the results. In all specifications, the Hansen’s p-values—robust to both heteroscedasticity and autocorrelation—support the validity of the over-identification restriction and confirm the exogeneity of instrument subsets. A glance at columns A–E of Table 2 reveals strong persistence in income inequality (Gini index), as indicated by the relatively large and statistically significant coefficient of the lagged dependent variable. This coefficient captures historical inertia in the model, Consistent with findings in previous studies (Woo et al., 2013; Azevedo et al., 2014). Our analysis focuses on the connection between direct and indirect taxes and income inequality. At this stage, we do not include additional control variables in the regressions presented in Table 2. These will be incorporated in subsequent sections as part of the robustness checks.

Column A of Table 2 presents the regression results for the full sample of countries. The coefficients for both direct and indirect taxes are statistically significant: the former show a negative correlation, and the latter exhibit a positive one, confirming that, on average, increases in direct tax revenue are linked to reductions in inequality, whereas increases in indirect tax revenue tend to exacerbate it. Specifically, a 1 % increase in GDP from direct taxes is associated with a 0.117 decrease in the Gini index, while a 1 % increase in GDP from indirect taxes corresponds to a 0.061 increase in the inequality measure.

We further divide the sample into OECD countries and non-OECD countries—shown in columns B and D, respectively—to capture potential differences between the two groups. The results are broadly consistent with those for the full sample; however, the main difference is that direct taxes are not statistically significant for OECD countries. This finding remains unchanged when we introduce an interactive dummy variable for the group, as shown in Column C. For non-OECD countries, both direct and indirect tax revenues are statistically and economically significant, as we can see in column D. In contrast, in Column E, indirect taxes are not statistically significant. The inverse relationship between direct taxes and inequality may reflect the progressive structure of the tax systems of the analyzed countries. With a progressive tax system, increases in direct tax revenue would yield a more significant redistributive effect and, thus, lower inequality (Lambert, 2001; Muinelo‐Gallo and Roca‐Sagalés, 2013).

Comparing the significant results of columns C and E, we find that a 1 % increase in GDP from indirect tax revenues raises the Gini index of OECD countries by 0.0442—equivalent to an approximate 0.15 % increase in the group’s average Gini. In contrast, a 1 % increase in GDP from direct tax revenues in non-OECD countries reduces the Gini index by 0.173, representing a decline of almost 0.5 % in the average Gini for that group. This result contrasts with the findings of Goñi et al. (2011) who—based on a comparison of Gini coefficients before and after direct taxes—concluded that the redistributive impact of direct taxes is much stronger in European countries than in Latin America. For the fifteen European countries analyzed, direct taxation reduced the Gini coefficient of household incomes by an average of five percentage points, while for Latin American countries, the reduction was just one percentage point on average. Similarly, Lustig et al. (2014) in their study of several Latin American countries found variation in the redistributive power of direct (personal income) taxes: Uruguay and Mexico experienced reductions in the Gini index of 2.8 and 2.6 percent, respectively, while Brazil and Peru saw more modest declines of 1.9 and 1.2 percent, respectively. Even so, the results in Table 2 indicate that increasing the share of direct tax revenue in total tax collection in Latin American and other developing regions could contribute meaningfully to reducing income inequality. In other words, there appears to be scope for tax systems of non-OECD countries to prioritize the distributive issue.

IV. Robustness check

This section examines the robustness of our main findings using alternative data sources and variables. We implement three strategies, described below. First, we re-estimate the baseline regressions from Table 2 using the Gini index provided by the World Bank as an alternative measure of inequality. The results are shown in Table 3. Overall, the findings are consistent with those in Table 2: increases in direct tax revenues are associated with reductions in income inequality, while higher indirect tax revenue is linked to increases in the Gini index for OECD and non-OECD countries.

