CC BY-NC-ND
82
Original article
DOI: 10.17323/1999-5431-2023-0-5-147-164
ASSESSMENT OF THE IMPACT OF CORRUPTION AND CIRCULAR DEBT ON ENERGY CRISIS IN PAKISTAN
Seema Arif1, Alam Khan2, Dilawar Khan3
1 M.Phil., Research Student, Department of Economics,
Kohat University of Science & Technology, Kohat, 26000, Pakistan; arifseema8@gmail.com; ORCID: https://orcid.org/0000-0002-9222-7848
2 Dr., Assistant Professor, Department of Economics,
Kohat University of Science & Technology, Kohat, 26000, Pakistan; alamkhan@kust.edu.pk; ORCID: https://orcid.org/0000-0001-8728-0516; (Corresponding Author)
3 Dr., Professor, Department of Economics,
Kohat University of Science & Technology, Kohat, 26000, Pakistan; dilawar@kust.edu.pk; ORCID: https://orcid.org/0000-0001-5872-267X
Abstract. The main purpose of this study was to explore the impact of corruption and circular debt on energy crises in the case of Pakistan. For this purpose, the study used annual time series data and employed an ARDL F-bounds test approach to achieve the determined objectives. The findings of the ARDL model reported that corruption, circular debt, and population have a positive impact, while industrial GDP and distribution and transmission contribute negatively to short- and long-term energy crises. Consequently, this study helps policy makers to focus on the issue of corruption and circular debt in the energy sector of Pakistan.
Keywords: corruption, circular debt, energy crisis, ARDL bounds testing.
For citation: Seema Arif, Alam Khan and Dilawar Khan (2023) Assessment of the impact of corruption and circular debt on energy crisis in Pakistan', Public Administration Issue, 5 (Special Issue I, electronic edition), pp. 147-164 (in English). DOI: 10.17323/19995431-2023-0-5-147-164.
JEL Classifications: G38, H11, L94.
Acknowledgement: This is part of the M.Phil research work submitted to Kohat University of Science & Technology.
Introduction
The economy is the backbone of a country's progress. The economy can remain healthy if industrial, agricultural, and domestic energy demand is sufficiently met by the government. It cannot be overemphasized that energy is one of the most important factors which fulfilled human needs. The world in the 21st century faces so many oppositions when it comes to energy, such as the availability of sufficient, affordable, and environmentally friendly energy. To overcome the problem of poverty, improve human well-being and raise the standard of living, there are achievable and sufficient energy needs. The South Asia region is rich in natural resources. The energy production of this region depends on the natural resources of coal, oil, natural gas and hydroelectric energy (Iqbal et al., 2019; Nawaz et al., 2013). Natural resources are drained with every tick of the clock, so the chance of turning natural resources into electrical energy becomes low. The energy sector is underdeveloped in Pakistan due to poor planning, supervision and management (Faheem, 2016). Electricity is one of the important sources for providing social services such as education, health, water purification and sanitation and medicine refrigeration. Electricity plays a vital role in generating income opportunities. Electricity is a service of monopolistic quality and social importance due to the problem of governance in the electricity sector. In many countries, only one company is responsible for providing electricity, so customers cannot switch to another company. Some countries have made their energy market competitive to overcome the governance problem in the electricity sector (Halpern et al., 2008). The crisis is not only the gap between demand and supply, but also these emerging crises due to poor governance, outdated technologies, and administrative problems, as well as circular debt damaged the supply chain (Heli et al., 2017). The energy sector that holds monopoly power is subject to corruption. Strong relationship exists between energy efficiency and corruption; greater corruptibility will reduce energy efficiency (Sekarfi and Sghaier, 2018; Qazi et al., 2020). The electricity sector in developing countries is mainly under public control. Public authorities use electricity as an instrument to achieve social, economic and political goals. The public sector holds in concessionaires, which leads to corruption and inefficiency in generation and distribution (Jamil and Ahmad, 2013). Corruption is pervading the electricity sector and imposes significant costs. Rehman and Deyuan (2018) examined the relationship between economic growth, population growth, energy use and access to electricity in Pakistan using the ARDL bounds testing approach.
