DS608_Final_Project_JChhabria

The Physical Bank Branch - Asset or Liability?

A lot has been said about the utility of bank branches over the years - as a way to bank the unbanked, extend small business credit, aid in educating people about financial literacy. Several studies since the Great Financial Crisis have shown a clear increase in adoption of digital banking for the more standard tasks that most consumers deal with - checking balances, depositing checks, paying bills etc. Coupled with bank mergers, this has had a clear impact on branch networks across the USA.

I chose to study this data to explore the relatonship between branch networks and deposits. With the growing advent of Fintech, it was expected that Bank CEOs would take a serious look at the cost benefit ratio for bank branches - both the tangibles as well as intangibles such as brand building.

I used bank branch data from the website of the Federal Deposit Insurance Corporation (FDIC) to show the distribution of physical bank branches across the USA. I focused on the branch distributions of the 5 largest national banks and pulled in historical data to show how their branch network has been changing in the past 10-11 years.

The links to the data sources can be found at: https://www7.fdic.gov/sod/dynaDownload.asp?barItem=6 http://www2.census.gov/programs-surveys/popest/datasets/2010-2019/national/totals/nst-est2019-alldata.csv

# Import libraries
library(dplyr)

## 
## Attaching package: 'dplyr'

## The following objects are masked from 'package:stats':
## 
##     filter, lag

## The following objects are masked from 'package:base':
## 
##     intersect, setdiff, setequal, union

library(ggplot2)
library(kableExtra)

## 
## Attaching package: 'kableExtra'

## The following object is masked from 'package:dplyr':
## 
##     group_rows

library(stringr)
library(tidyverse)

## -- Attaching packages ----------------------------------------------------------- tidyverse 1.3.0 --

## v tibble  3.0.3     v purrr   0.3.4
## v tidyr   1.1.2     v forcats 0.5.0
## v readr   1.3.1

## -- Conflicts -------------------------------------------------------------- tidyverse_conflicts() --
## x dplyr::filter()          masks stats::filter()
## x kableExtra::group_rows() masks dplyr::group_rows()
## x dplyr::lag()             masks stats::lag()

library(readr)
library(viridis)

## Loading required package: viridisLite

library(wesanderson)
library(hrbrthemes)

## NOTE: Either Arial Narrow or Roboto Condensed fonts are required to use these themes.

##       Please use hrbrthemes::import_roboto_condensed() to install Roboto Condensed and

##       if Arial Narrow is not on your system, please see https://bit.ly/arialnarrow

library(gridExtra)

## 
## Attaching package: 'gridExtra'

## The following object is masked from 'package:dplyr':
## 
##     combine

library(plotly)

## 
## Attaching package: 'plotly'

## The following object is masked from 'package:ggplot2':
## 
##     last_plot

## The following object is masked from 'package:stats':
## 
##     filter

## The following object is masked from 'package:graphics':
## 
##     layout

library(usmap)
library(rgdal)

## Loading required package: sp

## rgdal: version: 1.5-18, (SVN revision 1082)
## Geospatial Data Abstraction Library extensions to R successfully loaded
## Loaded GDAL runtime: GDAL 3.0.4, released 2020/01/28
## Path to GDAL shared files: C:/Users/juggy/Documents/R/win-library/3.5/rgdal/gdal
## GDAL binary built with GEOS: TRUE 
## Loaded PROJ runtime: Rel. 6.3.1, February 10th, 2020, [PJ_VERSION: 631]
## Path to PROJ shared files: C:/Users/juggy/Documents/R/win-library/3.5/rgdal/proj
## Linking to sp version:1.4-2
## To mute warnings of possible GDAL/OSR exportToProj4() degradation,
## use options("rgdal_show_exportToProj4_warnings"="none") before loading rgdal.

library(leaflet)
library(maps)

## 
## Attaching package: 'maps'

## The following object is masked from 'package:purrr':
## 
##     map

library(ggiraph)

# Perform housekeeping tasks
options(scipen=999)
defaultW <- getOption("warn")
options(warn = -1)
setwd('C:/Jagdish/MastersPrograms/CUNY/DS608KnowledgeAndVisualAnalytics/Final_Project')

# Clear objects that are not needed
#rm(top5_2015,top5_2016,top5_2017,top5_2018)
#rm(usa_branches)
gc()

##           used (Mb) gc trigger  (Mb) max used  (Mb)
## Ncells 1228861 65.7    2401055 128.3  2401055 128.3
## Vcells 2053724 15.7    8388608  64.0  3605704  27.6

I loaded the usa map data.

