#Downloading needed libraries
# library(tidyverse)
# library(broom)
# library(haven)
# library(AER)
# library(sandwich)
# library(lmtest)
# library(car)
# library(stargazer)
library(dplyr)
# library(ggplot2)
# library(GGally)
library(psych)
# library(corrplot)
library(readxl)
library(writexl)
library(kableExtra)
#
# #TS data
# library(quantmod)
# library(tseries)
# library(forecast)
#
# #TS graphs
# library(quantmod)
# library(tseries)
# library(forecast)
#
# library(TSstudio)
# library(reshape2)
# library(plotly)
#
# #Histograms
# library(hrbrthemes)
# library(viridis)
# library(forcats)
#
# #Spatial analysis
# # geo
# library(sp)
# library(spdep)
# library(rgdal)
# library(maptools)
# library(rgeos)
# library(sf)
# library(spatialreg)
#
# # mass retype()
# library(hablar)
#
# # multicollinearity test omcdiag imcdiag
# library(mctest)
#
# library(GGally)
#
# # Panel
#
# library(plm)
# library(lmtest)
# library(xtable)library(tidyverse)## Warning: пакет 'tidyverse' был собран под R версии 4.1.3## -- Attaching packages --------------------------------------- tidyverse 1.3.1 --## v ggplot2 3.3.5 v purrr 0.3.4
## v tibble 3.1.6 v stringr 1.4.0
## v tidyr 1.2.0 v forcats 0.5.1
## v readr 2.1.2## Warning: пакет 'tidyr' был собран под R версии 4.1.3## -- Conflicts ------------------------------------------ tidyverse_conflicts() --
## x ggplot2::%+%() masks psych::%+%()
## x ggplot2::alpha() masks psych::alpha()
## x dplyr::filter() masks stats::filter()
## x kableExtra::group_rows() masks dplyr::group_rows()
## x dplyr::lag() masks stats::lag()library(plm)##
## Присоединяю пакет: 'plm'## Следующие объекты скрыты от 'package:dplyr':
##
## between, lag, leadlibrary(sandwich)
library(lmtest)## Загрузка требуемого пакета: zoo##
## Присоединяю пакет: 'zoo'## Следующие объекты скрыты от 'package:base':
##
## as.Date, as.Date.numericlibrary(xtable)
library(stargazer)##
## Please cite as:## Hlavac, Marek (2022). stargazer: Well-Formatted Regression and Summary Statistics Tables.## R package version 5.2.3. https://CRAN.R-project.org/package=stargazerwide_data <- read_xlsx('C:/Users/Kolya/Documents/New_studies/Term_paper/Data/Raw data/Common_dataset_new.xlsx')
### LONG vs.WIDE format
long_data <- wide_data %>% pivot_longer(cols = !region,
names_to = c(".value", "year"),
names_pattern = "([A-Za-z]+)_(\\d+)")
# ### second variant - smth is wrong
# data<-reshape(wide_data, direction = 'long', idvar = 'region', varying = 2:ncol(wide_data), sep = '_')
#
# ### third variant
# datalong <- melt(wide_data, id.vars = 'region') #region var
# datalong <- cbind(datalong, colsplit(datalong$variable, '_', c('var', 'year')))#cells for variables
# datalong <- dcast(datalong, region + year ~ var, value.var = 'value') #assignment of panel structure# Descriptive statistics
number <- function(x, na.rm = TRUE){return(sum(!is.na(x)))}
#not good for many observations
desc <- long_data %>%
select(-region) %>%
# select(year, pop, crimes) %>%
group_by(year) %>%
summarise(across(where(is.numeric),
list(n = number, mean = mean, median = median, sd = sd, min = min, max = max),
na.rm = TRUE)) %>%
pivot_longer(!year, names_to = "name", values_to = "value") %>%
separate(name, c("variable", "statistic"), sep = "_") %>%
pivot_wider(names_from = statistic, values_from = value) %>%
arrange(variable, year) %>%
select(variable, year, n, mean, median, sd, min, max)
descdesc %>% xtable()#good for many observations - doesn't work
# desc_2 <- long_data %>% #beautiful table
# # select(totexp, posexp, suppins, phylim, actlim, totchr, age, female, income) %>%
# summarise(across(everything(),
# list(n = number, mean = mean, median = median, sd = sd, min = min, max = max),
# na.rm = TRUE)) %>%
# pivot_longer(everything(), names_to = "name", values_to = "value") %>%
# separate(name, c("variable", "statistic"), sep = "_") %>%
# pivot_wider(names_from = statistic, values_from = value) %>%
# arrange(variable) %>%
# select(variable, n, mean, median, sd, min, max)
long_data %>% summary()## region year educ GRP
## Length:948 Length:948 Min. : 13.0 Min. : 9034
## Class :character Class :character 1st Qu.: 280.0 1st Qu.: 132914
## Mode :character Mode :character Median : 366.0 Median : 275439
## Mean : 375.9 Mean : 611286
## 3rd Qu.: 449.5 3rd Qu.: 579415
## Max. :1254.0 Max. :16538190
## NA's :5
## LabPr Unemp Wage EmpPerWF
## Min. : 84.0 Min. : 0.800 Min. : 81.6 Min. :0.0002465
## 1st Qu.:100.9 1st Qu.: 5.100 1st Qu.: 98.7 1st Qu.:0.0087923
## Median :103.2 Median : 6.500 Median :103.0 Median :0.0129767
## Mean :103.2 Mean : 7.438 Mean :104.1 Mean :0.0146494
## 3rd Qu.:105.7 3rd Qu.: 8.400 3rd Qu.:108.9 3rd Qu.:0.0179141
## Max. :123.9 Max. :57.800 Max. :132.4 Max. :0.0997954
##
## Inv V Rv Uv
## Min. : 6525 Min. :0.8476 Min. :0.0000 Min. :0.8476
## 1st Qu.: 36619 1st Qu.:2.6594 1st Qu.:0.3174 1st Qu.:2.2831
## Median : 56129 Median :3.3077 Median :0.5030 Median :2.7238
## Mean : 73588 Mean :3.3179 Mean :0.5827 Mean :2.7352
## 3rd Qu.: 82885 3rd Qu.:3.8525 3rd Qu.:0.7372 3rd Qu.:3.1534
## Max. :623234 Max. :7.2797 Max. :4.7498 Max. :5.0489
## long_data %>% summary()## region year educ GRP
## Length:948 Length:948 Min. : 13.0 Min. : 9034
## Class :character Class :character 1st Qu.: 280.0 1st Qu.: 132914
## Mode :character Mode :character Median : 366.0 Median : 275439
## Mean : 375.9 Mean : 611286
## 3rd Qu.: 449.5 3rd Qu.: 579415
## Max. :1254.0 Max. :16538190
## NA's :5
## LabPr Unemp Wage EmpPerWF
## Min. : 84.0 Min. : 0.800 Min. : 81.6 Min. :0.0002465
## 1st Qu.:100.9 1st Qu.: 5.100 1st Qu.: 98.7 1st Qu.:0.0087923
## Median :103.2 Median : 6.500 Median :103.0 Median :0.0129767
## Mean :103.2 Mean : 7.438 Mean :104.1 Mean :0.0146494
## 3rd Qu.:105.7 3rd Qu.: 8.400 3rd Qu.:108.9 3rd Qu.:0.0179141
## Max. :123.9 Max. :57.800 Max. :132.4 Max. :0.0997954
##
## Inv V Rv Uv
## Min. : 6525 Min. :0.8476 Min. :0.0000 Min. :0.8476
## 1st Qu.: 36619 1st Qu.:2.6594 1st Qu.:0.3174 1st Qu.:2.2831
## Median : 56129 Median :3.3077 Median :0.5030 Median :2.7238
## Mean : 73588 Mean :3.3179 Mean :0.5827 Mean :2.7352
## 3rd Qu.: 82885 3rd Qu.:3.8525 3rd Qu.:0.7372 3rd Qu.:3.1534
## Max. :623234 Max. :7.2797 Max. :4.7498 Max. :5.0489
## tex_desc_stat <- long_data[, c(3:length(names(long_data)))] %>% describe() #good, but not LaTeX
tex_desc_stat <- tex_desc_stat[, c(3:5, 8:9 )]
tex_desc_stat %>% xtable()# kbl(tex_desc_stat, format = "text")#Specifications
###Pannel models
#Specifications V
grp_V <- log(GRP) ~ V + log(Inv) + educ + Wage
labpr_V <- LabPr ~ V + log(Inv) + educ + Wage
emp_V <- EmpPerWF ~ V + log(Inv) + educ + Wage
unemp_V <- Unemp ~ V + log(Inv) + educ + Wage
#Specifications RV, UV
grp_RV_UV <- log(GRP) ~ Rv + Uv + log(Inv) + educ + Wage
labpr_RV_UV <- LabPr ~ Rv + Uv + log(Inv) + educ + Wage
emp_RV_UV <- EmpPerWF ~ Rv + Uv + log(Inv) + educ + Wage
unemp_RV_UV <- Unemp ~ Rv + Uv + log(Inv) + educ + Wage
###Extra models with time-effects
#Specifications V, t
grp_V_t<- log(GRP) ~ V + log(Inv) + educ + Wage + factor(year)
labpr_V_t <- LabPr ~ V + log(Inv) + educ + Wage + factor(year)
emp_V_t <- EmpPerWF ~ V + log(Inv) + educ + Wage + factor(year)
unemp_V_t <- Unemp ~ V + log(Inv) + educ + Wage + factor(year)
#Specifications RV, UV, t
grp_RV_UV_t<- log(GRP) ~ Rv + Uv + log(Inv) + educ + Wage + factor(year)
labpr_RV_UV_t <- LabPr ~ Rv + Uv + log(Inv) + educ + Wage + factor(year)
emp_RV_UV_t <- EmpPerWF ~ Rv + Uv + log(Inv) + educ + Wage + factor(year)
unemp_RV_UV_t <- Unemp ~ Rv + Uv + log(Inv) + educ + Wage + factor(year)#Simple models for value-added growth V, without time-effects
# grp_V_time <- log(grp_V) ~ V + log(Inv) + educ + Wage + factor(year)
ols_plm_grp_V <- plm(grp_V,
data = long_data,
index = c("region", "year"),
model = "pooling")
cov_ols_plm_grp_V <- vcovHC(ols_plm_grp_V, type = "HC0")
se_ols_plm_grp_V <- sqrt(diag(cov_ols_plm_grp_V))
coeftest(ols_plm_grp_V, df = Inf, vcov = cov_ols_plm_grp_V)##
## z test of coefficients:
##
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 3.79025361 1.91630511 1.9779 0.0479404 *
## V 0.44536099 0.08196861 5.4333 5.532e-08 ***
## log(Inv) 0.84284763 0.14381909 5.8605 4.616e-09 ***
## educ 0.00138918 0.00093661 1.4832 0.1380202
## Wage -0.02371371 0.00711836 -3.3313 0.0008643 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1fe_ind_grp_V <- plm(grp_V,
data = long_data,
index = c("region", "year"),
effect = "individual",
model = "within")
cov_fe_ind_grp_V <- vcovHC(fe_ind_grp_V, type = "HC0")
se_fe_ind_grp_V <- sqrt(diag(cov_fe_ind_grp_V))
coeftest(fe_ind_grp_V, df = Inf, vcov = cov_fe_ind_grp_V)##
## z test of coefficients:
##
## Estimate Std. Error z value Pr(>|z|)
## V -0.02826253 0.01762951 -1.6031 0.1089
## log(Inv) 0.51869449 0.03001834 17.2793 < 2.2e-16 ***
## educ -0.00192108 0.00028119 -6.8320 8.372e-12 ***
## Wage -0.00955683 0.00109580 -8.7213 < 2.2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1# fixef(fe_ind_grp_V) #re_grp_V <- plm(grp_V,
data = long_data,
index = c("region", "year"),
model = "random")
cov_re_grp_V <- vcovHC(re_grp_V, type = "HC0")
se_re_grp_V <- sqrt(diag(cov_re_grp_V))
coeftest(re_grp_V, df = Inf, vcov = cov_re_grp_V)##
## z test of coefficients:
##
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 8.54160693 0.39062712 21.8664 < 2.2e-16 ***
