OLS versus quantile regression in extreme distributions

Abstract

Financial data mostly have fat tail and an analyst is much concerned about the tail part. Most of the study in finance extensible uses linear regression but when it comes to tail analysis it becomes ineffective. So, the present study tries to address the same by using Quantile regression in the tail analysis to study the value effect in 10 portfolios formed from BSE 500 stocks based on P/B ratio. The study result clearly indicates that Quantile regression estimates give more comprehensive and vibrant picture of the unpredictable effect of the predictors on the response variables.