# Load USA States map data
usa_map<-map_data("usa")
states<-map_data("state")

# Rename the region column in the states data and change case from lower to camel
states<-states%>%rename("Region"="region")
states$Region<-str_to_title(states$Region,locale="en")

I loaded the data files for bank data as well as state populations.

# Read in data files
# Load branch count and deposit data for USA and filter for years 2010 and later
usa_branches<-read_csv("usa_bank_data.csv")

## Parsed with column specification:
## cols(
##   Year = col_double(),
##   Region = col_character(),
##   Number_of_Banks = col_double(),
##   Number_of_Branches = col_double(),
##   Branches_Pct_Change = col_double(),
##   New_Charters = col_double(),
##   Total_Deposits = col_number(),
##   Deposits_Pct_Change = col_double(),
##   Population = col_double(),
##   Small_Business_Loans = col_number(),
##   Total_Business_Loans = col_number(),
##   Small_Loan_Share = col_double()
## )

usa_branches<-subset(usa_branches,Year>=2010)
usa_branches<-subset(usa_branches,Year<2020)

# Load in branch count for all states and filter for years 2010 and later
region_branches<-read_csv("states_bank_data.csv")

## Parsed with column specification:
## cols(
##   Year = col_double(),
##   Region = col_character(),
##   Number_of_Banks = col_double(),
##   Number_of_Branches = col_double(),
##   New_Charters = col_double()
## )

region_branches<-subset(region_branches,Year>=2010)

# Load branch count and deposits for top banks
top_banks<-read_csv("top_3_banks.csv")

## Parsed with column specification:
## cols(
##   Year = col_double(),
##   Bank = col_character(),
##   Branches = col_double(),
##   Deposits = col_double()
## )

top_banks<-subset(top_banks,Year<2020)
#top_banks<-merge(y=usa_branches[,c("Year","Number_of_Branches", "Total_Deposits")],x=top_banks,by=c("Year"),all.x = TRUE)

# Load the Census population data from 2010 to 2019
census_population_data<-read_csv('census_states_population_2019.csv')

## Parsed with column specification:
## cols(
##   Region_Number = col_character(),
##   State_Number = col_double(),
##   Region = col_character(),
##   `2010` = col_double(),
##   `2011` = col_double(),
##   `2012` = col_double(),
##   `2013` = col_double(),
##   `2014` = col_double(),
##   `2015` = col_double(),
##   `2016` = col_double(),
##   `2017` = col_double(),
##   `2018` = col_double(),
##   `2019` = col_double()
## )

# Convert from a wide to a long format
pop_hist<-census_population_data%>%gather(Year,Population,4:13)

# Add a new column for population in ten thousands
pop_hist$Population_10000<-round(pop_hist$Population/10000,0)

# Remove rows related to regions and columns for Region Number and State Number
pop_hist<-pop_hist[!pop_hist$Region %in% c('Northeast Region','Midwest Region', 'South Region','West Region'),]
pop_hist<-subset(pop_hist,select=-c(Region_Number,State_Number))

# Filter the top banks data frame to exclude rows for "United States"
top_banks<-top_banks[!top_banks$Bank %in% c('United States'),]

I calculated the branch count per 10,000 people.

