用R进行面板数据（Panel Data）分析

library(foreign) library(plm)

BankData <- read.dta("http://www.obhrm.net/data/franzese.dta", convert.factors=FALSE)

olsmodel2 <- lm(inf ~ cbilag1 + glag1 + tlag1 + flag1 + infalag1 + ulag1 + cwb, data=BankData, na.action=na.omit) summary(olsmodel2)

pdim(BankData, index=c("ctry", "year"))

1. 1. Fixed effects model

pmodel4 <- plm(inf ~ cbilag1 + glag1 + tlag1 + flag1 + infalag1 + ulag1 + cwb, data=BankData, index=c("ctry", "year"), na.action=na.omit, model="within") summary(pmodel4)

1. 1. F test for fixed effects

pFtest(pmodel4,olsmodel2)

1. 1. Random effects model

pmodel5 <- plm(inf ~ cbilag1 + glag1 + tlag1 + flag1 + infalag1 + ulag1 + cwb, data=BankData, index=c("ctry", "year"), na.action=na.omit, model="random")

1. 1. Crash because of negative estimated variance of individual effects

1. 1. Use different method of calculating variance of random effects
   2. random.method options only work on balanced panels

pmodel5 <- plm(inf ~ cbilag1 + glag1 + tlag1 + flag1 + infalag1 + ulag1 + cwb, data=BankData, index=c("ctry", "year"), na.action=na.omit, model="random", random.method="walhus") summary(pmodel5)

1. 1. Breusch-Pagan test

plmtest(pmodel5, effect="individual", type="bp")

1. 1. Hausman test

phtest(pmodel4, pmodel5)

代码来源：http://www.polsci.ucsb.edu/faculty/glasgow/ps207/ps207_class1.r