第一讲 导论与经济增长

樊潇彦 复旦大学经济学院 中级宏观

Ch01_导论与经济增长.rar3855.4kB

1. 准备工作

# 安装程序包
install.packages("dplyr")          # 读取和整理数据
install.packages("tidyr")
install.packages("data.table")
install.packages("foreign")
install.packages("readstata13")
install.packages("haven")
install.packages("ggplot2")        # 作图
install.packages("ggrepel")  
install.packages("dygraphs")
install.packages("plotrix")
install.packages("lubridate")      # 处理时序数据
install.packages("zoo")
install.packages("mFilter")
# 调用程序包
library(dplyr)
library(tidyr)
library(data.table)  
library(foreign)
library(readstata13)
library(haven)       # 有空字符时，用write_dta生成.dta
library(ggplot2)
library(ggrepel)
library(dygraphs)    # 动态作图
library(plotrix)     # 双轴作图
library(lubridate)   # 可以用year()、month()等提取日期
library(zoo)         # 处理时序数据
library(mFilter)     # HP滤波
# 设定工作目录
setwd("D:\\...")     # 设定工作目录
rm(list=ls())        # 清内存

2. 总量趋势与结构

# 数据读取和整理
gdp=fread("GDP.csv",header=T)                # 读取数据
str(gdp)                                     # 查看数据结构
gdp_var=gdp%>%select(varid,varname,var)      # 指标一览
gdp=gdp%>%
  select(-varid,-varname)%>%                 # 选择除varid和varname之外的所有指标
  gather(year,value,-var)%>%                 # 将列指标1952-2015按年整合
  mutate(year=as.numeric(year))%>%           # 将字符型指标转换为数字型
  filter(!is.na(value))%>%                   # 选取value不为空值的样本
  arrange(var,year)%>%                       # 按var和year排序
  spread(var,value)                          # 按var展开为列指标
head(gdp)                                    # 查看前6个样本

2.1 数据分解

# 计算对数实际人均GDP并分解
gdppa=gdp%>%
  select(year,gdppa,gdppaid)%>%
  mutate(gdppaid=ifelse(year==1952,100,gdppaid)) %>%
  mutate(realgdppa_1952=gdp$gdppa[gdp$year==1952]*cumprod(gdppaid/100))%>%
  mutate(realgdppa_2015=realgdppa_1952/realgdppa_1952[gdp$year==2015]*gdppa[gdp$year==2015])
gdppa_hp=hpfilter(log(gdppa$realgdppa_2015),drift=F,    # 用hp滤波提取趋势和周期
                  freq=100,type="lambda")               # 年度数据lambda=100，季度和月度分别为1600和14400
gdppa_hp=data.frame(year=gdppa$year, trend=gdppa_hp$trend,cycle=gdppa_hp$cycle)
twoord.plot(lx = gdppa_hp$year, ly = gdppa_hp$trend, 
            rx = gdppa_hp$year,  ry = gdppa_hp$cycle, 
            main = '对数实际人均GDP：1952-2015', xlab = '', 
            ylab = '趋势', rylab = '周期', 
            type = c('line','line')) 