# Combine states branch data with population data to calculate branches per 10000 people
branches_per_cap<-merge(y=region_branches,x=pop_hist,by=c("Year","Region"),all.x = TRUE)
#state_per_cap_branch<-dplyr::left_join(states, branches_per_cap, by='Region')
branches_per_cap$Branches_Per_Capita<-round(branches_per_cap$Number_of_Branches/branches_per_cap$Population_10000,2)

# Filter the data for USA rows only
#usa_total<-branches_per_cap%>%filter(Region=="United States")
#usa_total<-merge(y=usa_total,x=usa_branches,by=c("Year"),all.x = TRUE)
usa_branches$Population_10000<-round(usa_branches$Population/10000,0)
usa_branches$Branches_Per_Capita<-round(usa_branches$Number_of_Branches/usa_branches$Population_10000,2)
usa_branches$Deposits_Per_Branch<-round(usa_branches$Total_Deposits/usa_branches$Number_of_Branches,2)
usa_branches$Deposits_Per_Capita<-round(usa_branches$Total_Deposits/usa_branches$Population_10000,2)
#usa_branches

# Filter branch data for all of USA
usa_total<-branches_per_cap%>%filter(Region=="United States")
#kable(branches_per_cap)%>%kable_styling()
#head(branches_per_cap)

# Use 2019 for branch density map - filter data for 2019
state_per_cap_branch<-dplyr::left_join(states, branches_per_cap, by='Region')
state_pcb_2019<-state_per_cap_branch %>% filter(Year==2019)

I plot the annual trend in branch count across all banks in the USA

# Plot annual trend in branch network across all of USA
ggplot(data=usa_total,aes(x=Year,y=Number_of_Branches,group=1))+  geom_line(color="red")+geom_point()+labs(title="Trend in Bank Branches") 

# Create a variable for fill color
usa_branches<-usa_branches%>%mutate(Branches_Change_Positive=Branches_Pct_Change>0)

I plot the % change in bank branches from 2010 to 2019

# Plot % change in yearly Branch counts across all US Banks
plot1<-ggplot(data=usa_branches,aes(fill=Branches_Pct_Change>0, y=Branches_Pct_Change, x=Year))+ geom_bar(stat="identity",size=0.1,color="black",alpha=.7)+scale_x_continuous(name="Year",breaks=c(2010,2011,2012,2013,2014,2015,2016,2017,2018,2019))+scale_y_continuous(name="Branches Percent Change", limits=c(-0.025,+0.025),breaks=c(-0.02,-0.01,0,0.01,0.02))+labs(x="Year",y="Branches Percent Change", title = "Percentage change in Branch network across all US Banks", subtitles = "")+guides(fill = FALSE)+theme(axis.title = element_text(color="black",size=10,face=2),axis.line=element_line(size = 1,colour="black",linetype=1),axis.text = element_text(angle =90,color="black",size=10, face=2))+annotate("text",x=2010,y=0.01,size=3,label="Great Financial Crisis",color='red')

plot1

I plot the proportion of branches for the top 5 banks and all other banks.

p1<-ggplot(data=top_banks,aes(x=Year,y=Branches,group=Bank))+geom_col(aes(fill=Bank),width=0.7)+scale_x_continuous(breaks = unique(top_banks$Year))+theme(axis.text.x=element_text(angle=0, vjust=1))+geom_text(aes(label=Branches),position='stack',size=2.5, vjust=1)+labs(x="Year",y="Number of Branches")+ggtitle("Branch Count for Top Banks in USA")+scale_fill_brewer(palette="PiYG")
  
p1

# Filter out all other banks to leave only the top 5 banks in the dataset
top3_banks<-subset(top_banks,Bank!="All other banks")

# Create a bar plot for branches for top banks for 2020
p2<-ggplot(top3_banks,aes(fill=Bank, y=Branches, x=Year))+ geom_bar(position="dodge", stat="identity",size=0.1,color="black",alpha=.7)+scale_x_continuous(name="Year",breaks=c(2010,2011,2012,2013,2014,2015,2016,2017,2018,2019))+scale_y_continuous(name="Branch count",breaks=c(0,1000,2000,3000,4000,5000,6000))+labs(title = "Yearly trend in Branch network for top 5 banks in the USA")