## V -0.01925275 0.01725434 -1.1158 0.2645
## log(Inv) 0.52759795 0.03139337 16.8060 < 2.2e-16 ***
## educ -0.00185355 0.00028993 -6.3931 1.625e-10 ***
## Wage -0.00985371 0.00110186 -8.9428 < 2.2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Tests for model for value-added growth V, without time-effects
#Ftest - between fixed effects and pooled OLS
#H0 - pooled OLS are better than fixed effects (because all individual effects are the same)
#H1 - fixed effects are better than pooled OLS ones
pFtest(fe_ind_grp_V, ols_plm_grp_V)##
## F test for individual effects
##
## data: grp_V
## F = 415.29, df1 = 78, df2 = 860, p-value < 2.2e-16
## alternative hypothesis: significant effects#fixed effects are better, as expected#plmtest - between random effects and pooled OLS
#H0 - pooled OLS are better than random effects (because all random effects are the same - on penal effect)
#H1 - random effects are better than pooled OLS ones
plmtest(ols_plm_grp_V, type = c('bp'))##
## Lagrange Multiplier Test - (Breusch-Pagan) for unbalanced panels
##
## data: grp_V
## chisq = 3500.2, df = 1, p-value < 2.2e-16
## alternative hypothesis: significant effects#random effects are better, as expected#phtest - between fixed effects and random effects
#H0 - random effects are better than fixed effects (random coef. are more efficient)
#H1 - fixed effects are better than random effects (fixed coef. are more efficient)
phtest(fe_ind_grp_V, re_grp_V)##
## Hausman Test
##
## data: grp_V
## chisq = 19.137, df = 4, p-value = 0.0007387
## alternative hypothesis: one model is inconsistent#random effects are then fixed ones - unexpected
#fixed effects are better than random effectsstargazer(ols_plm_grp_V, fe_ind_grp_V, re_grp_V, se = list(se_ols_plm_grp_V, se_fe_ind_grp_V, se_re_grp_V), type = "text", omit = c('Constant'), keep.stat = c("n"), font.size = "tiny") #, font.size = "tiny"##
## ==========================================
## Dependent variable:
## -----------------------------
## log(GRP)
## (1) (2) (3)
## ------------------------------------------
## V 0.445*** -0.028 -0.019
## (0.082) (0.018) (0.017)
##
## log(Inv) 0.843*** 0.519*** 0.528***
## (0.144) (0.030) (0.031)
##
## educ 0.001 -0.002*** -0.002***
## (0.001) (0.0003) (0.0003)
##
## Wage -0.024*** -0.010*** -0.010***
## (0.007) (0.001) (0.001)
##
## ------------------------------------------
## Observations 943 943 943
## ==========================================
## Note: *p<0.1; **p<0.05; ***p<0.01#Extra models with time-effects, GRP, V
#extra
fe_ind_grp_V_t <- plm(grp_V_t,
data = long_data,
index = c("region", "year"),
effect = "twoways",
model = "within")
cov_fe_ind_grp_V_t <- vcovHC(fe_ind_grp_V_t, type = "HC0")
se_fe_ind_grp_V_t <- sqrt(diag(cov_fe_ind_grp_V_t))
coeftest(fe_ind_grp_V_t, df = Inf, vcov = cov_fe_ind_grp_V_t)##
## z test of coefficients:
##
## Estimate Std. Error z value Pr(>|z|)
## V -1.5075e-02 1.2860e-02 -1.1722 0.2411
## log(Inv) 1.3465e-01 2.7433e-02 4.9083 9.189e-07 ***
## educ 3.1351e-05 1.7391e-04 0.1803 0.8569
## Wage 8.8206e-05 6.4925e-04 0.1359 0.8919
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1# fixef(fe_ind_grp_V) ##extra
fe_ind_grp_V_t_i <- plm(grp_V_t,
data = long_data,
index = c("region", "year"),
effect = "individual",
model = "within")
cov_fe_ind_grp_V_t_i <- vcovHC(fe_ind_grp_V_t_i, type = "HC0")
se_fe_ind_grp_V_t_i <- sqrt(diag(cov_fe_ind_grp_V_t_i))
coeftest(fe_ind_grp_V_t_i, df = Inf, vcov = cov_fe_ind_grp_V_t_i)##
## z test of coefficients:
##
## Estimate Std. Error z value Pr(>|z|)
## V -1.5075e-02 1.2860e-02 -1.1722 0.2411
## log(Inv) 1.3465e-01 2.7433e-02 4.9083 9.189e-07 ***
## educ 3.1351e-05 1.7391e-04 0.1803 0.8569
## Wage 8.8206e-05 6.4925e-04 0.1359 0.8919
## factor(year)2007 1.7206e-01 1.5365e-02 11.1978 < 2.2e-16 ***
## factor(year)2008 3.2938e-01 2.1779e-02 15.1236 < 2.2e-16 ***
## factor(year)2009 3.3322e-01 2.3864e-02 13.9634 < 2.2e-16 ***
## factor(year)2010 4.7301e-01 2.8011e-02 16.8864 < 2.2e-16 ***
## factor(year)2011 6.3252e-01 3.4951e-02 18.0973 < 2.2e-16 ***
## factor(year)2012 7.1949e-01 3.8282e-02 18.7946 < 2.2e-16 ***
## factor(year)2013 7.9431e-01 4.1341e-02 19.2133 < 2.2e-16 ***
## factor(year)2014 8.8176e-01 4.4068e-02 20.0087 < 2.2e-16 ***
## factor(year)2015 9.8542e-01 4.5988e-02 21.4278 < 2.2e-16 ***
## factor(year)2016 1.1065e+00 5.0512e-02 21.9046 < 2.2e-16 ***
## factor(year)2017 1.1599e+00 5.2468e-02 22.1071 < 2.2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1# fixef(fe_ind_grp_V) #stargazer(fe_ind_grp_V_t, fe_ind_grp_V_t_i, se = list(se_fe_ind_grp_V_t, se_fe_ind_grp_V_t_i), type = "text", omit = c('Constant'), keep.stat = c("n"), font.size = "tiny") #, font.size = "tiny"##
## =============================================
## Dependent variable:
## ----------------------------
## log(GRP)
## (1) (2)
## ---------------------------------------------
## V -0.015 -0.015
## (0.013) (0.013)
##
## log(Inv) 0.135*** 0.135***
## (0.027) (0.027)
##
## educ 0.00003 0.00003
## (0.0002) (0.0002)
##
## Wage 0.0001 0.0001
## (0.001) (0.001)
##
## factor(year)2007 0.172***
## (0.015)
##
## factor(year)2008 0.329***
## (0.022)
##
## factor(year)2009 0.333***
## (0.024)
##
## factor(year)2010 0.473***
## (0.028)
##
## factor(year)2011 0.633***
## (0.035)
##
## factor(year)2012 0.719***
## (0.038)
##
## factor(year)2013 0.794***
## (0.041)
##
## factor(year)2014 0.882***
## (0.044)
##
## factor(year)2015 0.985***
## (0.046)
##
## factor(year)2016 1.106***
## (0.051)
##
## factor(year)2017 1.160***
## (0.052)
##
## ---------------------------------------------
## Observations 943 943
## =============================================
## Note: *p<0.1; **p<0.05; ***p<0.01#no changes#pFtest(fixed.time, fixed)
#H0 - time effects are unnecessary
pFtest(fe_ind_grp_V_t_i, fe_ind_grp_V)##
## F test for individual effects
##
## data: grp_V_t
## F = 174.24, df1 = 11, df2 = 849, p-value < 2.2e-16
## alternative hypothesis: significant effects# time effects are needed
#plmtest(fixed, c("time"), type=("bp"))
#H0 - time effects are unnecessary
plmtest(fe_ind_grp_V, c("time"), type=("bp"))##
## Lagrange Multiplier Test - time effects (Breusch-Pagan) for unbalanced
## panels
##
## data: grp_V
## chisq = 92.081, df = 1, p-value < 2.2e-16
## alternative hypothesis: significant effects# time effects are needed
#so, fe_ind_grp_V_t is the best model (twoways or individual- we are free to choose)#Simple models for value-added growth (RV, UV), without time-effects
ols_plm_grp_RV_UV <- plm(grp_RV_UV,
data = long_data,
index = c("region", "year"),
model = "pooling")
cov_ols_plm_grp_RV_UV <- vcovHC(ols_plm_grp_RV_UV, type = "HC0")
se_ols_plm_grp_RV_UV <- sqrt(diag(cov_ols_plm_grp_RV_UV))
coeftest(ols_plm_grp_RV_UV, df = Inf, vcov = cov_ols_plm_grp_RV_UV)##
## z test of coefficients:
##
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 3.49326554 1.92311226 1.8165 0.0692991 .
## Rv 0.11828061 0.13068814 0.9051 0.3654336
## Uv 0.65160115 0.14227131 4.5800 4.650e-06 ***
## log(Inv) 0.86977182 0.14140460 6.1509 7.702e-10 ***
## educ 0.00109896 0.00099407 1.1055 0.2689331
## Wage -0.02622851 0.00694625 -3.7759 0.0001594 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1fe_ind_grp_RV_UV <- plm(grp_RV_UV,
data = long_data,
index = c("region", "year"),
effect = "individual",
model = "within")
cov_fe_ind_grp_RV_UV <- vcovHC(fe_ind_grp_RV_UV, type = "HC0")
se_fe_ind_grp_RV_UV <- sqrt(diag(cov_fe_ind_grp_RV_UV))
coeftest(fe_ind_grp_RV_UV, df = Inf, vcov = cov_fe_ind_grp_RV_UV)##
## z test of coefficients:
##
## Estimate Std. Error z value Pr(>|z|)
## Rv 0.08323036 0.03724341 2.2348 0.025433 *
## Uv -0.09431496 0.03512445 -2.6852 0.007249 **
## log(Inv) 0.50744927 0.02934671 17.2915 < 2.2e-16 ***
## educ -0.00183886 0.00028028 -6.5607 5.355e-11 ***
## Wage -0.00901557 0.00105597 -8.5377 < 2.2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1# fixef(fe_ind_grp_RV_UV) #re_grp_RV_UV <- plm(grp_RV_UV,
data = long_data,
index = c("region", "year"),
model = "random")
cov_re_grp_RV_UV <- vcovHC(re_grp_RV_UV, type = "HC0")
se_re_grp_RV_UV <- sqrt(diag(cov_re_grp_RV_UV))
coeftest(re_grp_RV_UV, df = Inf, vcov = cov_re_grp_RV_UV)##
## z test of coefficients:
##
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 8.68711908 0.38741763 22.4231 < 2.2e-16 ***
## Rv 0.08724609 0.03551553 2.4566 0.01403 *
## Uv -0.08252838 0.03466634 -2.3806 0.01728 *
## log(Inv) 0.51678021 0.03052683 16.9287 < 2.2e-16 ***
## educ -0.00177547 0.00028932 -6.1367 8.425e-10 ***
## Wage -0.00933205 0.00105954 -8.8076 < 2.2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Tests for model for value-added growth (RV, UV), without time-effects
#Ftest - between fixed effects and pooled OLS
#H0 - pooled OLS are better than fixed effects (because all individual effects are the same)
#H1 - fixed effects are better than pooled OLS ones
pFtest(fe_ind_grp_RV_UV, ols_plm_grp_RV_UV)##
## F test for individual effects
##
## data: grp_RV_UV
## F = 416.3, df1 = 78, df2 = 859, p-value < 2.2e-16
## alternative hypothesis: significant effects#fixed effects are better, as expected#plmtest - between random effects and pooled OLS
#H0 - pooled OLS are better than random effects (because all random effects are the same - on penal effect)
#H1 - random effects are better than pooled OLS ones
plmtest(ols_plm_grp_RV_UV, type = c('bp'))##
