Task 1

The codes for download Fame-French

# base <- 
# "http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/ftp/"
# 
# factor <- "Global_3_Factors"
# 
# format<- "_CSV.zip"
# 
# full_url <-glue(base,factor,format,sep ="")
# 
# # download the file
# download.file(full_url,temp,quiet = TRUE)

pathDataSet <- "./datasets/"
# corredsponding to the stock of Microsfot ^MSFT
myData <- as_tibble(read.csv(paste0(pathDataSet,"MSFT.csv")))
myData$Date <- as.Date(myData$Date)

FFData <- read_csv(file = paste0(pathDataSet,"F-F_Research_Data_Factors_daily.CSV"),
                   skip=5,
                   col_types = cols(
                     `Mkt-RF` = col_double(),
                     SMB = col_double(),
                     HML = col_double(),
                     RF2 = col_double()))

## Warning: The following named parsers don't match the column names: Mkt-RF,
## SMB, HML, RF2

## Warning: 2 parsing failures.
##   row      col  expected                           actual                                             file
## 24537 19260701 a double  Copyright 2019 Kenneth R. French './datasets/F-F_Research_Data_Factors_daily.CSV'
## 24537 NA       5 columns 1 columns                        './datasets/F-F_Research_Data_Factors_daily.CSV'

# rename the column names
colnames(FFData) <- c("Date","Mkt.RF","SMB","HML","RF")
FFData$Date <- ymd(parse_date_time(FFData$Date, "%Y%m%d"))
# FFData$Date <- as.Date(FFData$Date,
#                        format = "%yy%mm%dd",
#                        origin = "19820509")
########## > Error in `$<-.data.frame`(`*tmp*`, Date, value = c(NA, NA, NA, NA, NA, : replacement has 1214 rows, data has 1212

# change the colnames
head(myData)

## # A tibble: 6 x 7
##   Date        Open  High   Low Close Adj.Close   Volume
##   <date>     <dbl> <dbl> <dbl> <dbl>     <dbl>    <int>
## 1 2011-01-03  28.0  28.2  27.9  28.0      22.6 53443800
## 2 2011-01-04  27.9  28.2  27.8  28.1      22.6 54405600
## 3 2011-01-05  27.9  28.0  27.8  28        22.6 58998700
## 4 2011-01-06  28.0  28.8  27.9  28.8      23.2 88026300
## 5 2011-01-07  28.6  28.7  28.2  28.6      23.1 73762000
## 6 2011-01-10  28.3  28.4  28.0  28.2      22.8 57573600

myPrice <- myData$`Adj.Close`
# tail(FFData)

logReturn <- function(price){
  
}

# cat(colnames(myData),"\n\n",colnames(FFData))



myDataNew <- merge(x = FFData,
                   y = myData,
                   by = "Date")
# clean the unecessary OCHL data excep adj.close price
dropList <- c("Open","High","Low","Close","Volume","Adj.Close")
myDataNew <- myDataNew %>% 
  mutate(logReturn = log(Adj.Close) - log(dplyr::lag(Adj.Close)))

# omit the missing values
myDataNew <- na.omit(myDataNew)
myDataNew <- select(myDataNew,-dropList)
rbind(head(myDataNew), tail(myDataNew))

##            Date Mkt.RF   SMB   HML   RF    logReturn
## 2    2011-01-04  -0.26 -1.38  0.11 0.00  0.003923630
## 3    2011-01-05   0.59  0.61  0.14 0.00 -0.003209246
## 4    2011-01-06  -0.15 -0.06 -0.33 0.00  0.028865002
## 5    2011-01-07  -0.21 -0.25 -0.30 0.00 -0.007662793
## 6    2011-01-10  -0.02  0.55 -0.16 0.00 -0.013376030
## 7    2011-01-11   0.39  0.12  0.22 0.00 -0.003905257
## 2006 2018-12-20  -1.62 -0.24  0.78 0.01 -0.021248303
## 2007 2018-12-21  -2.17 -0.57  0.83 0.01 -0.032845762
## 2008 2018-12-24  -2.55  0.98 -0.45 0.01 -0.042634859
## 2009 2018-12-26   5.06 -0.01 -1.07 0.01  0.066077834
## 2010 2018-12-27   0.78 -0.68 -0.09 0.01  0.006146463
## 2011 2018-12-28  -0.03  0.75  0.24 0.01 -0.007838476

# capm regression
capmRegres <- "logReturn - RF ~ Mkt.RF"
capmLM <- lm(formula = capmRegres,
           data=myDataNew)


# Fama-French regression
ffRegres <- "logReturn - RF ~ Mkt.RF + SMB + HML"
ffLM <- lm(formula = ffRegres,
           data=myDataNew)

library(summarytools)