p2

# Plot % change in yearly deposits and deposit levels across all US Banks
plot2<-ggplot(data=usa_branches,aes(y=Deposits_Pct_Change, x=Year))+ geom_line(size=0.1,color="blue",alpha=.7)+scale_x_continuous(name="Year",breaks=c(2010,2011,2012,2013,2014,2015,2016,2017,2018,2019))+scale_y_continuous(name="Deposits Percent Change", limits=c(0,0.15))+geom_point()+geom_text(aes(label=paste("$",round(Total_Deposits/1000000000000,1),"Tln")),color="blue",size=3,vjust="outward",hjust="outward",show.legend=TRUE)+labs(x="Year",y="Deposits Percent Change", title = "Total Bank Deposits in $ Trillions across all US Banks", subtitles = "")+guides(fill = FALSE)+theme(axis.title = element_text(color="black",size=10,face=2,hjust=.2),axis.line=element_line(size = 1,colour="black",linetype=1),axis.text = element_text(angle =90,color="black",size=10, face=2))+annotate("text",x=2010,y=0.13,hjust=.2,size=3,label="Great Financial Crisis",color='red')

plot2

#grid.arrange(plot1, plot2,nrow=2,ncol=1,heights=c(6,8))

# Create a line plot for yearly deposits for top banks
top3_banks<-subset(top_banks,Bank!="All other banks")
top3_banks$Deposits<-top3_banks$Deposits/1000000

p1<-ggplot(data=top3_banks,aes(x=Year,y=Deposits,color=Bank))+geom_line()+scale_x_continuous(name="Year",breaks=c(2010,2011,2012,2013,2014,2015,2016,2017,2018,2019))+scale_y_continuous(name="Deposits in $Millions",breaks=c(0,200,400,600,800,1000,1200,1400,1600))+labs(title = "Yearly trend in Deposits ($ million) for top 5 banks in the USA")

p1

# Filter data for branch count and small business loans
loans<-usa_branches[,(names(usa_branches)%in%c("Year","Small_Business_Loans","Total_Business_Loans"))]
loans<-loans%>%gather(Loan_Type,Loan_Amount_Mln,2:3)

Next we check whether there is a trend in small business credit as a proportion of total bank loans, the using small multiples approach.

# Check share of small business loans as a share of total bank loans in millions
ggplot(data=loans,aes(x="",y=Loan_Amount_Mln,fill=Loan_Type))+ 
geom_bar(stat="identity",position=position_fill())+
coord_polar(theta="y")+
facet_wrap(~Year)+
theme(axis.title.x=element_blank(),
axis.title.y=element_blank())+ 
theme(legend.position='bottom')

I load the detailed branch data from the FDIC website.

# Load the Bank Branches data for previous years
#Source:
#all2020<-read.csv('ALL_2020.csv')
all2020<-read_csv('ALL_2020.csv')

## Parsed with column specification:
## cols(
##   .default = col_character(),
##   YEAR = col_double(),
##   BRNUM = col_double(),
##   CONSOLD = col_double(),
##   BRSERTYP = col_double(),
##   DEPSUMBR = col_number(),
##   BKMO = col_double(),
##   CNTYNUMB = col_double(),
##   CSABR = col_double(),
##   DIVISIONB = col_double(),
##   MSABR = col_double(),
##   METROBR = col_double(),
##   MICROBR = col_double(),
##   NECTABR = col_double(),
##   NECNAMB = col_logical(),
##   PLACENUM = col_double(),
##   SIMS_LATITUDE = col_double(),
##   SIMS_LONGITUDE = col_double(),
##   STNUMBR = col_double(),
##   RSSDHCR = col_double(),
##   RSSDID = col_double()
##   # ... with 14 more columns
## )

## See spec(...) for full column specifications.

all2019<-read_csv('ALL_2019.csv')

## Parsed with column specification:
## cols(
##   .default = col_character(),
##   YEAR = col_double(),
##   BRNUM = col_double(),
##   CONSOLD = col_double(),
##   BRSERTYP = col_double(),
##   DEPSUMBR = col_number(),
##   BKMO = col_double(),
##   CNTYNUMB = col_double(),
##   CSABR = col_double(),
##   DIVISIONB = col_double(),
##   MSABR = col_double(),
##   METROBR = col_double(),
##   MICROBR = col_double(),
##   NECTABR = col_double(),
##   NECNAMB = col_logical(),
##   SIMS_LATITUDE = col_double(),
##   SIMS_LONGITUDE = col_double(),
##   STNUMBR = col_double(),
##   RSSDHCR = col_double(),
##   RSSDID = col_double(),
##   UNIT = col_double()
##   # ... with 13 more columns
## )
## See spec(...) for full column specifications.

