THE DYNAMICS OF POVERTY IN PORTUGAL: A MULTINOMIAL LOGISTIC REGRESSION ANALYSIS

This dissertation uses a four-year panel of 4929 individuals from Portugal collected during 2017-2020 to classify individuals into three groups: the chronically poor; the transitorily poor; and the never poor. Our aim is to identify key determinants which distinguish the chronically poor from the transitorily poor. After choosing well-defined explanatory variables, a multinomial logistic regression model was used to identify the causes of chronic and transitory poverty. Furthermore, in order to quantify the effect of each single explanatory variable, we have calculated the respective average marginal effects. In addition to identifying the main factors that are causing poverty in general in Portugal, evidence is provided that when using a multinomial logistic estimation model to estimate and understand chronic poverty we are able to identify characteristics that are more prevalent in causing chronic poverty, rather than transitory poverty. We conclude that a person's health or residential status, as well as whether he or she has completed primary education or is a single parent, are among the other distinguishing characteristics between the chronically and transitorily poor.