Economics Use the data in "fertil2" for this exercise. (i) Estimate the model children = Bo + Bage + Bage + Bseduc + BAelectric + Bsurban + Bespirit + Byprotest + Bgcatholic + u and report both the usual and robust standard errors. Compare the standard errors and specifically, are the robust standard errors always larger? (ii) We brushed over a subtle point in the previous part. The dummy variables spirit, protest, catholic are three religion dummy variables that are collected in the sample. Explain why we can include all three in this model without violating any assumptions of MLR. Hint: Think of what the base group is here. (iii) Conduct the White test for heteroskedasticity and conclude that there is heteroskedasticity in this model. (iv) Estimate the model by FGLS and compare the standard errors from this estimation against the usual and robust standard errors from part (i). Does the choice of standard errors change your findings with regards to significance of any of the variables? (v) Estimate the model with a modified form of FGLS where weights are constructed through a regression of the squared residuals on the OLS fitted values and the squared fitted values. Compare your findings. (vi) Are we concerned about the drawbacks of FGLS impacting estimates? Why not