Assessment of the Kuznets Environmental Hypothesis (Research Paper Sample)

ASSESSMENT OF THE KUZNETS ENVIRONMENTAL HYPOTHESIS
STUDENT NAME:
STUDENT ID:
DATE:
1.0: INTRODUCTION
The EKC hypothesis alleges that gases released to the environment will have an inverted u-shaped relation with economic growth. This means that at the beginning, as growth is fuelled in an economy, emissions will be rising, but will reach a pick and then start to fall as the economy grows (Dinda, 2004).
There has been a heated debate with regards to the hypothesis. Research, has been undertaken to test the hypothesis. Different researchers have agreed with the hypothesis, while others have refuted. Researchers in attempt to assess the hypothesis that deterioration is expected at the first stages of development, and quality is likely to be resort later has made use of pollutants like CO2, and have used GDP to show growth (Bo, 2011). In the advent of increased population, and adverse effects of climate change, the EKC hypothesis has been questioned. Some have alleged that it is even hard to determine the level at which degradation declines, and when quality is achieved. Different studies have concluded that the hypothesis would hold or fail to hold depending on the region of interest (Jalil and Rao, 2019). This questions has fuelled the current analysis, which makes use of CO2 emissions from 1970 to 2015.
2.0: METHODOLOGY
Current analysis is anchored on economic theory of diminishing and life cycle. Which stipulates that at the early stages of development, emissions will increase, but when economies of scale is achieved, deterioration is likely to reduce. The reduction is likely to resort the initial equilibrium state. This will mean an inverted u-shape in the relationship which would consequently satisfy the EKC hypothesis (Ajmi et.al, 2015).
Thus, the current analysis begins by assessing the trend and cycles in the series of emissions for the years under consideration (1970-2015). Application of Hodrick filtering and Hamilton filtering are applied to separate the trend and cycles in the series. Hodrick is used to smoothen the series by removing the short-term trends in a series. Research has shown that Hamilton is used as a better filtering to Hodrick as is able to show clear trends and makes a separation in data for the trend and cycles of the series (Cole et.al, 1997). In STATA, Hodrick filtering is possible, but for Hamilton we install following the command “ssc install hamiltonfilter”. Then, we generated the new variables for the Hodrick and Hamilton trends, and cycles. For Hodrick the command “tsfilter hp co2hodrick = co2emissionsmetrictonspercapita” for the cycles, and “tsfilter hp co2trend = co2emissionsmetrictonspercapita, trend(co2HP)” for the trend. Then, we plot the trend and the cycles of the new variables following “tsline var”. On the other hand, for the Hamilton, we use the command “hamiltonfilter co2emissionsmetrictonspercapita, stub(hamco2) frequency(yearly)”, and it will generate two variables (one for the cycles, and one for the trends), then we use “tsline var” to plot the series. After assessing the series, it was important to plot the persistence of the trend using the autocorrelation plot using the command “ac co2emissionsmetrictonspercapita, recast (line)”. Thus, the filtering is done to separate trends and cycles, and the autocorrelation will show the directional relations with difference in time.
The analysis progressed to stationarity tests making use of ADF, Phillip-Perron, and KPSS. The ADF and PP test has a hypothesis of unit root, while KPSS has a hypothesis of stationarity. To be able to undertake the ADF, we introduce lags, and if the p-value is less than the alpha, then we conclude that the series is stationary (Cole et.al, 1997). For effective time series analysis, use of natural logs is applied. If series is not stationary, time series analysis is not possible, and further test need be undertaken making use of first difference. According to Baek, (2015) series need to be stationary at level or first difference for effective time series analysis. In STATA common tests are ADF, and PP, thus we use command “ssc install kpss”, for installation the Kwiatkowski test. The test should be done both at level and with 1st difference. If, they are stationary at level, then level series is used for further analysis, and if 1st difference shows stationarity, then it is used for further analysis.
