Analysis of Minimum Distance Instrumental Variable Quantile Regression and Its Application on Analyzing Iranian Household Expenditure and Income Data

Hosseinzadeh, S.E.; Golalizadeh, Mousa

doi:10.22034/jr_iss.2024.729436

Analysis of Minimum Distance Instrumental Variable Quantile Regression and Its Application on Analyzing Iranian Household Expenditure and Income Data

Document Type : Original Article

Authors

S.E. Hosseinzadeh ¹

Mousa Golalizadeh ²

¹ Department of Statistics, Tarbiat Modares University, Tehran , Iran

² Department of Statistics Tarbiat Modares University Tehran,Iran

10.22034/jr_iss.2024.729436

Abstract

One of the most common statistical models to study relationships among variables in a scientific investigation is the linear (mean-based) regression that rests on assumptions like normal errors and no correlation between regressors and the error term. When the normality of error component is questionable, quantile regression is a good choice because it is more robust to outliers. If endogeneity is a concern-i.e., regressors correlate with the error- instrumental variables come to play. One the other hand, minimum distance method which minimizes the gap between observed and model-implied distributions is powerful tool to fit the models.

This paper lays out minimum-distance instrumental-variable quantile regression (MD-IVQR) and gives details on its relevant theoretical computation. Also, using Iranian household expenditure–income data from 1402, we compare MD-IVQR with conventional Quantile Regression method and show that it controls endogeneity and reveals distributional heterogeneity.

Subjects