Pakistan has been struggling with the problem of power shortage for several years. Electricity crisis occurs in the country, and then country initiative to find long-term effective solutions for power supply is strong as compared to now. The main purpose of this study is to gain knowledge about the outcomes of circular debt on Pakistan's electricity crisis. The study will help the policy makers to overcome the corruption problem in the electricity sector of Pakistan and help the manufacturers to mitigate the crisis by controlling population growth and line losses. This study is different from the previous study, because no previous study has examined the effect of circular debt, population growth and industrial growth
on electricity crisis in case of Pakistan. In addition, this study also focuses on distribution and transmission losses in the electricity sector. Therefore, this study helps the policy makers to address the issue of corruption and circular debt in the energy sector of Pakistan.
Literature review
Many studies have been done to explore the electricity crisis and its causes. Abbas (2015) discussed that the socio-economic growth of a country is affected by the performance of the energy sector. The energy sector has generated growth and development in the agricultural, industrial, and the defense sectors, as well as in the domestic sector. Sekrafi and Sghaier (2016) investigated how political stability, environmental quality, energy consumption, and corruption affected the economic growth of 13 MENA regions. They used a static and dynamic panel data approach over the period from 1984 to 2012. Their empirical results showed that an increase in corruption, directly affects environmental quality, economic growth and energy consumption, while economic growth is indirectly affected by corruption through environmental quality and energy consumption. Therefore, carbon emissions and energy consumption also affect economic growth. Furthermore, Imam et al. (2018) found no evidence regarding performance in technical, economic and welfare terms, and these reforms do not reduce or increase corruption. They conducted a study to estimate the impact of corruption on two aspects of electricity reform, having three indicators such as, revenue, access to electricity, and technical efficiency using a new panel dataset from 47 sub-Saharan African countries, for the period from 2002-2013. The study concluded that corruption mitigated technical efficiency and created a barrier to access to energy use and national income. They suggested that well-planned reforms not only influence the economic development, but also indirectly reduce negative institutional deficiencies in the form of corruption, both at the micro and macro levels.
In recent years, Pakistan is facing a drastic energy shortage. One of the reasons for this remarkable power outage, is the tackling in the use of resources. The situations caused not only an energy mix for electricity generation, but also circular debt has contributed to the crisis. The government of Pakistan has paid over one trillion rupees as Differential Tariff Subsidy (TDS) to protect people from the rising cost of generating electricity. The TDS not only overcomes the financial burden, but also the consequence of the loss of well-being. Ali and Badar (2010) examined the problem of circular debt in the energy sector of Pakistan. They presented an overview of the energy sector in the case of Pakistan. According to the study, they explained why circular debt arises in the energy sector. Their study identified the main reasons for the circular debt problem. Firstly, the consumer tariff was unsatisfactory to recover the increasing cost of electricity production and government did not compensate for the losses. The second reason highlighted by the researcher that PEPCO failed to recover dues from consumers. To solve the circular debt problem, the government should accept the cost of power subsidies in the budget and adjust power tariffs. Similarly, Qasim and Kotani (2014) addressed the shortage of electricity in the case of Pakistan. They studied the energy
shortage by using both an index comparing the demand for electricity, oil and gas and the public supply of electricity. They used a time series approach over the period 1971-2010. Their result suggested that price adjustment is not an effective policy to deal with power shortage, rather the government should improve the utilization rate and redirect the use of oil and gas for electricity generation. Awan et al. (2019) developed a different structure to estimate the impact of direct transfer mechanism of the Tariff Differential Subsidy (TDS) on social welfare from Pakistan perspective. The basic structure was to compare the welfare of poor households that paid the TDS directly, with the impact on the fiscal deficit after targeting the subsidies and evaluate the impact on circular debt. The study investigated these impacts by using the 2010-2011 Social Accounting Matrix (SAM) and the Computed General Equilibrium (CGE) model developed by IFPRI. Their results will not only help the policymakers to formulate a long term sustainable result of power outage but help to reduce negative social and economic impacts.
The remainder of the article includes "Data sources and Methodology" where data, data sources and empirical techniques are used to discover the impact of corruption and circular debt on the electricity crisis. The finding is presented in the "Results and discussion" section and the conclusions of the study are given in the "Conclusion" section of the article.
Methodology
Pakistan is a developing country where electricity gap is very high due to the obstruction in electricity supply and the increase in electricity demand. In this study, electricity gap was used as electricity crisis. Some of the problems like corruption and circular debt increase the demand for electricity. Therefore, it is necessary to analyze the impact of corruption and circular debt on Pakistan electricity crisis. In this study time series data over the selected time period from 1990 to2018 was used. The data were taken from the Pakistan Energy Year Books database of HDI (Pakistan Hydrocarbon Development Institute), the World Bank (2020) and Pakistan Economic Research. The Descriptions of the variables are given in Table 1.