## Lagrange Multiplier Test - (Breusch-Pagan) for unbalanced panels
##
## data: grp_RV_UV
## chisq = 3368.4, df = 1, p-value < 2.2e-16
## alternative hypothesis: significant effects#random effects are better, as expected#phtest - between fixed effects and random effects
#H0 - random effects are better than fixed effects (random coef. are more efficient)
#H1 - fixed effects are better than random effects (fixed coef. are more efficient)
phtest(fe_ind_grp_RV_UV, re_grp_RV_UV)##
## Hausman Test
##
## data: grp_RV_UV
## chisq = 18.563, df = 5, p-value = 0.002318
## alternative hypothesis: one model is inconsistent#random effects are then fixed ones - unexpected
#fixed effects are better than random effectsstargazer(ols_plm_grp_RV_UV, fe_ind_grp_RV_UV, re_grp_RV_UV, se = list(se_ols_plm_grp_RV_UV, se_fe_ind_grp_RV_UV, se_re_grp_RV_UV), type = "text", omit = c('Constant'), keep.stat = c("n"), font.size = "tiny") #, font.size = "tiny"##
## ==========================================
## Dependent variable:
## -----------------------------
## log(GRP)
## (1) (2) (3)
## ------------------------------------------
## Rv 0.118 0.083** 0.087**
## (0.131) (0.037) (0.036)
##
## Uv 0.652*** -0.094*** -0.083**
## (0.142) (0.035) (0.035)
##
## log(Inv) 0.870*** 0.507*** 0.517***
## (0.141) (0.029) (0.031)
##
## educ 0.001 -0.002*** -0.002***
## (0.001) (0.0003) (0.0003)
##
## Wage -0.026*** -0.009*** -0.009***
## (0.007) (0.001) (0.001)
##
## ------------------------------------------
## Observations 943 943 943
## ==========================================
## Note: *p<0.1; **p<0.05; ***p<0.01#Extra models with time-effects, GRP, RV, UV
#extra
fe_ind_grp_RV_UV_t <- plm(grp_RV_UV_t,
data = long_data,
index = c("region", "year"),
effect = "twoways",
model = "within")
cov_fe_ind_grp_RV_UV_t <- vcovHC(fe_ind_grp_RV_UV_t, type = "HC0")
se_fe_ind_grp_RV_UV_t <- sqrt(diag(cov_fe_ind_grp_RV_UV_t))
coeftest(fe_ind_grp_RV_UV_t, df = Inf, vcov = cov_fe_ind_grp_RV_UV_t)##
## z test of coefficients:
##
## Estimate Std. Error z value Pr(>|z|)
## Rv -1.5925e-02 2.8649e-02 -0.5559 0.5783
## Uv -1.4549e-02 1.9979e-02 -0.7282 0.4665
## log(Inv) 1.3465e-01 2.7423e-02 4.9100 9.109e-07 ***
## educ 3.1561e-05 1.7361e-04 0.1818 0.8557
## Wage 8.9917e-05 6.6014e-04 0.1362 0.8917
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1# fixef(fe_ind_grp_RV_UV) ##extra
fe_ind_grp_RV_UV_t_i <- plm(grp_RV_UV_t,
data = long_data,
index = c("region", "year"),
effect = "individual",
model = "within")
cov_fe_ind_grp_RV_UV_t_i <- vcovHC(fe_ind_grp_RV_UV_t_i, type = "HC0")
se_fe_ind_grp_RV_UV_t_i <- sqrt(diag(cov_fe_ind_grp_RV_UV_t_i))
coeftest(fe_ind_grp_RV_UV_t_i, df = Inf, vcov = cov_fe_ind_grp_RV_UV_t_i)##
## z test of coefficients:
##
## Estimate Std. Error z value Pr(>|z|)
## Rv -1.5925e-02 2.8649e-02 -0.5559 0.5783
## Uv -1.4549e-02 1.9979e-02 -0.7282 0.4665
## log(Inv) 1.3465e-01 2.7423e-02 4.9100 9.109e-07 ***
## educ 3.1561e-05 1.7361e-04 0.1818 0.8557
## Wage 8.9917e-05 6.6014e-04 0.1362 0.8917
## factor(year)2007 1.7202e-01 1.5464e-02 11.1239 < 2.2e-16 ***
## factor(year)2008 3.2934e-01 2.1829e-02 15.0875 < 2.2e-16 ***
## factor(year)2009 3.3334e-01 2.3929e-02 13.9304 < 2.2e-16 ***
## factor(year)2010 4.7305e-01 2.7943e-02 16.9288 < 2.2e-16 ***
## factor(year)2011 6.3275e-01 3.5155e-02 17.9988 < 2.2e-16 ***
## factor(year)2012 7.1970e-01 3.7929e-02 18.9750 < 2.2e-16 ***
## factor(year)2013 7.9450e-01 4.0740e-02 19.5018 < 2.2e-16 ***
## factor(year)2014 8.8201e-01 4.3465e-02 20.2923 < 2.2e-16 ***
## factor(year)2015 9.8573e-01 4.5603e-02 21.6155 < 2.2e-16 ***
## factor(year)2016 1.1068e+00 5.0007e-02 22.1335 < 2.2e-16 ***
## factor(year)2017 1.1603e+00 5.1912e-02 22.3513 < 2.2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1# fixef(fe_ind_grp_RV_UV) #stargazer(fe_ind_grp_RV_UV_t, fe_ind_grp_RV_UV_t_i, se = list(se_fe_ind_grp_RV_UV_t, se_fe_ind_grp_RV_UV_t_i), type = "text", omit = c('Constant'), keep.stat = c("n"), font.size = "tiny") #, font.size = "tiny"##
## =============================================
## Dependent variable:
## ----------------------------
## log(GRP)
## (1) (2)
## ---------------------------------------------
## Rv -0.016 -0.016
## (0.029) (0.029)
##
## Uv -0.015 -0.015
## (0.020) (0.020)
##
## log(Inv) 0.135*** 0.135***
## (0.027) (0.027)
##
## educ 0.00003 0.00003
## (0.0002) (0.0002)
##
## Wage 0.0001 0.0001
## (0.001) (0.001)
##
## factor(year)2007 0.172***
## (0.015)
##
## factor(year)2008 0.329***
## (0.022)
##
## factor(year)2009 0.333***
## (0.024)
##
## factor(year)2010 0.473***
## (0.028)
##
## factor(year)2011 0.633***
## (0.035)
##
## factor(year)2012 0.720***
## (0.038)
##
## factor(year)2013 0.795***
## (0.041)
##
## factor(year)2014 0.882***
## (0.043)
##
## factor(year)2015 0.986***
## (0.046)
##
## factor(year)2016 1.107***
## (0.050)
##
## factor(year)2017 1.160***
## (0.052)
##
## ---------------------------------------------
## Observations 943 943
## =============================================
## Note: *p<0.1; **p<0.05; ***p<0.01#no changes#pFtest(fixed.time, fixed)
#H0 - time effects are unnecessary
pFtest(fe_ind_grp_RV_UV_t_i, fe_ind_grp_RV_UV)##
## F test for individual effects
##
## data: grp_RV_UV_t
## F = 166.82, df1 = 11, df2 = 848, p-value < 2.2e-16
## alternative hypothesis: significant effects# time effects are needed
#plmtest(fixed, c("time"), type=("bp"))
#H0 - time effects are unnecessary
plmtest(fe_ind_grp_RV_UV, c("time"), type=("bp"))##
## Lagrange Multiplier Test - time effects (Breusch-Pagan) for unbalanced
## panels
##
## data: grp_RV_UV
## chisq = 100.51, df = 1, p-value < 2.2e-16
## alternative hypothesis: significant effects# time effects are needed
#so, fe_ind_grp_RV_UV_t is the best model (twoways or individual- we are free to choose)#Simple models for labour productivity V, without time-effects
ols_plm_labpr_V <- plm(labpr_V,
data = long_data,
index = c("region", "year"),
model = "pooling")
cov_ols_plm_labpr_V <- vcovHC(ols_plm_labpr_V, type = "HC0")
se_ols_plm_labpr_V <- sqrt(diag(cov_ols_plm_labpr_V))
coeftest(ols_plm_labpr_V, df = Inf, vcov = cov_ols_plm_labpr_V)##
## z test of coefficients:
##
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 71.56537228 4.05158322 17.6636 < 2e-16 ***
## V 0.32747799 0.18521070 1.7681 0.07704 .
## log(Inv) 0.31299588 0.24634909 1.2705 0.20389
## educ -0.00055153 0.00091585 -0.6022 0.54704
## Wage 0.26279519 0.02036891 12.9018 < 2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1fe_ind_labpr_V <- plm(labpr_V,
data = long_data,
index = c("region", "year"),
effect = "individual",
model = "within")
cov_fe_ind_labpr_V <- vcovHC(fe_ind_labpr_V, type = "HC0")
se_fe_ind_labpr_V <- sqrt(diag(cov_fe_ind_labpr_V))
coeftest(fe_ind_labpr_V, df = Inf, vcov = cov_fe_ind_labpr_V)##
## z test of coefficients:
##
## Estimate Std. Error z value Pr(>|z|)
## V 1.4289130 0.3345697 4.2709 1.947e-05 ***
## log(Inv) 0.6964646 0.4930316 1.4126 0.1578
## educ 0.0022717 0.0020270 1.1207 0.2624
## Wage 0.2354405 0.0291230 8.0843 6.250e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1# fixef(fe_ind_labpr_V) #re_labpr_V <- plm(labpr_V,
data = long_data,
index = c("region", "year"),
model = "random")
cov_re_labpr_V <- vcovHC(re_labpr_V, type = "HC0")
se_re_labpr_V <- sqrt(diag(cov_re_labpr_V))
coeftest(re_labpr_V, df = Inf, vcov = cov_re_labpr_V)##
## z test of coefficients:
##
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 71.37332895 4.19830217 17.0005 < 2e-16 ***
## V 0.42285094 0.19406768 2.1789 0.02934 *
## log(Inv) 0.31967281 0.25320279 1.2625 0.20676
## educ -0.00052371 0.00095752 -0.5469 0.58441
## Wage 0.26079756 0.02088377 12.4880 < 2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Tests for model for labour productivity V, without time-effects
#Ftest - between fixed effects and pooled OLS
#H0 - pooled OLS are better than fixed effects (because all individual effects are the same)
#H1 - fixed effects are better than pooled OLS ones
pFtest(fe_ind_labpr_V, ols_plm_labpr_V)##
## F test for individual effects
##
## data: labpr_V
## F = 1.6496, df1 = 78, df2 = 860, p-value = 0.0005703
## alternative hypothesis: significant effects#fixed effects are better, as expected#plmtest - between random effects and pooled OLS
#H0 - pooled OLS are better than random effects (because all random effects are the same - on penal effect)
#H1 - random effects are better than pooled OLS ones
plmtest(ols_plm_labpr_V, type = c('bp'))##
## Lagrange Multiplier Test - (Breusch-Pagan) for unbalanced panels
##
## data: labpr_V
## chisq = 5.6025, df = 1, p-value = 0.01793
## alternative hypothesis: significant effects#random effects are better, but not for sure#phtest - between fixed effects and random effects
#H0 - random effects are better than fixed effects (random coef. are more efficient)
#H1 - fixed effects are better than random effects (fixed coef. are more efficient)
phtest(fe_ind_labpr_V, re_labpr_V)##
## Hausman Test
##
## data: labpr_V
## chisq = 18.874, df = 4, p-value = 0.0008322
## alternative hypothesis: one model is inconsistent#random effects are then fixed ones - unexpected
#fixed effects are better than random effectsstargazer(ols_plm_labpr_V, fe_ind_labpr_V, re_labpr_V, se = list(se_ols_plm_labpr_V, se_fe_ind_labpr_V, se_re_labpr_V), type = "text", omit = c('Constant'), keep.stat = c("n"), font.size = "tiny") #, font.size = "tiny"##
