Data Panel with RStudio for FEB Universitas Jambi

RStudio untuk Penelitian Ekonomi dan Bisnis

Data Input merupakan file excel berformat (.xlsx) yang berisi data PDRB Perkapita, Indeks Pembangunan Manusia (IPM), dan Tingkat Penganggran Terbuka.

Import Data

library(openxlsx)
dt2 <- read.xlsx('datapanel.xlsx')

library(DT)
datatable(dt2)

library(plm)
library(ggplot2)

Plotting Data

ggplot(dt2,aes(x=Tahun,PDRB.Perkapita, color=Wilayah)) +
geom_line() + geom_point() +
theme_bw() + ggtitle('PDRB Perkapita Menurut Kabupaten/Kota di Prov. Jambi, 2017-2021')

ggplot(dt2,aes(x=Tahun,IPM, color=Wilayah)) +
geom_line() + geom_point() +
theme_bw() + ggtitle('IPM Menurut Kabupaten/Kota di Prov. Jambi, 2017-2021')

ggplot(dt2,aes(x=Tahun,TPT, color=Wilayah)) +
geom_line() + geom_point() +
theme_bw() + ggtitle('TPT Menurut Kabupaten/Kota di Prov. Jambi, 2017-2021') + ylab('TPT (%)')

ggplot(dt2,aes(x=Tahun,Miskin, color=Wilayah)) +
geom_line() + geom_point() +
theme_bw() + ggtitle('Persentase Penduduk Miskin Menurut Kabupaten/Kota di Prov. Jambi, 2017-2021') + ylab('Persentase Penduduk Miskin (%)')

Cek Korelasi Antar Variabel

library(GGally)
ggpairs(dt2[,3:6]) +
theme_bw()

Membuat Model Pooled Regression, Random Effect, dan Fixed Effect

Pooled Regression

pooled_ols_plm <- plm(IPM~log(PDRB.Perkapita)+TPT+Miskin+Gini.Ratio+Gini.Ratio, data = dt2,
index = c("Wilayah", "Tahun"), model = "pooling")
summary(pooled_ols_plm)

Pooling Model

Call:
plm(formula = IPM ~ log(PDRB.Perkapita) + TPT + Miskin + Gini.Ratio + 
    Gini.Ratio, data = dt2, model = "pooling", index = c("Wilayah", 
    "Tahun"))

Balanced Panel: n = 11, T = 5, N = 55

Residuals:
     Min.   1st Qu.    Median   3rd Qu.      Max. 
-5.216298 -1.530291 -0.062151  1.295367  5.054062 

Coefficients:
                    Estimate Std. Error t-value  Pr(>|t|)    
(Intercept)         62.42312    5.70811 10.9359 7.274e-15 ***
log(PDRB.Perkapita) -0.33219    0.96521 -0.3442  0.732165    
TPT                  0.99043    0.20219  4.8984 1.055e-05 ***
Miskin              -0.36687    0.13210 -2.7772  0.007698 ** 
Gini.Ratio          25.69335   12.62201  2.0356  0.047107 *  
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Total Sum of Squares:    711.69
Residual Sum of Squares: 284.99
R-Squared:      0.59956
Adj. R-Squared: 0.56753
F-statistic: 18.7159 on 4 and 50 DF, p-value: 1.85e-09

Fixed Effect Model

fe_model_plm <- plm(IPM~log(PDRB.Perkapita)+TPT+Miskin+Gini.Ratio+Gini.Ratio, data = dt2,
index = c("Wilayah", "Tahun"),
effect = "individual", model = "within")
summary(fe_model_plm)

Oneway (individual) effect Within Model

Call:
plm(formula = IPM ~ log(PDRB.Perkapita) + TPT + Miskin + Gini.Ratio + 
    Gini.Ratio, data = dt2, effect = "individual", model = "within", 
    index = c("Wilayah", "Tahun"))

Balanced Panel: n = 11, T = 5, N = 55

Residuals:
     Min.   1st Qu.    Median   3rd Qu.      Max. 
-0.915689 -0.241169 -0.071658  0.238912  1.008779 

Coefficients:
                     Estimate Std. Error t-value  Pr(>|t|)    
log(PDRB.Perkapita)  4.229359   0.782496  5.4050 3.233e-06 ***
TPT                  0.106246   0.069091  1.5378 0.1319777    
Miskin              -0.699710   0.194104 -3.6048 0.0008559 ***
Gini.Ratio          -3.621335   2.859948 -1.2662 0.2127576    
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Total Sum of Squares:    18.641
Residual Sum of Squares: 8.0556
R-Squared:      0.56786
Adj. R-Squared: 0.41661
F-statistic: 13.1407 on 4 and 40 DF, p-value: 6.3737e-07

fixef(fe_model_plm)

