Indonesia was governed by a centralized system. This era which began in 1966 is well known as New Order (Orde Baru). It last till 1998 when a reformed movement was emerged because Asian Financial Crisis hit Indonesia economy very badly. Since then, new era come to born namely Reformed Era. Today, Indonesia government is a decentralized system. The system allows cities, regencies, and provincies to have some degree of freedom to organize their economic activity. This means, economic activities become more sporadic than before. Hence, there are new source of economic growth at city, regency, and provincial level. At the other hand, according to Evironmental Kuznets Curve Hypothesis that the environmental degredation tend to increase at the beginning of economic growth. However, when it passes a certain level of income, the economic growth allows environmental remediation. This research aims is to check whether the impact of decentralization reduce air pollution or not.

Data and Method

This analysis employs Ordinary Least Square method. There are 4 variables will be included in the model, air pollution, economic growth, trade, and decentralization as dummy. For air pollution the GHG per capita will be used as an indicator. GDP per capita is an indicator for economic growth. Then, trade openness which is calculated as trade share of GDP is used for trade indicator. Decentralization take into place effectively since January 1, 2001. Data were imported from https://edgar.jrc.ec.europa.eu/country_profile/IDN, https://data.worldbank.org/indicator/NY.GDP.PCAP.KN?end=2019&locations=ID&start=1990, and https://data.worldbank.org/indicator/NE.TRD.GNFS.ZS?end=2019&locations=ID&start=1990

library(readxl)
df <- read_excel("D:/KARIR/R Portofolio/Regression/dataghgindonesia.xlsx")
df$decentralization = as.factor(df$decentralization)
head(df)
## # A tibble: 6 x 5
##    year ghgpercap tradeopenness gdppercap decentralization
##   <dbl>     <dbl>         <dbl>     <dbl> <fct>           
## 1  1970      1.78          28.7  7018993. No              
## 2  1971      1.75          31.1  7315288. No              
## 3  1972      1.76          35.4  7627129. No              
## 4  1973      1.83          39.5  8033962. No              
## 5  1974      1.83          50.4  8397281. No              
## 6  1975      1.79          44.5  8647029. No
str(df)
## tibble [49 x 5] (S3: tbl_df/tbl/data.frame)
##  $ year            : num [1:49] 1970 1971 1972 1973 1974 ...
##  $ ghgpercap       : num [1:49] 1.78 1.75 1.76 1.83 1.83 1.79 1.8 1.94 2.03 2.02 ...
##  $ tradeopenness   : num [1:49] 28.7 31.1 35.4 39.5 50.4 ...
##  $ gdppercap       : num [1:49] 7018993 7315288 7627129 8033962 8397281 ...
##  $ decentralization: Factor w/ 2 levels "No","Yes": 1 1 1 1 1 1 1 1 1 1 ...

Plotting Data

This step is useful to see whether there is quadratic relation between GDP per capita and GHG per capita. According to ] Enviromental Kuznets Curve, economic growth and pollution have a reverse u-shape curve. But, as we can see, it is not exist in the data.

plot(log(df$gdppercap), log(df$ghgpercap))

The Model with Decentralization as Dummy Variable

ols <- lm(log(ghgpercap) ~ log(gdppercap) + tradeopenness + decentralization, data = df)
summary(ols)
## 
## Call:
## lm(formula = log(ghgpercap) ~ log(gdppercap) + tradeopenness + 
##     decentralization, data = df)
## 
## Residuals:
##       Min        1Q    Median        3Q       Max 
## -0.080098 -0.021991 -0.004393  0.034156  0.092662 
## 
## Coefficients:
##                       Estimate Std. Error t value Pr(>|t|)    
## (Intercept)         -6.2900098  0.3700763 -16.997  < 2e-16 ***
## log(gdppercap)       0.4296514  0.0232388  18.489  < 2e-16 ***
## tradeopenness        0.0003034  0.0006427   0.472 0.639151    
## decentralizationYes  0.0928801  0.0222828   4.168 0.000138 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.04538 on 45 degrees of freedom
## Multiple R-squared:  0.9689, Adjusted R-squared:  0.9668 
## F-statistic: 467.6 on 3 and 45 DF,  p-value: < 2.2e-16

Heteroskedasticity Test

df$resi <- ols$residuals
library(ggplot2)
## Warning: package 'ggplot2' was built under R version 4.1.3
ggplot(data = df, aes(y = resi, x = log(ghgpercap))) + geom_point(col = 'blue') + geom_abline(slope = 0)

From the plot, it seems no evident of heteroskedasticity. But, it is better to conduct formal test using Breusch-Pagan Test.

library(lmtest)
## Loading required package: zoo
## 
## Attaching package: 'zoo'
## The following objects are masked from 'package:base':
## 
##     as.Date, as.Date.numeric
bptest(ols)
## 
##  studentized Breusch-Pagan test
## 
## data:  ols
## BP = 6.9472, df = 3, p-value = 0.0736

Since the P-value > 0.05, we do not reject null hypothesis or there is no heteroskedasticity.

Autocorrelation Test

bgtest(ols)
## 
##  Breusch-Godfrey test for serial correlation of order up to 1
## 
## data:  ols
## LM test = 29.503, df = 1, p-value = 5.583e-08

Because P-value < 0.05, there is autocorrelation problem. To overcome it, we use Newey West HAC Robust Standard Errors.

library(sandwich)
## Warning: package 'sandwich' was built under R version 4.1.3
coeftest(ols, vcov = NeweyWest(ols))
## 
## t test of coefficients:
## 
##                        Estimate  Std. Error t value  Pr(>|t|)    
## (Intercept)         -6.29000983  0.76294426 -8.2444 1.524e-10 ***
## log(gdppercap)       0.42965141  0.04635325  9.2691 5.304e-12 ***
## tradeopenness        0.00030339  0.00118573  0.2559   0.79922    
## decentralizationYes  0.09288011  0.05019681  1.8503   0.07084 .  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

As we can see, after we have handled autocorrelation problem, decentralization variable become insignificant in 5% level, but still significant in 10% level.

Conclusion

Reformed Era is characterized with decentralize system. In this era, there are new soruces of economic growth whether in city or provincal level. Air pollution is still in increasing trend. GDP per capita effected ghg per capita by 0.43. The decentralize system also contribute to ghg per capita but significant at 10% level. Therefore, Evironmental Kuznets Curve Hypothesis is not exist and decentralization impact is not different from centralization.