Andryan Setyadharma, FEB Ilmu Ekonomi (2021) A New Insight of Specification Error In Regression: Excluding Education Variable From a Model. In: International Conference Advanced in Applied Mathematics (ICAAM 2021).

	PDF - Published Version Download (1MB)
	PDF - Published Version Download (4MB)

Abstract

The aim of this study is to show specification error in a model that exclude a relevant independent variable. Literature suggests that human capital features, such as education, must be considered in studies of social problems, such as environmental degradation. Therefore, the absence of education variable in a multiple regression equation model that explain the determinants of environmental degradation may result in model misspecification. Firstly, this study constructs a model with four variables, i.e.: unemployment level (ULEVEL), poor areas (POOR_AREAS), income inequality (INEQ) and access to electricity (ELEC), that may affect environmental quality in Indonesia. Secondly, this study examines the second model when education variable is included in the model. This study uses panel data regression method from 33 provinces in Indonesia during 2012 to 2018. The results of the first model show that all four control variables are statistically significant. However, ELEC’s coefficient has unexpected sign. When education variable is added to the second model, ELEC’s coefficient has expected sign. In conclusion, education variable is theoretically and empirically important variable to explain the changes in the quality of environment in Indonesia and education variable should be used in the model to avoid specification error in a model.

Actions (login required)

View Item