rm(list=ls())
library(RPostgres)
library(data.table)
library(lfe)
library(stargazer)
library(dplyr)
library(ggplot2)
library(fst)
library(stringr)
library(stringi)
library(lubridate)
gc()## used (Mb) gc trigger (Mb) max used (Mb)
## Ncells 2303637 123.1 4717485 252 2879160 153.8
## Vcells 3946913 30.2 8388608 64 6553565 50.0regsample <- readRDS("C:/Users/dratnadiwakara2/Downloads/hmda_cleaned_all_mergers_10052023.rds")
gc()## used (Mb) gc trigger (Mb) max used (Mb)
## Ncells 2863703 153.0 4717485 252.0 2879160 153.8
## Vcells 7093428629 54118.6 10217129105 77950.6 7093438937 54118.7regsample[,refinance:=ifelse(purposeofloan==3,1,0)]
regsample[,race_eth:=ifelse(applicantrace1==3,"1. black","0. other")]
regsample[,race_eth:=ifelse(applicantethnicity==1,"2. hispanic",ifelse(is.na(applicantethnicity),NA,race_eth))]
regsample[,race:=ifelse(applicantrace1==3,"1. black","0. other")]
regsample[,bank_county:=paste(BHCRSSD,countycode)]
setorder(regsample,bank_county)
regsample <- regsample[asofdate>=2000]
gc()## used (Mb) gc trigger (Mb) max used (Mb)
## Ncells 2906285 155.3 4717485 252.0 2994690 160.0
## Vcells 6821762624 52046.0 14960971469 114143.2 14860395258 113375.9r <- list()
gc()## used (Mb) gc trigger (Mb) max used (Mb)
## Ncells 2906816 155.3 4717485 252.0 3523758 188.2
## Vcells 6821785993 52046.1 14960971469 114143.2 14938539536 113972.1r[[1]] <- felm(approved~msinc13+msinc46+suc_share+nationwide_msacq13+nationwide_msacq46+log(applicantincome)+log(amountofloan)+factor(race)+suc_share+refinance|bank_county+asofdate,data=regsample[typeofloan==1 ])
gc()## used (Mb) gc trigger (Mb) max used (Mb)
## Ncells 2949686 157.6 4717485 252.0 3523758 188.2
## Vcells 11606124276 88547.8 21543974914 164367.5 21510692611 164113.6stargazer(r,type="text")##
## ================================================
## Dependent variable:
## ---------------------------
## approved
## ------------------------------------------------
## msinc13 0.447***
## (0.006)
##
## msinc46 0.969***
## (0.007)
##
## suc_share -0.171***
## (0.002)
##
## nationwide_msacq13 -0.0005***
## (0.00001)
##
## nationwide_msacq46 0.108***
## (0.001)
##
## log(applicantincome) 0.074***
## (0.0001)
##
## log(amountofloan) 0.019***
## (0.0001)
##
## factor(race)1. black -0.102***
## (0.0001)
##
## refinance -0.100***
## (0.0001)
##
## ------------------------------------------------
## Observations 167,573,916
## R2 0.080
## Adjusted R2 0.080
## Residual Std. Error 0.451 (df = 167493809)
## ================================================
## Note: *p<0.1; **p<0.05; ***p<0.01r <- list()
gc()## used (Mb) gc trigger (Mb) max used (Mb)
## Ncells 3068525 163.9 4717485 252.0 4717485 252.0
## Vcells 6914157090 52750.9 21543974914 164367.5 21510692611 164113.6r[[1]] <- felm(approved~factor(msinc13Q)+factor(msinc46Q)+suc_share+nationwide_msacq13+nationwide_msacq46+log(applicantincome)+log(amountofloan)+factor(race)+suc_share+refinance|bank_county+asofdate,data=regsample[typeofloan==1 ])
gc()## used (Mb) gc trigger (Mb) max used (Mb)
## Ncells 3069185 164.0 4717485 252.0 4717485 252.0
## Vcells 12444177970 94941.6 21543974914 164367.5 21510890467 164115.1stargazer(r,type="text")##
## ================================================
## Dependent variable:
## ---------------------------
## approved
## ------------------------------------------------
## factor(msinc13Q)1 0.028***
## (0.0003)
##
## factor(msinc13Q)2 0.018***
## (0.0003)
##
## factor(msinc13Q)3 0.014***
## (0.0003)
##
## factor(msinc13Q)4 0.023***
## (0.0003)
##
## factor(msinc46Q)1 0.014***
## (0.0004)
##
## factor(msinc46Q)2 0.023***
## (0.0004)
##
## factor(msinc46Q)3 0.051***
## (0.0003)
##
## factor(msinc46Q)4 0.043***
## (0.0003)
