Time-Series-and-Panel-Data-Analysis-in-R-A-Coursera-Project/Panel_Data_Analysis_with_R.R at main · jaishreejoshita/Time-Series-and-Panel-Data-Analysis-in-R-A-Coursera-Project · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
## Panel Data Analysis with R

## 1. Set working directory

## 2. Load the library

library(plm)
library(knitr)
library(broom)
library(tidyverse)
library(stargazer)
library(lmtest)
library(gplots)

## 3. Import Dataset (STATA version)
# Rental Data from Wooldridge
#  Indexed by city and year

rental_p <- pdata.frame(RENTAL, index = c("city", "year"))

## 4. Display the data

str(RENTAL)
str(rental_p)
head(RENTAL)
head(rental_p)

#  Model lrent~y90+lpop+lavginc+pctstu

## 5. OLS using lm

ols <- lm(lrent~y90+lpop+lavginc+pctstu, data =RENTAL)
summary(ols)


## 6. OLS using plm

pooled <- plm(lrent~y90+lpop+lavginc+pctstu, data=RENTAL, model="pooling", index=c("city", "year"))
summary(pooled)

#OR use this format

pooled2 <- plm(lrent~y90+lpop+lavginc+pctstu, data=rental_p, model="pooling")
summary(pooled2)


## Results table

stargazer(pooled, type='text')

## 7. Test for heteroscedasticity
res <- residuals(ols)
yhat <- fitted(ols)
plot(RENTAL$cpctstu, res, xlab = "%students", ylab = "Residuals")
plot(yhat, res, xlab = "Fitted_Values", ylab = "Residuals")

## 8. Fixed Effects
# Includes within-entity effects

fe <- plm(lrent~y90+lpop+lavginc+pctstu, data=rental_p, model="within")
summary(fe)

# Show fixed effects for all 64 cities

fixef(fe)

## 9. Test for FE vs OLS
# Ho: OLS is better than FE, reject at p < 0.05

pFtest(fe, ols)

stargazer(fe, type = 'text')

## 10. Random Effects

# Includes both the within-entity and between-entity effects

re <- plm(lrent~y90+lpop+lavginc+pctstu, data=rental_p, model="random")
summary(re)

## 11. FE VS RE

## Hausman Test Ho: RE is preferred, Ha: FE is preferred (p < 0.05)

phtest(fe, re)

# Beautify / Tabulate result

kable(tidy(phtest(fe, re)), caption = "Hausman endogeneity test")

## 12. Breusch Pagan Lagrange Multiplier Test Ho: No panel effect, i.e., OLS is better. Ha: RE is better at p <0.05

plmtest(pooled, type =c('bp'))

#plmtest(ols, type=c("bp")) - use PLM package


## 13. Test for cross-sectional dependence [NOTE: Suitable only for macro panels with long time series] [Not suitable for RENTAL dataset]
# Breusch-Pagan LM test of independence and Pasaran CD test, Ho: There is no cross-sectional dependence

pcdtest(fe, test =("lm"))
pcdtest(fe, test =c("cd"))

## 14. Testing for serial correlation [NOTE: Suitable only for macro panels with long time series] [Not suitable for RENTAL dataset]
# Ho: There is no serial correlation

pbgtest(fe)

## 15. Breusch - Pagan test for heteroscedasticity Ho: Homoscedasticity Ha: Heteroscedasticity

bptest(lrent~y90+lpop+lavginc+pctstu+factor(city), data=rental_p, studentize = F)