# Make a list of columns of interest
cols<-c('ASSET', 'CHARTER', 'CITYBR', 'CNTRYNA', 'DEPDOM', 'DEPSUM', 'DEPSUMBR', 'NAMEBR', 'NAMEFULL', 'REGAGNT', 'RSSDID', 'SIMS_LATITUDE', 'SIMS_LONGITUDE', 'SPECDESC', 'SPECGRP', 'STALPBR', 'STNAMEBR', 'UNINUMBR', 'YEAR', 'ZIPBR')

# Filter the data frame to select only the columns of interest
all2020<-all2020[ ,which((names(all2020) %in% cols)==TRUE)]
all2019<-all2019[ ,which((names(all2019) %in% cols)==TRUE)]

# Derive the top 5 banks by number of branches over the past 11 years
kable(all2020%>%group_by(NAMEFULL)%>%summarise(n=n())%>%arrange(desc(n))%>%top_n(5))%>%kable_styling()

## `summarise()` ungrouping output (override with `.groups` argument)

## Selecting by n

NAMEFULL	n
Wells Fargo Bank, National Association	5410
Jpmorgan Chase Bank, National Association	4979
Bank Of America, National Association	4253
Truist Bank	2921
U.S. Bank National Association	2774

kable(all2019%>%group_by(NAMEFULL)%>%summarise(n=n())%>%arrange(desc(n))%>%top_n(5))%>%kable_styling()

## `summarise()` ungrouping output (override with `.groups` argument)
## Selecting by n

NAMEFULL	n
Wells Fargo Bank, National Association	5569
JPMorgan Chase Bank, National Association	5024
Bank of America, National Association	4336
U.S. Bank National Association	2979
PNC Bank, National Association	2398

We can see that the same banks show up in the list of top banks by branch count, except for a new entrant in 2020 - Truist which was the result of a merger between SunTrust Bank and BB&T. So going forward, we will focus on these 6 banks: Wells Fargo, JP Morgan Chase, Bank of America, Truist, US Bank and PNC.

Surprisingly Citibank which has a big banking brand does not show up in this list. This is because it has a much smaller branch network compared to the big 3 banks. It is actually between positions 10 to 20 as far as branch count is concerned.

# Create a dataframe with the bank short names
short<-read.table(file="bank_short_names.txt",header=TRUE,sep="|")

# Rename the columns to more intuitive names
# Convert case for the state column to allow it to be used in a merge
# Create a shortlist of top 5 bank by branch count for each year
# Filter the data and create a new dataset for each year with branch data only for top 5 banks

# 2020
#all2020$region<-tolower(all2020$STNAMEBR)
#all2020<-subset(all2020,select=-c(state))
all2020<-all2020%>%rename(state_code=STALPBR, region=STNAMEBR, total_assets=ASSET, city=CITYBR, domestic_deposits=DEPDOM, total_deposits=DEPSUM, branch_deposits=DEPSUMBR, branch=NAMEBR, bank_name=NAMEFULL,regulator=REGAGNT, specialization=SPECDESC, branch_id=RSSDID,zipcode=ZIPBR, location_id=UNINUMBR, spec_grp=SPECGRP, branch_lat=SIMS_LATITUDE, branch_lon=SIMS_LONGITUDE)

all2020$region<-tolower(all2020$region)
shortlist<-all2020%>%group_by(bank_name)%>%summarise(n=n())%>%arrange(desc(n))%>%top_n(5)

