Aim/purpose – This paper aims to examine with new empirical evidence the joint relationships between violence, income inequality, and real income per capita in a simultaneous equation framework using a worldwide sample at the country level.
Design/methodology/approach – To examine the several simultaneous relationships between the variables, this study uses the Seemingly Unrelated Regression (SUR) and Three-Stage Least Squares (3SLS) with two-way fixed effects on a linear system of regression equations. The data used for analysis are sourced from the World Bank, the SWIID inequality database, and the Penn World Table. The final sample for the estimations includes 110 countries in the period between 1994 and 2019.
Findings – Based on the estimations, the results confirm a strong positive relationship between violence and income inequality. Conversely, a negative but non-robust relationship exists between violence and real income per capita. Additionally, the findings show that human capital based on years of schooling plays a critical role in reducing both inequality and violence.
Research implications/limitations – The negative relationship between income and violence is sensitive to the sample size. The institutional framework characterized by high levels of democracy does not ensure by itself a reduction in violence. The SUR model is limited to the endogeneity of the variables. Instruments selected for the 3SLS are based on previous lags of the endogenous variables, no external instruments were used. Data availability also compromises extending the estimations with a greater number of controls.
Originality/value/contribution – This study considers the explicit joint simultaneous endogenous behavior of income inequality, violence, and real income in a worldwide sample, which contrasts most of the traditional individual-type analysis of previous studies with limited samples. Furthermore, it provides evidence of the importance of human capital and the existence of the non-robust relationships between income and violence.