Focusing only on the statistically significant results reported in Table 3, the coefficients of direct and indirect taxes consistently retain their negative and positive signs, respectively, across all specifications. That is, an increase in the share of direct taxation is associated with a drop in the Gini index. In contrast, indirect taxation correlates with a rise in the index. Diagnostic tests confirm the robustness of the models: as the Hansen tests support the validity of the instruments, and the Arellano-Bond test indicates no evidence of second-order autocorrelation in the error terms. While the estimated coefficients in Table 2 are, on average, smaller in magnitude than those in Table 3, it is important to note that these differences are not directly comparable, as they are based on alternative data sources for the Gini index.

In Table 3 (Column A), the coefficient for direct taxes is negative and statistically significant at the 5 % level, indicating that a 1 % increase in total revenue from direct taxes correlates with a decrease in the Gini index for the full sample of OECD and non-OECD countries by 0.18. Conversely, a 1 % increase in indirect tax revenue corresponds to a 0.25 rise in the Gini index. Column B, which focuses on OECD countries, yields results similar to those in Column B of Table 2; however, the association between indirect taxes and inequality is now statistically significant at the 1 % level—stronger than the 10 % level observed previously. In column C, the interaction terms for developed countries (Direct OECD and Indirect OECD) are both economically and statistically significant, although with different significance levels compared to Column C in Table 2. Focusing on the non-OECD countries, Column D shows that the coefficients for both direct and indirect taxes lose statistical significance relative to Table 2. Meanwhile, Column E presents broadly similar to its counterpart in Table 2, with the exception that the coefficient for Indirect non-OECD is now statistically significant.

As a second robustness check, presented in Table 4, we continue to use the World Bank Gini index but adopt an alternative measure of tax revenue. Specifically, we express direct and indirect tax revenues as a percentage of total revenue, rather than as a percentage GDP, as was done in tables 2 and 3. In addition, in line with the literature on fiscal incidence (Martinez-Vazquez, 2008; Martinez-Vazquez et al., 2014), or tax and expenditure incidence, that conceptualizes and evaluates how the revenue and expenditure sides of government budgets affect the distribution of income among households, we include an additional control variable: the total government expenditure. This addition is further supported by Mahler and Jesuit (2018), who emphasize the significant distributive role of public spending. This is because the literature indicates that at least some types of government spending can reduce the level of income inequality across various countries and regions (for example Goñi et al., 2011; Lustig et al., 2014; Martinez-Vazquez et al., 2014; Mahler and Jesuit, 2018; Lustig, 2016 and Fernandes et al., 2019).

When we compare the magnitudes of the coefficients for the groups of countries, the results in Table 4 suggest that the negative correlation of indirect taxes on inequality is more significant for non-OECD countries¹⁰. This means that a rise in indirect taxes aggravates inequality considerably more than in the OECD countries. For example, comparing columns B and D, a 1 % increase in the ratio of indirect taxes to total revenue is association with a 0.13 increase in the Gini index for non-OECD countries, whereas the same increase is associated with only a 0.08 rise in inequality in OECD countries.

In the third robustness check, presented in Table 5, we run the base regressions shown in Table 2 with total government expenditure and other control variables that appear in the literature as factors that can affect the degree of income inequality. Specifically, we include measures of human capital, welfare, and exchange rates, all sourced from the Penn World Tables. The human capital variable is an index of human capital per person, based on years of schooling and returns to education. For welfare, we use the welfare-relevant TFP at constant national prices. The exchange rate variable corresponds to national currency per USD. The dependent variable in columns A-D is the Gini index from Swiid, while in columns E-J, we use the Gini from the World Bank.

In the first column of Table 5, we examine the relationship between direct and indirect taxes, total expenditure, and welfare using the SWIID Gini index for the full sample of countries. Compared the results in Table 2, we find that the coefficient for direct taxes loses statistical significance when additional controls are introduced. However, the variable for total government expenditure becomes statistically significant and has a positive sign. This indicates that, in this specification, higher government spending may be linked to greater income inequality. Nevertheless, this result is not robust, as the coefficient for government expenditure in Column E has the opposite sign. By contrast, the welfare measure is significant at the 5 % level and exhibits a negative sign.

