source('~/Desktop/Empirical Research/Income Distribution/Gini Ratio/Hierarchical Levels/BGH/BGH Annual Gini.R')
g = aggregate(d, list(d$Year), FUN = ineq)
View(g)
boxplot(g$Pay)
View(g)
boxplot(g, ylim = c(0.1,0.3))
boxplot(g$Pay, ylim = c(0.1,0.3))
boxplot(g$Pay, ylim = c(0.15,0.25))
d = read_csv("BGH.csv")
d = data.frame(Year = d$year, Level = d$level, Pay = d$salary)
d = na.omit(d)
setwd("~/Desktop/Empirical Research/Income Distribution/Gini Ratio/Hierarchical Levels/Empirical/BGH")
d = read_csv("BGH.csv")
d = data.frame(Year = d$year, Level = d$level, Pay = d$salary)
d = na.omit(d)
setwd("~/Desktop/Empirical Research/Income Distribution/Gini Ratio/Hierarchical Levels/Empirical/BGH")
library(readr)
library(ineq)
d = read_csv("BGH.csv")
d = data.frame(Year = d$year, Level = d$level, Pay = d$salary)
d = na.omit(d)
test = subset(d, d$Year = 85)
test = subset(d, d$Year == 85)
test = subset(test, test$Level == 1)
hist(log(test))
hist(log(test$Pay))
library(readr)
library(ineq)
d = read_csv("BGH.csv")
d = data.frame(Year = d$year, Level = d$level, Pay = d$salary)
d = na.omit(d)
d$Year = paste("19", d$Year, sep = "")
d$Year = as.numeric(d$Year)
y = sort(unique(d$Year))
l = length(y)
final = NULL
source('~/Desktop/Empirical Research/Income Distribution/Gini Ratio/Hierarchical Levels/Empirical/Case Studies/BGH/BGH Level Analysis.R')
d = read_csv("BGH.csv")
d = read_csv("BGH.csv")
d = read.csv("BGH.csv")
source('~/Desktop/Empirical Research/Income Distribution/Gini Ratio/Hierarchical Levels/Empirical/Case Studies/BGH/BGH Level Analysis.R')
d = read_csv("BGH.csv")
d = data.frame(Year = d$year, Level = d$level, Pay = d$salary)
d = na.omit(d)
d$Year = paste("19", d$Year, sep = "")
d$Year = as.numeric(d$Year)
View(d)
d = read_csv("BGH.csv")
d = data.frame(Year = d$year, Level = d$level, Pay = d$salary)
View(d)
d = read_csv("BGH.csv")
d = data.frame(Year = d$year, Level = d$level, Pay = d$salary)
d = na.omit(d)
d$Year = paste("19", d$Year, sep = "")
d$Year = as.numeric(d$Year)
library(data.table)
d = data.table(Year = d$year, Level = d$level, Pay = d$salary)
d = read_csv("BGH.csv")
d = data.table(Year = d$year, Level = d$level, Pay = d$salary)
d = read_csv("BGH.csv")
d = data.table(Year = d$year, Level = d$level, Pay = d$salary)
d = na.omit(d)
d$Year = paste("19", d$Year, sep = "")
d$Year = as.numeric(d$Year)
View(final)
View(final)
source('~/Desktop/Empirical Research/Income Distribution/Gini Ratio/Hierarchical Levels/Empirical/Case Studies/BGH/BGH Level Analysis.R')
summary(d$Pay)
plot(d$Year, d$Pay, log = "y")
plot(d$Level, d$Pay, log = "y")
plot(d$Level, d$Pay)
source('~/Desktop/Empirical Research/Income Distribution/Gini Ratio/Hierarchical Levels/Empirical/Case Studies/BGH/BGH Level Analysis.R')
source('~/Desktop/Empirical Research/Income Distribution/Gini Ratio/Hierarchical Levels/Empirical/Case Studies/BGH/BGH Level Analysis.R')
source('~/Desktop/Empirical Research/Income Distribution/Gini Ratio/Hierarchical Levels/Empirical/Case Studies/BGH/BGH Level Analysis.R')
source('~/Desktop/Empirical Research/Income Distribution/Gini Ratio/Hierarchical Levels/Empirical/Case Studies/BGH/BGH Level Analysis.R')
plot(d$Level, d$Pay)
plot(d$Level, d$Pay, log = "y")
View(final)
