3. Use the data in LOANAPP.dta for this question. The binary variable to be explained is approved , which is equal to one if mortgage loan to an individual was approved. The dataset includes mortgage applicants who were asked to self-identify as either white ( i ), black ( i ), or Hispanic ( hispan ). To test for racial discrimination against applicants of color in the mortgage loan market, you regress approved on the dummy for & , plus other factors: approved;= Bo+ B, white,+ other factors+e; a. If there is discrimination against applicants of color in loan decisions, and all appropriate factors have been controlled for, what would you expect the sign of , to be? b. Regress approve on & , in a simple bivariate regression. Interpret the coefficient on i . Is it statistically significant? Is it meaningfully large? c. Now control for a rich set of individual level characteristic, whose omission from the regression might bias the regression in b. Specifically, add the variables: -housing expense (as fraction of income) hrat -other debt obligations (as a fraction of income) obrat , -loan amount as a fraction of the home price loanprc , -local unemployment rate unem , - female - married -number of dependents dep , -whether applicant has more than 12 years of education sch , -whether there's a cosigner cosign , -whether applicant has any delinquent accounts chist , -whether application has ever filed for bankruptcy pubrec , -whether applicant has 1-2 late payments mortlat 1 , -whether applicant has more than 2 late payments mortlat 2 , and -if the local area vacancy rate exceeds the city's median vr Is there still evidence of discrimination in lending decisions? And what do you infer from the change in the estimated coefficient on