## ==========================================
## Dependent variable:
## -----------------------------
## LabPr
## (1) (2) (3)
## ------------------------------------------
## V 0.327* 1.429*** 0.423**
## (0.185) (0.335) (0.194)
##
## log(Inv) 0.313 0.696 0.320
## (0.246) (0.493) (0.253)
##
## educ -0.001 0.002 -0.001
## (0.001) (0.002) (0.001)
##
## Wage 0.263*** 0.235*** 0.261***
## (0.020) (0.029) (0.021)
##
## ------------------------------------------
## Observations 943 943 943
## ==========================================
## Note: *p<0.1; **p<0.05; ***p<0.01#Extra models with time-effects, LabPr, V
#extra
fe_ind_labpr_V_t <- plm(labpr_V_t,
data = long_data,
index = c("region", "year"),
effect = "twoways",
model = "within")
cov_fe_ind_labpr_V_t <- vcovHC(fe_ind_labpr_V_t, type = "HC0")
se_fe_ind_labpr_V_t <- sqrt(diag(cov_fe_ind_labpr_V_t))
coeftest(fe_ind_labpr_V_t, df = Inf, vcov = cov_fe_ind_labpr_V_t)##
## z test of coefficients:
##
## Estimate Std. Error z value Pr(>|z|)
## V 0.04240686 0.35244760 0.1203 0.904229
## log(Inv) 2.38871058 0.57460650 4.1571 3.223e-05 ***
## educ 0.00067508 0.00275862 0.2447 0.806676
## Wage 0.08859887 0.03249204 2.7268 0.006395 **
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1# fixef(fe_ind_labpr_V) ##extra
fe_ind_labpr_V_t_i <- plm(labpr_V_t,
data = long_data,
index = c("region", "year"),
effect = "individual",
model = "within")
cov_fe_ind_labpr_V_t_i <- vcovHC(fe_ind_labpr_V_t_i, type = "HC0")
se_fe_ind_labpr_V_t_i <- sqrt(diag(cov_fe_ind_labpr_V_t_i))
coeftest(fe_ind_labpr_V_t_i, df = Inf, vcov = cov_fe_ind_labpr_V_t_i)##
## z test of coefficients:
##
## Estimate Std. Error z value Pr(>|z|)
## V 0.04240686 0.35244760 0.1203 0.904229
## log(Inv) 2.38871058 0.57460650 4.1571 3.223e-05 ***
## educ 0.00067508 0.00275862 0.2447 0.806676
## Wage 0.08859887 0.03249204 2.7268 0.006395 **
## factor(year)2007 0.65066374 0.53406386 1.2183 0.223100
## factor(year)2008 -3.09888938 0.60790207 -5.0977 3.438e-07 ***
## factor(year)2009 -8.68869374 0.88373777 -9.8318 < 2.2e-16 ***
## factor(year)2010 -3.35542115 0.65000047 -5.1622 2.441e-07 ***
## factor(year)2011 -1.78934048 0.84549057 -2.1163 0.034316 *
## factor(year)2012 -4.85969134 0.80684756 -6.0231 1.711e-09 ***
## factor(year)2013 -5.69473355 0.93111669 -6.1160 9.594e-10 ***
## factor(year)2014 -5.03152341 1.01797700 -4.9427 7.706e-07 ***
## factor(year)2015 -6.54828492 0.99398798 -6.5879 4.461e-11 ***
## factor(year)2016 -6.44513062 0.99438505 -6.4815 9.080e-11 ***
## factor(year)2017 -5.47892463 0.94687522 -5.7863 7.194e-09 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1# fixef(fe_ind_labpr_V) #stargazer(fe_ind_labpr_V_t, fe_ind_labpr_V_t_i, se = list(se_fe_ind_labpr_V_t, se_fe_ind_labpr_V_t_i), type = "text", omit = c('Constant'), keep.stat = c("n"), font.size = "tiny") #, font.size = "tiny"##
## =============================================
## Dependent variable:
## ----------------------------
## LabPr
## (1) (2)
## ---------------------------------------------
## V 0.042 0.042
## (0.352) (0.352)
##
## log(Inv) 2.389*** 2.389***
## (0.575) (0.575)
##
## educ 0.001 0.001
## (0.003) (0.003)
##
## Wage 0.089*** 0.089***
## (0.032) (0.032)
##
## factor(year)2007 0.651
## (0.534)
##
## factor(year)2008 -3.099***
## (0.608)
##
## factor(year)2009 -8.689***
## (0.884)
##
## factor(year)2010 -3.355***
## (0.650)
##
## factor(year)2011 -1.789**
## (0.845)
##
## factor(year)2012 -4.860***
## (0.807)
##
## factor(year)2013 -5.695***
## (0.931)
##
## factor(year)2014 -5.032***
## (1.018)
##
## factor(year)2015 -6.548***
## (0.994)
##
## factor(year)2016 -6.445***
## (0.994)
##
## factor(year)2017 -5.479***
## (0.947)
##
## ---------------------------------------------
## Observations 943 943
## =============================================
## Note: *p<0.1; **p<0.05; ***p<0.01#no changes#pFtest(fixed.time, fixed)
#H0 - time effects are unnecessary
pFtest(fe_ind_labpr_V_t_i, fe_ind_labpr_V)##
## F test for individual effects
##
## data: labpr_V_t
## F = 25.455, df1 = 11, df2 = 849, p-value < 2.2e-16
## alternative hypothesis: significant effects# time effects are needed
#plmtest(fixed, c("time"), type=("bp"))
#H0 - time effects are unnecessary
plmtest(fe_ind_labpr_V, c("time"), type=("bp"))##
## Lagrange Multiplier Test - time effects (Breusch-Pagan) for unbalanced
## panels
##
## data: labpr_V
## chisq = 1140.2, df = 1, p-value < 2.2e-16
## alternative hypothesis: significant effects# time effects are needed
#so, fe_ind_labpr_V_t is the best model (twoways or individual- we are free to choose)#Simple models for labour productivity (RV, UV), without time-effects
ols_plm_labpr_RV_UV <- plm(labpr_RV_UV,
data = long_data,
index = c("region", "year"),
model = "pooling")
cov_ols_plm_labpr_RV_UV <- vcovHC(ols_plm_labpr_RV_UV, type = "HC0")
se_ols_plm_labpr_RV_UV <- sqrt(diag(cov_ols_plm_labpr_RV_UV))
coeftest(ols_plm_labpr_RV_UV, df = Inf, vcov = cov_ols_plm_labpr_RV_UV)##
## z test of coefficients:
##
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 70.95923239 3.90448072 18.1738 < 2.2e-16 ***
## Rv -0.34007900 0.23505281 -1.4468 0.147947
## Uv 0.74840529 0.24796636 3.0182 0.002543 **
## log(Inv) 0.36794700 0.23273487 1.5810 0.113885
## educ -0.00114385 0.00090255 -1.2674 0.205030
## Wage 0.25766258 0.02060487 12.5049 < 2.2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1fe_ind_labpr_RV_UV <- plm(labpr_RV_UV,
data = long_data,
index = c("region", "year"),
effect = "individual",
model = "within")
cov_fe_ind_labpr_RV_UV <- vcovHC(fe_ind_labpr_RV_UV, type = "HC0")
se_fe_ind_labpr_RV_UV <- sqrt(diag(cov_fe_ind_labpr_RV_UV))
coeftest(fe_ind_labpr_RV_UV, df = Inf, vcov = cov_fe_ind_labpr_RV_UV)##
## z test of coefficients:
##
## Estimate Std. Error z value Pr(>|z|)
## Rv 1.4334676 0.4694984 3.0532 0.002264 **
## Uv 1.4262146 0.5042567 2.8284 0.004679 **
## log(Inv) 0.6960052 0.4701780 1.4803 0.138793
## educ 0.0022750 0.0021546 1.0559 0.291011
## Wage 0.2354626 0.0300514 7.8353 4.676e-15 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1# fixef(fe_ind_labpr_RV_UV) #re_labpr_RV_UV <- plm(labpr_RV_UV,
data = long_data,
index = c("region", "year"),
model = "random")
cov_re_labpr_RV_UV <- vcovHC(re_labpr_RV_UV, type = "HC0")
se_re_labpr_RV_UV <- sqrt(diag(cov_re_labpr_RV_UV))
coeftest(re_labpr_RV_UV, df = Inf, vcov = cov_re_labpr_RV_UV)##
## z test of coefficients:
##
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 70.78946028 4.04656803 17.4937 < 2.2e-16 ***
## Rv -0.20054176 0.26652928 -0.7524 0.451799
## Uv 0.79296876 0.25461874 3.1143 0.001844 **
## log(Inv) 0.37022105 0.24056836 1.5389 0.123818
## educ -0.00106240 0.00096166 -1.1048 0.269266
## Wage 0.25680886 0.02113893 12.1486 < 2.2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Tests for model for labour productivity (RV, UV), without time-effects
#Ftest - between fixed effects and pooled OLS
#H0 - pooled OLS are better than fixed effects (because all individual effects are the same)
#H1 - fixed effects are better than pooled OLS ones
pFtest(fe_ind_labpr_RV_UV, ols_plm_labpr_RV_UV)##
## F test for individual effects
##
## data: labpr_RV_UV
## F = 1.5855, df1 = 78, df2 = 859, p-value = 0.001428
## alternative hypothesis: significant effects#fixed effects are better, as expected#plmtest - between random effects and pooled OLS
#H0 - pooled OLS are better than random effects (because all random effects are the same - on penal effect)
#H1 - random effects are better than pooled OLS ones
plmtest(ols_plm_labpr_RV_UV, type = c('bp'))##
## Lagrange Multiplier Test - (Breusch-Pagan) for unbalanced panels
##
## data: labpr_RV_UV
## chisq = 3.7675, df = 1, p-value = 0.05226
## alternative hypothesis: significant effects#random effects are better, but not for sure#phtest - between fixed effects and random effects
#H0 - random effects are better than fixed effects (random coef. are more efficient)
#H1 - fixed effects are better than random effects (fixed coef. are more efficient)
phtest(fe_ind_labpr_RV_UV, re_labpr_RV_UV)##
## Hausman Test
##
## data: labpr_RV_UV
## chisq = 20.452, df = 5, p-value = 0.001028
## alternative hypothesis: one model is inconsistent#random effects are then fixed ones - unexpected
#fixed effects are better than random effectsstargazer(ols_plm_labpr_RV_UV, fe_ind_labpr_RV_UV, re_labpr_RV_UV, se = list(se_ols_plm_labpr_RV_UV, se_fe_ind_labpr_RV_UV, se_re_labpr_RV_UV), type = "text", omit = c('Constant'), keep.stat = c("n"), font.size = "tiny") #, font.size = "tiny"##
## ==========================================
## Dependent variable:
## -----------------------------
## LabPr
## (1) (2) (3)
## ------------------------------------------
## Rv -0.340 1.433*** -0.201
## (0.235) (0.469) (0.267)
##
## Uv 0.748*** 1.426*** 0.793***
## (0.248) (0.504) (0.255)
##
## log(Inv) 0.368 0.696 0.370
## (0.233) (0.470) (0.241)
##
## educ -0.001 0.002 -0.001
## (0.001) (0.002) (0.001)
##
## Wage 0.258*** 0.235*** 0.257***
## (0.021) (0.030) (0.021)
##
## ------------------------------------------
## Observations 943 943 943
## ==========================================
## Note: *p<0.1; **p<0.05; ***p<0.01#Extra models with time-effects, LabPr, RV, UV
#extra
