-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathmain_regressions.R
222 lines (153 loc) · 6.19 KB
/
main_regressions.R
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
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
# This script generates all the regressions and tables in the final paper
library(readr)
library(tidyverse)
library(lubridate)
library(stargazer)
library(fixest)
library(lfe)
library(data.table)
##############
## Load data
##############
df <-read_csv("merged_data0524.csv")
sumdf <- read_csv("summary.csv")
df <- df %>%
mutate(after = ifelse(date >ymd("2022-04-28"),1,0),
no_regional_info = ifelse(region_info,0,1),
week = floor_date(date, unit = "week"))
# summary stat
stargazer(as.data.table(df))
##############
## PART I: RDD
##############
# looking at the day before and after the policy change
df_day <- df %>%
filter(date > ymd("2022-04-26") & date < ymd("2022-04-30")) %>%
mutate(after = ifelse(date > ymd("2022-04-28"),1,0),
days_away = as.numeric(date - ymd("2022-04-28")))
df_day_plot <- df_day %>%
group_by(days_away)%>%
summarise(avg_score =mean(score, na.rm =T))
ggplot(df_day_plot, aes(x = days_away, y = avg_score))+
geom_point()+
geom_vline(xintercept = 0,color ="red")+
theme_bw()
# looking at the week before and after the policy change
df_week <- df %>%
filter(date > ymd("2022-04-20") & date < ymd("2022-05-06"))%>%
mutate(after = ifelse(date > ymd("2022-04-28"),1,0),
days_away = as.numeric(date - ymd("2022-04-28")))
df_week_plot <- df_week %>%
group_by(days_away)%>%
summarise(avg_score =mean(score, na.rm =T))
p1<-ggplot(df_week_plot, aes(x = days_away, y = avg_score))+
geom_point()+
geom_vline(xintercept = 0,color ="red")+
theme_bw()+
labs(x = "Days From IP Revealing",
y = "Average Sentiment Score")
ggsave("D:/ecma31320/rdd_illustration.png", plot=p1, width = 8, height = 4)
# looking at the month before and after the policy change
df_month <- df %>%
filter(date > ymd("2022-04-01") & date < ymd("2022-05-31"))%>%
mutate(after = ifelse(date > ymd("2022-04-28"),1,0),
days_away = as.numeric(date - ymd("2022-04-28")))
df_month_plot <- df_month %>%
group_by(days_away)%>%
summarise(avg_score =mean(score, na.rm =T))
ggplot(df_month_plot, aes(x = days_away, y = avg_score))+
geom_point()+
geom_vline(xintercept = 0,color ="red")+
theme_bw()
mod_rd1 <- lm(score ~ after + days_away, data = df_day)
summary(mod_rd1)
mod_rd2 <- lm(score ~ after + days_away, data = df_week)
summary(mod_rd2)
mod_rd3 <- lm(score ~ after + days_away + I(days_away^2), data = df_week)
summary(mod_rd3)
mod_rd4 <- lm(score ~ after + days_away, data = df_month)
summary(mod_rd4)
mod_rd5 <- lm(score ~ after + days_away + I(days_away^2), data = df_month)
summary(mod_rd5)
stargazer(mod_rd1, mod_rd2, mod_rd3, mod_rd4, mod_rd5)
##############
## PART II: DID
##############
df_with_control <- inner_join(sumdf, df)
# April 22 + May 22
df2 <- df_with_control %>%
filter(date > ymd("2022-03-31") & date < ymd("2022-05-31"))%>%
mutate(after = ifelse(date >ymd("2022-04-28"),1,0),
no_regional_info = ifelse(region_info,0,1))
# Apr 22 + Apr 23
df3 <- df_with_control %>%
filter(date < ymd("2022-04-29") | date > ymd("2022-05-31"))%>%
mutate(after = ifelse(date >ymd("2022-04-28"),1,0),
no_regional_info = ifelse(region_info,0,1))
mod01 <- lm(score~ after+no_regional_info + after*no_regional_info, data = df2)
mod02 <- lm(score~ after+no_regional_info + after*no_regional_info, data = df3)
summary(mod01)
summary(mod02)
mod03 <- lm(score~ after+no_regional_info + after*no_regional_info + rank, data = df2)
mod04 <- lm(score~ after+no_regional_info + after*no_regional_info + rank, data = df3)
summary(mod03)
summary(mod04)
stargazer(mod01, mod02, mod03, mod04, mod05, mod6)
stargazer(mod01, mod03, mod02, mod04, title="Regression Results",
column.labels=c("Apr 22 + May 22","Apr 22 + Apr 23"),
column.separate = c(2,2),
omit.stat="f", no.space=TRUE)
##############
## PART III: Robustness checks
##############
# RD with regional controls
mod_rd01 <- lm(score ~ after + days_away + no_regional_info + no_regional_info*after, data = df_day)
summary(mod_rd01)
mod_rd02 <- lm(score ~ after + days_away+ no_regional_info + no_regional_info*after, data = df_week)
summary(mod_rd02)
mod_rd03 <- lm(score ~ after + days_away + I(days_away^2)+ no_regional_info + no_regional_info*after, data = df_week)
summary(mod_rd03)
mod_rd04 <- lm(score ~ after + days_away+ no_regional_info + no_regional_info*after, data = df_month)
summary(mod_rd04)
mod_rd05 <- lm(score ~ after + days_away + I(days_away^2)+ no_regional_info + no_regional_info*after, data = df_month)
summary(mod_rd05)
stargazer(mod_rd01, mod_rd02, mod_rd03, mod_rd04, mod_rd05)
# rdd placebo test
df_week3 <- df %>%
filter(date > ymd("2023-04-15") & date < ymd("2023-04-30"))%>%
mutate(after = ifelse(date > ymd("2023-04-22"),1,0),
days_away = as.numeric(date - ymd("2023-04-22")))
mod_rd23 <- lm(score ~ after + days_away, data = df_week3)
summary(mod_rd23)
mod_rd24 <- lm(score ~ after + days_away + I(days_away^2), data = df_week3)
summary(mod_rd24)
mod_rd25 <- lm(score ~ after + days_away + no_regional_info + no_regional_info*after, data = df_week3)
summary(mod_rd25)
mod_rd26 <- lm(score ~ after + days_away + I(days_away^2)+ no_regional_info + no_regional_info*after, data = df_week3)
summary(mod_rd26)
stargazer(mod_rd23, mod_rd24, mod_rd25, mod_rd26)
# daily estimate
df22 <- df2 %>%
mutate(date = factor(date))
mod07 <- lm(score~ date+no_regional_info + date*no_regional_info + rank, data = df22)
summary(mod07)
mod07$coefficients[59:113]
dates <-unique(arrange(df2,date)$date)
dates <- dates[-1]
plotdf <- data.frame("date"= dates, "estimate" = mod07$coefficients[59:113])
ggplot(plotdf, aes(x = date,y = estimate))+
geom_vline(xintercept = ymd("2022-04-28"),color ="red")+
geom_point()+
geom_smooth(method = 'glm')+
theme_bw()+
labs(y = "Daily Estimate")
# clustered standard errors
clu01 <- felm(score~ after+no_regional_info + after*no_regional_info|0|0|week, data = df2)
clu02 <- felm(score~ after+no_regional_info + after*no_regional_info+rank|0|0|week, data = df2)
summary(clu01)
summary(clu02)
clu03 <- felm(score ~ after + days_away|0|0|week, data = df_week)
clu04 <- felm(score ~ after + days_away + I(days_away^2)|0|0|week, data = df_week)
summary(clu03)
summary(clu04)
stargazer(clu03, clu04, clu01, clu02)