Column B presents the effect of other controls on income inequality for all countries. In addition to indirect taxes and welfare, the exchange rate is statistically significant at the 5 % level, and has a positive sign, despite the magnitude of the coefficient being shallow. Column B also shows that human capital is statistically significant with a positive sign.

Columns C and D, Table 5, confirm that the connections of direct and indirect tax revenue for OECD and non-OECD countries remain generally consistent, even after introducing three control variables: total expenditure, human capital, and welfare. While the results changed slightly for the OECD group, Column C, the direct and indirect taxes are not significant, but most controls are. For the non-OECD group, Column D, the direct and indirect taxes are significant.

In the regressions presented in Table 5 using the Word Bank Gini index as a dependent variable, columns E to J, the results are more similar to those in Table 2. Although only the indirect tax revenue is significant for the whole group, for the OECD and NOECD groups, both revenues are significant and have the expected signs.

In general, the results of the robustness checks (Tables 3, 4, and 5) indicate that while direct taxes are associated with a reduction in income inequality, indirect tax revenues are associated with a widening income imbalance. Our results are robust in indicating the link between indirect taxation and raised income inequality. Increasing government revenues through this form of taxation is a way of weakening the distributive role of the tax system.

These findings are in line with the literature. For example, Weller (2007) uses cross-country data from 1981 to 2002 and finds positive effects of progressive taxation on income distribution, as do Duncan and Sabirianova (2016). Martinez-Vazquez et al. (2014) find that progressive personal income taxes and corporate income taxes—that is, direct taxes—reduce income inequality. In this regard, a one percentage point increase in the share of progressive personal income taxes to GDP leads to a 0.1 percentage point reduction in income inequality, and an increase of one percentage point in the share of general sales tax in GDP increases income inequality by around 0.5 percentage points.

Our results are also supported in part by other studies. For example, Woo et al. (2013) found in a sample of 48 advanced and emerging market economies that the coefficients of indirect taxes are significant and of the expected positive sign, a one percentage point of potential GDP increase in indirect taxes is associated with a 0.4-0.9 percent rise in inequality. However, when the authors investigate this effect exclusively for the OECD country sample, the coefficients of indirect taxes become insignificant. In contrast, the coefficients for individual income taxes generally exhibit a positive sign and are significant in both the full sample and the OECD sub-sample. Muinelo‐Gallo and Roca‐Sagalés (2013), using panel data of 21 high-income OECD countries during the 1972-2006 period and estimating two structural equations systems, find that the effect of direct taxes on inequality is negative and significant in all estimations—which may reflect the progressive structure of the tax systems of the analyzed countries—while indirect taxes have no significant effects on inequality. These results reinforce the observations by Ciminelli et al. (2017) and Mahler and Jesuit (2018) that indirect taxation can help to structure a distributive tax system due to its ability to provide resources to governments. However, our results contrast with such works as they present statistical evidence that indirect taxation has a regressive association with income distribution, even, or mainly, for OECD countries.

Finally, we recognize that one of the main concerns when using the GMM estimator in the system is the proliferation of instruments, which can compromise the validity of the results by weakening tests such as the Hansen test and introducing overfitting in the estimates. To mitigate this problem, as indicated in Section 4, we adopted the strategy of collapsing the instrument matrix, as Roodman (2009) recommended, reducing the number of instruments by creating a single one for each variable and lag. We, too, limited the number of lags used as instruments, prioritizing robustness over statistical efficiency. Despite these precautions, we acknowledge that the number of instruments remains relatively high in some specifications, especially considering the moderate size of our sample (52 countries, 609 observations). Although the results presented in our paper are consistent with the literature and robust across different specifications, they must be interpreted cautiously, given the possibility of residual bias associated with the overidentification of instruments.