Table 1
Description of the data series
Variables Definitions Measurement units Data sources
EC Electricity crisis Billion KW/h World Bank (2020)
CPI Corruption perception index Index The Global Economy (2020)
CD Circular debt Billion PKRs Energy Year Books (2020)
POP Population Number of persons World Bank (2020)
IGDP Industrial gross domestic product PKRs World Bank (2020)
DT Distribution and transmission losses % of electric power output World Bank (2020)
Sources: Compiled by the authors.
This study employed augmented Dickey and Fuller (1981) and Phillips and Peron (1988) test to examine the data series for stationarity. Using ADF test for checking the autocorrelation in the error term by taking the lagged difference terms of the dependent variable. While Phillips and Peron (1988) used nonpara-metric statistical methods to deal with the autocorrelation in the error term without taking lags of difference terms. The ADF and PP tests are unit root tests for checking stationarity of the variables. The ADF test could be used with autocorrelation. The ADF test is used for more complex models, while the Dickey Fuller test is used for simple models and this test is also more powerful. In the ADF and PP tests, we can have no constants, or we can have constants, or we can have these constants and a tendency to agree with the different possibilities. The ADF hypothesis for the unit root test is as follows:
H0 = y = 1 (There is unit root)
H1 = y < 1 (Series is stationary)
The ADF and PP tests used three basic regression models
(1)
The above equations y denotes the slope coefficient of the lag value of the taken variables and denotes the slope coefficients of the descriptive variables, a0 indicates the intercept term, [30 indicates the slope coefficient of the trend; Ht indicates the error terms at time t, and summation i indicates the length of the lags starting from 1 up to p. Phillips and Peron (1988) developed a number of unit root tests that have become widely used in the analysis of financial time series. The PP test for unit roots differs from the ADF tests mostly in the way it handles autocorrelation and heteroscedasticity in the error terms. The PP tests ignore autocorrelation in the test regression. The test regression for the PP tests is as follows:
Where [30 represents the constant terms, while y denotes the slope coefficients, P1 indicates that the coefficient of the trend variable, and et represents the error term at time t. The PP test has the advantage over the ADF test that it is robust to the general form of heteroskedasticity in the error term /j.t. The second advantage of PP is that researchers do not need to specify a lag length for the test regression.
To achieve the proposed objectives, this study used time series data for analysis and estimations of the results; E-views version 10 was used. In this study the ARDL model was used to determine the long-run cointegration between dependent and independent variables, across different levels of integration (Pesaran et al., 2001, Pesaran and Shin, 1999). The reparametrized result provides the short-run dynamics and long-run cointegration of the considered variables. The Pesaran et al. (1996) and Pesaran and Shin (1995) suggested Autoregressive Distributed Lag (ARDL) model for co-integration for a long run co-integration, ignoring weather the taken variables are stationary at level, stationary at 1st difference, or a combination of both. All variables are transformed into logarithmic form, we used log-log model which will reduces the sharpness of the time series data:
p q r
AlogECt = y0 + ^ Yu MogECt_i + ^ y2j MogCPlt_} + ^ y3k MogCDt_k
i=l j=0 k=0 s t u
+ ^ Y^logPOP^ + ^ Ysm MoglGDPt_m ^ Yen MogDTt_n
m=0
n=0
+ n1logCPIt_1 + n2logCDt_1 + n3logPOPt_1 + n4logIGDPt_1 + nslogDTt.
+ ^ ECTt_! + et
(7)
Where A is for differentiation. If there is at least one non-zero coefficient, the alternative is accepted, and rejected the null hypothesis; and it is concluded that the relationship is long-term. We used the F-bound co-integration test. If the calculated Wald statistics which follows the F statistic and the value of the F statistics is less than the lower bound, we concluded that there is no co-integration. If Wald's F statistic is greater than the upper bound of Narayan (2004) or Pesran et al. (2001) at 5%, we conclude that there is a long run relationship. Finally, if the value of the F statistic is between the lower and upper limits, the result is inconclusive.
The study also employed other diagnostic tests to check the predictability, reliability, and constancy of the parameters used in the F-bound test approach. The tests used in this study include the White test to identify the heteroscedasticity problem (White, 1980), the Jarque-Bera (JB) test for the normal distribution of the data (Jarque and Bera, 1987), and the Breusch-Pagan-Godfrey LM test to detect the serial correlation problem (Breusch, 1978; Godfrey, 1978). The research checks the stability of the short-run and long-run coefficients of the ARDL model; we used CUMSUM (Cumulative Sum) of the residuals (Brown et al., 1975).
