rm(list=ls())
library(data.table)
library(fst)
library(RSQLite)
library(DBI)
library(dplyr)
library(lfe)
library(stargazer)
library(stringr)
library(ggplot2)
library(readxl)
library(stringi)
library(zoo)
source('C:/Users/dratnadiwakara2/Documents/OneDrive - Louisiana State University/functions.R')
fit_felm <- function(formula, data, sample_frac,drop_dv_outliers=TRUE) {
dep_var_name <- as.character(formula[[2]])
if(drop_dv_outliers==TRUE) {
quantiles <- quantile(data[[dep_var_name]], c(0.01, 0.99), na.rm = TRUE)
subset_data <- data[data[[dep_var_name]] > quantiles[1] & data[[dep_var_name]] < quantiles[2], ]
} else {
subset_data = data
}
subset_data <- subset_data[sample(nrow(subset_data), size = nrow(subset_data)*sample_frac), ]
felm(formula, data = subset_data)
}
hmda_link_entity <- fread("C:/Users/dratnadiwakara2/Documents/OneDrive - Louisiana State University/Raw Data/HMDA/HMDA lender/hmdapan2017.csv",select=c("code","hmprid",paste0("RSSD",12:17)))#,paste0("ENTITY",15:16)
hmda_link_entity[,hmda_id:=paste0(code,"-",hmprid)]
hmda_link_entity[,c("code","hmprid"):=list(NULL)]
hmda_link_entity <- melt(hmda_link_entity,id.vars = c("hmda_id"))
hmda_link_entity <- data.table(hmda_link_entity)
hmda_link_entity <- hmda_link_entity[!is.na(value)]
hmda_link_entity[,year:=as.character(variable)]
hmda_link_entity[,year:=substr(year,5,6)]
setnames(hmda_link_entity,"value","RSSD")
hmda_link_entity[,variable:=NULL]
hmda_link_entity[,year:=as.numeric(year)]
hmda_link_entity[,year:=ifelse(year>85,1900+year,2000+year)]
hmda_link_2021 <- fread("C:/Users/dratnadiwakara2/Documents/OneDrive - Louisiana State University/Raw Data/HMDA/HMDA lender/hmdapan2021.csv",select=c("LEI","RSSD21","CODE21","HMPRID"))
hmda_link_2021[,hmda_id:=paste0(CODE21,"-",HMPRID)]
hmda_link_2021 <- hmda_link_2021[,c("LEI","RSSD21","hmda_id")]
gse_limit_files <- list.files("C:/Users/dratnadiwakara2/Documents/OneDrive - Louisiana State University/Raw Data/GSE Limits",full.names = T,pattern = ".xls")
gse_limits <- list()
i=1
for(fl in gse_limit_files) {
yr <- substr(fl,117,120)
if(yr<=2014) {
temp <- read_xls(fl,sheet = 1,skip=2,col_names = F)
} else{
temp <- read_xlsx(fl,sheet = 1,skip=2,col_names = F)
}
names(temp) <- c("statefips","countyfips","countyname","cbsa","statecode","gse_limit","gse_limit_2","gse_limit_3","gse_limit_4")
temp <- data.table(temp)
temp[,year:=yr]
gse_limits[[i]] <- temp
i=i+1
}
gse_limits <- rbindlist(gse_limits)
gse_limits[,gse_limit:=floor(gse_limit/1000)]
gse_limits[,year:=as.numeric(year)]
gse_limits[,county:=paste0(statefips,countyfips)]
gse_limits <- gse_limits[,c("county","year","gse_limit")]
setorder(gse_limits,county,year)
gse_limits[,gse_limit_1:=lag(gse_limit),by=county]
gse_limits[,gse_limit_2:=lead(gse_limit),by=county]
gse_limits[,limit_change:=gse_limit-gse_limit_1]
gse_limits_yr <- gse_limits[,.(gse_limit=median(gse_limit,na.rm=T),limit_change=median(limit_change,na.rm=T)),by=year]
setnames(gse_limits_yr,"gse_limit","gse_limit_yr")
ggplot(gse_limits_yr, aes(x=year, y=gse_limit_yr)) +
geom_line(color="dodgerblue",size=2) +
geom_point(size=3,color="dodgerblue4")+
scale_x_continuous(breaks = unique(gse_limits_yr$year))+
labs(x="",y="GSE limit ($'000)",title = "GSE Limit History")+
theme_minimal()
ggplot(gse_limits_yr, aes(x=year, y=limit_change)) +
geom_col(filll="dodgerblue",size=2) +
# geom_point(size=3,color="dodgerblue4")+
scale_x_continuous(breaks = unique(gse_limits_yr$year))+
labs(x="",y="GSE limit Change ($'000)",title = "GSE Limit History")+
theme_minimal()
hmda_con <- dbConnect(RSQLite::SQLite(), "C:/Users/dratnadiwakara2/Downloads/HMDA/hmda.db")
yrs <- as.character(2012:2017)
hmda <- list()
i=1
for(yr in yrs) {
print(yr)
hmda[[i]] <- data.table(dbGetQuery(hmda_con,
paste0("select
agencycode,
respondentid,
asofdate,
purposeofloan,
typeofpurchaser,
amountofloan,
censustract,
applicantincome,
typeofloan,
actiontaken,
denialreason1,
denialreason2,
denialreason3
from lar_",yr,"
where
actiontaken in (1,3,6) and
propertytype==1")))
i=i+1
}
## [1] "2012"
## [1] "2013"
## [1] "2014"
## [1] "2015"
## [1] "2016"
## [1] "2017"
hmda <- rbindlist(hmda)
hmda <- data.table(hmda)
hmda[,county:=substr(censustract,1,5)]
hmda <- merge(hmda,gse_limits,by.x=c("asofdate","county"),by.y=c("year","county"),all.x=T)
hmda <- merge(hmda,gse_limits_yr,by.x="asofdate",by.y="year")
hmda[,gse_limit:=ifelse(is.na(gse_limit),gse_limit_yr,gse_limit)]
hmda[,amountofloan:=as.numeric(amountofloan)]
hmda[,dist_gse_limit:=amountofloan-gse_limit]
hmda[,jumbo_this_year:=ifelse(amountofloan>gse_limit,1,0)]
hmda[,jumbo_last_year:=ifelse(amountofloan>gse_limit_1,1,0)]
hmda[,jumbo_next_year:=ifelse(amountofloan>gse_limit_2,1,0)]
hmda[,hmda_id:=paste0(agencycode,"-",respondentid)]
hmda <- merge(hmda,hmda_link_entity,by.x=c("asofdate","hmda_id"),by.y=c("year","hmda_id"))
