Introduction

This study investigates whether recent empirical findings support the theory of political business cycles in Latin America and examines whether income inequality varies according to the ruling party’s ideology. The political business cycle theory, particularly its partisan dimension, suggests that left wing and right wing parties adopt distinct policy approaches. Left-wing governments generally prioritize redistributive policies to reduce income inequality, while right-wing governments emphasize economic stability, often through inflation control (^{Hibbs, 1977}; ^{Alesina, 1987}; ^{Huber and Stephens, 2012}).

The research tests this theory using a panel dataset of 13 Latin American countries from 1994 to 2013, encompassing 260 observations. Employing robust causal inference techniques and statistical models, the study explores the relationship between political cycles and income inequality, contributing empirical evidence to the Political Science literature and assessing the relevance of classic theories in the Latin American context.

The paper is structured into five sections: a review of traditional political business cycle models with a focus on income inequality, hypotheses and variable definitions, empirical strategy and statistical modeling, presentation of results, and a conclusion summarizing the key findings.

Party Political Cycles and the Place of Income Inequality

The allocation of public resources is key in political science, influencing institutions and serving as political power (^{Dahl, 1996}; ^{Lijphart, 2012}). Research examines how preferences shape distribution, linking it to economic policy cycles (^{Downs, 1999}; ^{Nordhaus, 1975}; ^{Alesina, 1987}). ^{Hibbs (1977)} introduced ideology, leading to ‘partisan models’ exploring its effect on policies, though debates persist on its impact on macroeconomic and social variables.

Income inequality undermines political equality, weakening democracy (^{Lijphart, 2012}; ^{O’Donnell, 1999}; ^{Huntington, 1993}). Often viewed through an economic lens, its role in politics is underexplored. ^{Kerstenetzky (2002)} refuted ‘trickle-down’ growth, showing structural inequalities perpetuate poverty, erode social cohesion, and harm democracy. Redistributive policies are vital to address these issues.

This study adopts the CPO perspective, asserting that party ideology shapes resource allocation. Left-wing parties emphasize redistribution, while right-wing parties prioritize inflation control. Fiscal policy is central to implementing these ideologies. Figure 1 outlines a framework linking party cycles to income inequality, guiding the analysis.

Source: Elaborated by the authors.

Figure 1 Theoretical approach and the ‘place’ of inequality. 

Research increasingly views income inequality as politically driven, shaped by electoral power dynamics. Studies highlight government policies, particularly through party political cycles, as key influences on inequality.

^{Bradley et al. (2003)} found left-wing parties in post-industrial democracies reduced inequality via progressive taxes, transfers, and union support. ^{Morley (2001)} attributed Latin America’s inequality to colonial legacies of concentrated power and limited access to education and infrastructure. ^{Huber et al. (2006)} showed left-wing governments in Latin America and the Caribbean (1970-2000) significantly reduced wage disparities through redistributive policies.

^{Huber and Stephens (2012)} emphasized that structural barriers weakened leftist parties, limiting redistribution. Effective policies, such as conditional cash transfers and universal programs funded by taxes, reduced inequality but require middle-class support for sustainability.

These studies, while valuable, overlook critical political processes, recent shifts in power dynamics, and economic growth over the past two decades, limiting their ability to address contemporary income inequality in Latin America.

We acknowledge that the temporal scope of this study (1994-2013) did not explicitly account for the impact of the economic boom driven by raw material exports to China-a factor that may have influenced leftist governments’ resource distribution policies across the region. As noted by ^{Lustig (2018)} in her research on inequality, periods of economic growth can enhance state capacity to reduce social disparities. However, this dynamic is far from uniform, as illustrated by the case of Venezuela, where the long-standing presence of a leftwing government did not prevent persistently high levels of inequality. Given that social inequality is a multidimensional and highly complex phenomenon, we recognize this limitation and suggest that future research further explore the interplay between economic cycles and redistributive policies, in order to gain a more comprehensive understanding of how these variables affect the context under study.

Table 1 identifies five key categories influencing income inequality: ideological, economic, social, demographic, and institutional factors, summarized in Figure 2.

Source: Authors’ elaboration based on ^{Morley (2001)}, ^{Bradley et. al. (2003)}, ^{Huber et. al. (2006)}, and ^{Huber and Stephens (2012)}.

