Comparative Analysis of Heteroscedastic and Homoscedastic OLS Models

This study considered foreign direct investment FDI as response variable while, gross domestic product GDP , inflation and exchange rate were the predictor variables. The data were obtained from the Central Bank of Nigeria Statistical Bulletin spanning from 1970 to 2019. The study aimed at comparing heteroscedatic and homoscedastic OLS modes. Our findings revealed that the predictor variables in the heteroscedastic OLS model were not significant and were able to account for about 44 of the variation in the response variable. The diagnosis of the fitted regression model using BreuschPagan test showed that the assumption of homoscedasticity was violated. To address the problem of heteroscedasticity, all the variables were converted to log form to stabilise the variance. Our results from the now homoscedstic model revealed that all the predictor variables were significant and able to account for about 82 of the variation in the response variable. Therefore, our study established that when the assumption of homoscedasticity is violated, the model parameters become inefficient, the standard errors biased and the t statistics and the p values no more valid. On the other hand, this study evidently proved that homoscedastic OLS model provide better estimates than heteroscedastic OLS model. AkpensuenShiaondo Henry | Joel Simon | Alhaji Abdullahi Gwani | Joshua Hassan Jemna "Comparative Analysis of Heteroscedastic and Homoscedastic OLS Models" Published in International Journal of Trend in Scientific Research and Development (ijtsrd), ISSN: 2456-6470, Volume-4 | Issue-6 , October 2020, URL: https://www.ijtsrd.com/papers/ijtsrd33553.pdf Paper Url: https://www.ijtsrd.com/mathemetics/other/33553/comparative-analysis-of-heteroscedastic-and-homoscedastic-ols-models/akpensuenshiaondo-henry