Results and discussion
The study focused on the impact of corruption and circular debt on Pakistan electricity crisis. For this purpose, the study used time series data and selected time was from 1990 to2018. The section begins with the descriptive statistics of the variables. Table 2 shows the descriptive statistics of variables used in the study. It also provides the basic characteristics of the variables. Table 2 shows basic data information
including data standard deviation, maximum, minimum, mean, and median of the data. The mean data value shows the mean of the data. The median shows the mid value of the data. The standard deviation shows the value spread from its mean value.
Table 2
Descriptive statistics of the data series
EC CPI CD POP IGDP DT
Mean 5.50 0.88 14.0 4.43 3.22 3.3
Median 5.8 0.9 17.1 4.44 3.2 3.13
Maximum 6.17 1.2 20.7 4.9 3.9 5.33
Minimum 2.9 0.7 6.5 4.05 2.6 2.8
Std. Dev. 1.0 0.14 6.0 0.23 0.4 0.7
Sources: Compiled by the authors based on their own calculations (-hereinafter, unless otherwise noted).
Figure 1 represents the trends of the data series. Trends show the impact of time on the data. Trends can be positive or negative. In the figure, most variables show an increasing trend. The values of each data series increase over time. Therefore, the energy crisis shows a fluctuating trend with values increasing in some phases and decreasing in others. Overall, we can conclude that most economic variables increase over time.
Figure 1: Trends of variables used in the model.
The unit root is the problem in time series data where the variance, mean, and covariance of the data are not constant. Before applying the co-integration technique, we make our data stationary. To eliminate the stationarity problem, we use two tests, which are the Augmented Dickey Fuller (ADF) (Dickey and Fuller, 1984) and Phillips and Peron (1988). For these tests, we can use three types of equations (i) taking without intercept and trend (ii) with intercept and without trend (iii) with intercept and trend. Table 3 displays the results of the unit root tests. According to the results, all variables in the ADF and PP tests are stationary at the first difference.
Table 3
Unit root tests results
Variables ADF PP
Without Constant Constant and Trend Without constant and Trend Constant Constant and Trend
EC -0.832 -0.770 -1.20 -0.848 -0.857 -1.096
CPI Level -1.115 -0.808 -2.654 -1.627 -0.333 -2.356
CD -0.602 -1.176 -1.971 -0.805 -0.333 -2.083
POP -0.924 -0.207 -3.803 -2.25 -0.506 -2.77
IGDP -2.232 -0.506 -3.155 -1.954 -2.02 -1.97
DT -0.085 -1.815 -3.155 -0.167 -1.096 -2.22
EC -5.489* -5.514* -5.945* -5.490* -5.525* -6.137*
CPI -4.483* -4.414* -4.95* -4.431* -4.718* -5.175*
CD 1st -6.004* -6.138* -6.004* -5.942* -5.731* -6.02*
POP Difference -21.17* -3.160* -5.349* -13.60* -4.206* -5.36*
IGDP -6.47* -4.97* -4.166* -5.90* -4.644* -4.10*
DT -4.670* -5.25* -4.166* -4.716* -6.15* -4.45*
Note: * T-stat. shows significance.
The ARDL model requirements are selected on the basis of model selection criteria which include five indicators, namely, the Schwarz-Bayes information criterion (SBC / BIC), log-likelihood, Hannan-Quinn criterion (HQ), adjusted R-squared, and Akaike information criterion (AIC). The model selection criteria define the most appropriate model for the study. The most appropriate model is selected on the basis of length of the lags. In the case of the adjusted R-squared, the model with the highest value is preferred. In the case of other selection criteria, e.g., AIC, log likelihood, SBC / BIC, and HQ criteria, the model with the lowest value is preferred. In this study, the best model is selected on the basis of AIC (Khan and Ullah, 2019), that contains a value of -1.51. The best model is used in the research has the specification (1, 0, 0, 0, 0, 0). We chose the top 20 models in Table 4.
