Introduction

• Article 4 of the Paris Agreement (2015) stipulates ‘Mitigation’; where each country is responsible, within their means, to adopt economic wide policies to combat climate change.
  
• Europe, the 2nd most densest populated continent in the world, is responsible for a large amount of CO2 emissions.
• But is there a shift in government policies, peoples behaviours, and businesses adoption of cleaner energy across European countries?
• The aim of this investigation is to address this question by looking at end-user consumption of fossil fuels and renewable energy.

Introduction Cont.

• Measured annually, the Eurostat website stores end-user consumption data for different sources of energy across European countries.
• End-User consumption of energy includes entities such as households, transportation, industry and agriculture.
• Does not include the energy sector itself.
• Consumption is measured in thousand tonnes of oil equivalent (TOE)
• The current investigation is only concerned with fossil fuel and renewable energy.

Problem Statement

• In line with the EU’s commitment to cleaner energy for the future, the current analysis aims to test if there’s been a change in fossil fuel and renewable end-user consumption across European countries.
• Specifically, this investigation will test if there is a significant shift in fossil fuel and renewable consumption between 2009 and 2018.
  
• With the use of RStudio and it’s functions, the presentation will include descriptive statistics, plots and paired-sample t-tests in order to address the research question.

Data

• The data used to test the research question was obtained from Eurostat. (https://ec.europa.eu/eurostat/home?)
• Titled ‘Final energy consumption by product [TEN00123]’
• The full dataset was downloaded as an .xlsx sheet and stored in the ‘R Working Directory’ file set for this project.
• Each table, split by sheet in Excel, has TOE data for each significant energy source from 2009-2018.
• Observations are of each European country and totals from Eu-Member/Euro-Area states.

Preprocessing

• The current investigation is only concerned with the following information:
  • Fossil Fuel and Renewable end-user consumption (sheets 4 and 10; converted to a double variable).
  • TOE data from 2009 and 2018 (converted to factors, to well-define the groups)
  • Values from singular countries (totals were excluded)
• Below is the code from the preprocessing.

Fossil_Fuel <- read_xlsx("Energy EU.xlsx", sheet = 4, skip = 9)
Fossil_Fuel <- Fossil_Fuel[-c(1:4, 45:47), c(1,2, 20)] #Dropped unused Rows and Columns
colnames(Fossil_Fuel)[1] <- "Country"
Fossil_Fuel <- Fossil_Fuel %>% gather('2009', '2018', key = 'Year', value = 'TOE')  
Fossil_Fuel$TOE <-as.double(Fossil_Fuel$TOE) 
Fossil_Fuel$TOE <-  round(Fossil_Fuel$TOE, digits = 2)
Fossil_Fuel$Year <- factor(Fossil_Fuel$Year, levels = c("2009", "2018"))

Renewables <- read_xlsx("Energy EU.xlsx", sheet = 10, skip = 9)
Renewables <- Renewables[-c(1:4, 45:47), c(1,2, 20)]
colnames(Renewables)[1] <- "Country"
Renewables <- Renewables %>% gather('2009', '2018', key = 'Year', value = 'TOE')
Renewables$TOE <-as.double(Renewables$TOE) 
Renewables$TOE <-  round(Renewables$TOE, digits = 2)
Renewables$Year <- factor(Renewables$Year, levels = c("2009", "2018"))

Descriptive Statistics and Visualisation

• The main two variables of interest are TOE and year.
• Below is a descriptive summary of the Fossil Fuel sample split by year.

Fossil_Sum <- Fossil_Fuel %>% group_by(Year) %>% summarise(Min = min(TOE , na.rm = TRUE),
     Q1 = quantile(TOE , probs = .25,na.rm = TRUE),
     Median = median(TOE , na.rm = TRUE),
     Q3 = quantile(TOE , probs = .75,na.rm = TRUE),
     Max = max(TOE , na.rm = TRUE),
     Mean = mean(TOE , na.rm = TRUE),
     SD = sd(TOE , na.rm = TRUE),
     n = n(),
     Missing = sum(is.na(TOE)))
kable(Fossil_Sum)

Descriptives for Renewable Sample

Renewable_Sum <- Renewables %>% group_by(Year) %>% summarise(Min = min(TOE , na.rm = TRUE),
        Q1 = quantile(TOE , probs = .25,na.rm = TRUE),
        Median = median(TOE , na.rm = TRUE),
        Q3 = quantile(TOE , probs = .75,na.rm = TRUE),
        Max = max(TOE , na.rm = TRUE),
        Mean = mean(TOE , na.rm = TRUE),
        SD = sd(TOE , na.rm = TRUE),
        n = n(),
        Missing = sum(is.na(TOE)))
kable(Renewable_Sum)

Identified two NAs from 2009, Bosnia and Georgia (in both energy samples) due to data not being available (referenced on Eurostat website). Decided to exclude these two countries for hypothesis testing.

Box-Plot of differences of scores

Renewable_D <- as.data.frame(Renewable_D)
fossil_D <- as.data.frame(fossil_D)
boxplot(fossil_D$fossil_D, Renewable_D$Renewable_D, at=c(1,2), names = c("Fossil Fuel", "Renewables"), ylab= "TOE Difference (2009-2018)")

Descriptives Interpretation.

• Several key pointers to take from the descriptive section:
• Average end-user consumption of fossil fuel products has gone down between 2009 to 2018.
  