V. Conclusion

Our objective in this paper was to empirically investigate how the composition of tax sources between direct and indirect taxes is correlated with the observed income inequality measured using the Gini index. Authors like Chu et al. (2004), Decoster et al. (2010), Bastagli et al. (2012), Woo et al. (2013), and Martinez-Vazquez et al. (2014) pointed out that when the share of direct taxes in revenue is higher than the share of indirect taxes, this leads to redistributive effects that can reduce income distribution inequality. This article provides new evidence about the association of direct and indirect taxes on inequality in a panel of 52 OECD and non-OECD over the 2000-2012 period. We observed a relatively higher dependence of the countries on indirect taxes than on direct taxes, which is even more significant in our second group of non-OECD countries. While the increase of revenue with direct taxes is related to lower levels of income inequality, the rise of revenue from indirect taxes is linked to higher income inequality.

When we compared these relationships for OECD and non-OECD groups, the results were more statistically significant for OECD countries. However, this does not weaken the ability of cross-national evidence to assist in the definition of redistributive policies in a specific economy, even if it is in development. In fact, the results presented here show how increasing the tax burden through indirect taxation can have negative distributional effects, even in developed countries. This likely reflects the extent to which income redistribution considerations influence the design of tax systems in different institutional contexts. In developing economies, taxation is often viewed primarily as a source of revenue for governments, whereas in developed countries, it plays a broader role within fiscal policy.

The evidence presented in this paper highlights the need to go beyond assessing how much revenue is raised through taxation and to consider how that revenue is collected. If, on the one hand, taxation can jeopardize work decisions and human and physical capital formation, on the other, it also serves as a policy tool for preventing income inequality from reaching levels that undermine sustainable economic growth in the long run. As our findings suggest, there is clear scope for developing countries to place greater emphasis on income in the design of their tax systems. This includes increasing the role of direct taxes within the overall structure of tax revenue.

Although our analysis provides empirical evidence on the importance of tax structure and its association with economic growth, it does not address taxation at the margin—for example, the threshold at which the redistributive effects may change. This limitation is closely related to the quality of institutions and the effectiveness of fiscal policy in each country. Likewise, as noted in the introduction, due to data constraints, our study does not account for differences in the progressivity of direct and indirect taxes across countries in the sample. These issues lie beyond the scope of this paper but offer promising directions for future research—for example, through case study analyses of specific economic regimes at different stages of economic development.

This study underscores the importance of the tax structure in shaping income distribution across countries. As such, distributional concerns must be integrated into broader debates on tax reform. More fundamentally, structuring a fiscal system that can address the trade-off between efficiency and equity is critical for sustaining long-term economic growth.

Acknowledgment

This research was financed by National Postdoctoral Program (PNPD-CAPES). We especially thank the anonymous reviewers of the journal for many insightful comments and suggestions.

Notas al pie

1. Stephenson (2018) examines how the tax structure affects inequality in only five European countries.↩︎
2. The work of Diamond and Mirrlees (1971) and Atkinson and Stiglitz (1976) supported the design of an optimal tax theory, which argues that indirect taxation is suboptimal and that redistribution should be pursued solely through direct income tax.↩︎
3. In turn, Adam et al. (2015) show how the income distribution structure affects the tax system.↩︎
4. In contrast, Ciminelli et al. (2017) analyzed the fiscal consolidation processes of 16 OECD countries, and they found that adjustments made via indirect tax increases contribute more to reducing income inequality than do direct taxes. This result was likely found due to the fact, as shown by Mahler and Jesuit (2018), that indirect taxes, being more regressive, provide more resources for income transfers capable of reducing income inequalities.↩︎
5. More detailed information on the GRD can be found in Prichard et al. (2014).↩︎
6. As underlined by Prichard et al. (2014), one of the significant challenges in constructing the GRD is the treatment of social contributions, as “both countries and data sources vary in whether social contributions are included in revenue statistics at all, and, in turn, whether they are classified as tax revenue, direct tax revenue, or neither of the two. (…) these classification choices can have a huge implication for interpreting the extractive capacity of states – and can lead to major inconsistencies in seeking to draw comparisons across countries, or even within countries over time” (p.12). As a result, they decided to exclude social contributions from the revenue categories.↩︎
7. Poterba (2007) highlights the endogeneity problem arising from the potential reverse causation between income inequality and taxation. Other potential sources of endogeneity include: omitted variable effects that are present in the error term; simultaneity bias and time-invariant unobservable variables, as well as the quality of fiscal institutions.↩︎
8. In contrast to the Sargan test the Hansen J statistic does not require homoscedasticity for testing overidentification.↩︎
9. As recommended by Roodman (2009), the p-values for the Hansen tests are above the minimum of 0.25.↩︎
10. Based on the theory developed by Meltzer and Richard (1981), this combination of a higher level of income inequality and a more significant share of indirect taxation in the non-OECD economies may indicate a deficiency in the democratic system of these countries. This is an important issue, but lies beyond the scope of this paper.↩︎