## Registered S3 method overwritten by 'pryr':
##   method      from
##   print.bytes Rcpp

## Warning in fun(libname, pkgname): couldn't connect to display ":0"

## system has no X11 capabilities, therefore only ascii graphs will be produced by dfSummary()

## 
## Attaching package: 'summarytools'

## The following objects are masked from 'package:Hmisc':
## 
##     label, label<-

## The following objects are masked from 'package:xtable':
## 
##     label, label<-

## The following object is masked from 'package:tibble':
## 
##     view

displayDataFrame <- as.data.frame(summarytools::descr(myDataNew))
displayDataFrame[14,] <- as.integer(displayDataFrame[14,])
knitr::kable(displayDataFrame[c(1,2,3,5,7,14),],
             caption = "Table 1: Summary Statistics (2011-2018)")

Table 1: Summary Statistics (2011-2018)
	HML	logReturn	Mkt.RF	RF	SMB
Mean	-0.0063184	0.0007372	0.0441194	1.407e-03	-0.0056020
Std.Dev	0.4843078	0.0145751	0.9458480	2.450e-03	0.5160314
Min	-1.6900000	-0.1210331	-6.9700000	0.000e+00	-1.9200000
Median	-0.0300000	0.0003824	0.0600000	0.000e+00	-0.0100000
Max	2.3900000	0.0994128	5.0600000	1.000e-02	3.6000000
N.Valid	2010.0000000	2010.0000000	2010.0000000	2.010e+03	2010.0000000

summ(capmLM)

Observations	2010
Dependent variable	logReturn - RF
Type	OLS linear regression

F(1,2008)	1634.67
R²	0.45
Adj. R²	0.45

	Est.	S.E.	t val.	p
(Intercept)	-0.00	0.00	-4.62	0.00
Mkt.RF	0.01	0.00	40.43	0.00
Standard errors: OLS

summ(ffLM)

Observations	2010
Dependent variable	logReturn - RF
Type	OLS linear regression

F(3,2006)	640.22
R²	0.49
Adj. R²	0.49

	Est.	S.E.	t val.	p
(Intercept)	-0.00	0.00	-5.18	0.00
Mkt.RF	0.01	0.00	42.99	0.00
SMB	-0.00	0.00	-9.07	0.00
HML	-0.00	0.00	-9.87	0.00
Standard errors: OLS

Task Two

Read the data

firmFinData <- read.csv(paste0(pathDataSet,"firm financial dataset.csv"))
head(firmFinData) %>% knitr::kable()

gvkey	datadate	fyear i	ndfmt c	onsol p	opsrc d	atafmt c	usip c	onm c	urcd	at	bkvlps	ceq	che	csho	dlc	dltt	ebitda	invt	ppent	sale c	ostat	prcc_f	sic
1000	19701231	1970	INDL	C	D	STD	32102	A & E PLASTIK PAK INC	USD	33.450	4.3107	10.544	1.660	2.446	12.378	0.917	5.386	10.199	8.876	45.335	I	10.000	3089
1000	19711231	1971	INDL	C	D	STD	32102	A & E PLASTIK PAK INC	USD	29.330	2.7987	8.381	2.557	2.995	2.857	10.318	2.512	8.187	7.639	47.033	I	5.750	3089
1000	19721231	1972	INDL	C	D	STD	32102	A & E PLASTIK PAK INC	USD	19.907	2.4194	7.021	2.027	2.902	0.000	7.000	4.109	4.743	7.013	34.362	I	5.125	3089
1000	19731231	1973	INDL	C	D	STD	32102	A & E PLASTIK PAK INC	USD	21.771	3.0165	8.567	1.357	2.840	0.000	7.000	4.514	5.683	7.408	37.750	I	1.750	3089
1000	19741231	1974	INDL	C	D	STD	32102	A & E PLASTIK PAK INC	USD	25.638	4.5781	9.843	1.338	2.150	0.500	7.000	4.820	7.177	5.691	50.325	I	2.125	3089
1000	19751231	1975	INDL	C	D	STD	32102	A & E PLASTIK PAK INC	USD	23.905	4.8808	10.240	1.679	2.098	1.251	4.592	6.364	6.227	7.263	51.192	I	4.375	3089

Data wrangling: combine these variables with the original dataset and clean up data

# focus on the year b/t 1975~2015
firmFinData <- firmFinData %>%
  dplyr::filter(.,fyear >= 1975) %>%
  dplyr::filter(.,fyear <= 2015)
# keep observations with SIC codes betwen 2000~3999(manufacturing firms)
firmFinData <- firmFinData %>%
  dplyr::filter(.,sic >= 2000) %>%
  dplyr::filter(.,sic <= 3999)

firmFinDataExtra <- firmFinData %>% 
  # Market leverage:
  mutate(mktLeverage = (dltt+dlc)/( dltt+dlc+ csho*prcc_f),
         # Cash holding:
         cashHolding = che/at,
         # Tangibility: 
         tangibility = (invt+ppent)/at,
         # Market to book ratio:
         mktoBookRatio  = prcc_f/bkvlps,
         # Size:
         size = log(sale),
         # Profitability
         profitability=ebitda/at)

