Research on factors affecting demand for life insurance typically uses cross-sectional data, where households that have insurance coverage are compared with those that do not. Few studies have tracked changes in life insurance consumption over time within the same household. A notable exception is Liebenberg et al. (2012). They used data for 1479 American households from the Survey of Consumer Finances (SCF) that was conducted in 1983 and 1989 and used a panel approach. However, their primary focus was to understand whether specific life events such as marriage, a new job, or having a new child affect insurance demand. In this study, we use a unique, large dataset comprising 34,885 households across 1503 villages and 971 urban neighbourhoods that were included in the Indian Human Development Survey (IHDS) conducted in 2004–2005 and 2011–2012. The availability of data for individual households from the two periods allows us to model changes in life insurance demand within the same household over time, which takes care of omitted variable bias and possible issues with endogeneity.
Separate logistic regression models are built for acquisition and discontinuation, and for urban and rural households. Economic factors, such as income, socio-economic status, changes in financial conditions, and financial inclusion (such as opening a bank account or taking a bank loan), as well as demographic factors, such as gender of the household head and an increase in family size, had statistically significant effects. The main difference between rural and urban households was that financial inclusion affected the former but not the latter. The probability of being insured increased with an increase in family size, and if the head of the household was male. We find that poor households that would benefit the most from insurance do not, in fact, have such coverage. Financial inclusion and education may improve insurance penetration among such households.
