A fundamental question in economics is whether inequality is good or bad for growth. Here, we postulate that the composite nature of overall inequality is probably causing the lack of consensus. Inequality is generated by factors beyond the individuals’ control (background, race, macroeconomic conditions) and by the individual willingness to exert effort. The cholesterol hypothesis suggests that the former, inequality of opportunity (IO), is growth-deterring while the latter, inequality of pure effort, is growth-enhancing. Inequality and poverty are closely interrelated aspects of the income distribution so any attempt to understand the effect of inequality on growth must also consider the influence of poverty. We build an overlapping generation model with human capital and poverty trap to derive a reduced form equation that is consistent with the cholesterol hypothesis. Later, we study the causal effect of inequality and IO on growth by using instrumental variables and System-GMM extended with instrumental variables. Results show that the effect of IO on growth is significantly negative and robust, while the estimated coefficient for overall inequality turns higher when IO is included in the estimation, which is consistent with the cholesterol hypothesis.