## Warning in log(sale): NaNs produced

# remove all the inf and -inf values 
# leave only the finite number
#firmFinDataExtra <- firmFinData %>% 
#  mutate_if(is.numeric, list(~na_if(., Inf))) %>% 
#  mutate_if(is.numeric, list(~na_if(., -Inf)))
#firmFinDataExtra[is.infinite(firmFinDataExtra)]<-NA # default method not implemented for type 'list
firmFinDataExtra[mapply(is.infinite, firmFinDataExtra)] <- NA

# exclude negative market leverage, tangibility, and market to book ratio measures;
#has.neg <- apply(firmFinDataExtra, 1, function(row) any(row < 0))
#firmFinDataExtra[has.neg] <- NA
######################################
#negIntoNA <- function(xx){
#  negRet <- ifelse(is.numeric(xx)<0,NA,xx)
#  return(negRet)
#}
#firmFinDataExtra <- mapply(negIntoNA, firmFinDataExtra)
firmFinDataExtra <- firmFinDataExtra %>%
  # exclude negative market leverage measure
  dplyr::filter(.,mktLeverage > 0) %>%
  # exclude negative tangibility measure
  dplyr::filter(.,tangibility > 0) %>%
  # exclude negative market to book ratio measures
  dplyr::filter(.,mktoBookRatio > 0) 

####################################
## clean out the missing value
####################################
firmFinDataExtra <- na.omit(firmFinDataExtra)

# firmFinDataExtra %>% summary()  # why there are NaNs produced
## summary(firmFinDataExtra)

Run ols

# The dependent variable(result) is Book leverage. The independent variables (causes) are Tangibility, Market to book ratio, Size and Profitability. 

regrRawData <- firmFinDataExtra %>%
  group_by(fyear) %>%
  summarise(avgMktLeverage=mean(mktLeverage),
            avgTangibility=mean(tangibility),
            avgMktoBookRatio=mean(mktoBookRatio),
            avgSize=mean(size),
            avgProfitablity=mean(profitability))

olsExpr <- "avgMktLeverage ~avgTangibility + avgMktoBookRatio + avgSize + avgProfitablity"
leverageLM <- lm(formula = olsExpr,
                 data = regrRawData)
#summary(leverageLM)
#stargazer(leverageLM,type="latex")  # produce tex codes for the table
#xtable(anova(leverageLM), type="LaTex")
summ(leverageLM,
     title = "Linear Regression")

Observations	41
Dependent variable	avgMktLeverage
Type	OLS linear regression

F(4,36)	38.91
R²	0.81
Adj. R²	0.79

	Est.	S.E.	t val.	p
(Intercept)	-0.90	0.22	-4.05	0.00
avgTangibility	1.61	0.30	5.37	0.00
avgMktoBookRatio	-0.00	0.00	-1.33	0.19
avgSize	0.10	0.02	4.77	0.00
avgProfitablity	-1.03	0.34	-3.00	0.00
Standard errors: OLS

Capital Structure Theory

The leverage is positive assoicative with tangibility.

Whether U.S. corporations hold too much cash for a while

medianCashHoldingOverall <- firmFinDataExtra %>%
  group_by(fyear) %>%
  summarise(medianCashHolding = median(cashHolding))

meanCashHoldingOverall <- firmFinDataExtra %>%
  group_by(fyear) %>%
  summarise(meanCashHolding = mean(cashHolding))

plot(medianCashHoldingOverall,
     main = "Ups and Downs of Median Cash Holding over years",
     xlab = "Fiscal Year",
     ylab = "Median Cash Holdings")

plot(meanCashHoldingOverall,
     main = "Ups and Downs of Mean Cash Holding over years",
     xlab = "Fiscal Year",
     ylab = "Mean Cash Holdings")

The cash holding goes up and down cyclically by year. But overall, there is trend that companies tend to hold more cash in my sample. The phenomena is more obvious in recent years.

There are many factors contributing the holding cashes more and more. Following are some guess:

From Richard Koo’s notion of “balance recession”, corporations prefer to holds cashes and do not invest and R&D more after The Great Recession experienced by the world in year 2008 (see Richard koo’s book) even in a (historically lowest )zero-interest rate and quantitative easing environment.
After several rounds of QE by Fed, big companies like Apple and Amazon and manufacturing firms in our sample accumulate too many cashes.
Also, unlike tech firms, a manufacturing firm few investment opportuniy which yields good return.

These factors over all make manufacturing hold more and more cashes at hand.