## `summarise()` ungrouping output (override with `.groups` argument)

## Selecting by n

top5_2020<-all2020[all2020$bank_name %in% shortlist$bank_name,]
top5_2020<-merge(top5_2020,short,by='bank_name')

# 2019
all2019<-all2019%>%rename(state_code=STALPBR, region=STNAMEBR, total_assets=ASSET, city=CITYBR, domestic_deposits=DEPDOM, total_deposits=DEPSUM, branch_deposits=DEPSUMBR, branch=NAMEBR, bank_name=NAMEFULL,regulator=REGAGNT, specialization=SPECDESC, branch_id=RSSDID,zipcode=ZIPBR, location_id=UNINUMBR, spec_grp=SPECGRP, branch_lat=SIMS_LATITUDE, branch_lon=SIMS_LONGITUDE)

all2019$region<-tolower(all2019$region)
shortlist<-all2019%>%group_by(bank_name)%>%summarise(n=n())%>%arrange(desc(n))%>%top_n(5)

## `summarise()` ungrouping output (override with `.groups` argument)
## Selecting by n

top5_2019<-all2019[all2019$bank_name %in% shortlist$bank_name,]
top5_2019<-merge(top5_2019,short,by='bank_name')

Next I plot state totals for bank branches and bank deposits and introduce the state population as a third variable in a bubble chart.

# Derive state-wise totals for deposits and branches and combine with population for 2019
#arrange((distinct(all2020, state)))
#length(unique(all2020$state_code))
pop_2019<-subset(pop_hist,pop_hist$Year==2019)
pop_2019<-pop_2019%>%rename("region"="Region")
#head(pop_2019)
state_2019<-all2019%>%group_by(region)%>%summarize(tot_deposits=sum(branch_deposits),tot_branches=n())

## `summarise()` ungrouping output (override with `.groups` argument)

state_2019$region<-str_to_title(state_2019$region,locale="en")
#head(state_2019)
state_2019<-merge(y=pop_2019,x=state_2019,by=c("region"),all.x = TRUE)
state_2019<-state_2019[complete.cases(state_2019),]
#state_2019

# Derive state-wise totals for deposits and branches and combine with population for 2019
pop_2019<-subset(pop_hist,pop_hist$Year==2019)
pop_2019<-pop_2019%>%rename("region"="Region")

state_2019<-all2019%>%group_by(region)%>%summarize(tot_deposits=sum(branch_deposits),tot_branches=n(),state_code=state_code)

## `summarise()` regrouping output by 'region' (override with `.groups` argument)

state_2019$region<-str_to_title(state_2019$region,locale="en")

state_2019<-merge(y=pop_2019,x=state_2019,by=c("region"),all.x = TRUE)
state_2019<-state_2019[complete.cases(state_2019),]
#state_2019

# Create a bubble chart showing branch count versus deposits for all the states for 2019 - the bubble size is governed by the state's population

p<-state_2019%>%ggplot(aes(x=tot_branches,y=round(tot_deposits/1000000,2),size=round(Population/10000000,2),fill=region))+
geom_point(alpha=0.9,shape=21,color="black",position=position_dodge(0.9))+
geom_text(aes(label=state_code),size=2,nudge_x=0.0,nudge_y=-0.15)+
scale_size(range = c(.1, 24), name="Population")+ scale_fill_manual(values=wes_palette(name="GrandBudapest2",n=51,type="continuous"))+
#scale_fill_viridis(discrete=TRUE, guide=FALSE, option="B") +
theme(legend.position="bottom")+
theme_classic()+
ylab("Deposits")+
xlab("Branches")+theme(legend.position = "none")

#p<-ggplotly(p, tooltip="label")
p

I now plot a chloropeth map shwing bank branch density by state for 2019 using the ggiraph package. I also create an interactive plot for this using R Shiny.