          BATANGHARI                BUNGO              KERINCI 
              60.039               58.011               61.034 
          KOTA JAMBI    KOTA SUNGAI PENUH             MERANGIN 
              67.932               59.263               60.167 
         MUARO JAMBI           SAROLANGUN TANJUNG JABUNG BARAT 
              55.089               59.798               55.303 
TANJUNG JABUNG TIMUR                 TEBO 
              53.402               58.382 

Random Effect Model

re_model_plm <- plm(IPM~log(PDRB.Perkapita)+TPT+Miskin+Gini.Ratio+Gini.Ratio, data = dt2,
index = c("Wilayah", "Tahun"),
effect = "individual", model = "random")
summary(re_model_plm)

Oneway (individual) effect Random Effect Model 
   (Swamy-Arora's transformation)

Call:
plm(formula = IPM ~ log(PDRB.Perkapita) + TPT + Miskin + Gini.Ratio + 
    Gini.Ratio, data = dt2, effect = "individual", model = "random", 
    index = c("Wilayah", "Tahun"))

Balanced Panel: n = 11, T = 5, N = 55

Effects:
                 var std.dev share
idiosyncratic 0.2014  0.4488 0.032
individual    6.1337  2.4766 0.968
theta: 0.9192

Residuals:
     Min.   1st Qu.    Median   3rd Qu.      Max. 
-1.026329 -0.229023 -0.048265  0.286270  1.515741 

Coefficients:
                    Estimate Std. Error z-value  Pr(>|z|)    
(Intercept)         61.65901    3.57213 17.2611 < 2.2e-16 ***
log(PDRB.Perkapita)  3.45872    0.82277  4.2038 2.625e-05 ***
TPT                  0.15978    0.07541  2.1189    0.0341 *  
Miskin              -0.71354    0.18079 -3.9469 7.918e-05 ***
Gini.Ratio          -2.58469    3.16024 -0.8179    0.4134    
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Total Sum of Squares:    23.163
Residual Sum of Squares: 12.598
R-Squared:      0.4561
Adj. R-Squared: 0.41258
Chisq: 41.928 on 4 DF, p-value: 1.7265e-08

Menentukan Model Terbaik

Pooled vs Random Effect

plmtest(pooled_ols_plm, effect = "individual", type = c("bp"))

    Lagrange Multiplier Test - (Breusch-Pagan)

data:  IPM ~ log(PDRB.Perkapita) + TPT + Miskin + Gini.Ratio + Gini.Ratio
chisq = 60.955, df = 1, p-value = 5.841e-15
alternative hypothesis: significant effects

Random Effect lebih baik dibanding Pooled

Random vs Fixed Effect

phtest(re_model_plm,fe_model_plm)

    Hausman Test

data:  IPM ~ log(PDRB.Perkapita) + TPT + Miskin + Gini.Ratio + Gini.Ratio
chisq = 9.8392, df = 4, p-value = 0.04323
alternative hypothesis: one model is inconsistent

Fixed Effect lebih baik dibanding Random Effect

Cek Serial Correlation dan Cross-sectional dependence

pcdtest(fe_model_plm, test = c("lm"))

    Breusch-Pagan LM test for cross-sectional dependence in
    panels

data:  IPM ~ log(PDRB.Perkapita) + TPT + Miskin + Gini.Ratio + Gini.Ratio
chisq = 100.99, df = 55, p-value = 0.0001561
alternative hypothesis: cross-sectional dependence

pcdtest(fe_model_plm, test = c("cd"))

    Pesaran CD test for cross-sectional dependence in
    panels

data:  IPM ~ log(PDRB.Perkapita) + TPT + Miskin + Gini.Ratio + Gini.Ratio
z = 3.3215, p-value = 0.0008954
alternative hypothesis: cross-sectional dependence

pbgtest(fe_model_plm)

    Breusch-Godfrey/Wooldridge test for serial correlation
    in panel models

data:  IPM ~ log(PDRB.Perkapita) + TPT + Miskin + Gini.Ratio + Gini.Ratio
chisq = 11.072, df = 5, p-value = 0.04997
alternative hypothesis: serial correlation in idiosyncratic errors

Cek Asumsi Klasik

library(lmtest)
library(tseries)

Tes Normalitas

Tes Autokorelasi

pdwtest(fe_model_plm)

    Durbin-Watson test for serial correlation in panel
    models

data:  IPM ~ log(PDRB.Perkapita) + TPT + Miskin + Gini.Ratio + Gini.Ratio
DW = 1.7505, p-value = 0.162
alternative hypothesis: serial correlation in idiosyncratic errors

Tes Heteroskedastisitas

bptest(fe_model_plm, studentize = F)

    Breusch-Pagan test

data:  fe_model_plm
BP = 9.0248, df = 4, p-value = 0.06048

Tes Multikoleniaritas

library(car)
vif(pooled_ols_plm)

log(PDRB.Perkapita)                 TPT              Miskin 
           1.174486            1.375883            1.160777 
         Gini.Ratio 
           1.329677