References

1. Acemoglu, D. (2009). Introduction to modern economic growth. Princeton University Press.
2. Acemoglu, D., & Robinson, J. (2012). Por que as nações fracassam: as origens do poder, da prosperidade e da pobreza. Elsevier.
3. Arellano, M., & Bover, O. (1995) Another Look at the Instrumental Variable Estimation of Error-Components Models. Journal of Econometrics, 68(1), 29-51.
4. Adam, A., Kammas, P., & Lapatinas, A. (2015). Income inequality and the tax structure: evidence from developed and developing countries. Journal of Comparative Economics- 43(1), 138-154.
5. Atkinson, A. B. (2015). Desigualdade: o que pode ser feito? LeYa.
6. Atkinson, A. B., & Stiglitz, J. E. (1976). The design of tax structure: direct versus indirect taxation. Journal of Public Economics. 6(1-2). 55-75.
7. Azevedo, J. P., David, A. C, Bastos, F.R., & Pineda, E. (2014). Fiscal Adjustment and Income Inequality: Sub-National Evidence from Brazil(Working Paper No. 6945). The World Bank. [http://documents.worldbank.org/curated/en/151341468236702574]
8. Bastagli, F., Coady, D., & Gupta, S. (2012). Income Inequality and Fiscal Policy(IMF Staff Discussion Note No. 12/08R). International Monetary Fund. [https://www.imf.org/external/pubs/ft/sdn/2012/sdn1208.pdf]
9. Berg, A., & Ostry, J. (2011). Inequality and unsustainable growth: Two sides of the same coin? (IMF Staff Discussion Note No. 11/08). International Monetary Fund. [https://www.imf.org/external/pubs/ft/sdn/2011/sdn1108.pdf]
10. Berg, A., Ostry, J. D., & Zettelmeyer, J. (2012). What makes growth sustained? Journal of Development Economics. 98(2). 149-166.
11. Blundell, R., & Bond, S. (1998). Initial Conditions and Moment Restrictions in Dynamic Panel Data Models’. Journal of Econometrics. 87(1), 115-143.
12. Bond, S. R. (2002). Dynamic Panel Data Models: a guide to micro data methods and practice. Portuguese Economic Journal, 1(2), 141-162.
13. Burman, L. E. (2012). Taxes and inequality. Tax Law Review, 66, 563.
14. Ciminelli, G., Ernst, E., Giuliodori, M., & Merola, R. (2017). The composition effects of tax-based consolidations on income inequality (Research Department Working Paper No. 19). International Labour Office. [https://www.ilo.org/publications/composition-effects-tax-based-consolidations-income-inequality]
15. Cingano, F. (2014). Trends in Income In equality and its impact on Economic Growth (OECD Social, Employment and Migration Working Papers No. 163). OECD Publishing. [https://www.oecd.org/en/publications/trends-in-income-inequality-and-its-impact-on-economic-growth_5jxrjncwxv6j-en.html]
16. Chu, K. Y., Davoodi, H., & Gupta, S. (2004). Income Distribution and Tax and Government Social-Spending Policies in Developing Countries. In G. A. Cornia (Ed.), Inequality, Growth, and Poverty in an Era of Liberalization and Globalization (pp.249). Oxford University Press on Demand.
17. Decoster, A., Loughrey, J., O`Donoghue, C., & Verwerft, D. (2010). How regressive are indirect taxes? A microsimulation analysis for five European countries. Journal of Policy Analysis and Management, 29(2), 326-350.
18. Diamond, P.A., & Mirrlees, J.A. (1971). Optimal taxation and public production. The American Economic Review, 61(1), 8-27.
19. Djankov, S., Ganser, T., Mcleish, C., Ramalho, L., & Shleifer, A. (2009). The effect of corporate taxes on investment and entrepreneurship. American Economics Journal, 4(2), 23-32.