Confirming stationarity either at level or 1st difference allows for Johansen test of cointegration. The JT test is undertaken if stationarity is confirmed, either at level of 1st difference. The test will show the number of possible relations that exist in the series Νίκα, 2018.). In STATA we use the command “vecrank lnCO2dif lnGDPdif lnEnergyif lngdp2dif, trend(constant) max”. The reporting is a default trace, but can specify reporting of max statistic. The rank column, will show the different possible equations in our model and thus the null hypothesis of no cointegration. If the trace or max statistic is higher than the critical, we will report existence of cointegration. Existence of the relations will qualify use of vector correction model (VECM), to assess for long-run relationships. If there is no cointegration, then only a short-run model is possible, making use of VAR model. Following the command “vec lnCO2dif lnGDPdif lnEnergyif lngdp2dif, trend(constant)”, we will look at the Johansen normalisation test, which is the long-term equation. Signs of the test should be inverted for interpretation, as p-value will be assessed to confirm significance (Jalil and Rao, 2019). The omitted variable is our variable of interest, and significant relation will indicate existence of long run relationship between the variable and CO2 to prove the EKC hypothesis.
3.0: RESULTS AND DISCUSSION
3.1: PLOTTING THE SERIES OF CO2
The results as shown on appendix 1 indicate that CO2 has a down ward trend. This is an indication that at the beginning, emissions are high, but as time progresses, CO2 emissions decline. This means that irrespective of economic growth, emissions or rather degradation will reduce. This could be as a result of improved production technologies, and increased environmental awareness. Since the trend indicates non-stationarity in the series, it becomes important to undertake filtering to have a clear series movement.
3.1.1: Series of CO2 after Hodrick-Prescott filtering
HP is a filtering technique that allows for smoothing to remove short run trends. As shown on appendix 3 (cycles), and appendix 4(trends), the series indicates fast cycles of upward and downward movement in the series. The trend is smooth, and downward sloping. This is an indication that there are small cycles between periods, and generally the series will be falling with time. Therefore, irrespective of whether a country achieves economic growth, emissions will be falling with time. That is, environmental degradation will not rise first, and later reach a turning point to start decreasing, rather it will maintain a steady decreasing trend.
3.1.2: Series of CO2 after Hamilton filtering
HF has been a test like the HP but research has qualified it to be better filtering technique as it makes use of the series as it is without smoothing which could eliminate critical points within the trend. It has thus been applied as an alternative to HP filtering (Onafowora and Owoye, 2014). The assessment calculation is known for separating trends and cycles within a series and thus a better method than HP. From the current analysis, the results as indicated in appendix 4 (trend) and 5 (cycles), the series of CO2 emissions decreases with time with up and down cycles. This series is an indication of non-stationarity which qualifies for time series test to assess the hypothesis of stationarity.
3.1.3: Autocorrelation in CO2 series
The autocorrelation analysis will show the correlation between a series of one period and that of another period. For instance, with lag 1, we show the correlation between the current emissions, and the emission observed 1 year before. Lag 2 will indicate correlation of CO2 emissions, with the emission recorded 2 time periods prior. As plotted and depicted on appendix 6, the correlation graph depict an inverse trend, an indication that the correlation decreases as time lag increases. It is more likely that closer time lags are more correlated. This is as indicated by the correlation points out of the shade region of 95% confidence interval.
3.2: STATIONARITY CHECK
3.2.1: Augmented dickey-fuller
The ADF is undertaken with lagged series. Without the lag it would be a Dickey-Fuller test. It is a test for unit root. The test could be done with or without trend, and at level or with 1st difference (Onafowora and Owoye, 2014). As shown on appendix 7, ADF without trend p=0.9975 which is higher than 0.05. With trend (appendix 8) p=0.995, again higher than 0.05. This is an indication that the hypothesis of unit root should hold, and thus the series is not stationary.
3.2.2: Phillip-Perron
Like the ADF test, the PP test will assess for existence of unit root. Again we look at the p-value. The test could be done with or without trend, and at level or with 1st difference (Onafowora and Owoye, 2014). The current study results indicated at appendix 8 show that p= 0.9969 without trend, and 0.9607 with trend, which are higher than 0.05. Thus, we confirm that there is existence of unit root in the series.
3.2.3: KPSS
The Kwiatkowski test like the ADF, and PP is used to test for stationarity. ...