fe_ind_labpr_RV_UV_t <- plm(labpr_RV_UV_t,
data = long_data,
index = c("region", "year"),
effect = "twoways",
model = "within")
cov_fe_ind_labpr_RV_UV_t <- vcovHC(fe_ind_labpr_RV_UV_t, type = "HC0")
se_fe_ind_labpr_RV_UV_t <- sqrt(diag(cov_fe_ind_labpr_RV_UV_t))
coeftest(fe_ind_labpr_RV_UV_t, df = Inf, vcov = cov_fe_ind_labpr_RV_UV_t)##
## z test of coefficients:
##
## Estimate Std. Error z value Pr(>|z|)
## Rv 0.45321644 0.42434893 1.0680 0.285508
## Uv -0.21164323 0.51386443 -0.4119 0.680438
## log(Inv) 2.39026774 0.56764171 4.2109 2.544e-05 ***
## educ 0.00057344 0.00273163 0.2099 0.833725
## Wage 0.08777297 0.03231192 2.7164 0.006599 **
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1# fixef(fe_ind_labpr_RV_UV) ##extra
fe_ind_labpr_RV_UV_t_i <- plm(labpr_RV_UV_t,
data = long_data,
index = c("region", "year"),
effect = "individual",
model = "within")
cov_fe_ind_labpr_RV_UV_t_i <- vcovHC(fe_ind_labpr_RV_UV_t_i, type = "HC0")
se_fe_ind_labpr_RV_UV_t_i <- sqrt(diag(cov_fe_ind_labpr_RV_UV_t_i))
coeftest(fe_ind_labpr_RV_UV_t_i, df = Inf, vcov = cov_fe_ind_labpr_RV_UV_t_i)##
## z test of coefficients:
##
## Estimate Std. Error z value Pr(>|z|)
## Rv 0.45321644 0.42434893 1.0680 0.285508
## Uv -0.21164323 0.51386443 -0.4119 0.680438
## log(Inv) 2.39026774 0.56764171 4.2109 2.544e-05 ***
## educ 0.00057344 0.00273163 0.2099 0.833725
## Wage 0.08777297 0.03231192 2.7164 0.006599 **
## factor(year)2007 0.66645560 0.53257443 1.2514 0.210794
## factor(year)2008 -3.07903864 0.60369549 -5.1003 3.391e-07 ***
## factor(year)2009 -8.74804826 0.89410172 -9.7842 < 2.2e-16 ***
## factor(year)2010 -3.37514020 0.65133017 -5.1819 2.196e-07 ***
## factor(year)2011 -1.89950619 0.85396601 -2.2243 0.026126 *
## factor(year)2012 -4.96017257 0.82216315 -6.0331 1.609e-09 ***
## factor(year)2013 -5.78902002 0.94260620 -6.1415 8.174e-10 ***
## factor(year)2014 -5.15598221 1.03625594 -4.9756 6.505e-07 ***
## factor(year)2015 -6.69719311 1.01949948 -6.5691 5.062e-11 ***
## factor(year)2016 -6.62933271 1.03702952 -6.3926 1.631e-10 ***
## factor(year)2017 -5.65652091 0.98485656 -5.7435 9.274e-09 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1# fixef(fe_ind_labpr_RV_UV) #stargazer(fe_ind_labpr_RV_UV_t, fe_ind_labpr_RV_UV_t_i, se = list(se_fe_ind_labpr_RV_UV_t, se_fe_ind_labpr_RV_UV_t_i), type = "text", omit = c('Constant'), keep.stat = c("n"), font.size = "tiny") #, font.size = "tiny"##
## =============================================
## Dependent variable:
## ----------------------------
## LabPr
## (1) (2)
## ---------------------------------------------
## Rv 0.453 0.453
## (0.424) (0.424)
##
## Uv -0.212 -0.212
## (0.514) (0.514)
##
## log(Inv) 2.390*** 2.390***
## (0.568) (0.568)
##
## educ 0.001 0.001
## (0.003) (0.003)
##
## Wage 0.088*** 0.088***
## (0.032) (0.032)
##
## factor(year)2007 0.666
## (0.533)
##
## factor(year)2008 -3.079***
## (0.604)
##
## factor(year)2009 -8.748***
## (0.894)
##
## factor(year)2010 -3.375***
## (0.651)
##
## factor(year)2011 -1.900**
## (0.854)
##
## factor(year)2012 -4.960***
## (0.822)
##
## factor(year)2013 -5.789***
## (0.943)
##
## factor(year)2014 -5.156***
## (1.036)
##
## factor(year)2015 -6.697***
## (1.019)
##
## factor(year)2016 -6.629***
## (1.037)
##
## factor(year)2017 -5.657***
## (0.985)
##
## ---------------------------------------------
## Observations 943 943
## =============================================
## Note: *p<0.1; **p<0.05; ***p<0.01#no changes#pFtest(fixed.time, fixed)
#H0 - time effects are unnecessary
pFtest(fe_ind_labpr_RV_UV_t_i, fe_ind_labpr_RV_UV)##
## F test for individual effects
##
## data: labpr_RV_UV_t
## F = 25.493, df1 = 11, df2 = 848, p-value < 2.2e-16
## alternative hypothesis: significant effects# time effects are needed
#plmtest(fixed, c("time"), type=("bp"))
#H0 - time effects are unnecessary
plmtest(fe_ind_labpr_RV_UV, c("time"), type=("bp"))##
## Lagrange Multiplier Test - time effects (Breusch-Pagan) for unbalanced
## panels
##
## data: labpr_RV_UV
## chisq = 1161.1, df = 1, p-value < 2.2e-16
## alternative hypothesis: significant effects# time effects are needed
#so, fe_ind_labpr_RV_UV_t is the best model (twoways or individual- we are free to choose)#Simple models for Empoyment V, without time-effects
ols_plm_emp_V <- plm(emp_V,
data = long_data,
index = c("region", "year"),
model = "pooling")
cov_ols_plm_emp_V <- vcovHC(ols_plm_emp_V, type = "HC0")
se_ols_plm_emp_V<- sqrt(diag(cov_ols_plm_emp_V))
coeftest(ols_plm_emp_V, df = Inf, vcov = cov_ols_plm_emp_V)##
## z test of coefficients:
##
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -6.5172e-02 1.3831e-02 -4.7121 2.452e-06 ***
## V 9.3934e-04 6.9341e-04 1.3547 0.1755
## log(Inv) 7.0705e-03 9.8017e-04 7.2136 5.449e-13 ***
## educ -1.6029e-06 3.9421e-06 -0.4066 0.6843
## Wage -7.3659e-07 5.4548e-05 -0.0135 0.9892
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1fe_ind_emp_V <- plm(emp_V,
data = long_data,
index = c("region", "year"),
effect = "individual",
model = "within")
cov_fe_ind_emp_V <- vcovHC(fe_ind_emp_V, type = "HC0")
se_fe_ind_emp_V <- sqrt(diag(cov_fe_ind_emp_V))
coeftest(fe_ind_emp_V, df = Inf, vcov = cov_fe_ind_emp_V)##
## z test of coefficients:
##
## Estimate Std. Error z value Pr(>|z|)
## V 1.7133e-03 8.4138e-04 2.0363 0.04172 *
## log(Inv) 5.9142e-03 1.1195e-03 5.2829 1.272e-07 ***
## educ -2.0663e-05 1.0542e-05 -1.9601 0.04998 *
## Wage 9.9593e-05 4.4418e-05 2.2422 0.02495 *
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1# fixef(fe_ind_GDP) #re_emp_V <- plm(emp_V,
data = long_data,
index = c("region", "year"),
model = "random")
cov_re_emp_V <- vcovHC(re_emp_V, type = "HC0")
se_re_emp_V <- sqrt(diag(cov_re_emp_V))
coeftest(re_emp_V, df = Inf, vcov = cov_re_emp_V)##
## z test of coefficients:
##
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -6.5691e-02 1.2724e-02 -5.1628 2.433e-07 ***
## V 1.6898e-03 6.1940e-04 2.7281 0.006371 **
## log(Inv) 6.5683e-03 8.9824e-04 7.3124 2.623e-13 ***
## educ -1.4756e-05 8.7369e-06 -1.6890 0.091228 .
## Wage 8.0667e-05 4.1269e-05 1.9546 0.050625 .
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Tests for model for Employment V, without time-effects
#Ftest - between fixed effects and pooled OLS
#H0 - pooled OLS are better than fixed effects (because all individual effects are the same)
#H1 - fixed effects are better than pooled OLS ones
pFtest(fe_ind_emp_V, ols_plm_emp_V)##
## F test for individual effects
##
## data: emp_V
## F = 11.42, df1 = 78, df2 = 860, p-value < 2.2e-16
## alternative hypothesis: significant effects#fixed effects are better, as expected#plmtest - between random effects and pooled OLS
#H0 - pooled OLS are better than random effects (because all random effects are the same - on penal effect)
#H1 - random effects are better than pooled OLS ones
plmtest(ols_plm_emp_V, type = c('bp'))##
## Lagrange Multiplier Test - (Breusch-Pagan) for unbalanced panels
##
## data: emp_V
## chisq = 1002.2, df = 1, p-value < 2.2e-16
## alternative hypothesis: significant effects#random effects are better, as expected#phtest - between fixed effects and random effects
#H0 - random effects are better than fixed effects (random coef. are more efficient)
#H1 - fixed effects are better than random effects (fixed coef. are more efficient)
phtest(fe_ind_emp_V, re_emp_V)##
## Hausman Test
##
## data: emp_V
## chisq = 13.761, df = 4, p-value = 0.008099
## alternative hypothesis: one model is inconsistent#random effects are then fixed ones - unexpected
#fixed effects are better than random effectsstargazer(ols_plm_emp_V, fe_ind_emp_V, re_emp_V, se = list(se_ols_plm_emp_V, se_fe_ind_emp_V, se_re_emp_V), type = "text", omit = c('Constant'), keep.stat = c("n"), font.size = "tiny") #, font.size = "tiny"##
## ===========================================
## Dependent variable:
## ------------------------------
## EmpPerWF
## (1) (2) (3)
## -------------------------------------------
## V 0.001 0.002** 0.002***
## (0.001) (0.001) (0.001)
##
## log(Inv) 0.007*** 0.006*** 0.007***
## (0.001) (0.001) (0.001)
##
## educ -0.00000 -0.00002** -0.00001*
## (0.00000) (0.00001) (0.00001)
##
## Wage -0.00000 0.0001** 0.0001*
## (0.0001) (0.00004) (0.00004)
##
## -------------------------------------------
## Observations 943 943 943
## ===========================================
## Note: *p<0.1; **p<0.05; ***p<0.01#Extra models with time-effects Emp, V
#extra
fe_ind_emp_V_t <- plm(emp_V_t,
data = long_data,
index = c("region", "year"),
effect = "twoways",
model = "within")
cov_fe_ind_emp_V_t <- vcovHC(fe_ind_emp_V_t, type = "HC0")
se_fe_ind_emp_V_t <- sqrt(diag(cov_fe_ind_emp_V_t))
coeftest(fe_ind_emp_V_t, df = Inf, vcov = cov_fe_ind_emp_V_t)##
## z test of coefficients:
##
## Estimate Std. Error z value Pr(>|z|)
## V 1.3111e-03 7.7438e-04 1.6930 0.09045 .
## log(Inv) 2.2306e-03 2.1646e-03 1.0305 0.30279
## educ 5.6591e-06 1.3799e-05 0.4101 0.68172
## Wage 8.9167e-05 5.0471e-05 1.7667 0.07728 .
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1# fixef(fe_ind_emp_V) ##extra
fe_ind_emp_V_t_i <- plm(emp_V_t,
data = long_data,
index = c("region", "year"),
effect = "individual",
model = "within")
cov_fe_ind_emp_V_t_i <- vcovHC(fe_ind_emp_V_t_i, type = "HC0")
se_fe_ind_emp_V_t_i <- sqrt(diag(cov_fe_ind_emp_V_t_i))