# Create the same chloropeth map using the ggiraph package to make it interactive
p0<-ggplot(data=state_per_cap_branch)+geom_polygon_interactive(aes(x=long,y=lat,fill=Branches_Per_Capita,group=group,tooltip=paste(Region,":",Branches_Per_Capita)),data_id=state_per_cap_branch$Region,color="white")+theme_classic()+xlab("Longitude")+ylab("Latitude")+labs(title="Bank Branches Per 10000 people in the USA for 2019") 

p1<-p0+labs(fill="Branch Density")+scale_fill_gradient(low="white",high="#CB454A")
p2<-p1+theme(legend.position="right",legend.background=element_rect(size=2,linetype='solid'),legend.text = element_text(size=6),legend.title=element_text(size=8),legend.box.margin = margin(0, 0, 0, 0, "cm"))

#,width=.7

girafe(code=print(p2),width_svg=9,height_svg=6,options=list(opts_sizing(rescale=FALSE)))

I drill down into Bank of America’s data. and create a leaflet map showing all the branches across the USA. I also create an interactive Shiny map allowing for selection of the state, and showing only the branches within that state along with a tooltip showing the branch deposits on mouse-over.

# Create a new dataset by filtering branches for top banks only
bankam_branches<-top5_2020%>%filter(bank=="Bank of America")
#wellsfargo_branches<-top5_2020%>%filter(bank=="Wells Fargo")
#jpmchase_branches<-top5_2020%>%filter(bank=="JPM Chase")
#truist_branches<-top5_2020%>%filter(bank=="Truist")
#usbank_branches<-top5_2020%>%filter(bank=="US Bank")
#pnc_branches<-top5_2020%>%filter(bank=="PNC")

# Create a new dataset by filtering branches for one state only
#MA_top5_2020<-top5_2020%>%filter(state_code=="MA")
#alabama.map<-states%>%filter(region=='alabama')
#alabama.branches<-top5_2020%>%filter(region=='alabama')
#alabama_top5_2020<-merge(alabama.map,alabama.branches,by='region')

# Create icons based on bank logos

bankamIcon<-icons("bankam.png",iconWidth=8,iconHeight=8)
wellsfargoIcon<-icons("wellsfargo.png",iconWidth=6,iconHeight=6)
jpmchaseIcon<-icons("jpmchase.jpg",iconWidth=8,iconHeight=8)
truistIcon<-icons("truist.png",iconWidth=8,iconHeight=8)
usbankIcon<-icons("usbank.jpg",iconWidth=8,iconHeight=8)
pncIcon<-icons("pnc.png",iconWidth=8,iconHeight=8)

# Generate branch density plot for Bank of America for the USA for 2020
#m%>%clearTiles()
#m<-leaflet(options=leafletOptions(minZoom=0,maxZoom=18))%>%setView(-96,37.8,4)

leaflet()%>%setView(-96,37.8,4)%>%addProviderTiles("CartoDB.Positron")%>%addMarkers(data=bankam_branches,lng=~branch_lon,lat=~branch_lat,icon=bankamIcon,label=~branch,popup=~paste("Deposits = $",branch_deposits))

#leaflet()%>%setView(-96,37.8,4)%>%addProviderTiles("CartoDB.Positron")%>%addCircleMarkers(data=bankam_branches,lng=~branch_lon,lat=~branch_lat,label=~branch, popup=~paste("$",branch_deposits), radius=~branch_deposits/total_deposits)

Conclusion

Based on the analysis and visualizations above, it can be seen that the physical bank branch is seeing a lot of headwinds since the past 11 years. While the number of banks across the country have declined by about 30%, the number of branches have declined by about 5%. This is probably on account of mergers and acquisitions. Given the advent of Fintech and the increasing penetration of digital banking, there has been a secular decline in branch networks as bank managements optimize the branch network. At the same time, the amount of bank deposits has been steadily climbing every year. This is probably due to the large amount of liquidity pumped into the system by the Federal Reserve and the fiscal policy.

All of the above data is pre-Covid. It is safe to assume that the data for next year would show a furher accentuation in decrease in bank branches and increase in bank deposits due to the effect of the pandemic.

In terms of further study, it would be interesting to repeat this analysis next year, as well as dig deeper into Citibank’s data to understand how it maintains a large asset and deposit base despite having far fewer branches.