hmda <- hmda[RSSD>0 & !is.na(RSSD)]
gc()
## used (Mb) gc trigger (Mb) max used (Mb)
## Ncells 2594406 138.6 4806374 256.7 4806374 256.7
## Vcells 1891748216 14432.9 4728287392 36074.0 4725499797 36052.8
yrs <- 2018:2022
hmda_post <- list()
i=1
for(yr in yrs) {
print(yr)
hmda_post[[i]] <- data.table(dbGetQuery(hmda_con,
paste0("select
lei,
asofdate,
purposeofloan,
typeofpurchaser,
amountofloan,
censustract,
actiontaken,
conforming_loan_limit,
combined_loan_to_value_ratio,
interest_rate,
rate_spread,
total_loan_costs,
debt_to_income_ratio,
typeofloan,
applicant_age,
applicantincome,
denialreason1,
denialreason2,
denialreason3
from lar_",yr,"
where
actiontaken in (1,3,6) and
derived_dwelling_category='Single Family (1-4 Units):Site-Built'")))
i=i+1
}
## [1] 2018
## [1] 2019
## [1] 2020
## [1] 2021
## [1] 2022
hmda_post <- rbindlist(hmda_post)
hmda_post <- data.table(hmda_post)
hmda_post[,county:=substr(censustract,1,5)]
hmda_post <- merge(hmda_post,gse_limits,by.x=c("asofdate","county"),by.y=c("year","county"),all.x=T)
hmda_post <- merge(hmda_post,gse_limits_yr,by.x="asofdate",by.y="year")
hmda_post[,gse_limit:=ifelse(is.na(gse_limit),gse_limit_yr,gse_limit)]
hmda_post[,amountofloan:=as.numeric(amountofloan)]
hmda_post[,dist_gse_limit:=amountofloan-gse_limit]
hmda_post[,jumbo_this_year:=ifelse(amountofloan>gse_limit,1,0)]
hmda_post[,jumbo_last_year:=ifelse(amountofloan>gse_limit_1,1,0)]
hmda_post[,jumbo_next_year:=ifelse(amountofloan>gse_limit_2,1,0)]
hmda_post <- merge(hmda_post,hmda_link_2021,by.x=c("lei"),by.y=c("LEI"))
hmda_post <- hmda_post[RSSD21>0 & !is.na(RSSD21)]
hmda_post[,purposeofloan:=ifelse(purposeofloan>10,3,purposeofloan)]
setnames(hmda_post,"RSSD21","RSSD")
gc()
## used (Mb) gc trigger (Mb) max used (Mb)
## Ncells 2837372 151.6 4806374 256.7 4806374 256.7
## Vcells 3852413835 29391.6 6808909844 51947.9 6785332717 51768.0
# hmda_1_summary <- hmda[actiontaken==1,.(no_loans=.N,total_loan_amt=sum(amountofloan)),
# by=.(RSSD,asofdate)]
#
# hmda_2_summary <- hmda_post[actiontaken==1,.(no_loans=.N,total_loan_amt=sum(amountofloan)),
# by=.(RSSD,asofdate)]
jumbo_100k_by_lender <- hmda[ actiontaken==1 & typeofloan==1 &
dist_gse_limit>10 &
asofdate %in% 2012:2016,
.(no_of_jumbo_loans=.N,
total_jumbo_loan_amt=sum(amountofloan)),
by=RSSD]
all_by_lender <- hmda[asofdate %in% 2012:2016 & actiontaken==1 & typeofloan==1 ,
.(all_no_of_loans=.N,
all_total_loan_amt=sum(amountofloan)),
by=RSSD]
jumbo_100k_by_lender <- merge(jumbo_100k_by_lender,all_by_lender,by="RSSD")
jumbo_100k_by_lender[,no_of_jumbo_loans_pct:=no_of_jumbo_loans*100/all_no_of_loans]
jumbo_100k_by_lender[,total_jumbo_loan_amt_pct:=total_jumbo_loan_amt*100/all_total_loan_amt]
HHI <- hmda[ actiontaken==1 & typeofloan==1 &
asofdate %in% 2012:2016,.(amountofloan=sum(amountofloan,na.rm=T)),
by=.(RSSD,county)]
HHI[, total_loan_by_rssd_asofdate := sum(amountofloan), by = .(RSSD)]
HHI[, market_share := amountofloan / total_loan_by_rssd_asofdate]
HHI[, squared_market_share := market_share^2, by = .(RSSD, county)]
HHI <- HHI[market_share>0, .(HHI = sum(squared_market_share),.N), by = .(RSSD)]
jumbo_100k_by_lender <- merge(jumbo_100k_by_lender,HHI,by="RSSD",all.x=T)
con_call <- dbConnect(RSQLite::SQLite(), "C:/Users/dratnadiwakara2/Documents/OneDrive - Louisiana State University/Raw Data/Call Reports/call_reports.db")
con_ubpr <- dbConnect(RSQLite::SQLite(), "C:/Users/dratnadiwakara2/Documents/OneDrive - Louisiana State University/Raw Data/Call Reports/ubpr.db")
data_periods <- apply(expand.grid(c("0331","0630","0930","1231"), 2012:2023), 1, paste, collapse="")
# data_periods <- c(data_periods,"03312023")
# data_periods <- apply(expand.grid(c("1231"), 2022), 1, paste, collapse="")
upbr <- list()
i=1
for (dp in data_periods) {
ubpr_dp <- as.Date(dp,"%m%d%Y")
ubpr_yr <- year(ubpr_dp)
ubpr_data <- dbGetQuery(con_ubpr,paste0("select
UBPR1754 htm_cost,
UBPR1771 htm_value,
UBPRE566 htm_assets_ubpr,
UBPRE630 roe,
UBPRE565 afs_assets_upbr,
UBPRKX40 nii_assets,
UBPRE023 nii_earnings_assets,
UBPRE002 interest_expense_assets_1,
UBPRD087 asset_growth,
UBPRE162 deposit_growth_qt,
UBPRE209 deposit_growth_yr,
UBPR1410 real_estate_loans,
UBPR1766 commercial_industrial_loans,
UBPR2122 total_loans,
UBPR2170 total_assets,
UBPR2746 cre_construction_land_dev_loans,
UBPRD214 re_loans_sfr,
UBPRD220 sfr_first_lein_ma,
UBPRE006 provisions_to_assets,
UBPRE397 re_loans_net_loss,
UBPRE401 sfr_net_loss,
UBPRE408 commercial_industrial_net_loss,
UBPRE410 loans_to_individuals_net_loss,
UBPR3210 total_equity,
UBPR0081+UBPR0071 cash,
UBPRD588 total_securities,
UBPRB559 mortgage_backed_securities,
UBPRB558 govt_agency_securities,
UBPR2200 total_deposits,
UBPRE660 risk_weighted_assets,
UBPRE678 interest_income_assets,
UBPRE679 interest_expense_assets_2,