Table 4
Model selection criteria
Model LogL AIC* BIC HQ Adj. R-sq Specification
91 28.19 -1.51 -1.02 -1.38 0.94 ARDL(1, 0, 0, 0, 0, 0)
59 28.94 -1.49 -0.95 -1.35 0.95 ARDL(3, 0, 0, 1, 0, 1)
27 29.933 -1.49 -0.90 -1.34 0.94 ARDL(4, 0, 0, 1, 0, 1)
83 28.71 -1.47 -0.93 -1.33 0.95 ARDL(2, 0, 1, 1, 0, 1)
19 30.69 -1.47 -0.83 -1.31 0.95 ARDL(4, 0, 1, 1, 0, 1)
89 28.57 -1.46 -0.92 -1.32 0.94 ARDL(2, 0, 0, 1, 1, 1)
17 31.50 -1.46 -0.77 -1.27 0.95 ARDL(4, 0, 1, 1, 1, 1)
75 28.49 -1.46 -0.91 -1.32 0.94 ARDL(2, 1, 0, 1, 0, 1)
123 26.38 -1.44 -1.01 -1.33 0.94 ARDL(1, 0, 0, 1, 0, 1)
51 29.33 -1.44 -0.85 -1.30 0.95 ARDL(3, 0, 1, 1, 0, 1)
81 29.29 -1.44 -0.85 -1.28 0.95 ARDL(2, 0, 1, 1, 1, 1)
43 29.09 -1.42 -0.83 -1.27 0.94 ARDL(3, 1, 0, 1, 0, 1)
73 29.05 -1.42 -0.83 -1.26 0.94 ARDL(2, 1, 0, 1, 1, 1)
25 30.05 -1.42 -0.78 -1.25 0.95 ARDL(4, 0, 0, 1, 1, 1)
11 29.99 -1.41 -0.77 -1.25 0.95 ARDL(4, 1, 0, 1, 0, 1)
121 26.94 -1.41 -0.92 -1.28 0.95 ARDL(1, 0, 0, 1, 1, 1)
57 28.94 -1.41 -0.82 -1.25 0.95 ARDL(3, 0, 0, 1, 1, 1)
115 26.92 -1.40 -0.92 -1.28 0.95 ARDL(1, 0, 1, 1, 0, 1)
67 28.85 -1.40 -0.82 -1.25 0.95 ARDL(2, 1, 1, 1, 0, 1)
To achieve the proposed objectives, time series data were used for the analysis and estimation of the results; Version 10 of E-views was used. In this research the ARDL bounds testing method, and the ARDL F-bounds co-integration method were used to determine the long-term correlation between series with different orders of integration (Pesaran et al., 2001; Pesaran and Shin, 1999). The reparameterization result provides the short-term dynamics and the long-term correlation of the variables considered. Pesaran and Shin (1995) and Pesaran et al. (1996) offered the ARDL model for cointegration or the F-bound technique for a long-run cointegration, ignoring whether the variables are integrated at all levels, intergraded at 1st difference, or a combination of both. It has previously said that the integration process (making data stationary by the unit root method) does not allow a simple Ordinary Least Squares (OLS)) approach to give long-term relationship. Therefore, it is also compulsory to make the data stationary, because the data with unit root problems give results that we cannot
rely on and this regression is called spurious regression. The spurious regression results are unreliable and predictable, and an efficient and effective decision cannot be made (Granger and Newbold, 1974). This approach includes two parts, the first part is called the F-bound test, which generally specifies the co-integration in the model and the other part is known as the ARDL model, which allows individual short- and long-term results to be obtained. Table 5 presents the results of the ARDL bounds test. The value of the F statistic is 5.68 > then the upper critical limit (4.25) at 5% and K = 5. We accept the alternative hypothesis against null hypothesis and conclude that there is long-term cointegration between Electricity Crisis (EC) with corruption, circular debt, population, industrial gross domestic product and transmission and distribution losses.
Table 5
Results of Bound test for co-integration
Test statistics Value K
F-statistics 5.68* 5
Critical Value Bounds
Significance I0 Bound I1 Bound
10% 2.75 3.79
5% 3.12 4.25
2.5% 3.49 4.67
1% 3.93 5.23
Note: * 5% shows significance.
Table 6 shows the results of the ARDL model, the variables with A sign show the short-term impacts, while the variables without A provide the long-term results. As the double log model was used the coefficients of the variables represent the short- and long-term elasticities, respectively. The selection of lags is based on the Akaike Information Criteria. AIC was used to determine the most appropriate length of lags in the model (Khan and Ullah, 2019). We compare the p-value to decide when the p-value < 0.05 (5%) or P < 0.01 (1%). We accept the alternative hypotheses and conclude that there is a short-term relationship. We also consider the t statistics > critical value t, so we accept that there is a short-term relationship for individual coefficients. Here we use the F values to combine the result, because we selected the model based on the AIC, if the F-statistic > F-tab statistics we accept the alternative hypothesis and conclude that there are short-term impacts.
