Nowcasting Official Poverty Statistics in the Philippines
Author: Manuel Leonard F. Albis, Josefina V. Almeda, Ph.D, Jessa S. Lopez, Roxanne Jean L. Elumbre & Dannela Jann B. Galias
Abstract:
Cognizant of the need to generate timely, relevant, and granular statistics for monitoring and policymaking, an alternative methodology that can measure poverty during periods where data from the Family Income and Expenditure Survey (FIES) is still being collected is essential. To address this need, this paper presents an alternative methodology through the use of Dynamic Factor Model (DFM) in state-space form to generate nowcasts of poverty incidence among families in the Philippines. A two-stage modeling approach was implemented, wherein dynamic factors were extracted from collected macroeconomic indicators and employment data. The extracted dynamic factors were then utilized to construct a poverty incidence model that could generate nowcasts of poverty incidence among families for 2019 to 2021. The constructed DFMs for both the macroeconomic indicators and labor information both yielded relatively low forecast errors (MAE and MAPE). Relatively high forecast accuracy was also observed for the predicted values of poverty incidences among families.
Keywords:Dynamic Factor Model (DFM), state-space form, poverty statistics, nowcasting, poverty incidence, factor analysis