From Description to Simulation: Using Survey Data to Approach the Empirical Puzzle of Financial Inclusion in Mexico

Financial inclusion, typically measured by bank account ownership rates, is seen as a driver of economic growth and a tool for reducing inequalities, enhancing financial security and opportunities to save and invest. In recent decades, significant progress has been made worldwide, largely due to the rise of mobile banking (Demirgüç-Kunt et al., 2018). This article examines the case of Mexico, where, despite favorable conditions for increased financial inclusion, the percentage of people with a bank account remains strikingly low: only around 50% of the population owns one (Demirgüç-Kunt et al., 2022; INEGI, 2022).

This overall percentage, however, masks significant inequalities across income groups: people in higher income quintiles are more likely to own a bank account than those in lower quintiles. Additionally, within these groups, there is considerable heterogeneity: individuals with similar incomes making divergent decisions regarding account ownership. Following an analytical sociology approach (Hedström & Bearman, 2009; Manzo, 2014), I take these patterns, shown by the National Survey on Financial Inclusion (ENIF), as an explanandum. Drawing on DiMaggio & Garip (2011, 2012) and Zhao & Garip (2021), I then postulate network externalities and homophily as potential mechanisms driving this outcome.

Building on Manzo’s (2013) model on educational choices, I develop an empirically calibrated agent-based model, where the probability of an agent opening an account increases with higher income and education levels (assigned based on ENIF) and as more contacts adopt this behavior (with network size derived from ISSP). By manipulating the presence of externalities and levels of homophily in network formation, I observe their effects on account ownership rates within the artificial population. Results show a particularly close match with the empirically observed pattern for an intermediate level of homophily when externalities are present, suggesting that income and education differences alone cannot fully explain the macroscopic structure of interest.