UBPR7414 delinquent_to_loans,
UBPR7400 personnel_expense_to_assets,
UBPRE090 assets_per_employee,
UBPRD667+UBPRD669 pastdue_nonaccrual_loans,
UBPRK437+UBPRK426 time_deposits,
UBPRA549+UBPRA555+UBPRA564+UBPRA570 assets_maturity_less_than_3_months,
UBPRA550+UBPRA556+UBPRA248+UBPRA565+UBPRA571+UBPRA247 assets_maturity_3_mn_1yr,
UBPRA551+UBPRA557+UBPRA561+UBPRA566+UBPRA572 assets_maturity_1_to_3_yr,
UBPRA552+UBPRA558+UBPRA562+UBPRA567+UBPRA573 assets_maturity_3_to_5_yr,
UBPRA553+UBPRA559+UBPRA568+UBPRA574 assets_maturity_5_to_15_yr,
UBPRA554+UBPRA560+UBPRA569+UBPRA575 assets_maturity_more_than_15_yr,
UBPRA549+UBPRA555 security_maturity_less_than_3_months,
UBPRA550+UBPRA556 security_maturity_3_mn_1yr,
UBPRA551+UBPRA557 security_maturity_1_to_3_yr,
UBPRA552+UBPRA558 security_maturity_3_to_5_yr,
UBPRA553+UBPRA559 security_maturity_5_to_15_yr,
UBPRA554+UBPRA560 security_maturity_more_than_15_yr,
UBPRA564 re_loans_maturity_less_than_3_months,
UBPRA570 other_loans_maturity_less_than_3_months,
UBPRA565 re_loans_maturity_3_mn_1yr,
UBPRA571 other_loans_maturity_3_mn_1yr,
UBPRA566 re_loans_maturity_1_to_3_yr,
UBPRA572 other_loans_maturity_1_to_3_yr,
UBPRA567 re_loans_maturity_3_to_5_yr,
UBPRA573 other_loans_maturity_3_to_5_yr,
UBPRA568 re_loans_maturity_5_to_15_yr,
UBPRA574 other_loans_maturity_5_to_15_yr,
UBPRA569 re_loans_maturity_more_than_15_yr,
UBPRA575 other_loans_maturity_more_than_15_yr,
UBPRD535 deposits_over_1_yr,
UBPRD536 deposits_over_3_yr,
UBPRE485+UBPRE486 mtg_delinq_pct,
UBPRE487+UBPRE488 mtg_delinq_pct_2,
UBPRD488 total_capital_ratio,
UBPR3833 unused_commitments_total_gt1yr,
UBPRE262 unused_commitments_assets_home_equity,
UBPRE263 unused_commitments_assets_credit_card,
UBPRE264+UBPRE265+UBPRE225 unused_commitments_assets_cre,
UBPRE218 unused_commitments_assets_sfr_construction,
UBPRE266 unused_commitments_assets_all_other,
UBPRE262+UBPRE263+UBPRE264+UBPRE265+UBPRE225+UBPRE218+UBPRE266 unused_commitments_assets_total,
IDRSSD
from ubpr_",ubpr_yr,"
where
data_period=",as.numeric(ubpr_dp),""))
call_1 <- dbGetQuery(con_call,paste0("select
RCON2200 deposits_domestic,
RCONF049 deposits_domestic_insured,
RCONF051 deposits_domestic_uninsured,
RCONF050 deposits_domestic_no_insured,
RCONF052 deposits_domestic_no_uninsured,
RCON6631 deposits_domestic_non_interest,
RCON2210 demand_deposits,
RCON1797+RCON5367+RCON5368+RCON1460+RCONF158 secured_by_residential_real_estate,
RCONF160+RCONF161+RCONF159 secured_by_nonresidential_real_estate,
RCONB538+RCONB539+RCONK137+RCONK207 consumer_loans,
RCON0010 cash_due_from_depositories,
RCON0090 balance_due_from_fed,
IDRSSD
from call_1_",dp," "))
call_2 <- dbGetQuery(con_call,paste0("select
FDIC_Certificate_Number,
Financial_Institution_Name,
Financial_Institution_State,
data_period,
RIADC017 data_processing_expenses,
RIAD4135 salaries_employee_benefits,
RIAD4150 no_of_full_time_employees,
IDRSSD
from call_2_",dp," "))
ubpr_data <- merge(ubpr_data,call_2,by="IDRSSD")
ubpr_data <- merge(ubpr_data,call_1,by="IDRSSD")
ubpr_data <- data.table(ubpr_data)
upbr[[i]] <- ubpr_data
i=i+1
}
upbr <- rbindlist(upbr,fill=T)
dbDisconnect(con_call)
dbDisconnect(con_ubpr)
upbr[,wa_maturiy_loans:=(re_loans_maturity_less_than_3_months*0.25+re_loans_maturity_3_mn_1yr*0.75+re_loans_maturity_1_to_3_yr*2+re_loans_maturity_3_to_5_yr*4+re_loans_maturity_5_to_15_yr*10+re_loans_maturity_more_than_15_yr*20 + other_loans_maturity_less_than_3_months*0.25+other_loans_maturity_3_mn_1yr*0.75+other_loans_maturity_1_to_3_yr*2+other_loans_maturity_3_to_5_yr*4+other_loans_maturity_5_to_15_yr*10+other_loans_maturity_more_than_15_yr*20)/(re_loans_maturity_less_than_3_months+re_loans_maturity_3_mn_1yr+re_loans_maturity_1_to_3_yr+re_loans_maturity_3_to_5_yr+re_loans_maturity_5_to_15_yr+re_loans_maturity_more_than_15_yr+other_loans_maturity_less_than_3_months+other_loans_maturity_3_mn_1yr+other_loans_maturity_1_to_3_yr+other_loans_maturity_3_to_5_yr+other_loans_maturity_5_to_15_yr+other_loans_maturity_more_than_15_yr)]
upbr[,wa_maturiy_assets:=(assets_maturity_less_than_3_months*0.25+assets_maturity_3_mn_1yr*0.75+assets_maturity_1_to_3_yr*2+assets_maturity_3_to_5_yr*4+assets_maturity_5_to_15_yr*10+assets_maturity_more_than_15_yr*20)/(assets_maturity_less_than_3_months+assets_maturity_3_mn_1yr+assets_maturity_1_to_3_yr+assets_maturity_3_to_5_yr+assets_maturity_5_to_15_yr+assets_maturity_more_than_15_yr)]
upbr[,wa_maturiy_security:=(security_maturity_less_than_3_months*0.25+security_maturity_3_mn_1yr*0.75+security_maturity_1_to_3_yr*2+security_maturity_3_to_5_yr*4+security_maturity_5_to_15_yr*10+security_maturity_more_than_15_yr*20)/(security_maturity_less_than_3_months+security_maturity_3_mn_1yr+security_maturity_1_to_3_yr+security_maturity_3_to_5_yr+security_maturity_5_to_15_yr+security_maturity_more_than_15_yr)]
