Static and dynamic panel models: Which is better? (Case study: Poverty data in Indonesia 2012-2019)

Abstract

Indonesia’s poverty rate continues to decline from year to year; however, the decline in the percentage of poor people is slowing down continuously, following the law of diminishing returns. Many factors influence poverty, including HDI, Gini ratio, and the open unemployment rate. This study model static and dynamic panel data on Indonesia’s poverty cases in 2012-2019 to explain the relationship between HDI, Gini ratio, and the unemployment rate to poverty. It also compares poverty modeling with static and dynamic panel data regression. The results show that the Gini ratio has a significant and positive relationship to poverty. HDI has a significant and negative relationship to poverty. However, there is insufficient evidence to suggest that unemployment affects poverty. Using static panel data regression, the selected model is a fixed effect model but still experiences violations of the classic autocorrelation assumption. The dynamic panel data model gives better results than the static panel data regression model when viewed from the r-square value and the number of variables that significantly affect. From the results of this study, it can be implemented that an increase in economic growth without an increase in equity has no impact on poverty. So, it needs extra efforts by the government in economic equality for all regions of Indonesia. Also, there is a need for additional actions by the government and the community to improve education, health, purchasing power, and community competitiveness.