2.2 结构分析

# 生产结构和支出结构
gdpstr=gdp%>%
  mutate(sec1=gdp1/gdp)%>%
  mutate(sec2=gdp2/gdp)%>%
  mutate(sec3=gdp3/gdp)%>%
  mutate(con=gdpexpcpri/gdpexp)%>%
  mutate(inv=gdpexpi/gdpexp)%>%
  mutate(gov=gdpexpcgov/gdpexp)%>%
  select(year,sec1:gov)%>%
  gather(var,share,-year)
ggplot(gdpstr[gdpstr$var %in% c("sec1","sec2","sec3"),],aes(year,share,color=var))+ 
  geom_line(size=1)+ labs(title="生产法GDP结构",x="",y="")+
  scale_colour_discrete(labels = c('第一产业','第二产业','第三产业'))+
  guides(color = guide_legend(title = NULL)) +  theme_bw()+ theme(legend.position = 'bottom') 
ggplot(gdpstr[gdpstr$var %in% c("con","inv","gov"),],aes(year, share, color=var))+ 
  geom_line(size=1)+ labs(title="支出法GDP结构",x="",y="")+ 
  scale_colour_discrete(labels = c('居民消费','政府支出','资本形成'))+
  guides(color = guide_legend(title = NULL)) +  theme_bw()+ 
  theme(legend.position = 'bottom') 
# 收入结构
gdpinc=fread("GDP_inc.csv",header=T)
gdpinc_var=gdpinc%>%  select(varid,reg,var)
reginfo=data.frame(reg=unique(gdpinc_var$reg))
str(reginfo)
reginfo=reginfo%>%
  mutate(reg=as.character(reg))%>%
  mutate(regno=31:1)%>%arrange(regno)%>%
  mutate(reg6=rep(c("华北","东北","华东","华南","西南","西北"),times=c(5,3,7,6,5,5)))%>%
  mutate(reg6no=rep(c(10,20,30,40,50,60),times=c(5,3,7,6,5,5)))
gdpinc=gdpinc%>%
  select(-varid)%>%
  gather(year,value,-reg,-var)%>%
  spread(var,value)
gdpreg_inc=gdpinc%>%                               # 计算分省的收入结构
  mutate(dep=`固定资产折旧`/`收入法`)%>%
  mutate(lab=`劳动者报酬`/`收入法`)%>%
  mutate(tax=`生产税净额`/`收入法`)%>%
  mutate(pro=`营业盈余`/`收入法`)%>%
  select(reg,year,dep:pro)%>%
  gather(var,share,-reg,-year)
a=gdpinc%>%                                        # 计算全国的
  group_by(year)%>%
  mutate(dep=sum(`固定资产折旧`)/sum(`收入法`))%>%
  mutate(lab=sum(`劳动者报酬`)/sum(`收入法`))%>%
  mutate(tax=sum(`生产税净额`)/sum(`收入法`))%>%
  mutate(pro=sum(`营业盈余`)/sum(`收入法`))%>%
  select(reg,year,dep:pro)%>%
  filter(reg=="北京")%>%
  mutate(reg="全国")%>%
  gather(var,share,-reg,-year)
a=data.frame(a)                                    # 将数据格式设定为data.frame
gdpinc=rbind(gdpreg_inc,a)                         # 省与全国合并
gdpinc=gdpinc%>%
  left_join(reginfo,by="reg")%>%
  mutate(regno=ifelse(reg=="全国",0,regno))%>%
  mutate(reg6no=ifelse(reg=="全国",0,reg6no))%>%
  mutate(reg6=ifelse(reg=="全国",reg,reg6))%>%
  arrange(var,year,regno)
ggplot(gdpinc[gdpinc$regno==0,],aes(year,share,group=var,color=var))+geom_line(size=1)+ 
  labs(title="收入法GDP结构",x="",y="")+ 
  scale_colour_manual(values= c("dep"="grey","lab"="blue","pro"="red","tax"="purple"),
                        labels = c('固定资产折旧','劳动者报酬','营业盈余','生产税净额'))+
  guides(color = guide_legend(title = NULL)) +  
  theme_bw()+ theme(legend.position = 'bottom') 

3. 区域增长与结构

gdpreg=fread("GDP_reg.csv",header=T)
gdpreg_var=gdpreg%>%select(varid,var,reg)

3.1 收敛

gdppareg=gdpreg%>%
  filter(var=="GDP指数")%>%
  select(-varid,-var)%>%
  gather(year,idx,-reg)%>%
  arrange(reg,year)%>%
  group_by(reg)%>%
  mutate(gdpidx=cumprod(idx/100))
gdppa1978=data.frame(regno=1:31,reg=gdpreg$reg[gdpreg$var=="人均GDP"],
                     gdppa=gdpreg$`1978`[gdpreg$var=="人均GDP"])
a=merge(gdppareg,gdppa1978,by="reg",all.x=T)%>%
  arrange(regno,year)%>% 
  filter(year==1978)%>%
  rename(gdpidx1978=gdpidx)%>%
  rename(gdppa1978=gdppa)%>%
  select(reg,regno,gdpidx1978,gdppa1978)
gdppareg=merge(gdppareg,a,by="reg",all.x=T)
gdppareg=gdppareg%>%
  mutate(gdppa=gdpidx/gdpidx1978*gdppa1978)%>%
  select(year,regno,reg,gdppa)%>%
  group_by(year)%>%
  mutate(gdppa_rel=gdppa/max(gdppa))%>%
  arrange(year,regno)
table(gdppareg$reg[gdppareg$gdppa_rel==1]) # 检查每年人均GDP最高的地区，全是上海
g_level=gdppareg%>%
  filter(year %in% c(1978,2015))%>%
  select(-gdppa)%>%
  spread(year,gdppa_rel)%>%
  mutate(rank78=n()+1-min_rank(`1978`))%>%
  mutate(rank15=n()+1-min_rank(`2015`))%>%
  mutate(top5=ifelse(rank78<=6 | rank15<=6, reg, ""))
ggplot(g_level,aes(`1978`,`2015`,color=regno))+geom_point()+
  geom_abline(intercept = 0, slope = 1, color="red")+
  geom_text(aes(`1978`,`2015`, label=top5, hjust=0.5, vjust=-0.5))+
  labs(title = "各地区实际人均GDP相对于上海的变化",x="1978年相对于上海",y="2015年相对于上海")+
  theme_bw()
g_gr=gdppareg%>%
  filter(year %in% c(1978,1996,1997,2015))%>%
  select(-gdppa_rel)%>%
  spread(year,gdppa)%>%
  mutate(gr7896=exp(log(`1996`/`1978`)/18)-1)%>%
  mutate(gr9715=exp(log(`2015`/`1997`)/18)-1)%>%
  mutate(rank1=n()+1-min_rank(gr7896))%>%
  mutate(rank2=n()+1-min_rank(gr9715))%>%
  mutate(top5=ifelse(rank1<=5 | rank2<=5, reg, ""))%>%
  select(regno,reg,gr7896,gr9715,top5)
ggplot(g_gr,aes(gr7896,gr9715,color=regno))+geom_point()+
  geom_abline(intercept = 0, slope = 1, color="red")+
  geom_text_repel(aes(label = reg))+
  ylim(0.06,0.15) +xlim(0.06,0.15)+
  labs(title = "各地区实际人均GDP增长率的变化",x="1978-1996年",y="1997-2015年")+
  theme_bw()
g_conv=gdppareg%>%
  filter(year %in% c(1978,2015))%>%
  select(-gdppa_rel)%>%
  spread(year,gdppa)%>%
  mutate(gr=exp(log(`2015`/`1978`)/37)-1)%>%
  mutate(rank=n()+1-min_rank(gr))%>%
  mutate(top5=ifelse(rank<=5, reg, ""))%>%
  select(regno,reg,`1978`,gr,top5)
ggplot(g_conv,aes(log(`1978`),gr,color=regno))+geom_point()+geom_smooth(method="lm")+
  geom_text_repel(aes(label = reg))+
  labs(title = "地区收敛（含直辖市）",x="1978年对数实际GDP",y="1978-2015年年均增长率")+
  theme_bw()
ggplot(g_conv[!g_conv$reg %in% c("北京","天津","上海","重庆"),],
       aes(log(`1978`),gr,color=regno))+geom_point()+geom_smooth(method="lm")+
  geom_text_repel(aes(label = reg))+
  labs(title = "地区收敛（不含直辖市）",x="1978年对数实际GDP",y="1978-2015年年均增长率")+
  theme_bw()
reg=lm(gr~log(`1978`),data=g_conv)  # OLS估计
summary(reg) 