Figure 2 Causal relations of income inequality 

Causal connections

This paper outlines the hypotheses developed within the traditional analytical framework of party political cycles. It incorporates the well-established dependent variables frequently analyzed in the literature on political party cycles-economic growth, inflation, and unemployment. Furthermore, it introduces a less-explored variable as a dependent variable: income inequality (see Table 2).

We have selected four dependent variables (DVs) for this analysis: DV1) Economic Growth, DV2) Inflation, DV3) Unemployment and DV4) Income Inequality. The first three variables are traditionally analyzed in studies testing the theory of political-economic cycles from a partisan perspective. In contrast, DV4 enables us to examine income inequality not merely as a ‘statistical control,’ as it is often treated in the literature, but as an outcome shaped by a political variable- namely, party ideology. The specific metrics used to measure these dependent variables are detailed below:

DV1: Economic Growth was measured using the annual percentage variation in the Gross Domestic Product (GDP) of the selected countries. Data for this variable were sourced from the Economic Commission for Latin America and the Caribbean (ECLAC) and the International Monetary Fund (IMF).

Dv₂ = Inflation was measured based on the annual percentage change in the consumer price index (CPI) of the selected countries. Data for this variable were also obtained from ECLAC and the IMF.

Dv₃ = Unemployment was assessed as the percentage of the labor force unemployed in the selected countries. The data for this variable were sourced from ECLAC and the IMF.

Dv₄ = Income Inequality was measured using the Gini Index, a widely recognized indicator in the literature. The Gini Index ranges from 0 to 100, providing a detailed assessment of income inequality. It is important to note that, in this analysis, the Gini coefficient is presented on a 0 to 100 scale rather than the 0 to 1 scale commonly used. Data for this variable were obtained from ECLAC and the World Bank.

For the independent variables, party ideology was identified as the primary variable of interest, supplemented by a set of control variables. Additionally, some dependent variables were treated as potential independent variables in accordance with the theoretical expectations of the models. The operationalization of these variables is detailed as follows:

Not all estimation models employed the full set of variables. Following the referenced literature, the model with economic growth as the dependent variable included control variables such as inflation, human capital, and trade openness. For the model using inflation as the dependent variable, the control variables considered were economic growth and trade openness. Similarly, in the model where unemployment served as the dependent variable, the control variables included economic growth, human capital, and trade openness. Finally, in the analysis of income inequality as the dependent variable, the estimation model incorporated economic growth, inflation, unemployment, and human capital as control variables (^{Hibbs, 1977}; ^{Ames, 1987}; ^{Alesina, 1987}; ^{Garrett, 1998}; ^{Borsani, 2003}; ^{Morley, 2001}; ^{Bradley et al., 2003}; ^{Huber et al., 2006}; ^{Huber and Stephens, 2012}).

A lagged dependent variable was included as an independent variable in all statistical models, with further details provided in later sections. The variable selection follows established approaches in similar empirical studies. Table 3 lists the variables, their parameterization, and data sources, while Table 4 summarizes the dependent and independent variables and the expected causal relationships.

The empirical strategy

Using panel data in this study enhanced inference by increasing the number of observations (^{Hsiao, 2007}). However, this approach required applying specific statistical procedures to address seasonal patterns or trends, ensuring accurate analysis of causal relationships between variables.

Chart 1 highlights the temporal trends of four dependent variables: Economic Growth (DV1), Inflation (DV2), Unemployment (DV3), and Income Inequality (DV4). Over the observed period, Unemployment and Income Inequality demonstrated a declining trend, whereas Economic Growth and Inflation exhibited relatively stable upward trajectories.

Source: Elaborated by the authors

Chart 1 Dependent variables in levels 

Statistical tests were conducted to assess the stationarity of the historical series for the dependent variables. The approach recommended in the literature to achieve stationarity involves applying successive differencing to the original series (i.e., the series in levels). Following the methodology proposed by ^{Morettin and Toloi (2006)}, the first difference of Z(t) is defined as:

With the second difference defined as:

Therefore, for the n-th difference of the series Z(t), the equation is defined as:

To determine the optimal number of lags, we employed the Akaike Information Criterion (AIC), Hannan-Quinn Information Criterion (HQIC), and Schwarz’s Bayesian Information Criterion (SBIC). The test results revealed that the historical series achieved stationarity with one lag, as shown in Annex 1A

Unit root tests were conducted on the historical series using a single lag. The

Augmented Dickey-Fuller (ADF) test and the Kwiatkowski-Phillips-SchmidtShin (KPSS) test were employed, with the latter serving as a complementary validation to the former. The results revealed that economic growth, unemployment, and income inequality exhibit stationarity at the first difference, whereas inflation achieves stationarity at the second difference (see Annexes 1B and 1C).

