-
Notifications
You must be signed in to change notification settings - Fork 4
/
KFW_CrossSectionResults_Median_Pre2001_MatchIt.R
198 lines (158 loc) · 9.69 KB
/
KFW_CrossSectionResults_Median_Pre2001_MatchIt.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
#-----------------------
#KFW 1 Cross-Sectional Model
#Treatment: Early Demarcated through PPTAL by April 2001 (vs. demarcated after April 2001)
#Outcome: Median NDVI change in level from 1995-2001, Median NDVI change in level from 2001-2010
#Using MatchIt package instead of SCI
#-----------------------
library(devtools)
devtools::install_github("itpir/SCI@master")
library(SCI)
library(stargazer)
loadLibs()
library(MatchIt)
library(rgeos)
library(maptools)
library(rgdal)
library(sp)
shpfile = "processed_data/kfw_analysis_inputs.shp"
dta_Shp = readShapePoly(shpfile)
#-------------------------------------------------
#-------------------------------------------------
#Pre-processing to create cross-sectional variable summaries
#-------------------------------------------------
#-------------------------------------------------
#Calculate NDVI Trends
dta_Shp$pre_trend_NDVI_mean <- timeRangeTrend(dta_Shp,"MeanL_[0-9][0-9][0-9][0-9]",1982,1995,"id")
dta_Shp$pre_trend_NDVI_max <- timeRangeTrend(dta_Shp,"MaxL_[0-9][0-9][0-9][0-9]",1982,1995,"id")
dta_Shp$pre_trend_NDVI_med <- timeRangeTrend(dta_Shp,"MedL_[0-9][0-9][0-9][0-9]",1982,1995,"id")
#NDVI Max Trends for 1995-2001
dta_Shp$post_trend_NDVI_95_01 <- timeRangeTrend(dta_Shp,"MaxL_[0-9][0-9][0-9][0-9]",1995,2001,"id")
dta_Shp@data["NDVILevelChange_95_01"] <- dta_Shp$MaxL_2001 - dta_Shp$MaxL_1995
#dta_Shp@data["NDVIslopeChange_95_01"] <- dta_Shp@data["post_trend_NDVI_95_01"] - dta_Shp@data["pre_trend_NDVI_max"]
#NDVI Max Trends for 2001-2010
dta_Shp$post_trend_NDVI_01_10 <- timeRangeTrend(dta_Shp,"MaxL_[0-9][0-9][0-9][0-9]",2001,2010,"id")
dta_Shp@data["NDVILevelChange_01_10"] <- dta_Shp$MaxL_2010 - dta_Shp$MaxL_2001
#dta_Shp@data["NDVIslopeChange_01_10"] <- dta_Shp@data["post_trend_NDVI_01_10"] - dta_Shp@data["pre_trend_NDVI_max"]
#NDVI Med Trends for 1995-2001
dta_Shp$post_trend_NDVI_95_01_Med <- timeRangeTrend(dta_Shp,"MedL_[0-9][0-9][0-9][0-9]",1995,2001,"id")
dta_Shp@data["NDVILevelChange_95_01_Med"] <- dta_Shp$MedL_2001 - dta_Shp$MedL_1995
#NDVI Med Trends for 2001-2010
dta_Shp$post_trend_NDVI_01_10_Med <- timeRangeTrend(dta_Shp,"MedL_[0-9][0-9][0-9][0-9]",2001,2010,"id")
dta_Shp@data["NDVILevelChange_01_10_Med"] <- dta_Shp$MedL_2010 - dta_Shp$MedL_2001
#Calculate Temp and Precip Pre and Post Trends
dta_Shp$pre_trend_temp_mean <- timeRangeTrend(dta_Shp,"MeanT_[0-9][0-9][0-9][0-9]",1982,1995,"id")
dta_Shp$pre_trend_temp_max <- timeRangeTrend(dta_Shp,"MaxT_[0-9][0-9][0-9][0-9]",1982,1995,"id")
dta_Shp$pre_trend_temp_min <- timeRangeTrend(dta_Shp,"MinT_[0-9][0-9][0-9][0-9]",1982,1995,"id")
dta_Shp$post_trend_temp_95_01 <- timeRangeTrend(dta_Shp,"MeanT_[0-9][0-9][0-9][0-9]",1995,2001,"id")
dta_Shp$post_trend_temp_01_10 <- timeRangeTrend(dta_Shp,"MeanT_[0-9][0-9][0-9][0-9]",2001,2010,"id")
dta_Shp$pre_trend_precip_mean <- timeRangeTrend(dta_Shp,"MeanP_[0-9][0-9][0-9][0-9]",1982,1995,"id")