• While renewable consumption has gone up.
• Although, the box-plot for the fossil-fuel variable showed that most scores fall close to zero, i.e, there was little change between end-user fossil fuel consumption during this period across most EU countries. However, the dots below the tail identify extreme scores (perhaps explaining the mean difference).
• It was decided to remove the most extreme value (Turkey) for hypothesis testing as it will bias the estimates of model parameters.
• As for Renewables, the box-plot identified 5 potential outliers. Using the same rationale as above, it was decided to exclude the United Kingdom and Turkey from hypothesis testing (highest and lowest points on the box-plot).

Hypothesis Testing

• To test the research question, paired t-test’s will be used to check if there is a significant difference between mean end-user consumption from 2009 to 2018, separately for fossil fuel and renewable products.
  
• The following mathematical equations are the hypotheses for both Fossil Fuel and Renewable end-user consumption:

\[H_0: \mu_Δ = 0 \]

\[H_A: \mu_Δ \ne 0\] - The assumption for paired t-test’s requires the differences in scores to be normally distributed

• Before testing this assumption, extreme outliers identified in the description section were removed.

Assumption Testing, Fossil Fuel

fossil_D <- as.data.frame(fossil_D)
fossil_D <- fossil_D[-c(35),]#Removed Turkey
qqPlot(fossil_D, dist="norm")

## [1] 21  3

Assumption Testing, Renewables

Renewable_D <- Renewable_D[-c(28,35),] #Removed UK and Turkey
Renewable_D <- as.data.frame(Renewable_D)
qqPlot(Renewable_D$Renewable_D, dist= "norm")

## [1] 27 10

Assumption Testing Cont

– The distribution of difference scores highlighted by the QQ-Plot for the Fossil Fuel sample displayed 5 potential sources of outliers. However, as the sample size >30 (n=37), it was decided to keep these extreme cases in, although the results should be interpreted with caution

• Likewise with the Fossil Fuel sample, the QQ-Plot for Renewable TOE consumption difference highlights the presence of extreme cases. The ‘s’ like shape on the plot show signs of non-normality. As this sample is also above 30 (n=36), it was decided to keep the remaining observations for hypothesis testing, although the results should be interpreted with caution.

Paired T-test, Fossil Fuel

summary(fossil_D)

##     Min.  1st Qu.   Median     Mean  3rd Qu.     Max. 
## -1038.20   -61.44   -15.82   -93.22    29.35   152.70

t.test(fossil_D, mu= 0, alternative = "two.sided")

## 
##  One Sample t-test
## 
## data:  fossil_D
## t = -2.1978, df = 36, p-value = 0.03448
## alternative hypothesis: true mean is not equal to 0
## 95 percent confidence interval:
##  -179.234386   -7.199669
## sample estimates:
## mean of x 
## -93.21703

Paired T-test, Renewables

summary(Renewable_D)

##   Renewable_D     
##  Min.   :-221.75  
##  1st Qu.:  20.29  
##  Median : 162.15  
##  Mean   : 611.20  
##  3rd Qu.: 812.28  
##  Max.   :2786.20

t.test(Renewable_D, mu= 0, alternative = "two.sided")

## 
##  One Sample t-test
## 
## data:  Renewable_D
## t = 4.3445, df = 35, p-value = 0.000114
## alternative hypothesis: true mean is not equal to 0
## 95 percent confidence interval:
##  325.5984 896.7944
## sample estimates:
## mean of x 
##  611.1964

Results

Fossil Fuels

• The mean difference of fossil fuel consumption from 2009 to 2018 across European countries was found to be -93.22 TOE (SD = 257.99). The paired-samples t-test found a statistically significant negative mean difference between the two time-points, t(df=36)=−2.20, p =.034, 95% [-179.23, -7.20].

Renewables

• The mean difference of renewable consumption from 2009 to 2018 across European countries was 611.20 TOE (SD = 844.09). The paired-samples t-test found a statistically significant positive mean difference between the two time-points, t(df=35)=4.34, p<.001, 95% [325.60, 896.79].

Discussion

• From the paired-sample t-test’s for fossil fuel and renewable end user consumption between 2009 to 2018, it was decided to reject both null hypotheses and conclude there was a significant difference in consumption over the nine years.
• Both results demonstrate a general shift in end user consumption across European countries. With fossil fuel consumption declining from 2009 to 2018 and renewable energy increasing in the same period.
• The study does come with limitations that can be addressed in future research;
  • The sample data didn’t control for other sources of energy or total end-user consumption across each country. For example, although the study found an increase use of renewable energies, it may in part be a result of an increase of end-user consumption overall.
  • Similarly, the data was highly variable. Larger countries understandably had higher end-user consumption overall and the differences between 2009 to 2018 for them were generally more profound. This may have exaggerated the mean of differences and/or minimized the effect of smaller countries. Again, controlling for total end-user consumption may mitigate this limitation.

Conclusion

• The current investigation aimed to compared the change of fossil fuel and renewable end-user consumption across European countries from 2009 to 2018.
• The study found a statistical difference for both energy sources between the two time points in question, with fossil fuel consumption declining and renewable consumption increasing.
• Although the findings suggest a change towards renewable energy across European countries, future research should address the limitations outlined in the previous slide by including other confounding and/or predictor variables to better address the original research question.

References

• https://ec.europa.eu/eurostat/home
• https://unfccc.int/process-and-meetings/the-paris-agreement/what-is-the-paris-agreement