3.2 结构

# 产出
gdpstrreg1=gdpreg%>%
  select(-varid)%>%
  filter(var %in% c("GDP","第一产业","第二产业","第三产业"))%>%
  gather(year,value,-reg,-var)%>%
  filter(year>=1978)%>%
  spread(var,value)%>%
  mutate(Sector1=`第一产业`/GDP)%>%
  mutate(Sector2=`第二产业`/GDP)%>%
  mutate(Sector3=`第三产业`/GDP)%>%
  select(year,reg,Sector1:Sector3)%>%
  gather(var,share,-reg,-year)%>%
  left_join(reginfo,by="reg")%>%
  mutate(year=as.numeric(year))%>%
  arrange(regno,year,var)
ggplot(gdpstrreg1,aes(year,share,color=reg6))+geom_point()+facet_wrap(~var,ncol=3)+
  scale_x_continuous(breaks=seq(1980, 2015, 5))+
  guides(color = guide_legend(title = NULL)) +  
  labs(title="各地区产业结构变动：1978-2015",x="",y="")+
  theme_bw()+ theme(legend.position = 'bottom') 
# 支出
gdpstrreg2=gdpreg%>%
  select(-varid)%>%
  filter(var %in% c("支出法","最终消费支出","资本形成总额"))%>%
  gather(year,value,-reg,-var)%>%
  filter(year>=1993 & year<=2014)%>%
  spread(var,value)%>%
  mutate(`最终消费`=`最终消费支出`/`支出法`)%>%
  mutate(`资本形成`=`资本形成总额`/`支出法`)%>%
  select(year,reg,`最终消费`,`资本形成`)%>%
  gather(var,share,-reg,-year)%>%
  left_join(reginfo,by="reg")%>%
  mutate(year=as.numeric(year))%>%
  arrange(regno,year,var)
ggplot(gdpstrreg2,aes(year,share,color=reg6))+geom_point()+facet_wrap(~var,ncol=2)+
  scale_x_continuous(breaks=seq(1995, 2015, 5))+
  guides(color = guide_legend(title = NULL))+   
  labs(title="各地区支出结构变动：1993-2014",x="",y="")+
  theme_bw()+ theme(legend.position = 'bottom') 
# 收入
gdpstrreg3=gdpinc%>%
  select(regno,year,var,share)%>%
  filter(regno>0)%>%
  spread(var,share)%>%
  rename(`固定资产折旧`=dep)%>%
  rename(`劳动者报酬`=lab)%>%
  rename(`营业盈余`=pro)%>%
  rename(`生产税净额`=tax)%>%
  gather(var,share,-regno,-year)%>%
  mutate(year=as.numeric(year))%>%
  left_join(reginfo,by="regno")
ggplot(gdpstrreg3,aes(year,share,color=reg6))+geom_point()+facet_wrap(~var,ncol=2)+
  scale_x_continuous(breaks=seq(1995, 2015, 5))+
  guides(color = guide_legend(title = NULL))+   
  labs(title="各地区收入结构变动：1993-2014",x="",y="")+
  theme_bw()+ theme(legend.position = 'bottom')