Hi, I am running panel data analysis using Eviews. My dataset consists of one dependent variable and 11 independent variables for six countries over 21 years. After conducting multicollinearity tests, I found that all VIF values are below 5. However, I discovered heteroscedasticity and my data is not normally distributed. I checked for endogeneity but found no issues with it. Nonetheless, cointegration is present and some variables are stationary at first difference. After conducting POLS, RANDOM, and FIXED effects, I ran the Hausman test, which showed that I should select fixed effects. However, due to cointegration and unit root, I decided to use dynamic panel data. Since T>N, I cannot use GMM, so I opted for FMOLS. Unfortunately, I obtained negative R squared and adjusted R squared. Can anyone suggest what I should do next to get results since the adjusted R squared is unusually low?
Thanks in advance.
This book might be useful: “Panel Data Analysis using EViews” by I. Gusti Ngurah Agung.