Learning Poverty: Measures and Simulations
João Pedro Azevedo
Abstract
COVID-19-related school closures are \n pushing countries off track from achieving their learning \n goals. This paper builds on the concept of learning poverty \n and draws on axiomatic properties from social choice \n literature to propose and motivate a distribution-sensitive \n measures of learning poverty. Numerical, empirical, and \n practical reasons for the relevance and usefulness of these \n complementary inequality sensitive aggregations for \n simulating the effects of COVID-19 are presented. In a \n post-COVID-19 scenario of no remediation and low mitigation \n effectiveness for the effects of school closures, the \n simulations show that learning poverty increases from 53 to \n 63 percent. Most of this increase seems to occur in \n lower-middle-income and upper-middle-income countries, \n especially in East Asia and the Pacific, Latin America, and \n South Asia. The countries that had the highest levels of \n learning poverty before COVID-19 (predominantly in Africa \n and the low-income country group) might have the smallest \n absolute and relative increases in learning poverty, \n reflecting how great the learning crisis was in those \n countries before the pandemic. Measures of learning poverty \n and learning deprivation sensitive to changes in \n distribution, such as gap and severity measures, show \n differences in learning loss regional rankings. Africa \n stands to lose the most. Countries with higher inequality \n among the learning poor, as captured by the proposed \n learning poverty severity measure, would need far greater \n adaptability to respond to broader differences in student needs.