dta_Shp$pre_trend_precip_max <- timeRangeTrend(dta_Shp,"MaxP_[0-9][0-9][0-9][0-9]",1982,1995,"id")
dta_Shp$pre_trend_precip_min <- timeRangeTrend(dta_Shp,"MinP_[0-9][0-9][0-9][0-9]",1982,1995,"id")
dta_Shp$post_trend_precip_95_01 <- timeRangeTrend(dta_Shp,"MeanP_[0-9][0-9][0-9][0-9]",1995,2001,"id")
dta_Shp$post_trend_precip_01_10 <- timeRangeTrend(dta_Shp,"MeanP_[0-9][0-9][0-9][0-9]",2001,2010,"id")
#-------------------------------------------------
#-------------------------------------------------
#Define the Treatment Variable and Population
#-------------------------------------------------
#-------------------------------------------------
#Make a binary to test treatment..
dta_Shp@data["TrtBin"] <- 0
dta_Shp@data$TrtBin[dta_Shp@data$demend_y <= 2001] <- 1
dta_Shp@data$TrtBin[(dta_Shp@data$demend_m > 4) & (dta_Shp@data$demend_y==2001)] <- 0
#Remove units that did not ever receive any treatment (within-sample test)
dta_Shp@data$NA_check <- 0
dta_Shp@data$NA_check[is.na(dta_Shp@data$demend_y)] <- 1
int_Shp <- dta_Shp[dta_Shp@data$NA_check != 1,]
dta_Shp <- int_Shp
table(dta_Shp@data$TrtBin)
#--------------------------
#Matching, with replacement
#--------------------------
#identify vars needed for psm model and analytic models (only these will be included in new matched dataset)
aVars <- c("reu_id","UF","TrtBin", "terrai_are","Pop_1990","Pop_2000", "MeanT_1995","MeanT_2001", "pre_trend_temp_mean",
"pre_trend_temp_min", "pre_trend_temp_max", "MeanP_1995", "MeanP_2001","pre_trend_precip_min",
"pre_trend_NDVI_mean", "pre_trend_NDVI_max","pre_trend_NDVI_med",
"MaxL_1995","MedL_1995","NDVILevelChange_95_01","NDVILevelChange_01_10",
"NDVILevelChange_95_01_Med","NDVILevelChange_01_10_Med",
"Slope","Elevation","Riv_Dist","Road_dist",
"pre_trend_precip_mean", "pre_trend_precip_max",
"post_trend_temp_95_01","post_trend_temp_01_10",
"post_trend_precip_95_01","post_trend_precip_01_10")
#propensity score model
#replace=TRUE to match with replacement
#exact="UF" restricts matches to same Brazilian state and discard="both" must accompany it
#use resulting weights in models to account for matching with replacement
psmModel <- matchit(TrtBin ~ terrai_are + Pop_1990 + MeanT_1995 + pre_trend_temp_mean + pre_trend_temp_min +
pre_trend_temp_max + MeanP_1995 + pre_trend_precip_min +
pre_trend_NDVI_mean + pre_trend_NDVI_max + pre_trend_NDVI_med+
Slope + Elevation + MedL_1995 + Riv_Dist + Road_dist +
pre_trend_precip_mean + pre_trend_precip_max,
data=dta_Shp@data[aVars],
method="nearest",replace=TRUE, exact="UF",discard="both")
print(summary(psmModel))
#create new dataset with matches
model_data<-match.data(psmModel)
#check states that were dropped out
summary(model_data$UF)
##create standardized dataset to produce standardized coefficients in models that are easy to output
#identify vars for inclusion in standardized dataset
#include all numeric variables from psm equation and that will be included in models
#exclude any id fields and weights created from matchit
