##############################
# Table 1
##############################

rm (list = ls())

library(foreign)
library(Hmisc)
library(plm)
library(stargazer)
library(effects)
library(ggplot2)
library(xtable)

# Load data
load("aggregate_data_2016aug15.RData")

# Explore data
dim(d)
names(d)
head(d)

# Histogram for outcome
qplot(d$volinc, geom="histogram", binwidth = 0.025) + labs(x="Change in Incumbent Vote Share", y="Frequency") +  theme_bw()  

# New variables
d$ageofdemocracy[d$country=="Nicaragua1990"] = 1 # 1 instead of 0 to compute log
d$log_inflation_lagged = log(d$inflation_lagged)
d$log_reserves_lagged = log(d$reserves_lagged)
d$log_ageofdemocracy = log(d$ageofdemocracy)

d$cleanelec = d$cleanelec*100
d$polcorruption = d$polcorruption*100

# New data sets 
d_hyper = d[d$inflation_lagged < 150,]
d$country[d$inflation_lagged > 150]

d_arg = d[d$country != "Argentina1995" & d$country != "Argentina1999",] 

d_growth = d[abs(d$gdpgrowth_lagged) < 10,]
d$country[(d$gdpgrowth_lagged) > 10]
d$country[(d$gdpgrowth_lagged) < -10]

#######################
# Descriptive
#######################

vars = c("volinc",
         "volb",
         "gdpgrowth_lagged",
         "inflation_lagged",
         "reserves_lagged",
         "ideologypres",
         "ageofdemocracy",
         "enp_con_lagged",
         "presvote_lagged",
         "ballotage",
         "reelection",
         "conc_elec",
         "comp_vote",
         "cleanelec",
         "partbarriers",
         "polcorruption",
         "partylinks")

d2 = d[vars] 
names(d2)

stargazer(d2,
          type = "text", 
          title="Descriptive statistics aggregate data", 
          digits=2, 
          rownames=FALSE, 
          align=TRUE, 
          omit.stat=c("LL","ser","f"), 
          covariate.labels=c("Change in Incumbent Vote Share",
                             "Type B Volatility",
                             "Lagged GDP Growth",
                             "Lagged log Inflation",
                             "Lagged log Reserves",
                             "Ideology Presidential Party",
                             "Log Age of Democracy",
                             "Effective Number of Parties",
                             "Lagged Incumbent Vote Share",
                             "Second Round",
                             "Reelection",
                             "Concurrent Election",
                             "Compulsory Vote",
                             "Clean Elections",
                             "Barriers to Parties",
                             "Political Corruption",
                             "Party Linkages"),
          no.space=TRUE,  
          multicolumn = TRUE,
          table.placement = "H")

#############################
# Regressions
#############################

# economic variables
t1 = lm(volinc ~ gdpgrowth_lagged  + log_inflation_lagged + log_reserves_lagged  +
                 fcountry + fdecade, data=d)                              
summary(t1)

# main specification
t2 = lm(volinc ~ gdpgrowth_lagged  + log_inflation_lagged + log_reserves_lagged  +
                 ideologypres + log_ageofdemocracy + enp_con_lagged + presvote_lagged +         
                 ballotage + reelection + conc_elec + comp_vote + 
                 cleanelec + partbarriers + polcorruption + partylinks +
                 fcountry + fdecade, data=d)                              
summary(t2)

# no hyper inflation
t3 = lm(volinc ~ gdpgrowth_lagged  + log_inflation_lagged + log_reserves_lagged + 
          ideologypres + log_ageofdemocracy + enp_con_lagged + presvote_lagged +          
          ballotage + reelection + conc_elec + comp_vote + 
          cleanelec + partbarriers + polcorruption + partylinks + 
          fcountry + fdecade, data=d_hyper)                              
summary(t3)

# no argentina 1990
t4 = lm(volinc ~ gdpgrowth_lagged  + log_inflation_lagged + log_reserves_lagged + 
          ideologypres + log_ageofdemocracy + enp_con_lagged + presvote_lagged +          
          ballotage + reelection + conc_elec + comp_vote + 
          cleanelec + partbarriers + polcorruption + partylinks + 
          fcountry + fdecade, data=d_arg)                              
summary(t4)

# no abs(growth) > 10
t5 = lm(volinc ~ gdpgrowth_lagged  + log_inflation_lagged + log_reserves_lagged +
          ideologypres + log_ageofdemocracy + enp_con_lagged + presvote_lagged +          
          ballotage + reelection + conc_elec + comp_vote + 
          cleanelec + partbarriers + polcorruption + partylinks + 
          fcountry + fdecade, data=d_growth)                              
summary(t5)

# volatility B
t6 = lm(volb ~ gdpgrowth_lagged  + log_inflation_lagged + log_reserves_lagged + 
               ideologypres + log_ageofdemocracy + enp_con_lagged + presvote_lagged +              
               ballotage + reelection + conc_elec + comp_vote + 
               cleanelec + partbarriers + polcorruption + partylinks + 
               fcountry + fdecade, data=d)                               
summary(t6)

###########
# Tables
###########

stargazer(t1,t2,t3,t4,t5,t6,title="Regression results aggregate level",
          align=TRUE, 
          type = "text",
          omit.stat=c("LL","ser","f"), 
          covariate.labels=c("Lagged GDP Growth",
                             "Lagged log Inflation",
                             "Lagged log Reserves",
                             "Ideology Presidential Party",
                             "Log Age of Democracy",
                             "Effective Number of Parties",
                             "Lagged Incumbent Vote Share",
                             "Second Round",
                             "Reelection",
                             "Concurrent Election",
                             "Compulsory Vote",
                             "Clean Elections",
                             "Barriers to Parties",
                             "Political Corruption",
                             "Party Linkages"),
          no.space=TRUE,  
          multicolumn = TRUE,
          table.placement = "H",
          #single.row = TRUE,
          column.labels   = c("Change in Incumbent Vote Share", "Type B Volatility"),
          column.separate = c(5, 1),
          dep.var.caption  = "Dependent Variables:",
          dep.var.labels.include = FALSE,
          omit = c("fcountry","fdecade"),
          add.lines = list(c("Country Fixed effects", "Yes", "Yes", "Yes", "Yes", "Yes", "Yes"),
                           c("Decade Fixed effects", "Yes", "Yes", "Yes", "Yes", "Yes", "Yes"),
                           c("Sample", "All countries", "All countries", "No hyperinflation", "No argentina '90", "No extreme growth", "All countries")))