coeftest(fe_ind_emp_V_t_i, df = Inf, vcov = cov_fe_ind_emp_V_t_i)##
## z test of coefficients:
##
## Estimate Std. Error z value Pr(>|z|)
## V 1.3111e-03 7.7438e-04 1.6930 0.09045 .
## log(Inv) 2.2306e-03 2.1646e-03 1.0305 0.30279
## educ 5.6591e-06 1.3799e-05 0.4101 0.68172
## Wage 8.9167e-05 5.0471e-05 1.7667 0.07728 .
## factor(year)2007 1.2508e-03 8.2954e-04 1.5078 0.13160
## factor(year)2008 -1.1994e-03 1.5033e-03 -0.7979 0.42494
## factor(year)2009 -9.8623e-04 1.8254e-03 -0.5403 0.58900
## factor(year)2010 1.0340e-03 1.8081e-03 0.5719 0.56739
## factor(year)2011 3.5225e-03 2.0945e-03 1.6818 0.09261 .
## factor(year)2012 5.8211e-03 2.5741e-03 2.2614 0.02374 *
## factor(year)2013 7.3621e-03 3.0431e-03 2.4193 0.01555 *
## factor(year)2014 9.0391e-03 3.6630e-03 2.4677 0.01360 *
## factor(year)2015 6.1871e-03 3.6361e-03 1.7015 0.08884 .
## factor(year)2016 7.6130e-03 3.8412e-03 1.9819 0.04749 *
## factor(year)2017 9.5064e-03 3.7157e-03 2.5585 0.01051 *
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1# fixef(fe_ind_emp_V) #stargazer(fe_ind_emp_V_t, fe_ind_emp_V_t_i, se = list(se_fe_ind_emp_V_t, se_fe_ind_emp_V_t_i), type = "text", omit = c('Constant'), keep.stat = c("n"), font.size = "tiny") #, font.size = "tiny"##
## =============================================
## Dependent variable:
## ----------------------------
## EmpPerWF
## (1) (2)
## ---------------------------------------------
## V 0.001* 0.001*
## (0.001) (0.001)
##
## log(Inv) 0.002 0.002
## (0.002) (0.002)
##
## educ 0.00001 0.00001
## (0.00001) (0.00001)
##
## Wage 0.0001* 0.0001*
## (0.0001) (0.0001)
##
## factor(year)2007 0.001
## (0.001)
##
## factor(year)2008 -0.001
## (0.002)
##
## factor(year)2009 -0.001
## (0.002)
##
## factor(year)2010 0.001
## (0.002)
##
## factor(year)2011 0.004*
## (0.002)
##
## factor(year)2012 0.006**
## (0.003)
##
## factor(year)2013 0.007**
## (0.003)
##
## factor(year)2014 0.009**
## (0.004)
##
## factor(year)2015 0.006*
## (0.004)
##
## factor(year)2016 0.008**
## (0.004)
##
## factor(year)2017 0.010**
## (0.004)
##
## ---------------------------------------------
## Observations 943 943
## =============================================
## Note: *p<0.1; **p<0.05; ***p<0.01#no changes#pFtest(fixed.time, fixed)
#H0 - time effects are unnecessary
pFtest(fe_ind_emp_V_t_i, fe_ind_emp_V)##
## F test for individual effects
##
## data: emp_V_t
## F = 7.1277, df1 = 11, df2 = 849, p-value = 1.142e-11
## alternative hypothesis: significant effects# time effects are needed
#plmtest(fixed, c("time"), type=("bp"))
#H0 - time effects are unnecessary
plmtest(fe_ind_emp_V, c("time"), type=("bp"))##
## Lagrange Multiplier Test - time effects (Breusch-Pagan) for unbalanced
## panels
##
## data: emp_V
## chisq = 58.784, df = 1, p-value = 1.76e-14
## alternative hypothesis: significant effects# time effects are needed
#so, fe_ind_emp_V_t is the best model (twoways or individual- we are free to choose)#Simple models for Employment (RV, UV), without time-effects
ols_plm_emp_RV_UV <- plm(emp_RV_UV,
data = long_data,
index = c("region", "year"),
model = "pooling")
cov_ols_plm_emp_RV_UV <- vcovHC(ols_plm_emp_RV_UV, type = "HC0")
se_ols_plm_emp_RV_UV<- sqrt(diag(cov_ols_plm_emp_RV_UV))
coeftest(ols_plm_emp_RV_UV, df = Inf, vcov = cov_ols_plm_emp_RV_UV)##
## z test of coefficients:
##
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -6.6191e-02 1.3692e-02 -4.8344 1.335e-06 ***
## Rv -1.8316e-04 1.1141e-03 -0.1644 0.8694
## Uv 1.6471e-03 1.0332e-03 1.5943 0.1109
## log(Inv) 7.1629e-03 9.8453e-04 7.2755 3.452e-13 ***
## educ -2.5989e-06 4.3925e-06 -0.5917 0.5541
## Wage -9.3670e-06 5.5666e-05 -0.1683 0.8664
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1fe_ind_emp_RV_UV <- plm(emp_RV_UV,
data = long_data,
index = c("region", "year"),
effect = "individual",
model = "within")
cov_fe_ind_emp_RV_UV <- vcovHC(fe_ind_emp_RV_UV, type = "HC0")
se_fe_ind_emp_RV_UV <- sqrt(diag(cov_fe_ind_emp_RV_UV))
coeftest(fe_ind_emp_RV_UV, df = Inf, vcov = cov_fe_ind_emp_RV_UV)##
## z test of coefficients:
##
## Estimate Std. Error z value Pr(>|z|)
## Rv 1.6988e-03 1.6455e-03 1.0324 0.30187
## Uv 1.7219e-03 2.0543e-03 0.8382 0.40192
## log(Inv) 5.9156e-03 9.9665e-04 5.9355 2.929e-09 ***
## educ -2.0674e-05 1.1847e-05 -1.7450 0.08098 .
## Wage 9.9523e-05 4.8829e-05 2.0382 0.04153 *
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1# fixef(fe_ind_GDP) #re_emp_RV_UV <- plm(emp_RV_UV,
data = long_data,
index = c("region", "year"),
model = "random")
cov_re_emp_RV_UV <- vcovHC(re_emp_RV_UV, type = "HC0")
se_re_emp_RV_UV <- sqrt(diag(cov_re_emp_RV_UV))
coeftest(re_emp_RV_UV, df = Inf, vcov = cov_re_emp_RV_UV)##
## z test of coefficients:
##
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -6.6183e-02 1.1520e-02 -5.7449 9.195e-09 ***
## Rv 1.3180e-03 1.4423e-03 0.9138 0.36080
## Uv 1.9128e-03 1.5689e-03 1.2192 0.22276
## log(Inv) 6.6057e-03 8.2874e-04 7.9708 1.577e-15 ***
## educ -1.5025e-05 9.8202e-06 -1.5300 0.12602
## Wage 7.8667e-05 4.5019e-05 1.7474 0.08057 .
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Tests for model for Employment (RV, UV), without time-effects
#Ftest - between fixed effects and pooled OLS
#H0 - pooled OLS are better than fixed effects (because all individual effects are the same)
#H1 - fixed effects are better than pooled OLS ones
pFtest(fe_ind_emp_RV_UV, ols_plm_emp_RV_UV)##
## F test for individual effects
##
## data: emp_RV_UV
## F = 11.348, df1 = 78, df2 = 859, p-value < 2.2e-16
## alternative hypothesis: significant effects#fixed effects are better, as expected#plmtest - between random effects and pooled OLS
#H0 - pooled OLS are better than random effects (because all random effects are the same - on penal effect)
#H1 - random effects are better than pooled OLS ones
plmtest(ols_plm_emp_RV_UV, type = c('bp'))##
## Lagrange Multiplier Test - (Breusch-Pagan) for unbalanced panels
##
## data: emp_RV_UV
## chisq = 992.84, df = 1, p-value < 2.2e-16
## alternative hypothesis: significant effects#random effects are better, as expected#phtest - between fixed effects and random effects
#H0 - random effects are better than fixed effects (random coef. are more efficient)
#H1 - fixed effects are better than random effects (fixed coef. are more efficient)
phtest(fe_ind_emp_RV_UV, re_emp_RV_UV)##
## Hausman Test
##
## data: emp_RV_UV
## chisq = 13.828, df = 5, p-value = 0.01674
## alternative hypothesis: one model is inconsistent#random effects are then fixed ones - unexpected
#fixed effects are better than random effectsstargazer(ols_plm_emp_RV_UV, fe_ind_emp_RV_UV, re_emp_RV_UV, se = list(se_ols_plm_emp_RV_UV, se_fe_ind_emp_RV_UV, se_re_emp_RV_UV), type = "text", omit = c('Constant'), keep.stat = c("n"), font.size = "tiny") #, font.size = "tiny"##
## ==========================================
## Dependent variable:
## -----------------------------
## EmpPerWF
## (1) (2) (3)
## ------------------------------------------
## Rv -0.0002 0.002 0.001
## (0.001) (0.002) (0.001)
##
## Uv 0.002 0.002 0.002
## (0.001) (0.002) (0.002)
##
## log(Inv) 0.007*** 0.006*** 0.007***
## (0.001) (0.001) (0.001)
##
## educ -0.00000 -0.00002* -0.00002
## (0.00000) (0.00001) (0.00001)
##
## Wage -0.00001 0.0001** 0.0001*
## (0.0001) (0.00005) (0.00005)
##
## ------------------------------------------
## Observations 943 943 943
## ==========================================
## Note: *p<0.1; **p<0.05; ***p<0.01#Extra models with time-effects, emp, RV, UV
#extra
fe_ind_emp_RV_UV_t <- plm(emp_RV_UV_t,
data = long_data,
index = c("region", "year"),
effect = "twoways",
model = "within")
cov_fe_ind_emp_RV_UV_t <- vcovHC(fe_ind_emp_RV_UV_t, type = "HC0")
se_fe_ind_emp_RV_UV_t <- sqrt(diag(cov_fe_ind_emp_RV_UV_t))
coeftest(fe_ind_emp_RV_UV_t, df = Inf, vcov = cov_fe_ind_emp_RV_UV_t)##
## z test of coefficients:
##
## Estimate Std. Error z value Pr(>|z|)
## Rv -4.4930e-04 1.6374e-03 -0.2744 0.78378
## Uv 2.3997e-03 1.9717e-03 1.2170 0.22359
## log(Inv) 2.2239e-03 2.1594e-03 1.0299 0.30308
## educ 6.0946e-06 1.3768e-05 0.4427 0.65801
## Wage 9.2706e-05 4.9632e-05 1.8679 0.06178 .
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1# fixef(fe_ind_emp_RV_UV) ##extra
fe_ind_emp_RV_UV_t_i <- plm(emp_RV_UV_t,
data = long_data,
index = c("region", "year"),
effect = "individual",
model = "within")
cov_fe_ind_emp_RV_UV_t_i <- vcovHC(fe_ind_emp_RV_UV_t_i, type = "HC0")
se_fe_ind_emp_RV_UV_t_i <- sqrt(diag(cov_fe_ind_emp_RV_UV_t_i))
coeftest(fe_ind_emp_RV_UV_t_i, df = Inf, vcov = cov_fe_ind_emp_RV_UV_t_i)##
## z test of coefficients:
##
## Estimate Std. Error z value Pr(>|z|)
## Rv -4.4930e-04 1.6374e-03 -0.2744 0.78378
## Uv 2.3997e-03 1.9717e-03 1.2170 0.22359
## log(Inv) 2.2239e-03 2.1594e-03 1.0299 0.30308
## educ 6.0946e-06 1.3768e-05 0.4427 0.65801
## Wage 9.2706e-05 4.9632e-05 1.8679 0.06178 .
## factor(year)2007 1.1831e-03 7.7788e-04 1.5210 0.12826
## factor(year)2008 -1.2845e-03 1.4295e-03 -0.8985 0.36889
## factor(year)2009 -7.3189e-04 2.0295e-03 -0.3606 0.71838
## factor(year)2010 1.1185e-03 1.8796e-03 0.5951 0.55177
## factor(year)2011 3.9946e-03 2.4290e-03 1.6446 0.10006
## factor(year)2012 6.2517e-03 2.8985e-03 2.1569 0.03101 *
## factor(year)2013 7.7661e-03 3.3581e-03 2.3126 0.02074 *
## factor(year)2014 9.5724e-03 3.9889e-03 2.3998 0.01640 *
## factor(year)2015 6.8252e-03 3.9799e-03 1.7149 0.08636 .
## factor(year)2016 8.4023e-03 4.2754e-03 1.9653 0.04938 *
## factor(year)2017 1.0267e-02 4.1234e-03 2.4900 0.01277 *
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1# fixef(fe_ind_emp_RV_UV) #stargazer(fe_ind_emp_RV_UV_t, fe_ind_emp_RV_UV_t_i, se = list(se_fe_ind_emp_RV_UV_t, se_fe_ind_emp_RV_UV_t_i), type = "text", omit = c('Constant'), keep.stat = c("n"), font.size = "tiny") #, font.size = "tiny"##
## =============================================
## Dependent variable:
## ----------------------------
## EmpPerWF
## (1) (2)
## ---------------------------------------------
## Rv -0.0004 -0.0004
## (0.002) (0.002)
##
## Uv 0.002 0.002
## (0.002) (0.002)
##
## log(Inv) 0.002 0.002
## (0.002) (0.002)
##
## educ 0.00001 0.00001
## (0.00001) (0.00001)
##
## Wage 0.0001* 0.0001*
## (0.00005) (0.00005)
##
## factor(year)2007 0.001
## (0.001)
##
## factor(year)2008 -0.001
## (0.001)
##
## factor(year)2009 -0.001
## (0.002)
##
## factor(year)2010 0.001
## (0.002)
##
## factor(year)2011 0.004
## (0.002)
##
## factor(year)2012 0.006**
## (0.003)
##
## factor(year)2013 0.008**
## (0.003)
##
## factor(year)2014 0.010**
## (0.004)
##
## factor(year)2015 0.007*
## (0.004)
##
## factor(year)2016 0.008**
## (0.004)
##
## factor(year)2017 0.010**
## (0.004)
##
## ---------------------------------------------
## Observations 943 943
## =============================================
## Note: *p<0.1; **p<0.05; ***p<0.01#no changes#pFtest(fixed.time, fixed)
#H0 - time effects are unnecessary
pFtest(fe_ind_emp_RV_UV_t_i, fe_ind_emp_RV_UV)##
## F test for individual effects
##
## data: emp_RV_UV_t
## F = 7.4005, df1 = 11, df2 = 848, p-value = 3.374e-12
## alternative hypothesis: significant effects# time effects are needed
#plmtest(fixed, c("time"), type=("bp"))
#H0 - time effects are unnecessary
plmtest(fe_ind_emp_RV_UV, c("time"), type=("bp"))##
## Lagrange Multiplier Test - time effects (Breusch-Pagan) for unbalanced
## panels
##
## data: emp_RV_UV
## chisq = 61.765, df = 1, p-value = 3.87e-15
## alternative hypothesis: significant effects# time effects are needed
#so, fe_ind_emp_RV_UV_t is the best model (twoways or individual- we are free to choose)#Simple models for Unemployment V, without time-effects
ols_plm_unemp_V <- plm(unemp_V,
data = long_data,
index = c("region", "year"),
model = "pooling")
cov_ols_plm_unemp_V <- vcovHC(ols_plm_unemp_V, type = "HC0")
se_ols_plm_unemp_V <- sqrt(diag(cov_ols_plm_unemp_V))
coeftest(ols_plm_unemp_V, df = Inf, vcov = cov_ols_plm_unemp_V)##
## z test of coefficients:
##
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 42.7122588 11.0498708 3.8654 0.0001109 ***