# upbr[,interest_deposits:=total_assets*interest_expense_assets/total_deposits]
upbr[,time_deposit_frac:=time_deposits/total_deposits]
upbr[,deposits_over_1_yr_frac:=deposits_over_1_yr/total_deposits]
upbr[,deposits_over_3_yr_frac:=deposits_over_3_yr/total_deposits]
upbr[,pastdue_nonaccrual_loans_to_loans:=pastdue_nonaccrual_loans*100/total_loans]
upbr[,re_loans_sfr_assets:=re_loans_sfr/total_assets]
upbr[,data_processing_expenses_nii:=((data_processing_expenses*100)/total_assets)/nii_assets]
upbr[,ci_assets:=commercial_industrial_loans/total_assets]
upbr[,uninsured_deposits_assets:=deposits_domestic_uninsured/total_assets]
upbr_2016 <- upbr[data_period=="12312016"]
upbr_2016 <- merge(upbr_2016,jumbo_100k_by_lender,by.x="IDRSSD",by.y="RSSD")
upbr_2016[,re_loan_frac_assets:=sfr_first_lein_ma*100/total_assets]
upbr_2016[,jumbo_frac_assets:=total_jumbo_loan_amt*100/(total_assets)]
upbr_2016[,re_loan_dependance:=ntile(re_loan_frac_assets,100)]
upbr_2016[,jumbo_dependance:=ntile(jumbo_frac_assets,4)]
upbr_2016[,jumbo_dependance:=ifelse(jumbo_frac_assets>15 & jumbo_frac_assets<25,5,
ifelse(jumbo_frac_assets>=25,6,jumbo_dependance))]
upbr_2016[,jumbo_7590:=ntile(jumbo_frac_assets,100)]
upbr_2016[,jumbo_7590:=ifelse(jumbo_7590<75,1,ifelse(jumbo_7590<90,2,3))]
upbr_2016[,loans_assets:=total_loans/total_assets]
upbr_2016[,deposits_assets:=total_deposits/total_assets]
upbr_2016[,securities_assets:=total_securities/total_assets]
ggplot(upbr_2016[jumbo_frac_assets<50],aes(x=jumbo_frac_assets,fill=factor(jumbo_7590)))+geom_histogram(alpha=0.5,bins=60)
high_jumbo_cutoff <- 4:6
library(DescTools)
round_values <- function(x) {
ifelse(x > 100, round(x, 0), ifelse(x > 10, round(x, 1), round(x, 2)))
}
upbr_2016[!is.na(jumbo_7590) ,.N,by=jumbo_7590]
## jumbo_7590 N
## 1: 1 2684
## 2: 2 540
## 3: 3 396
columns_to_include <- c('total_assets','loans_assets','deposits_assets','re_loan_frac_assets','jumbo_frac_assets','securities_assets','nii_assets','interest_expense_assets_1',"HHI","N") #,'deposits_state_hhi','deposits_college_hhi','deposits_income_hhi'
# result_table <- bank_wa_mean_data[!is.na(bank_size), lapply(.SD, function(x) list(Mean = mean(x, na.rm = TRUE))), by = bank_size, .SDcols = columns_to_include]
result_table <- upbr_2016[!is.na(jumbo_7590) , lapply(.SD, function(x) {
x_winsorized <- Winsorize(x, probs = c(0.05, 0.95), na.rm = TRUE)
list(Winsorized_Mean = mean(x_winsorized, na.rm = TRUE))
}), by = jumbo_7590, .SDcols = columns_to_include]
setorder(result_table,jumbo_7590)
transposed_result_table <- t(result_table)
colnames(transposed_result_table) <- unlist(transposed_result_table[1,])
transposed_result_table <- transposed_result_table[-1,]
transposed_result_table <- cbind(rownames(transposed_result_table), transposed_result_table)
transposed_result_table <- as.data.table(transposed_result_table)
transposed_result_table <- data.frame(lapply(transposed_result_table, function(x) unlist(x, use.names = FALSE)))
transposed_result_table <- data.table(transposed_result_table)
numeric_cols <- names(transposed_result_table)[sapply(transposed_result_table, is.numeric)]
transposed_result_table[, (numeric_cols) := lapply(.SD, round_values), .SDcols = numeric_cols]
# print(transposed_result_table)
stargazer(transposed_result_table,summary=F,type="text")
##
## ======================================================
## V1 X1 X2 X3
## ------------------------------------------------------
## 1 total_assets 743,364 903,420 1,061,157
## 2 loans_assets 0.660 0.740 0.770
## 3 deposits_assets 0.840 0.830 0.820
## 4 re_loan_frac_assets 15.400 21.400 28.600
## 5 jumbo_frac_assets 2.980 9.370 23.600
## 6 securities_assets 0.200 0.130 0.110
## 7 nii_assets 3.380 3.450 3.340
## 8 interest_expense_assets_1 0.390 0.460 0.510
## 9 HHI 0.350 0.340 0.380
## 10 N 32.900 49.300 74.300
## ------------------------------------------------------
hmda[,jumbo:=ifelse(dist_gse_limit>1,1,ifelse(dist_gse_limit< -1,0,NA))]
hmda_post[,jumbo:=ifelse(dist_gse_limit>1,1,ifelse(dist_gse_limit< -1,0,NA))]
hmda_1_summary <- hmda[actiontaken==1 & purposeofloan %in% c(1,3) & typeofloan==1 ,.(no_loans=.N,total_loan_amt=sum(amountofloan)),
by=.(RSSD,asofdate,jumbo)]
hmda_2_summary <- hmda_post[actiontaken==1 & purposeofloan %in% c(1,3) & typeofloan==1 ,.(no_loans=.N,total_loan_amt=sum(amountofloan)),
by=.(RSSD,asofdate,jumbo)]
hmda_summary <- rbind(hmda_1_summary,hmda_2_summary)
hmda_summary <- merge(hmda_summary,upbr_2016[,c("IDRSSD","jumbo_dependance","re_loan_frac_assets","jumbo_frac_assets","total_assets","jumbo_7590")],
by.x="RSSD",by.y="IDRSSD",all.x=T)
hmda_summary[,new_loans_assets:=total_loan_amt/total_assets]
hmda_summary[, normalized_amount := total_loan_amt / total_loan_amt[asofdate == 2016], by = .(RSSD,jumbo)]