20. Duncan, D., & Sabirianova Peter, K. (2016). Unequal inequalities: Do progressive taxes reduce income inequality? International Tax and Public Finance, 23(4), 762-783.
21. Fernandes, C. B. S., Cunha, M. S., & Vasconcelos, M. R. (2019). Giving with one hand and taking away with the other. Estudios económicos, 36(73), 43-70.
22. Gobetti, S. W. (2022). Novas tendências para uma boa reforma da tributação da renda. In: M. Pires (Org), Progressividade tributária e crescimento econômico (pp. 15-49. FGV IBRE.
23. Goñi, E., López, J.H., & Servén, L. (2011). Fiscal Redistribution and Income Inequality in Latin America. World Development, 39(9), 1558-1569.
24. Gornick, J. C., & Jäntti, M. (2014). Income Inequality: Economic Disparities and the Middle Class in Affluent Countries. Stanford University Press.
25. Joumard, I., Pisu, M., & Bloch, D. (2012). Tackling income inequality: the role of taxes and transfers. OECD Journal: Economic Studies, 2012(1), 1-36.
26. Kennedy, T., Smyth, R., Valadkhani, A., & Chen, G. (2017). Does income inequality hinder economic growth? New evidence using Australian taxation statistics. Economic Modelling, 65, 119-128.
27. Kuznets, S. (1955). Economic growth and income inequality. American Economic Review, 45(1), 1-28.
28. Lambert, P. J. (2001). The Distribution and Redistribution of Income (3rd ed.). Manchester University Press.
29. Lee, D. J., & Son, J. C. (2016). Economic growth and income inequality: evidence from dynamic panel investigation. Global Economic Review, 45(4), 331-358.
30. Lustig, N. (2016). Fiscal policy, inequality and the poor in the developing world (Working Paper No. 164). Center for Global Development. [https://ideas.repec.org/p/cgd/wpaper/441.html]
31. Lustig, N., Pessino, C., & Scott, J. (2014). The impact of taxes and social spending on inequality and poverty in Argentina, Bolivia, Brazil, Mexico, Peru, and Uruguay: Introduction to the special issue. Public Finance Review, 42(3), 287-303.
32. Mahler, V. A., & Jesuit, D. K. (2018). Indirect taxes and government inequality reduction: a cross-national analysis of the developed world. Journal of Income Distribution, 16(2), 1-26.
33. Martinez-Vazquez, J. (2008). The impact of budgets on the poor: Tax and expenditure benefit incidence analysis. In B. Moreno-Dodson & Q. (Eds.), Public finance for poverty reduction: Concepts and case studies from Africa and Latin America (pp. 113-162). World Bank.
34. Martinez-Vazquez, J., Vulovic, V., & Liu, Y. (2011). Direct versus indirect taxation: trends, theory, and economic significance. In E. Albi & J. Martinez-Vazquez (Eds.), The Elgar guide to tax systems (pp. 37-92). Edward Elgar Publishing.
35. Martinez-Vazquez, J.,, B. Moreno Dodson., &V. Vulovic. (2014). The impact of tax and expenditures policies on income distribution: evidence from a large panel of countries. Review of Public Economics, 200(4/2012), 95-130.
36. Meltzer, A. H., & Richard, S. F. (1981). A rational theory of the size of government. Journal of Political Economy, 89, 914-927.
37. Muinelo-Gallo, L., & Roca-Sagalés, O. (2013). Joint determinants of fiscal policy, income inequality and economic growth. Economic Modelling, 30, 814-824.
38. North, D. C. (1990). Institutions, institutional change and economic performance. Cambridge University Press.
39. Ostry J., Berg, A., & Tsangarides, C. (2014). Redistribution, inequality, and growth (IMF Staff Discussion Note No.14/02). International Monetary Fund. [https://www.imf.org/external/pubs/ft/sdn/2014/sdn1402.pdf]