## V -0.8177594 0.7727464 -1.0583 0.2899412
## log(Inv) -2.8379332 0.9040166 -3.1392 0.0016938 **
## educ -0.0058946 0.0030306 -1.9450 0.0517714 .
## Wage 0.0068294 0.0382779 0.1784 0.8583962
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1fe_ind_unemp_V <- plm(unemp_V,
data = long_data,
index = c("region", "year"),
effect = "individual",
model = "within")
cov_fe_ind_unemp_V <- vcovHC(fe_ind_unemp_V, type = "HC0")
se_fe_ind_unemp_V <- sqrt(diag(cov_fe_ind_unemp_V))
coeftest(fe_ind_unemp_V, df = Inf, vcov = cov_fe_ind_unemp_V)##
## z test of coefficients:
##
## Estimate Std. Error z value Pr(>|z|)
## V -0.1905456 0.1341484 -1.4204 0.1554887
## log(Inv) -1.7786968 0.5296942 -3.3580 0.0007852 ***
## educ 0.0041326 0.0020303 2.0355 0.0418032 *
## Wage -0.0383888 0.0132867 -2.8893 0.0038614 **
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1# fixef(fe_ind_GDP) #re_unemp_V <- plm(unemp_V,
data = long_data,
index = c("region", "year"),
model = "random")
cov_re_unemp_V <- vcovHC(re_unemp_V, type = "HC0")
se_re_unemp_V <- sqrt(diag(cov_re_unemp_V))
coeftest(re_unemp_V, df = Inf, vcov = cov_re_unemp_V)##
## z test of coefficients:
##
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 31.5462034 6.4585369 4.8844 1.037e-06 ***
## V -0.2629284 0.1313009 -2.0025 0.0452324 *
## log(Inv) -1.8982017 0.5589742 -3.3959 0.0006841 ***
## educ 0.0033206 0.0021656 1.5333 0.1251928
## Wage -0.0356207 0.0139916 -2.5459 0.0109006 *
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Tests for model for Unemployment V, without time-effects
#Ftest - between fixed effects and pooled OLS
#H0 - pooled OLS are better than fixed effects (because all individual effects are the same)
#H1 - fixed effects are better than pooled OLS ones
pFtest(fe_ind_unemp_V, ols_plm_unemp_V)##
## F test for individual effects
##
## data: unemp_V
## F = 70.489, df1 = 78, df2 = 860, p-value < 2.2e-16
## alternative hypothesis: significant effects#fixed effects are better, as expected#plmtest - between random effects and pooled OLS
#H0 - pooled OLS are better than random effects (because all random effects are the same - on penal effect)
#H1 - random effects are better than pooled OLS ones
plmtest(ols_plm_unemp_V, type = c('bp'))##
## Lagrange Multiplier Test - (Breusch-Pagan) for unbalanced panels
##
## data: unemp_V
## chisq = 3556.7, df = 1, p-value < 2.2e-16
## alternative hypothesis: significant effects#random effects are better, as expected#phtest - between fixed effects and random effects
#H0 - random effects are better than fixed effects (random coef. are more efficient)
#H1 - fixed effects are better than random effects (fixed coef. are more efficient)
phtest(fe_ind_unemp_V, re_unemp_V)##
## Hausman Test
##
## data: unemp_V
## chisq = 17.541, df = 4, p-value = 0.001517
## alternative hypothesis: one model is inconsistent#random effects are then fixed ones - unexpected
#fixed effects are better than random effectsstargazer(ols_plm_unemp_V, fe_ind_unemp_V, re_unemp_V, se = list(se_ols_plm_unemp_V, se_fe_ind_unemp_V, se_re_unemp_V), type = "text", omit = c('Constant'), keep.stat = c("n"), font.size = "tiny") #, font.size = "tiny"##
## ==========================================
## Dependent variable:
## -----------------------------
## Unemp
## (1) (2) (3)
## ------------------------------------------
## V -0.818 -0.191 -0.263**
## (0.773) (0.134) (0.131)
##
## log(Inv) -2.838*** -1.779*** -1.898***
## (0.904) (0.530) (0.559)
##
## educ -0.006* 0.004** 0.003
## (0.003) (0.002) (0.002)
##
## Wage 0.007 -0.038*** -0.036**
## (0.038) (0.013) (0.014)
##
## ------------------------------------------
## Observations 943 943 943
## ==========================================
## Note: *p<0.1; **p<0.05; ***p<0.01#Extra models with time-effects, unemp, V
#extra
fe_ind_unemp_V_t <- plm(unemp_V_t,
data = long_data,
index = c("region", "year"),
effect = "twoways",
model = "within")
cov_fe_ind_unemp_V_t <- vcovHC(fe_ind_unemp_V_t, type = "HC0")
se_fe_ind_unemp_V_t <- sqrt(diag(cov_fe_ind_unemp_V_t))
coeftest(fe_ind_unemp_V_t, df = Inf, vcov = cov_fe_ind_unemp_V_t)##
## z test of coefficients:
##
## Estimate Std. Error z value Pr(>|z|)
## V -0.0079013 0.1925919 -0.0410 0.96728
## log(Inv) -1.3707419 0.7435901 -1.8434 0.06527 .
## educ -0.0043559 0.0039747 -1.0959 0.27312
## Wage 0.0048148 0.0152795 0.3151 0.75267
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1# fixef(fe_ind_emp) ##extra
fe_ind_unemp_V_t_i <- plm(unemp_V_t,
data = long_data,
index = c("region", "year"),
effect = "individual",
model = "within")
cov_fe_ind_unemp_V_t_i <- vcovHC(fe_ind_unemp_V_t_i, type = "HC0")
se_fe_ind_unemp_V_t_i <- sqrt(diag(cov_fe_ind_unemp_V_t_i))
coeftest(fe_ind_unemp_V_t_i, df = Inf, vcov = cov_fe_ind_unemp_V_t_i)##
## z test of coefficients:
##
## Estimate Std. Error z value Pr(>|z|)
## V -0.0079013 0.1925919 -0.0410 0.9672750
## log(Inv) -1.3707419 0.7435901 -1.8434 0.0652691 .
## educ -0.0043559 0.0039747 -1.0959 0.2731207
## Wage 0.0048148 0.0152795 0.3151 0.7526742
## factor(year)2007 -0.5758939 0.2769010 -2.0798 0.0375454 *
## factor(year)2008 0.4571022 0.5711218 0.8004 0.4235031
## factor(year)2009 2.1544683 0.5844126 3.6866 0.0002273 ***
## factor(year)2010 1.2880364 0.5775992 2.2300 0.0257486 *
## factor(year)2011 0.5222539 0.6967844 0.7495 0.4535439
## factor(year)2012 -0.4438067 0.6642352 -0.6681 0.5040398
## factor(year)2013 -0.6320303 0.6411442 -0.9858 0.3242386
## factor(year)2014 -1.1290164 0.5825699 -1.9380 0.0526241 .
## factor(year)2015 -0.8593051 0.6461787 -1.3298 0.1835757
## factor(year)2016 -0.8784373 0.7186209 -1.2224 0.2215589
## factor(year)2017 -1.1836659 0.7341363 -1.6123 0.1068914
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1# fixef(fe_ind_emp) #stargazer(fe_ind_unemp_V_t, fe_ind_unemp_V_t_i, se = list(se_fe_ind_unemp_V_t, se_fe_ind_unemp_V_t_i), type = "text", omit = c('Constant'), keep.stat = c("n"), font.size = "tiny") #, font.size = "tiny"##
## =============================================
## Dependent variable:
## ----------------------------
## Unemp
## (1) (2)
## ---------------------------------------------
## V -0.008 -0.008
## (0.193) (0.193)
##
## log(Inv) -1.371* -1.371*
## (0.744) (0.744)
##
## educ -0.004 -0.004
## (0.004) (0.004)
##
## Wage 0.005 0.005
## (0.015) (0.015)
##
## factor(year)2007 -0.576**
## (0.277)
##
## factor(year)2008 0.457
## (0.571)
##
## factor(year)2009 2.154***
## (0.584)
##
## factor(year)2010 1.288**
## (0.578)
##
## factor(year)2011 0.522
## (0.697)
##
## factor(year)2012 -0.444
## (0.664)
##
## factor(year)2013 -0.632
## (0.641)
##
## factor(year)2014 -1.129*
## (0.583)
##
## factor(year)2015 -0.859
## (0.646)
##
## factor(year)2016 -0.878
## (0.719)
##
## factor(year)2017 -1.184
## (0.734)
##
## ---------------------------------------------
## Observations 943 943
## =============================================
## Note: *p<0.1; **p<0.05; ***p<0.01#no changes#pFtest(fixed.time, fixed)
#H0 - time effects are unnecessary
pFtest(fe_ind_unemp_V_t_i, fe_ind_unemp_V)##
## F test for individual effects
##
## data: unemp_V_t
## F = 14.825, df1 = 11, df2 = 849, p-value < 2.2e-16
## alternative hypothesis: significant effects# time effects are needed
#plmtest(fixed, c("time"), type=("bp"))
#H0 - time effects are unnecessary
plmtest(fe_ind_unemp_V, c("time"), type=("bp"))##
## Lagrange Multiplier Test - time effects (Breusch-Pagan) for unbalanced
## panels
##
## data: unemp_V
## chisq = 23.745, df = 1, p-value = 1.1e-06
## alternative hypothesis: significant effects# time effects are needed
#so, fe_ind_unemp_V_t is the best model (twoways or individual- we are free to choose)#Simple models for Unemployment (RV, UV), without time-effects
ols_plm_unemp_RV_UV <- plm(unemp_RV_UV,
data = long_data,
index = c("region", "year"),
model = "pooling")
cov_ols_plm_unemp_RV_UV <- vcovHC(ols_plm_unemp_RV_UV, type = "HC0")
se_ols_plm_unemp_RV_UV <- sqrt(diag(cov_ols_plm_unemp_RV_UV))
coeftest(ols_plm_unemp_RV_UV, df = Inf, vcov = cov_ols_plm_unemp_RV_UV)##
## z test of coefficients:
##
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 44.5573476 11.4950675 3.8762 0.0001061 ***
## Rv 1.2142830 0.6145211 1.9760 0.0481567 *
## Uv -2.0990614 1.3061676 -1.6070 0.1080460
## log(Inv) -3.0052043 0.9484889 -3.1684 0.0015327 **
## educ -0.0040916 0.0025103 -1.6299 0.1031153
## Wage 0.0224530 0.0425869 0.5272 0.5980347
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1fe_ind_unemp_RV_UV <- plm(unemp_RV_UV,
data = long_data,
index = c("region", "year"),
effect = "individual",
model = "within")
cov_fe_ind_unemp_RV_UV <- vcovHC(fe_ind_unemp_RV_UV, type = "HC0")
se_fe_ind_unemp_RV_UV <- sqrt(diag(cov_fe_ind_unemp_RV_UV))
coeftest(fe_ind_unemp_RV_UV, df = Inf, vcov = cov_fe_ind_unemp_RV_UV)##
## z test of coefficients:
##
## Estimate Std. Error z value Pr(>|z|)
## Rv 0.0942017 0.2939824 0.3204 0.7486401
## Uv -0.3592402 0.2213856 -1.6227 0.1046557
## log(Inv) -1.8074166 0.5396089 -3.3495 0.0008096 ***
## educ 0.0043426 0.0020655 2.1025 0.0355137 *
## Wage -0.0370065 0.0132800 -2.7866 0.0053259 **
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1# fixef(fe_ind_GDP) #re_unemp_RV_UV <- plm(unemp_RV_UV,
data = long_data,