The KPSS test statistics, being lower than the critical values, corroborated the findings of the ADF test, providing no statistical evidence to reject the null hypothesis of stationarity for the series with the specified lags. Chart 2 displays the stationary dependent variables, expressed in their differenced forms.

Source: Elaborated by the authors

Chart 2 Dependent variables in difference 

Results and Discussion

This study’s theoretical framework argues that party ideology influences macroeconomic variables such as economic growth, inflation, and unemployment. Left-wing governments prioritize economic growth to reduce unemployment, even at the cost of higher inflation, while centerand right-wing governments focus on controlling inflation to enhance social well-being (^{Hibbs, 1977}; ^{Alesina, 1987}).

The analysis tests these claims through hypotheses H1, H2, and H3, examining the impact of party ideology on traditional partisan political cycle variables. Additionally, it explores the effect of party ideology on income inequality (H4). Regression models emphasize the role of left-wing governments, with centerand right-wing governments serving as the reference category.

Table 7 presents the estimation models, highlighting coefficients and statistical significance, offering robust evidence for the hypotheses.

Model [1] in Table 7 tests hypothesis H1, showing that party ideology is statistically significant and supports theoretical expectations. During left-wing governments, economic growth increased by 1.07 units compared to centeror right-wing administrations. Among control variables, the lagged dependent variable, human capital, and trade openness were also significant. Past economic performance negatively affected current growth, while higher human capital added 2.03 units to growth. Greater trade openness contributed a modest 0.013 units, aligning with findings by ^{Acemoglu et al. (2015)} and ^{Acs et al. (2014)}. Inflation, however, was not statistically significant, indicating no measurable impact on growth in this dataset.

Model [2], testing H2, shows that while party ideology aligns with expectations (inflation tends to rise under left-wing governments), it is not statistically significant, suggesting no clear link between ideology and inflation. Among control variables, the lagged dependent variable and economic growth were significant but unexpectedly negatively correlated with inflation. Trade openness was not significant, indicating no measurable effect on inflation in this dataset.

Model [3], testing H3, showed the expected sign for partisan ideology but no statistically significant association with unemployment. The only significant variable was GDP growth, confirming that higher economic growth correlates with lower unemployment. Human capital, trade openness, and the lagged dependent variable were not significant, leaving their effects inconclusive.

Model [4] extends party political cycle theories by incorporating income inequality as a dependent variable tied to party ideology, testing H4. Fixed-effects model results indicate that left-wing governments reduce wage inequality compared to centerand right-wing governments. The variables in this model demonstrated causal directions consistent with predictions in the literature, although unemployment was not statistically significant. Other control variables, however, showed significance at varying levels: inflation at p < 1%, economic growth and human capital at p < 5%, and the lagged dependent variable (past income inequality) at p < 10%.

Conclusions

This study analyzed the influence of party ideology on four key variables: economic growth, inflation, unemployment, and particularly income inequality, aiming to test the partisan theory of political economy cycles in Latin America from 1994 to 2013. Specifically, it sought to determine whether party political cycles explain the observed trend of declining income inequality in the region. By treating income inequality as a dependent variable shaped by political decisions, the paper shifts focus from the traditional view of inequality influencing politics to exploring how ideology impacts inequality, contributing to Brazilian political science.

The study built hypotheses based on the theory of party political cycles, proposing causal links between party ideology and macroeconomic variables, and notably, between ideology and income inequality. The findings suggest that income inequality is not merely an economic by-product but also a result of political decisions. This approach highlights the role of social choices and conflicts of interest in shaping public policies, offering the possibility of altering inequality patterns over time.

The results provide initial evidence that party ideology influences income inequality trajectories. However, the study did not examine specific mechanisms used by left-wing governments to reduce inequality, such as redistributive programs or poverty reduction measures. These questions present a promising research agenda for future studies, emphasizing the potential of political action to transform this persistent social issue.