# stvars <- c("TrtBin", "terrai_are","Pop_1990","Pop_2000" ,"MeanT_1995","MeanT_2001", "pre_trend_temp_mean",
# "pre_trend_temp_min", "pre_trend_temp_max", "MeanP_1995","MeanP_2001", "pre_trend_precip_min",
# "pre_trend_NDVI_mean", "pre_trend_NDVI_max","pre_trend_NDVI_med",
# "NDVILevelChange_95_01_Med","NDVILevelChange_01_10_Med",
# "MaxL_1995","MedL_1995",
# "Slope","Elevation","Riv_Dist","Road_dist",
# "pre_trend_precip_mean", "pre_trend_precip_max",
# "NDVILevelChange_95_01","NDVILevelChange_01_10","post_trend_temp_95_01","post_trend_temp_01_10",
# "post_trend_precip_95_01","post_trend_precip_01_10")
#
# model_data_st<- model_data
# model_data_st[stvars]<-lapply(model_data_st[stvars],scale)
#--------------
#Analytic Models
#--------------
##Early Models, Outcome: 1995-2001 Median
#Create dataset with some common names for stargazer
model_data_early <- model_data
colnames(model_data_early)[(colnames(model_data_early)=="Pop_1990")] <- "Pop_B"
colnames(model_data_early)[(colnames(model_data_early)=="MeanT_1995")] <- "MeanT_B"
colnames(model_data_early)[(colnames(model_data_early)=="MeanP_1995")] <- "MeanP_B"
colnames(model_data_early)[(colnames(model_data_early)=="post_trend_temp_95_01")] <- "post_trend_temp"
colnames(model_data_early)[(colnames(model_data_early)=="post_trend_precip_95_01")] <- "post_trend_precip"
#ModelEarly2, treatment effect + weights, 1995-2001 Median
ModelEarly2_Med <- lm(NDVILevelChange_95_01_Med ~ TrtBin, data=model_data, weights=(weights))
#ModelEarly3, treatment effect + weights + covars, 1995-2001 Median
ModelEarly3_Med<-lm(NDVILevelChange_95_01_Med~TrtBin +pre_trend_NDVI_med + MedL_1995 + terrai_are+Pop_B+
MeanT_B + post_trend_temp+
MeanP_B + post_trend_precip+
Slope+Elevation+Riv_Dist+Road_dist,
data=model_data_early,
weights=(weights))
##Late Models
#Create dataset with some common names for stargazer
model_data_late<-model_data
colnames(model_data_late)[(colnames(model_data_late)=="Pop_2000")] <- "Pop_B"
colnames(model_data_late)[(colnames(model_data_late)=="MeanT_2001")] <- "MeanT_B"
colnames(model_data_late)[(colnames(model_data_late)=="MeanP_2001")] <- "MeanP_B"
colnames(model_data_late)[(colnames(model_data_late)=="post_trend_temp_01_10")] <- "post_trend_temp"
colnames(model_data_late)[(colnames(model_data_late)=="post_trend_precip_01_10")] <- "post_trend_precip"
#ModelLate, treatment effect + weights + covars, 2001-2010 Median
ModelLate_Med<-lm(NDVILevelChange_01_10_Med~TrtBin+ pre_trend_NDVI_med + MedL_1995+terrai_are+Pop_B+
MeanT_B+post_trend_temp+
MeanP_B + post_trend_precip+
Slope + Elevation + Riv_Dist + Road_dist,
data=model_data_late,
weights=(weights))
#-------------
#Stargazer
#-------------
stargazer(ModelEarly2_Med, ModelEarly3_Med,ModelLate_Med,
keep=c("TrtBin", "pre_trend_NDVI_med","MedL_1995", "terrai_are","Pop_B","MeanT_B","post_trend_temp","MeanP_B",
"post_trend_precip","Slope","Elevation","Riv_Dist","Road_dist"),
covariate.labels=c("Treatment (Early Demarcation)", "Pre-Trend NDVI", "Baseline NDVI", "Area (hectares)","Baseline Population Density",
"Baseline Temperature", "Temperature Trends","Baseline Precipitation","Precipitation Trends",
"Slope", "Elevation", "Distance to River", "Distance to Road"),
dep.var.labels=c("Median NDVI 1995-2010"," Median NDVI 2001-2010"),
title="Regression Results", type="html", omit.stat=c("f","ser"), align=TRUE)