The literature on growth convergence has focused to a great extent on the role of initial incomes as a primary determinant of long-term growth outcomes. Expanded versions of growth models have used other explanators to unpack the growth process. In this paper we add to the literature in two significant ways: (a) we use socioeconomic variables that are sometimes overlooked in explaining growth (such as, political stability and political alliance, social heterogeneity, and demographic distribution), and (b) we demonstrate that earlier analyses may be overlooking the problem of normality and endogeneity in regression models (and we provide alternate methods like instrumental variable and distribution dynamics to control for these). In this paper we analyze the per capita income growth at the subnational level in India for the period 1981–82 to 2010–11 using an expanded growth framework. We find that initial incomes, the ratio of working age group to total population, political stability and alliance, and the extent of development expenditure play a positive and significant role in predicting growth. We also find that, contrary to popular belief, the presence of marginalized groups—namely Scheduled Castes and Scheduled Tribes—have not been a hindrance to growth of per capita incomes in states. Our findings on the influence of social institutions may have significant implications for a public policy of affirmative action in India. The results on the impact of development expenditure on growth is also important for states seeking to increase their growth rates through policy intervention.