hmda_summary[,log_total_loan_amt:=log(1+total_loan_amt)]
hmda_summary[,high_jumbo_assets:=ifelse(jumbo_dependance %in% high_jumbo_cutoff,1,0)]
hmda_summary[,high_jumbo_assets2:=ifelse(jumbo_dependance %in% 5:6,1,0)]
hmda_summary_plot <- hmda_summary[asofdate<2022]
# hmda_summary_plot[,high_jumbo_assets:=ifelse(jumbo_dependance %in% high_jumbo_cutoff,1,0)]
hmda_summary_plot <- hmda_summary_plot[,.(normalized_amount=mean(normalized_amount,na.rm=T)),by=.(asofdate,jumbo_7590,jumbo)]
g1 <- ggplot(hmda_summary_plot[jumbo==1 & !is.na(jumbo_7590)],
aes(x=asofdate,y=normalized_amount,color=factor(jumbo_7590)))+
geom_line()+
theme(legend.position="bottom")+ggtitle("Jumbo")
g2 <- ggplot(hmda_summary_plot[jumbo==0 & !is.na(jumbo_7590)],
aes(x=asofdate,y=normalized_amount,color=factor(jumbo_7590)))+
geom_line()+
theme(legend.position="bottom")+ggtitle("Non-Jumbo")
library(gridExtra)
grid.arrange(g1,g2,nrow=1)
r <- list()
r[[1]] <- felm(log_total_loan_amt~log(jumbo_frac_assets+0.00001)*I(asofdate>2017)|RSSD+asofdate|0|RSSD,data=hmda_summary[asofdate %in% 2012:2021 & jumbo==1 & total_assets<010e7])
r[[2]] <- felm(log_total_loan_amt~factor(jumbo_7590)*I(asofdate>2017)|RSSD+asofdate|0|RSSD,data=hmda_summary[asofdate %in% 2012:2021 & jumbo==1 & total_assets<010e7])
r[[3]] <- felm(log_total_loan_amt~log(jumbo_frac_assets+0.00001)*I(asofdate>2017)|RSSD+asofdate|0|RSSD,data=hmda_summary[asofdate %in% 2012:2021 & jumbo==0 & total_assets<10e7])
r[[4]] <- felm(log_total_loan_amt~factor(jumbo_7590)*I(asofdate>2017)|RSSD+asofdate|0|RSSD,data=hmda_summary[asofdate %in% 2012:2021 & jumbo==0 & total_assets<10e7])
stargazer(r,type="text",no.space = T,column.labels = c("New Purchases","Refinacnes"),column.separate = c(2,2),omit.stat = "ser")
##
## =========================================================================================
## Dependent variable:
## ---------------------------------------
## log_total_loan_amt
## New Purchases Refinacnes
## (1) (2) (3) (4)
## -----------------------------------------------------------------------------------------
## log(jumbo_frac_assets + 1e-05)
## (0.000) (0.000)
## factor(jumbo_7590)2
## (0.000) (0.000)
## factor(jumbo_7590)3
## (0.000) (0.000)
## I(asofdate > 2017)
## (0.000) (0.000) (0.000) (0.000)
## log(jumbo_frac_assets + 1e-05):I(asofdate > 2017) -0.149*** -0.061***
## (0.019) (0.018)
## factor(jumbo_7590)2:I(asofdate > 2017) -0.230*** -0.072*
## (0.055) (0.043)
## factor(jumbo_7590)3:I(asofdate > 2017) -0.266*** -0.233***
## (0.061) (0.056)
## -----------------------------------------------------------------------------------------
## Observations 23,452 23,452 27,139 27,139
## R2 0.835 0.835 0.907 0.907
## Adjusted R2 0.805 0.805 0.893 0.893
## =========================================================================================
## Note: *p<0.1; **p<0.05; ***p<0.01
r <- list()
r[[1]] <- felm(log_total_loan_amt~high_jumbo_assets*factor(asofdate)|RSSD+asofdate|0|RSSD,data=hmda_summary[jumbo==1])
r[[2]] <- felm(log_total_loan_amt~high_jumbo_assets*factor(asofdate)|RSSD+asofdate|0|RSSD,data=hmda_summary[jumbo==0])
g1 <- coef_plot_1reg(r[[1]],"high_jumbo_assets:factor(asofdate)",2012)+ggtitle("log(Total loan volume), Jumbo")
g2 <- coef_plot_1reg(r[[2]],"high_jumbo_assets:factor(asofdate)",2012)+ggtitle("log(Total loan volume), Non-jumbo")
grid.arrange(g1,g2,nrow=1)
hmda_merged <- rbind(hmda[actiontaken %in% c(1,3) & asofdate %in% 2012:2022,
c("actiontaken","amountofloan","applicantincome","asofdate","purposeofloan","dist_gse_limit","RSSD","county","typeofloan")],
hmda_post[actiontaken %in% c(1,3) & asofdate %in% 2012:2022,
c("actiontaken","amountofloan","applicantincome","asofdate","purposeofloan","dist_gse_limit","RSSD","county","typeofloan")])
hmda_merged <- merge(hmda_merged,
upbr_2016[,c("IDRSSD","jumbo_dependance","re_loan_frac_assets","jumbo_frac_assets","total_assets","jumbo_7590")],
by.x="RSSD",by.y="IDRSSD",all.x=T)
hmda_merged[,county_year:=paste(county,asofdate)]
# hmda_merged[,bank_county:=paste(county,RSSD)]
hmda_merged <- hmda_merged[applicantincome>0 & amountofloan>0]
hmda_merged[,loan_to_income:=amountofloan/applicantincome]
hmda_merged[,high_jumbo_assets:=ifelse(jumbo_dependance %in% high_jumbo_cutoff,1,0)]
hmda_merged[,high_jumbo_assets2:=ifelse(jumbo_dependance %in% 5:6,1,0)]
hmda_merged[,approved:=ifelse(actiontaken==1,1,ifelse(actiontaken==3,0,NA))]
# hmda_merged[,jumbo_dependance_cat:=ifelse(jumbo_dependance<4,1,jumbo_dependance)]
hmda_merged_jumbo <- hmda_merged[dist_gse_limit>10 & dist_gse_limit<750]
hmda_merged_jumbo[,loan_amount_bin:=ntile(amountofloan,10),by=county_year]
hmda_merged_jumbo[,county_year_amt_bin:=paste(county_year,loan_amount_bin)]
hmda_merged_jumbo[,bank_county:=paste(county,RSSD)]
r <- list()
