Patterns of poverty among elderly Americans: a latent class Markov model

ABSTRACT

This article studies poverty persistence and the role of social security programmes on poverty among elderly in the US. We use a Latent Markov model to disentangle unobserved heterogeneity and state dependence. Because of its dynamic nature, unobserved heterogeneity is modelled to vary over time. This allows to capture different latent states of poverty that change over time. Result indicates the existence of three unobserved types evolving over time according to their propensity to be poor. Moreover, a strong persistence in poverty especially for women, individuals living alone and ethnic minorities is found. Finally, the estimates indicate that giving social assistance tends to reduce poverty.

KEYWORDS:

• Poverty persistence
• older Americans
• latent Markov model
• social security programmes

JEL CLASSIFICATION:

• I32
• C23
• J14

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ A set of Matlab functions for the estimation are available upon request.