The coefficient of Corruption Perception Index (CPI) is 1.36; the probability value of CPI is 0.001, which is significant at 5%. Therefore, a 1% increase in CPI causes a 1.3% increase in the electricity crisis in the short term, keeping other factors constant. The value of CPI coefficient 1.3% > 1% is elastic in the short term reveals that when a 1% corruption increases it will bring a 1.3% increase
in the electricity crisis. The CPI coefficient is highly influential in the long term. The long-term CPI coefficient is 0.973, which is significant at 1%. The long-term elasticity is also less than 1, which shows that CPI is also long-term inelastic. A1% increase in CPI will bring a 0.973% increase in the electricity crisis in Pakistan. The circular debt (CD) coefficient is 0.02, containing 0.02 probability values, which is significant at 5%. CD is inelastic in the short term, i.e., 0.02 <1. The results showed that a 1% increase in CD will bring a 0.02% increase in the electricity crisis in the short term, keeping other factors constant. The CD result is different in the long run. The long-term coefficient is 0.014, which is significant as it contains the p-value 0.01. Long-term elasticity is inelastic, which is less than 1%, i.e., 0.014 <1%. A 1% increase in circular debt in Pakistan leads to a 0.014% increase in electricity crisis in energy sector. The population (POP) coefficient is 0.995 with a probability value of 0.001, which is significant at 5%. Population is inelastic in the short term i.e., 0.995 <1% reveals that population growth is less affected by electricity consumption. Statistics have shown that a 1% increase in population leads to a 0.995% increase in the electricity crisis, keeping other factors constant. Pakistan's population growth has the greatest long-term impact.
The increase in population is directly and significantly related to electricity consumption as stated by Jones (1991), Shiu (2004), Shehbaz (2012), and Imran and Naseem (2015) in the case of Pakistan. The long-term coefficient is 27.23, containing a probability of 0.0004, which is significant at 5%. The population is inelastic in the long term, that is, 27.23 > 1% showed that population growth has a great influence on the electricity crisis. A 1% increase in population leads to a 27.2% increase in Pakistan's electricity crisis (Heli et al., 2017). The industrial gross domestic product is one of the important factors causing the electricity crisis. The transport coefficient of the result -0.44 with a p value of 0.5, which is not significant at 5% and at 1%. The short-term elasticity of the IGDP is inelastic less than 1, that is, -0.44 <1. The results revealed that a 1% increase in industrial growth leads to a 0.44% reduction in the electricity crisis, keeping other factors constant. The result of industrial growth is different in the long run. The IGDP coefficient is -0.32 with a significance of 5% having a probability value of 0.03. The long-term IGDP is inelastic at a 1%, i.e., -0.32 > 1. A 1% increase in industrial growth will ease the -0.32% electricity crisis while keeping other factors constant. The consistent study was employed by Imoro Brainmah (2012) in Ghana, Nwankwo (2013) in Nigeria, and Kiran (2016) in Pakistan (Juneju and Khoso, 2018). Distribution and transmission losses in the energy sector have less impact compared to the long term. The distribution and transmission losses coefficient (DT) () is 2.07 and significant because the p-value is 0.001. A 1% increase in DT leads to a a 2.07% change in the electricity crisis. In the short term, distribution and line losses at each point cut off electricity (Bhatti et al., 2015). In the long run, DT causes a crisis. The coefficient is 1.5, which is significant as it contains the p-value of 0.001. DT is elastic in the long run. The small change in DT means Pakistan's line losses because of technical or non-technical causes of load shedding in the country. In the long run, a 1% increase in DT leads to a 1.5% increase in the electricity crisis, keeping other
factors constant (Bhatti et al., 2015). The error correction term (ECT) discovers the dynamic stability of a model in the study. Dynamic stability means that the short-term imbalance in a model automatically leads to long-term equilibrium. The ECT also shows by period (e.g., by year, by month, etc. that depends on the period of the data collected) the model's fit effect and specifies the time period in which the model will enter an equilibrium position. For dynamic stability, the ECT coefficient must be negative and significant. In our results, the value of the ECT coefficient is -1.40, which satisfies the condition of a negative values of the coefficient. The probability value of 0.001 of the ECT coefficient is highly significant at 1%, showing that our model is dynamically stable. The value of the ECT coefficient shows that there is a 140% adjustment effect in the model indicating that the short-term equilibrium is adjusted by 0.71% in one year.