index = c("region", "year"),
model = "random")
cov_re_unemp_RV_UV <- vcovHC(re_unemp_RV_UV, type = "HC0")
se_re_unemp_RV_UV <- sqrt(diag(cov_re_unemp_RV_UV))
coeftest(re_unemp_RV_UV, df = Inf, vcov = cov_re_unemp_RV_UV)##
## z test of coefficients:
##
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 32.1183492 6.6679175 4.8168 1.458e-06 ***
## Rv 0.1155232 0.2731488 0.4229 0.6723454
## Uv -0.4922722 0.2402307 -2.0492 0.0404461 *
## log(Inv) -1.9409097 0.5737423 -3.3829 0.0007173 ***
## educ 0.0035726 0.0021777 1.6406 0.1008825
## Wage -0.0336379 0.0142364 -2.3628 0.0181366 *
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Tests for model for Unemployment (RV, UV), without time-effects
#Ftest - between fixed effects and pooled OLS
#H0 - pooled OLS are better than fixed effects (because all individual effects are the same)
#H1 - fixed effects are better than pooled OLS ones
pFtest(fe_ind_unemp_RV_UV, ols_plm_unemp_RV_UV)##
## F test for individual effects
##
## data: unemp_RV_UV
## F = 67.908, df1 = 78, df2 = 859, p-value < 2.2e-16
## alternative hypothesis: significant effects#fixed effects are better, as expected#plmtest - between random effects and pooled OLS
#H0 - pooled OLS are better than random effects (because all random effects are the same - on penal effect)
#H1 - random effects are better than pooled OLS ones
plmtest(ols_plm_unemp_RV_UV, type = c('bp'))##
## Lagrange Multiplier Test - (Breusch-Pagan) for unbalanced panels
##
## data: unemp_RV_UV
## chisq = 3449, df = 1, p-value < 2.2e-16
## alternative hypothesis: significant effects#random effects are better, as expected#phtest - between fixed effects and random effects
#H0 - random effects are better than fixed effects (random coef. are more efficient)
#H1 - fixed effects are better than random effects (fixed coef. are more efficient)
phtest(fe_ind_unemp_RV_UV, re_unemp_RV_UV)##
## Hausman Test
##
## data: unemp_RV_UV
## chisq = 22.935, df = 5, p-value = 0.0003473
## alternative hypothesis: one model is inconsistent#random effects are then fixed ones - unexpected
#fixed effects are better than random effectsstargazer(ols_plm_unemp_RV_UV, fe_ind_unemp_RV_UV, re_unemp_RV_UV, se = list(se_ols_plm_unemp_RV_UV, se_fe_ind_unemp_RV_UV, se_re_unemp_RV_UV), type = "text", omit = c('Constant'), keep.stat = c("n"), font.size = "tiny") #, font.size = "tiny"##
## ==========================================
## Dependent variable:
## -----------------------------
## Unemp
## (1) (2) (3)
## ------------------------------------------
## Rv 1.214** 0.094 0.116
## (0.615) (0.294) (0.273)
##
## Uv -2.099 -0.359 -0.492**
## (1.306) (0.221) (0.240)
##
## log(Inv) -3.005*** -1.807*** -1.941***
## (0.948) (0.540) (0.574)
##
## educ -0.004 0.004** 0.004
## (0.003) (0.002) (0.002)
##
## Wage 0.022 -0.037*** -0.034**
## (0.043) (0.013) (0.014)
##
## ------------------------------------------
## Observations 943 943 943
## ==========================================
## Note: *p<0.1; **p<0.05; ***p<0.01#Extra models with time-effects, unemp, RV, UV
#extra
fe_ind_unemp_RV_UV_t <- plm(unemp_RV_UV_t,
data = long_data,
index = c("region", "year"),
effect = "twoways",
model = "within")
cov_fe_ind_unemp_RV_UV_t <- vcovHC(fe_ind_unemp_RV_UV_t, type = "HC0")
se_fe_ind_unemp_RV_UV_t <- sqrt(diag(cov_fe_ind_unemp_RV_UV_t))
coeftest(fe_ind_unemp_RV_UV_t, df = Inf, vcov = cov_fe_ind_unemp_RV_UV_t)##
## z test of coefficients:
##
## Estimate Std. Error z value Pr(>|z|)
## Rv 0.4982158 0.2759034 1.8058 0.07096 .
## Uv -0.3208908 0.3037499 -1.0564 0.29077
## log(Inv) -1.3688235 0.7439311 -1.8400 0.06577 .
## educ -0.0044811 0.0039943 -1.1219 0.26191
## Wage 0.0037973 0.0151760 0.2502 0.80242
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1# fixef(fe_ind_emp) ##extra
fe_ind_unemp_RV_UV_t_i <- plm(unemp_RV_UV_t,
data = long_data,
index = c("region", "year"),
effect = "individual",
model = "within")
cov_fe_ind_unemp_RV_UV_t_i <- vcovHC(fe_ind_unemp_RV_UV_t_i, type = "HC0")
se_fe_ind_unemp_RV_UV_t_i <- sqrt(diag(cov_fe_ind_unemp_RV_UV_t_i))
coeftest(fe_ind_unemp_RV_UV_t_i, df = Inf, vcov = cov_fe_ind_unemp_RV_UV_t_i)##
## z test of coefficients:
##
## Estimate Std. Error z value Pr(>|z|)
## Rv 0.4982158 0.2759034 1.8058 0.0709555 .
## Uv -0.3208908 0.3037499 -1.0564 0.2907713
## log(Inv) -1.3688235 0.7439311 -1.8400 0.0657702 .
## educ -0.0044811 0.0039943 -1.1219 0.2619135
## Wage 0.0037973 0.0151760 0.2502 0.8024183
## factor(year)2007 -0.5564383 0.2809202 -1.9808 0.0476171 *
## factor(year)2008 0.4815583 0.5795909 0.8309 0.4060533
## factor(year)2009 2.0813436 0.5685319 3.6609 0.0002513 ***
## factor(year)2010 1.2637426 0.5727740 2.2064 0.0273592 *
## factor(year)2011 0.3865298 0.6724930 0.5748 0.5654458
## factor(year)2012 -0.5675995 0.6439334 -0.8815 0.3780706
## factor(year)2013 -0.7481911 0.6317096 -1.1844 0.2362583
## factor(year)2014 -1.2823496 0.5954980 -2.1534 0.0312867 *
## factor(year)2015 -1.0427599 0.6717461 -1.5523 0.1205874
## factor(year)2016 -1.1053741 0.7587159 -1.4569 0.1451437
## factor(year)2017 -1.4024644 0.7799962 -1.7980 0.0721707 .
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1# fixef(fe_ind_emp) #stargazer(fe_ind_unemp_RV_UV_t, fe_ind_unemp_RV_UV_t_i, se = list(se_fe_ind_unemp_RV_UV_t, se_fe_ind_unemp_RV_UV_t_i), type = "text", omit = c('Constant'), keep.stat = c("n"), font.size = "tiny") #, font.size = "tiny"##
## =============================================
## Dependent variable:
## ----------------------------
## Unemp
## (1) (2)
## ---------------------------------------------
## Rv 0.498* 0.498*
## (0.276) (0.276)
##
## Uv -0.321 -0.321
## (0.304) (0.304)
##
## log(Inv) -1.369* -1.369*
## (0.744) (0.744)
##
## educ -0.004 -0.004
## (0.004) (0.004)
##
## Wage 0.004 0.004
## (0.015) (0.015)
##
## factor(year)2007 -0.556**
## (0.281)
##
## factor(year)2008 0.482
## (0.580)
##
## factor(year)2009 2.081***
## (0.569)
##
## factor(year)2010 1.264**
## (0.573)
##
## factor(year)2011 0.387
## (0.672)
##
## factor(year)2012 -0.568
## (0.644)
##
## factor(year)2013 -0.748
## (0.632)
##
## factor(year)2014 -1.282**
## (0.595)
##
## factor(year)2015 -1.043
## (0.672)
##
## factor(year)2016 -1.105
## (0.759)
##
## factor(year)2017 -1.402*
## (0.780)
##
## ---------------------------------------------
## Observations 943 943
## =============================================
## Note: *p<0.1; **p<0.05; ***p<0.01#no changes#pFtest(fixed.time, fixed)
#H0 - time effects are unnecessary
pFtest(fe_ind_unemp_RV_UV_t_i, fe_ind_unemp_RV_UV)##
## F test for individual effects
##
## data: unemp_RV_UV_t
## F = 15.074, df1 = 11, df2 = 848, p-value < 2.2e-16
## alternative hypothesis: significant effects# time effects are needed
#plmtest(fixed, c("time"), type=("bp"))
#H0 - time effects are unnecessary
plmtest(fe_ind_unemp_RV_UV, c("time"), type=("bp"))##
## Lagrange Multiplier Test - time effects (Breusch-Pagan) for unbalanced
## panels
##
## data: unemp_RV_UV
## chisq = 27.991, df = 1, p-value = 1.219e-07
## alternative hypothesis: significant effects# time effects are needed
#so, fe_ind_unemp_RV_UV_t is the best model (twoways or individual- we are free to choose)#Stargazer common table
stargazer(fe_ind_grp_V_t, fe_ind_labpr_V_t, fe_ind_emp_V_t, fe_ind_unemp_V_t, fe_ind_grp_RV_UV_t, fe_ind_labpr_RV_UV_t,fe_ind_emp_RV_UV_t, fe_ind_unemp_RV_UV_t,
se = list(se_fe_ind_grp_V_t, se_fe_ind_labpr_V_t, se_fe_ind_emp_V_t, se_fe_ind_unemp_V_t, se_fe_ind_grp_RV_UV_t, se_fe_ind_labpr_RV_UV_t, se_fe_ind_emp_RV_UV_t, se_fe_ind_unemp_RV_UV_t),
type = "text", omit = c('Constant'), keep.stat = c("n"), font.size = "tiny", title = 'Fixed effects and time effects models', add.lines = list(c('model', 'FE + TE', 'FE + TE', 'FE + TE', 'FE + TE', 'FE + TE', 'FE + TE', 'FE + TE', 'FE + TE')), header=FALSE) #, font.size = "tiny"##
## Fixed effects and time effects models
## ====================================================================================
## Dependent variable:
## -----------------------------------------------------------------------
## log(GRP) LabPr EmpPerWF Unemp log(GRP) LabPr EmpPerWF Unemp
## (1) (2) (3) (4) (5) (6) (7) (8)
## ------------------------------------------------------------------------------------
## V -0.015 0.042 0.001* -0.008
## (0.013) (0.352) (0.001) (0.193)
##
## Rv -0.016 0.453 -0.0004 0.498*
## (0.029) (0.424) (0.002) (0.276)
##
## Uv -0.015 -0.212 0.002 -0.321
## (0.020) (0.514) (0.002) (0.304)
##
## log(Inv) 0.135*** 2.389*** 0.002 -1.371* 0.135*** 2.390*** 0.002 -1.369*
## (0.027) (0.575) (0.002) (0.744) (0.027) (0.568) (0.002) (0.744)
##
## educ 0.00003 0.001 0.00001 -0.004 0.00003 0.001 0.00001 -0.004
## (0.0002) (0.003) (0.00001) (0.004) (0.0002) (0.003) (0.00001) (0.004)
##
## Wage 0.0001 0.089*** 0.0001* 0.005 0.0001 0.088*** 0.0001* 0.004
## (0.001) (0.032) (0.0001) (0.015) (0.001) (0.032) (0.00005) (0.015)
##
## ------------------------------------------------------------------------------------
## model FE + TE FE + TE FE + TE FE + TE FE + TE FE + TE FE + TE FE + TE
## Observations 943 943 943 943 943 943 943 943
## ====================================================================================
## Note: *p<0.1; **p<0.05; ***p<0.01