r[[1]] <- felm(approved~high_jumbo_assets*I(asofdate>2016)+log(applicantincome)+log(amountofloan)|county_year_amt_bin+RSSD|0|county, data=hmda_merged_jumbo[purposeofloan==1 & total_assets < 1e9])
r[[2]] <- felm(approved~factor(jumbo_7590)*I(asofdate>2016)+log(applicantincome)+log(amountofloan)|county_year_amt_bin+RSSD|0|county, data=hmda_merged_jumbo[purposeofloan==1 & total_assets < 1e9])
r[[3]] <- felm(approved~log(0.01+jumbo_frac_assets)*I(asofdate>2016)+log(applicantincome)+log(amountofloan)|county_year_amt_bin+RSSD|0|county, data=hmda_merged_jumbo[purposeofloan==1 & total_assets < 1e9])
r[[4]] <- felm(approved~high_jumbo_assets*I(asofdate>2016)+log(applicantincome)+log(amountofloan)|county_year_amt_bin+RSSD|0|county, data=hmda_merged_jumbo[purposeofloan==3 & total_assets < 1e9])
r[[5]] <- felm(approved~factor(jumbo_7590)*I(asofdate>2016)+log(applicantincome)+log(amountofloan)|county_year_amt_bin+RSSD|0|county, data=hmda_merged_jumbo[purposeofloan==3 & total_assets < 1e9])
r[[6]] <- felm(approved~log(0.01+jumbo_frac_assets)*I(asofdate>2016)+log(applicantincome)+log(amountofloan)|county_year_amt_bin+RSSD|0|county, data=hmda_merged_jumbo[purposeofloan==3 & total_assets < 1e9])
stargazer(r,type="text",omit.stat = "ser")
##
## ============================================================================================================
## Dependent variable:
## -----------------------------------------------------------
## approved
## (1) (2) (3) (4) (5) (6)
## ------------------------------------------------------------------------------------------------------------
## high_jumbo_assets
## (0.000) (0.000)
##
## factor(jumbo_7590)2
## (0.000) (0.000)
##
## factor(jumbo_7590)3
## (0.000) (0.000)
##
## log(0.01 + jumbo_frac_assets)
## (0.000) (0.000)
##
## I(asofdate > 2016)
## (0.000) (0.000) (0.000) (0.000) (0.000) (0.000)
##
## log(applicantincome) 0.082*** 0.082*** 0.082*** 0.130*** 0.130*** 0.130***
## (0.002) (0.002) (0.002) (0.002) (0.002) (0.002)
##
## log(amountofloan) -0.105*** -0.105*** -0.105*** -0.177*** -0.177*** -0.179***
## (0.011) (0.011) (0.011) (0.018) (0.019) (0.018)
##
## high_jumbo_assets:I(asofdate > 2016) -0.044*** -0.077***
## (0.004) (0.004)
##
## factor(jumbo_7590)2:I(asofdate > 2016) -0.045*** -0.087***
## (0.004) (0.006)
##
## factor(jumbo_7590)3:I(asofdate > 2016) -0.042*** -0.066***
## (0.004) (0.004)
##
## log(0.01 + jumbo_frac_assets):I(asofdate > 2016) -0.009*** -0.025***
## (0.001) (0.002)
##
## ------------------------------------------------------------------------------------------------------------
## Observations 810,667 810,667 810,667 686,656 686,656 686,656
## R2 0.190 0.190 0.189 0.246 0.246 0.245
## Adjusted R2 0.082 0.082 0.081 0.137 0.137 0.136
## ============================================================================================================
## Note: *p<0.1; **p<0.05; ***p<0.01
r <- list()
r[[1]] <- felm(approved~high_jumbo_assets*factor(asofdate)+log(applicantincome)+log(amountofloan)|county_year_amt_bin+RSSD|0|county, data=hmda_merged_jumbo[purposeofloan==1 & asofdate %in% 2014:2021 & typeofloan==1])
r[[2]] <- felm(approved~high_jumbo_assets*factor(asofdate)+log(applicantincome)+log(amountofloan)|county_year_amt_bin+RSSD|0|county, data=hmda_merged_jumbo[purposeofloan==3 & asofdate %in% 2014:2021 & typeofloan==1 ])
g1 <- coef_plot_1reg(r[[1]],"high_jumbo_assets:factor(asofdate)",2014)+ggtitle("Jumbo, Approval; New purchases")
g2 <- coef_plot_1reg(r[[2]],"high_jumbo_assets:factor(asofdate)",2014)+ggtitle("Jumbo, Approval; Refinances")
grid.arrange(g1,g2,nrow=1)
r <- list()
r[[1]] <- felm(loan_to_income~high_jumbo_assets*factor(asofdate)+log(applicantincome)|county_year_amt_bin+bank_county|0|county, data=hmda_merged_jumbo[purposeofloan==1 & asofdate %in% 2014:2021 & typeofloan==1 & loan_to_income>1 & loan_to_income<10])
r[[2]] <- felm(loan_to_income~high_jumbo_assets*factor(asofdate)+log(applicantincome)|county_year_amt_bin+bank_county|0|county, data=hmda_merged_jumbo[purposeofloan==3 & asofdate %in% 2014:2021 & typeofloan==1 & loan_to_income>1 & loan_to_income<10])
g1 <- coef_plot_1reg(r[[1]],"high_jumbo_assets:factor(asofdate)",2014)+ggtitle("Jumbo, Loan-to-income; New purchases")
g2 <- coef_plot_1reg(r[[2]],"high_jumbo_assets:factor(asofdate)",2014)+ggtitle("Jumbo, Loan-to-income; Refinances")
grid.arrange(g1,g2,nrow=1)
hmda_merged_conforming <- hmda_merged[dist_gse_limit< -10 & dist_gse_limit> -110]
hmda_merged_conforming[,loan_amount_bin:=ntile(amountofloan,10),by=county_year]
hmda_merged_conforming[,county_year_amt_bin:=paste(county_year,loan_amount_bin)]
r <- list()
r[[1]] <- felm(approved~high_jumbo_assets*factor(asofdate)+log(applicantincome)+log(amountofloan)|county_year_amt_bin+RSSD|0|county, data=hmda_merged_conforming[purposeofloan==1 & asofdate %in% 2014:2021 & typeofloan==1 ])