source('~/Desktop/Empirical Research/Income Distribution/Gini Ratio/Hierarchical Levels/Empirical/Case Studies/BGH/BGH Level Analysis.R')
source('~/Desktop/Empirical Research/Income Distribution/Gini Ratio/Hierarchical Levels/Empirical/Case Studies/BGH/Level Analysis.R')
source('~/Desktop/Empirical Research/Income Distribution/Gini Ratio/Hierarchical Levels/Empirical/Case Studies/BGH/Level Analysis.R')
test = d[,list(
Year = head(Year, 1),
Level = head(Level,1)
),
by = .(Year, Level)       ]
View(test)
View(test)
test = d[,list(
N = length(Pay),
Mean = mean(Pay)
),
by = .(Year, Level)       ]
View(test)
test = d[,list(
N = length(Pay),
Mean = mean(Pay),
Gini = ineq(Pay)
),
by = .(Year, Level)       ]
View(test)
list(
N = length(Pay),
Mean = mean(Pay),
Gini = ineq(Pay)*N
)
test = d[,list(
N = length(Pay),
Mean = mean(Pay),
Gini = ineq(Pay)*length(Pay)/(length(Pay) - 1)
),
by = .(Year, Level)       ]
View(test)
?ifelse
gini.adjust = function(x){
ifelse(length(x) == 1, 0, ineq(x)*length(x)/(length(x) -1))
}
test = d[,list(
N = length(Pay),
Mean = mean(Pay),
Gini = gini.adjust
),
by = .(Year, Level)       ]
test = d[,list(
N = length(Pay),
Mean = mean(Pay),
Gini = gini.adjust(Pay)
),
by = .(Year, Level)       ]
View(test)
mean(test$Gini)
mean(final$Gini)
test = d[,list(
N = length(Pay),
Mean = mean(Pay),
Gini = gini.adjust(Pay)
Source = "BGH"
),
by = .(Year, Level)       ]
test = d[,list(
N = length(Pay),
Mean = mean(Pay),
Gini = gini.adjust(Pay),
Source = "BGH"
),
by = .(Year, Level)       ]
View(test)
plot(final$Gini, test$Gini)
result = d[,list(
N = length(Pay),
Mean = mean(Pay),
Gini = gini.adjust(Pay),
Source = "BGH"
),
by = .(Year, Level)       ]
result = result[order(Year, Level), ]
View(result)
plot(final$Gini, result$Gini)
sd(1)
sd(5)
plot(result$Mean, final$Mean)
d = read.csv("BGH.csv")
d = data.table(Year = d$year, Level = d$level, Pay = d$salary) %>% na.omit()
d$Year = paste("19", d$Year, sep = "")
d$Year = as.numeric(d$Year)
source('~/Desktop/Empirical Research/Income Distribution/Gini Ratio/Hierarchical Levels/Empirical/Case Studies/BGH/Level Analysis.R')
source('~/Desktop/Empirical Research/Income Distribution/Gini Ratio/Hierarchical Levels/Empirical/Case Studies/BGH/Level Analysis.R')
source('~/Desktop/Empirical Research/Income Distribution/Gini Ratio/Hierarchical Levels/Empirical/Case Studies/BGH/Level Analysis.R')
source('~/Desktop/Empirical Research/Income Distribution/Gini Ratio/Hierarchical Levels/Empirical/Case Studies/BGH/Level Analysis.R')
source('~/Desktop/Empirical Research/Income Distribution/Gini Ratio/Hierarchical Levels/Empirical/Case Studies/BGH/Level Analysis.R')
source('~/Desktop/Empirical Research/Income Distribution/Gini Ratio/Hierarchical Levels/Empirical/Case Studies/BGH/Promotions Analysis.R')
source('~/Desktop/Empirical Research/Income Distribution/Gini Ratio/Hierarchical Levels/Empirical/Case Studies/BGH/Promotions Analysis.R')
source('~/Desktop/Empirical Research/Income Distribution/Gini Ratio/Hierarchical Levels/Empirical/Case Studies/Dohmen/Data/Dohmen Firm Gini.R')
source('~/Desktop/human_capital/Supplementary Material/Data/Case Studies/BGH/BGH_human_capital.R')
source('~/Desktop/human_capital/Supplementary Material/Data/Case Studies/BGH/BGH_human_capital.R')
