Panel Data Analysis

Question Imagine that you have been employed as an economist of riksbank Sweden and you have been asked to investigate on “the factors that influence financial performance for Swedish monetary financial institutions using panel data between the periods 2010-2014”. Your research question is what are the factors that influence performance in the Swedish financial sector? To achieve this objective, performance which is also known as profitability is measure using returns on Assets (ROA) is given with a series of explanatory factors which are either macro or external factors and micro or internal balance sheet factors. Data for all these variables are provided. Now you are expected to perform a regression analysis using panel data between 2010-2014 and model which factors affects profitability (ROA). The model is specified as follows ROA_(it )═ β_0 + 〖β_1 Intr〗_it + 〖β_2 Inflr〗_it + 〖β_3 LnGDP/Cap〗_it + 〖β_4 Gearingr〗_it + 〖β_5 Liquidityr〗_it +β_6 ProfMargr_it +〖β_7 [login to view URL]〗_it + 〖β_8 [login to view URL]〗_it + Ԑit. The model indicates the external and internal factors that influence financial performance for Swedish monetary institutions. Their ‘a priori theoretical expectations’ are as follows: β_1<0,β_2<0,β_(3 )˃0,β_(4 )˃0 ,β_5 ˃ 0 β_6 ˃0,β_7 ˃0 ,β_8 <0 Now perform the following tests for the following model (Note that in the model above the GDP is natural log (ln)) Descriptive statistics (summarised) for panel data showing the Mean, etc Test for Multicollinearity using the VIF for the data set Check for the linear relation using the correlation coefficient Now perform the Hausman test and conclude if it is a fe or re model, motivate why Check for Heteroskedasticity using the White test and Breusch pagan, what did you conclude from your results? Check for Auto correlation using the DW stat and Wooldridge test, what did you conclude from your results? Perform the GLS Perform the FGLS Now check for SE robustness from the above results and then decide which model is the best. I.e. A model with correction for standard errors and a model without correction for standard errors. The aim is to see if the results would be the same and it is another way to check for reliability. NB (Is there any test you can perform to achieve this result) Check this link [login to view URL] [login to view URL] From the data, is it possible to use natural log (Ln) for all the variables both dependent and independent. If it is possible based on the data, I think it is the best to use but I was thinking such principle does not occur for negative and zero figures. If it is possible to do that then I think it is the best OR you can use any of your expertise to produce the best results. Summary This work is to perform a panel regression and present the results without correction for standard errors and with corrections for heteroskedasticity and autocorrelation. Please check this link to have an understanding of what could be done. The aim is to ensure the robustness of the result is achieved by correcting for SE. You must read the following before doing the Job [login to view URL] [login to view URL] [login to view URL]