Table 6
Results of ARDL model
Variables Short run Elasticities
Coefficient t-statistics p-value
ALCPI 1.365* 2.33 0.001
ALCD 0.02 1.16 0.20
ALPOP 0.995* 3.866 0.0012
ALIGDP - 0.44 - 0.7 0.540
ALDT 2.07* 11.3 0.001
ECTt-i -1.4* - 14.9 0.001
Long run Elasticities
LCPI 0.973* 2.431 0.026
LCD 0.014* 1.190 0.0128
LPOP 27.231* 4.380 0.0004
LIGDP -0.320* -0.625 0.0365
LDT 1.5* 17.34 0.001
Constant -99.440* -3.991 0.0009
Log likelihood 28.195
AIC -1.5163
BIC -1.025
HQ -1.38608
Adj. R2 0.94
ARDL (1,0,0,0,0,0)
Note: * shows significance.
In this study, different diagnostic tests were used to check the stability, reliability, and predictability of the series. Table 7 shows the results of the diagnostic tests. The White test (White, 1980) was employed to check the heteroscedasticity problem. The heteroscedasticity problem indicates that the variance of the error terms is not constant. It follows the chi-square (x2) distribution. The rule for making decision is, if the probability value of the x2 statistic is less than 0.05, we accept the alternative hypothesis H} which states that there is a heteroskedasticity problem in the model. If the probability value of the x2 statistic is greater than 0.05, we accept the null hypothesis H0 showing that there is homoscedasticity (no heteroscedasticity problem) in the data. In the results of this study, the probability value of the x2 statistic is 0.98, which is greater than 0.05. This means that we accept the null hypothesis H0 which states that homoscedasticity exists, i.e., there is no heteroscedasticity problem in the model. The Breusch-Pagan-Godfrey LM test (Breusch, 1978; Godfrey, 1978) is employed to check the serial correlation problem. The serial correlation also shows the existence of a relationship between the residuals. In this case, the correlation must exist between the error terms of two consecutive periods. The chi-square distribution (x2) follows. Table 7 shows the results of the Breusch-Pagan-Godfrey LM test. The decision rule is that, if the probability value of x2-statistic is less than 0.05, we accept the alternative hypothesis Hj, which shows the existence of the problem of serial correlation whereas, if the probability value of x2-statistic is greater than 0.05, we accept the null hypothesis H0, which indicates that there is no serial correlation problem in the data. The result of our study indicates that, the probability value of x2-statistic is 0.45 which is greater than critical value at 5% (i.e., 0.45 > 0.05). It expresses that we will accept the null hypothesis H0, which designates that the model does not have the problem of serial correlation. The study also used the Jarque-Bera test (Jarque and Bera, 1987) for testing normal distribution of the data. The normal distribution exposes that the data are equally distributed on both sides of their mean value. The Jarque-Bera test follows the F-distribution. The decision rule is that, if the probability value of the F-statistic is less than critical value at 5% (< 0.05), then we accept the alternative hypothesis Hj showing that the data are not normally distributed and if the probability value of the F-statistic is greater than critical value at 5% (> 0.05), then we accept the null hypothesis H0 showing that the data are normally distributed. The outcomes of this study, the probability value of F-statistic is which is greater than critical value at 5% (i.e., 0.46 > 0.05). This shows that we will accept the null hypothesis H0, which states that our data are normally distributed.
Table 7
Results of Diagnostic Test
Test Type Statistics Value DF Probability
White Test X2-statistic 0.57 23,1 0.98
Breusch-Pagan-Godfrey LM test X2-tatistic 0.82 23,1 0.45
Jarque-Bera test F-statistic 1.54 - 0.46
Figure 2 explains the results of the CUSUM test, which is used to test the stability of the model (Brown et al., 1975). If the results of both used techniques are significant at 5%, we accept the alternative hypothesis H which states that the parameters are stable if the results of both tests are insignificant at 5%, this specifies that the acceptance of the null hypothesis H0 showing that the parameters are unstable. Figure 2 shows that the results of the CUSUM test are significant at 5%. Therefore, we reject the null hypothesis and accept the alternative hypothesis H1, which proves that the parameters are stable.