r[[2]] <- felm(approved~high_jumbo_assets*factor(asofdate)+log(applicantincome)+log(amountofloan)|county_year_amt_bin+RSSD|0|county, data=hmda_merged_conforming[purposeofloan==3 & asofdate %in% 2014:2021 & typeofloan==1 ])
g1 <- coef_plot_1reg(r[[1]],"high_jumbo_assets:factor(asofdate)",2014)+ggtitle("Conforming, Approval, New Purchase")
g2 <- coef_plot_1reg(r[[2]],"high_jumbo_assets:factor(asofdate)",2014)+ggtitle("Conforming, Approval, Refinance")
grid.arrange(g1,g2,nrow=1)
r <- list()
r[[1]] <- felm(loan_to_income~high_jumbo_assets*factor(asofdate)+log(applicantincome)|county_year_amt_bin+RSSD|0|county, data=hmda_merged_conforming[purposeofloan==1 & asofdate %in% 2014:2021 & typeofloan==1 & loan_to_income>1 & loan_to_income<5])
r[[2]] <- felm(loan_to_income~high_jumbo_assets*factor(asofdate)+log(applicantincome)|county_year_amt_bin+RSSD|0|county, data=hmda_merged_conforming[purposeofloan==3 & asofdate %in% 2014:2021 & typeofloan==1 & loan_to_income>1 & loan_to_income<5])
g1 <- coef_plot_1reg(r[[1]],"high_jumbo_assets:factor(asofdate)",2014)+ggtitle("Conforming, Loan-to-income, New Purchase")
g2 <- coef_plot_1reg(r[[2]],"high_jumbo_assets:factor(asofdate)",2014)+ggtitle("Conforming, Loan-to-income, Refinances")
grid.arrange(g1,g2,nrow=1)
upbr_merged <- merge(upbr,upbr_2016[,c("IDRSSD","jumbo_dependance","re_loan_frac_assets",
"jumbo_frac_assets","jumbo_7590")],
by.x="IDRSSD",by.y="IDRSSD")
upbr_merged[,data_period:=as.Date(data_period,format="%m%d%Y")]
upbr_merged[,qtrs:=((year(data_period) - 2017) * 4) + quarter(data_period) -4 ]
upbr_merged[,re_loan_frac_assets:=sfr_first_lein_ma*100/total_assets]
upbr_merged[,unrealized_loss_equity:=(htm_value-htm_cost)/total_equity]
upbr_merged[,equity_assets:=total_equity/total_assets]
upbr_merged[,loans_assets:=total_loans/total_assets]
upbr_merged[,other_loans_assets:=(total_loans-sfr_first_lein_ma)/total_assets]
upbr_merged[,other_loans:=(total_loans-sfr_first_lein_ma)]
upbr_merged[,real_estate_loans_assets:=real_estate_loans/total_assets]
upbr_merged[,commercial_industrial_loans:=commercial_industrial_loans/total_assets]
upbr_merged[,cre_construction_land_dev_loans_assets:=cre_construction_land_dev_loans/total_assets]
upbr_merged[,mbs_assets:=mortgage_backed_securities/total_assets]
upbr_merged[,govt_assets:=govt_agency_securities/total_assets]
upbr_merged[,sfr_first_lein_ma_assets:=sfr_first_lein_ma/total_assets]
upbr_merged[,deposits_assets:=total_deposits/total_assets]
upbr_merged[,securities_assets:=total_securities/total_assets]
upbr_merged[,cash_assets:=cash/total_assets]
upbr_merged[,log_total_assets:=log(total_assets)]
upbr_merged[,log_maturity_security:=log(wa_maturiy_security)]
upbr_merged[,log_maturity_assets:=log(wa_maturiy_assets)]
upbr_merged[,log_salaries_employee_benefits:=log(1+salaries_employee_benefits)]
upbr_merged[,log_no_of_full_time_employees:=log(1+no_of_full_time_employees)]
upbr_merged[,rwa_assets:=risk_weighted_assets/total_assets]
upbr_merged[,high_jumbo_assets:=ifelse(jumbo_dependance %in% high_jumbo_cutoff,1,0)]
dep_var <- "mtg_delinq_pct"
formula <- as.formula(paste( dep_var,"~ high_jumbo_assets * factor(qtrs) + log(total_assets) + equity_assets | IDRSSD + data_period | 0 | IDRSSD"))
lower_bound <- quantile(upbr_merged$mtg_delinq_pct, probs = 0.01, na.rm = TRUE)
upper_bound <- quantile(upbr_merged$mtg_delinq_pct, probs = 0.99, na.rm = TRUE)
filtered_data <- upbr_merged %>%
filter((mtg_delinq_pct > lower_bound & mtg_delinq_pct < upper_bound) & qtrs <= 16 & qtrs>= -12)
coef_plot_1reg(felm(formula, data = filtered_data[qtrs <= 16 & qtrs>= -12]),"high_jumbo_assets:factor(qtrs)",-12)+ggtitle(dep_var)
dep_var <- "mtg_delinq_pct"
formula <- as.formula(paste( dep_var,"~ high_jumbo_assets * factor(qtrs) + log(total_assets) + equity_assets | IDRSSD + data_period | 0 | IDRSSD"))
lower_bound <- quantile(upbr_merged$mtg_delinq_pct, probs = 0.01, na.rm = TRUE)
upper_bound <- quantile(upbr_merged$mtg_delinq_pct, probs = 0.99, na.rm = TRUE)
filtered_data <- upbr_merged %>%
filter((mtg_delinq_pct > lower_bound & mtg_delinq_pct < upper_bound) & qtrs <= 16 & qtrs>= -12 & total_assets>1e6 & total_assets<1e7)
coef_plot_1reg(felm(formula, data = filtered_data[qtrs <= 16 & qtrs>= -12]),"high_jumbo_assets:factor(qtrs)",-12)+ggtitle(dep_var)
run_bank_fe_regression <- function(dep_var_name,min_qt,max_qt,sample_frac) {
formula <- as.formula(paste(dep_var_name, "~ high_jumbo_assets * factor(qtrs) + log(total_assets) + equity_assets | IDRSSD + data_period | 0 | IDRSSD"))
lower_bound <- quantile(upbr_merged[[dep_var_name]], probs = 0.01, na.rm = TRUE)
upper_bound <- quantile(upbr_merged[[dep_var_name]], probs = 0.99, na.rm = TRUE)
filtered_data <- upbr_merged %>%
filter((.data[[dep_var_name]] > lower_bound & .data[[dep_var_name]] < upper_bound) & qtrs <= max_qt & qtrs >= min_qt)
