Comparative Analysis of the Impact of Some Economic Variables on Inflation Rates in Nigeria using Robust Regression

This study examines the impact of Economic Variables (FDI, Money Supply, GDP per Capita, Population Growth and Real Interest rate) on Inflation rate in Nigeria over the period of 2006 to 2023 using OLS regression and Robust Regression (M-estimation, Bi-square Estimation, Iteratively Re-weighted Least Squares (IRLS), S-Estimation). It was found out that M-estimation and Iteratively Re-weighted Least Squares (IRLS) give similar result as Bi-square robust estimation method. From the results of OLS regression it was found that Money Supply shows a positive significant impact on inflation rate, Real Interest Rate shows a negative significant impact on inflation rate while FDI, GDP per capita and population growth shows insignificant negative impacts on inflation rate. Robust regression estimation method using bi-square estimation method indicates that money supply shows significant positive impacts on inflation rates; population growth rate shows insignificant positive impacts on inflation rates while GDP per capita and real interest rate shows negative significant impact on inflation rate; FDI shows insignificant negative impact on inflation rate. From the results of S-robust regression analysis method it is seen that FDI, GDP per capita, population growth rates and real interest rates have negative and significant impacts on inflation rates whereas money supply has a positive but insignificant impact on inflation rates in Nigeria. Also the nature of impact of the variables in the model does not change but there is a reduction in the standard error of the residual from 0.21 in OLS and 0.13 in S- Estimation model to 0.084 in bi-square regression; R^2 has increased from 77.2% in OLS to 90.3% in Bi-Square regression; S-estimation model has the highest R^2 of 96.6%. Comparing the three measures, bi-square, M and IRLS estimation methods has been considered since a higher R^2 with a higher standard error of residual may indicate over fitting problems.