40. Persson, T., & Tabellini, G. (1994). Is Inequality Harmful for Growth? The American Economic Review, 84(3), 600-621.
41. Pickering, A., & Rajput, S. (2018). Inequality and the composition of taxes. International Tax and Public Finance, 25(4), 1001-1028.
42. Piketty, T. (2014). Capital in the Twenty-First Century. The Belknap Press of Harvard University Press.
43. Poterba, J. M. (2007). Income inequality and income taxation. Journal of Policy Modeling, 29, 623-633.
44. Prichard, W., Goodall, A., & Cobham, A. (2014). The ICTD Government Revenue Dataset (ICTD Working Paper No. 19).International Centre for Tax and Development. [https://www.ictd.ac/publication/the-ictd-government-revenue-dataset/]
45. Roodman, D. (2009). How to do Xtabond2: An Introduction to Difference and System GMM in Stata. Stata Journal, 9(1), 86-136.
46. Stephenson, A. V. (2018). The impact of personal income tax structure on income inequality for Belgium, Bulgaria, Germany, Lithuania, and Poland: A comparison of flat and graduated income tax structures. Atlantic Economic Journal, 46(4), 405-417.
47. Stiglitz, J. E., & Rosengard, J. K. (2015). Economics of the Public Sector. W.W. Norton & Company.
48. Stiglitz, J. E. (2017). El Precio de La Desigualdad: el 1% de la población tiene lo que el 99% necesita. Taurus Pensamiento.
49. UNU-WIDER (2018). GRD–Government Revenue Dataset [Data Set]. UNI-WIDER. https://www.wider.unu.edu/project/government-revenue-dataset
50. Vartia, L. (2008). How do taxes affect investment and productivity? An industry-level analysis of OECD countries (OECD Economics Department Working Paper No. 656). Organisation for Economic Co-operation and Development. [https://www.oecd.org/content/dam/oecd/en/publications/reports/2008/12/how-do-taxes-affect-investment-and-productivity_g17a1c35/230022721067.pdf]
51. Weller, C. E. (2007). The Benefits of Progressive Taxation in Economic Development. Review of Radical Political Economics, 39(3), 368-376.
52. Windmeijer, F. (2005). A Finite Sample Correction for the Variance of Linear Efficient Two-Step GMM Estimators. Journal of Econometrics, 126(1), 25-51.
53. Wintoki, M. B., Linck, J. S., & Netter, J. M. (2012). Endogeneity and the dynamics of internal corporate governance. Journal of financial economics, 105(3), 581-606.
54. Woo, J., Bova, M. E., Kinda, M. T., & Zhang, M. Y. S. (2013). Distributional consequences of fiscal consolidation and the role of fiscal policy: What do the data say? (IMF Working Paper No. 13/195). International Monetary Fund. [https://www.imf.org/en/Publications/WP/Issues/2016/12/31/Distributional-Consequences-of-Fiscal-Consolidation-and-the-Role-of-Fiscal-Policy-What-Do-40942]
55. Yamada, K., & Kawaguchi, D. (2015). The changing and unchanged nature of inequality and seniority in Japan. Journal of Economic Inequality, 13(1), 129-153.
56. Yellen, J. (2014, October 17). Perspectives on Inequality and Opportunity from the Survey of Consumer Finances [Speech]. Federal Reserve Bank of Boston. https://www.federalreserve.gov/newsevents/speech/yellen20141017a.htm
57. Yokoyama, I., Kodama, N., & Higuchi, Y. (2019). What happened to wage inequality in Japan during the last 25 years? Evidence from the FFL decomposition method. Journal of the Japanese and International Economies, 51, 87-98.

Appendix