rssd_list <- unique(filtered_data$IDRSSD)
rssd_sample <- sample(rssd_list,size=length(rssd_list)*sample_frac,replace = F)
plot_result <- coef_plot_1reg(felm(formula, data = filtered_data[IDRSSD %in% rssd_sample]), "high_jumbo_assets:factor(qtrs)", min_qt) + ggtitle(dep_var_name)
return(plot_result)
}
dependent_vars <- c("loans_assets", "re_loans_sfr_assets", "other_loans_assets",
"nii_assets", "securities_assets","interest_expense_assets_2",
"personnel_expense_to_assets","assets_per_employee",
"wa_maturiy_assets","log_no_of_full_time_employees","total_capital_ratio",
"roe","unused_commitments_assets_total","uninsured_deposits_assets",
"unused_commitments_assets_cre","unused_commitments_assets_sfr_construction",
"ci_assets","deposits_assets")
for(dv in dependent_vars) {
print(run_bank_fe_regression(dv,-7,8,1))
}
# creates a data.table with county-level house prices (zhvi) for years 2015 and 2022
zrd <- fread("C:/Users/dratnadiwakara2/Documents/OneDrive - Louisiana State University/Raw Data/Zillow Research Data/County_zhvi_uc_sfr_tier_0.33_0.67_sm_sa_month.csv")
zrd[,fips:=paste0(str_pad(StateCodeFIPS,2,pad="0"),str_pad(MunicipalCodeFIPS,3,pad="0"))]
zrd[,c("RegionID","RegionName","SizeRank","StateName","RegionType","State","Metro","StateCodeFIPS","MunicipalCodeFIPS")] <- NULL
zrd <- melt(zrd,id.vars = c("fips"))
zrd$month <- as.Date(as.character(zrd$variable),origin = "1970-01-01")
zrd$variable <- NULL
names(zrd) <- c("county","zhvi","month")
zrd[,asofdate:=year(month)]
zrd <- zrd[,.(zhvi=mean(zhvi,na.rm=T)),by=.(county,asofdate)]
zrd[, rank := frank(-zhvi, ties.method = "min"), by = asofdate]
hmda_county <- hmda[actiontaken==1 & purposeofloan %in% c(1) & typeofloan==1 & dist_gse_limit>10,.(amountofloan=sum(amountofloan,na.rm=T)),
by=.(RSSD,asofdate,county)]
temp <- hmda_post[actiontaken==1 & purposeofloan %in% c(1) & typeofloan==1 & dist_gse_limit>10,.(amountofloan=sum(amountofloan,na.rm=T)),
by=.(RSSD,asofdate,county)]
hmda_county <- rbind(hmda_county,temp)
hmda_county[, total_loan_by_rssd_asofdate := sum(amountofloan), by = .(RSSD, asofdate)]
hmda_county[, market_share := amountofloan / total_loan_by_rssd_asofdate]
hmda_county[, squared_market_share := market_share^2, by = .(RSSD, asofdate, county)]
hmda_county <- merge(hmda_county, zrd[,c("rank","county","asofdate","zhvi")],by=c("county","asofdate"))
# Step 3: Aggregate to calculate HHI for each RSSD-asofdate
HHI <- hmda_county[market_share>0, .(HHI = sum(squared_market_share),.N,zhvi=mean(zhvi),rank=mean(rank)), by = .(RSSD, asofdate)]
HHI <- merge(HHI,upbr_2016[,c("IDRSSD","jumbo_dependance","re_loan_frac_assets","jumbo_frac_assets","total_assets","jumbo_7590")],
by.x="RSSD",by.y="IDRSSD",all.x=T)
HHI[,high_jumbo_assets:=ifelse(jumbo_dependance %in% high_jumbo_cutoff,1,0)]
# set.seed(1982)
rssd_list <- unique(HHI$RSSD)
rssd_sample <- sample(rssd_list,size=length(rssd_list)*1,replace = F)
r <- list()
r[[1]] <- felm(HHI~factor(jumbo_7590)*I(asofdate>2016)|RSSD+asofdate|0|RSSD,data=HHI[RSSD %in% rssd_sample])
r[[2]] <- felm(log(N)~factor(jumbo_7590)*I(asofdate>2016)|RSSD+asofdate|0|RSSD,data=HHI[RSSD %in% rssd_sample])
# r[[3]] <- felm(log(rank)~factor(jumbo_7590)*I(asofdate>2016)|RSSD+asofdate|0|RSSD,data=HHI[RSSD %in% rssd_sample])
stargazer(r,type="text",no.space = T,column.labels = c("New Purchases","Refinacnes"),omit.stat = "ser")
##
## ===================================================================
## Dependent variable:
## ----------------------------
## HHI log(N)
## New Purchases Refinacnes
## (1) (2)
## -------------------------------------------------------------------
## factor(jumbo_7590)2
## (0.000) (0.000)
## factor(jumbo_7590)3
## (0.000) (0.000)
## I(asofdate > 2016)
## (0.000) (0.000)
## factor(jumbo_7590)2:I(asofdate > 2016) 0.043*** -0.075**
## (0.010) (0.030)
## factor(jumbo_7590)3:I(asofdate > 2016) 0.073*** -0.135***
## (0.010) (0.034)
## -------------------------------------------------------------------
## Observations 20,901 20,901
## R2 0.647 0.842
## Adjusted R2 0.578 0.811
## ===================================================================
## Note: *p<0.1; **p<0.05; ***p<0.01
r[[3]] <-felm(HHI~high_jumbo_assets*factor(asofdate)|RSSD+asofdate|0|RSSD,data=HHI[RSSD %in% rssd_sample])
r[[4]] <-felm(log(N)~high_jumbo_assets*factor(asofdate)|RSSD+asofdate|0|RSSD,data=HHI[RSSD %in% rssd_sample])
# r[[6]] <-felm(log(rank)~high_jumbo_assets*factor(asofdate)|RSSD+asofdate|0|RSSD,data=HHI[RSSD %in% rssd_sample])
coef_plot_1reg(r[[3]],"high_jumbo_assets:factor(asofdate)",2012)+ggtitle("HHI, Jumbo")
coef_plot_1reg(r[[4]],"high_jumbo_assets:factor(asofdate)",2012)+ggtitle("No counties, Jumbo")
# coef_plot_1reg(r[[6]],"high_jumbo_assets:factor(asofdate)",2012)