##
## suc_share -0.125***
## (0.002)
##
## nationwide_msacq13 -0.001***
## (0.00001)
##
## nationwide_msacq46 0.100***
## (0.001)
##
## log(applicantincome) 0.074***
## (0.0001)
##
## log(amountofloan) 0.019***
## (0.0001)
##
## factor(race)1. black -0.102***
## (0.0001)
##
## refinance -0.100***
## (0.0001)
##
## ------------------------------------------------
## Observations 167,573,916
## R2 0.080
## Adjusted R2 0.080
## Residual Std. Error 0.451 (df = 167493803)
## ================================================
## Note: *p<0.1; **p<0.05; ***p<0.01r <- list()
gc()## used (Mb) gc trigger (Mb) max used (Mb)
## Ncells 3068558 163.9 4717485 252.0 4717485 252.0
## Vcells 6914157593 52750.9 21543974914 164367.5 21510890467 164115.1r[[1]] <- felm(approved~msinc13*factor(race_eth)+msinc46*factor(race_eth)+suc_share+nationwide_msacq13+nationwide_msacq46+log(applicantincome)+log(amountofloan)+suc_share+refinance|bank_county+asofdate,data=regsample[typeofloan==1 ])
gc()## used (Mb) gc trigger (Mb) max used (Mb)
## Ncells 3069246 164.0 4717485 252.0 4717485 252.0
## Vcells 10678751468 81472.5 21543974914 164367.5 21525095365 164223.5stargazer(r,type="text")##
## ===============================================================
## Dependent variable:
## ---------------------------
## approved
## ---------------------------------------------------------------
## msinc13 0.975***
## (0.008)
##
## factor(race_eth)1. black -0.118***
## (0.0002)
##
## factor(race_eth)2. hispanic -0.062***
## (0.0002)
##
## msinc46 1.121***
## (0.013)
##
## suc_share -0.068***
## (0.002)
##
## nationwide_msacq13 -0.0003***
## (0.00001)
##
## nationwide_msacq46 0.284***
## (0.003)
##
## log(applicantincome) 0.068***
## (0.0001)
##
## log(amountofloan) 0.003***
## (0.0001)
##
## refinance -0.110***
## (0.0001)
##
## msinc13:factor(race_eth)1. black -0.021
## (0.021)
##
## msinc13:factor(race_eth)2. hispanic -0.574***
## (0.022)
##
## factor(race_eth)1. black:msinc46 -0.040
## (0.039)
##
## factor(race_eth)2. hispanic:msinc46 -0.095**
## (0.038)
##
## ---------------------------------------------------------------
## Observations 112,404,872
## R2 0.074
## Adjusted R2 0.073
## Residual Std. Error 0.453 (df = 112342276)
## ===============================================================
## Note: *p<0.1; **p<0.05; ***p<0.01r <- list()
gc()## used (Mb) gc trigger (Mb) max used (Mb)
## Ncells 3068579 163.9 4717485 252.0 4717485 252.0
## Vcells 6914158149 52750.9 21543974914 164367.5 21525095365 164223.5r[[1]] <- felm(approved~msinc13*low_income+msinc46*low_income+suc_share+nationwide_msacq13+nationwide_msacq46+low_income+log(amountofloan)+factor(race)+suc_share+refinance|bank_county+asofdate,data=regsample[typeofloan==1 ])
gc()## used (Mb) gc trigger (Mb) max used (Mb)
## Ncells 3069235 164.0 4717485 252.0 4717485 252.0
## Vcells 11941456932 91106.1 21543974914 164367.5 21527274282 164240.1stargazer(r,type="text")##
## ================================================
## Dependent variable:
## ---------------------------
## approved
## ------------------------------------------------
## msinc13 0.533***
## (0.008)
##
## low_income -0.078***
## (0.0001)
##
## msinc46 1.031***
## (0.008)
##
## suc_share -0.183***
## (0.002)
##
## nationwide_msacq13 -0.0005***
## (0.00001)
##
## nationwide_msacq46 0.106***
## (0.001)
##
## log(amountofloan) 0.030***
## (0.0001)
##
## factor(race)1. black -0.106***
## (0.0001)
##
## refinance -0.102***
## (0.0001)
##
## msinc13:low_income -0.216***
## (0.011)
##
## low_income:msinc46 -0.154***
## (0.011)
##
## ------------------------------------------------
## Observations 167,573,916
## R2 0.078
## Adjusted R2 0.077
## Residual Std. Error 0.452 (df = 167493807)
## ================================================
## Note: *p<0.1; **p<0.05; ***p<0.01rm(r)
r <- list()
