Non-parametric analysis of poverty duration using repeated cross section: an application for Peru

Using repeated cross-section annual data for Peru spanning 2002–2011 and non-parametric duration analysis, our estimates support the hypothesis that both stay in and exit from poverty (non-poverty) depends on the duration and sequence of poverty (non-poverty) spells. We find that longer periods in poverty reduce the probability of leaving poverty and, conversely, longer periods spent out of poverty reduce the chance of falling back into poverty. Also, we show that, at least in the last decade (of high economic growth), the probability of staying in poverty was lower than staying in non-poverty and the probability of re-entering in poverty was higher than re-entering in non-poverty, being both differences growing with the number of accumulated spells. Past experiences of poverty and non-poverty seem to be essential to predict the future status of poverty.