Overview
Here is the website where you will find the data:
https://www.consumerfinance.gov/data-research/hmda/historic-data/
Select all records for "Records included" and plain language labels and HDMA codes for "variable description". The codebook for the datasets are given on the right of the website under the "Data Dictionaries" section.
The question of interest in this project is to see if there is racial discrimination in home mortgage lending. The project is a little bit more challenging than the first one. You will need to download, clean, and write a regression model yourself and therefore, documenting your code in a .do file is very important.
As before, the paper should 5-6 pages. Include an introduction, brief literature review, econometric specification/model, data description, a regression table (create columns including/excluding various fixed effects), a summary statistics table, explanation of regression results, and a conclusion.
Submit your .do file and log files with the project.
use state of Connecticut , years: 2016 and 2017 .*For 2016:
import excel
tab denial_reason_name_1
tab denial_reason_name_1 denial_reason_1
tab denial_reason_name_1, gen(den_r)
tab action_taken_name
gen approved=0
replace approved=1 if action_taken==1
replace approved=. if (action_taken==2| action_taken==4|action_taken==5| action_taken ==6| action_taken==7)
tab applicant_race_name_1
gen race_aa=0
replace race_aa=1 if applicant_race_1==3
replace race_aa=. if applicant_race_1==7
reg approved race_aa
reg approved race_aa i.county_code , r cluster( county_code)
predict pr, xb
histogram pr
reg approved race_aa i.county_code i.applicant_sex applicant_income_000s , r cluster( county_code)
sum race_aa
probit approved race_aa hud_median_family_income
probit approved i.race_aa hud_median_family_income
logit approved i.race_aa hud_median_family_income
logit approved i.race_aa hud_median_family_income, r cluster(county_code)
summarize approved race_aa county_code hud_median_family_income
*For 2017:
tab denial_reason_name_1
tab denial_reason_name_1 denial_reason_1
tab denial_reason_name_1, gen(den_r)
tab action_taken_name
gen approved=0
replace approved=1 if action_taken==1
replace approved=. if (action_taken==2| action_taken==4|action_taken==5| action_taken ==6| action_taken==7)
tab applicant_race_name_1
gen race_aa=0
replace race_aa=1 if applicant_race_1==3
replace race_aa=. if applicant_race_1==7
reg approved race_aa
reg approved race_aa i.county_code , r cluster( county_code)
predict pr, xb
histogram pr
reg approved race_aa i.county_code i.applicant_sex applicant_income_000s , r cluster( county_code)
sum race_aa
probit approved race_aa hud_median_family_income
probit approved i.race_aa hud_median_family_income
logit approved i.race_aa hud_median_family_income
logit approved i.race_aa hud_median_family_income, r cluster(county_code)
summarize approved race_aa county_code hud_median_family_income