source('~/Desktop/human_capital/Supplementary Material/Data/Case Studies/BGH/BGH_human_capital.R')
d1 = data.table(log.norm.pay = log(d$pay.norm) , level = d$level, edyrs = d$edyrs, age = d$age, yratco = d$yratco)
d1 = na.omit(d1)
library(data.table)
library(magrittr)
library(ppcor)
d = fread("BGH.csv", stringsAsFactors = F)
d$Year = paste("19", d$year, sep = "")
d$Year = as.numeric(d$Year)
d$level = as.numeric(d$level)
# remove BGH outlier years
remove = c(1986, 1987, 1988)
d = subset(d, !Year %in% remove)
# total pay = salary + bonus
d$salary[is.na(d$salary)] = 0
d$bonus[is.na(d$bonus)] = 0
d$pay = d$salary + d$bonus
d = d[pay != 0]
d = d[order(Year, pay)]
pay.norm = d[, pay/mean(pay), by = Year]
d$pay.norm = pay.norm$V1
d1 = data.table(log.norm.pay = log(d$pay.norm) , level = d$level, edyrs = d$edyrs, age = d$age, yratco = d$yratco)
d1 = na.omit(d1)
# multivariate regression on all human capital
k.regress = lm(d1$log.norm.pay ~ d1$edyrs + d1$age +  d1$yratco)
# partial correlation
pcor(d1)
p = pcor(d1)
p = pcor(d1)
p = p$estimate
p
p = p$estimate %>% data.table()
p = pcor(d1)
p = p$estimate %>% data.table()
View(p)
source('~/Desktop/human_capital/Supplementary Material/Data/Case Studies/BGH/BGH_human_capital.R')
source('~/Desktop/human_capital/Supplementary Material/Data/Case Studies/BGH/BGH_human_capital.R')
source('~/Desktop/human_capital/Supplementary Material/Data/Case Studies/BGH/BGH_human_capital.R')
View(d1)
source('~/Desktop/human_capital/Supplementary Material/Data/Case Studies/BGH/BGH_human_capital.R')
library(data.table)
library(magrittr)
library(ppcor)
d = fread("BGH.csv", stringsAsFactors = F)
d$Year = paste("19", d$year, sep = "")
d$Year = as.numeric(d$Year)
d$level = as.numeric(d$level)
# remove BGH outlier years
remove = c(1986, 1987, 1988)
d = subset(d, !Year %in% remove)
# total pay = salary + bonus
d$salary[is.na(d$salary)] = 0
d$bonus[is.na(d$bonus)] = 0
d$pay = d$salary + d$bonus
d = d[pay != 0]
d = d[order(Year, pay)]
pay.norm = d[, pay/mean(pay), by = Year]
d$pay.norm = pay.norm$V1
plot(d$rating, d$pay)
plot(d$rating, d$pay, log = "y")
hist(d$rating)
library(data.table)
library(magrittr)
library(ppcor)
d = fread("BGH.csv", stringsAsFactors = F)
d$Year = paste("19", d$year, sep = "")
d$Year = as.numeric(d$Year)
d$level = as.numeric(d$level)
# remove BGH outlier years
remove = c(1986, 1987, 1988)
d = subset(d, !Year %in% remove)
unique(d$year)
sort(unique(d$year))
library(data.table)
library(magrittr)
library(ppcor)
d = fread("BGH.csv", stringsAsFactors = F)
d$Year = paste("19", d$year, sep = "")
d$Year = as.numeric(d$Year)
d$level = as.numeric(d$level)
# remove BGH outlier years
remove = c(1986, 1987, 1988)
d = subset(d, !Year %in% remove)
# total pay = salary + bonus
d$salary[is.na(d$salary)] = 0
d$bonus[is.na(d$bonus)] = 0
d$pay = d$salary + d$bonus
d = d[pay != 0]
d = d[order(Year, pay)]
pay.norm = d[, pay/mean(pay), by = Year]
d$pay.norm = pay.norm$V1
d1 = data.table(log.norm.pay = log(d$pay.norm) , level = d$level, edyrs = d$edyrs, age = d$age, yratco = d$yratco)
d1 = na.omit(d1)
# partial correlation
p = pcor(d1)
p = p$estimate %>% data.table()
fwrite(p, "BGH_partial_cor.csv" )
# multivariate regression on all human capital
k.regress = lm(d1$log.norm.pay ~ d1$edyrs + d1$age +  d1$yratco)
a = coef(k.regress)