gc()## used (Mb) gc trigger (Mb) max used (Mb)
## Ncells 3068755 163.9 4717485 252.0 4717485 252.0
## Vcells 6914160002 52750.9 21543974914 164367.5 21527274282 164240.1r[[1]] <- felm(approved~msinc13+msinc46+suc_share+nationwide_msacq13+nationwide_msacq46+log(applicantincome)+log(amountofloan)+factor(race_eth)+suc_share+refinance|bank_county+asofdate,data=regsample[typeofloan==2 ])
gc()## used (Mb) gc trigger (Mb) max used (Mb)
## Ncells 3069385 164.0 4717485 252.0 4717485 252.0
## Vcells 7302123197 55710.8 21543974914 164367.5 21527274282 164240.1r[[2]] <- felm(approved~msinc13*factor(race_eth)+msinc46*factor(race_eth)+suc_share+nationwide_msacq13+nationwide_msacq46+log(applicantincome)+log(amountofloan)+suc_share+refinance|bank_county+asofdate,data=regsample[typeofloan==2 ])
gc()## used (Mb) gc trigger (Mb) max used (Mb)
## Ncells 3070054 164.0 4717485 252.0 4717485 252.0
## Vcells 7743116954 59075.3 21543974914 164367.5 21527274282 164240.1r[[3]] <- felm(approved~msinc13*low_income+msinc46*low_income+suc_share+nationwide_msacq13+nationwide_msacq46+log(amountofloan)+factor(race_eth)+suc_share+refinance|bank_county+asofdate,data=regsample[typeofloan==2 ])
# gc()
# r[[2]] <- felm(approved~msinc13+msinc46|bank_cbsa+asofdate,data=regsample[pred_share_1>0])
# gc()
# r[[3]] <- felm(approved~msinc13+msinc46+log(applicantincome)|bank_cbsa+asofdate+purposeofloan,data=regsample[pred_share_1>0 & purposeofloan==1 & asofdate>2010])
gc()## used (Mb) gc trigger (Mb) max used (Mb)
## Ncells 3070715 164.0 4717485 252.0 4717485 252.0
## Vcells 8157590204 62237.5 21543974914 164367.5 21527274282 164240.1stargazer(r,type="text")##
## =====================================================================================================
## Dependent variable:
## -----------------------------------------------------------------
## approved
## (1) (2) (3)
## -----------------------------------------------------------------------------------------------------
## msinc13 -0.652*** -0.552*** -0.250***
## (0.032) (0.036) (0.053)
##
## low_income -0.040***
## (0.0003)
##
## msinc46 -1.047*** -0.964*** -0.739***
## (0.050) (0.056) (0.082)
##
## suc_share -0.033*** -0.034*** -0.028***
## (0.006) (0.006) (0.006)
##
## nationwide_msacq13 0.002*** 0.002*** 0.002***
## (0.00003) (0.00003) (0.00003)
##
## nationwide_msacq46 -0.159*** -0.157*** -0.167***
## (0.014) (0.014) (0.014)
##
## log(applicantincome) 0.082*** 0.082***
## (0.0003) (0.0003)
##
## log(amountofloan) 0.005*** 0.005*** 0.042***
## (0.0003) (0.0003) (0.0003)
##
## factor(race_eth)1. black -0.067*** -0.067*** -0.070***
## (0.0004) (0.0004) (0.0004)
##
## factor(race_eth)2. hispanic -0.022*** -0.021*** -0.029***
## (0.0004) (0.0004) (0.0004)
##
## refinance -0.275*** -0.275*** -0.275***
## (0.0003) (0.0003) (0.0003)
##
## msinc13:factor(race_eth)1. black 0.035
## (0.071)
##
## msinc13:factor(race_eth)2. hispanic -1.147***
## (0.096)
##
## factor(race_eth)1. black:msinc46 0.099
## (0.120)
##
## factor(race_eth)2. hispanic:msinc46 -0.980***
## (0.148)
##
## msinc13:low_income -0.568***
## (0.060)
##
## low_income:msinc46 -0.429***
## (0.092)
##
## -----------------------------------------------------------------------------------------------------
## Observations 13,260,180 13,260,180 13,260,180
## R2 0.123 0.123 0.119
## Adjusted R2 0.121 0.121 0.116
## Residual Std. Error 0.428 (df = 13223661) 0.428 (df = 13223657) 0.429 (df = 13223659)
## =====================================================================================================
## Note: *p<0.1; **p<0.05; ***p<0.01