12
2002 2004 2006 2008 2010 2012 2014 2016
-CUSUM -----5% Significance
Figure 2: Result of CUSUM test
Conclusions and recommendations
The aim of this research was to examine the impact of corruption on the electricity crisis in Pakistan. This study used annual time series data from 1990 to 2018. The study used electricity crisis as the dependent variable, the electricity crisis is measured by the difference of electric power consumption and electric power production in KWh, the five independent variables namely corruption, circular debt, population and distribution and transmission losses. At first, the ADF test and PP test were used to examine the problem of unit root and made series stationary. Both tests were applied to three types of equations such as (i) without intercept and trend, (ii) with intercept and without trend (iii) with intercept and trend. Based on the ADF test, decided that all variables are stationary at 1st difference i.e., I (1) or at level. The results of the ADF without intercept and trend were stationary at the first difference having different lags, as well as the results of the ADF with intercept were also stationary at the first difference while the variable POP (Population) was stationary at level having intercept and trend factor. To eliminate the problem of unit root, the study checks short run and long run impact of variables using the ARDL F- bound test. The outcome of ARDL reveals that corruption and circular debt have positive and significant
impact on electricity crisis. corruption has high influence on the electricity sector in Pakistan. The amount of circular debt in the power sector rises per year due to non-payment to IPPs the electricity shortage arises in each sector in Pakistan. The corruption in the power sector and bribery causes less amount of electricity crisis in the short run. In addition, the circular debt also affects the electricity crisis. Corruption in the power sectors highly influenced all areas of the power sector. Many reasons cause the electricity crisis. This study used population growth as the cause of the crisis, when the result had a positive and significant impact on the power sector in both the short and long term. With industrial growth more machinery was used for operation, so the electricity demand increased. Industrial growth has a negative and significant impacts on the electricity crisis which means that when the industry expanded the supply of electricity decreases, leading to a reduction in the production. The industrial gross domestic product carries negative and significant impacts electricity crisis, both in the long and short term. Distribution and transmission losses in Pakistan were also one of the important determinants that caused the electricity crisis. In the short run, the distribution and transmission losses of power lines have a strong impact on the supply of electricity to the population. While in the long run the distribution and transmission losses have less impact as compared to short run. The distribution and transmission losses are of technical and non-technical nature, the result of this study showed that both types of losses cause electricity crisis. The result of the ARDL F-bound test also shows that all the variables used in the model are co-integrated. The proper lags length in the ARDL model was determined based on AIC. After testing co-integration in the model, to investigate the reliability, stability and predictability of parameters, some of the diagnostic tests were used. The findings of the White test showed that the selected ARDL model is free from the problem of heteroscedasticity. The Breusch-Pagan-Godfrey LM test confirmed that the model is free from problem of autocorrelation/ serial correlation. The results of the Jarque-Bera test confirmed that the data employed in this study are normally distributed. The study also used CUSUM test to check the stability of model's parameters. The CUSUM test confirmed that parameters are stable.
The results of the study proved that corruption is one of the reasons for load shedding and electricity outage in Pakistan. This study concluded that circular debt is one of the main drivers of Pakistan's electricity crisis, which could be traced back to the area where payment disruption arises and to reduce the debt burden of the electricity sector. Some recommendations based on the findings of this study can be suggested:
1. Strict government policies against corruption can be a vital in the energy sector. Bribery is also a part of corruption. It also needs considerable steps to eradicate corruption.
2. Circular debt is an extra burden on the economy. Each institution in the electricity sector should make sure to clear its dues.
3. The introduction of electrically efficient equipment will reduce the energy crisis. Rapid population growth will not be a major cause of the energy crisis. It will also allow the commercial sector to use energy more efficiently.
4. The efficient use of electricity through the introduction of advanced devices will encourage the industrial sector to contribute more to Pakistan's GDP.
5. New electricity production projects are needed to overcome the energy crisis. CPEC project will be one of the government's major initiatives to overcome the problem of Pakistan's electricity crisis. It is important to note that green projects will be more beneficial for both people and economy.
REFERENCES
1. Abbas, M.N. (2015) Energy crisis in Pakistan. Naval Postgraduate School Monterey United States.
2. Ali, S.S. and Badar, S. (2010) 'Dynamics of circular debt in Pakistan and its resolution', Lahore Journal of Economics, 15.
3. Awan, H.S., Samad, G. and Faraz, N. (2019) Electricity subsidies and welfare analysis: The perspective of Pakistan. Pakistan Institute of Development Economics.
4. Bhatti, S.S., Lodhi, M.U.U., Haq, S., Gardezi, E.S.N.M., Javaid, E.M.A., Raza, M.Z. and Lo-dhi, M.I.U. (2015) 'Electric power transmission and distribution losses overview and minimization in Pakistan, International Journal of Scientific and Engineering Research, 6(4), pp. 1108-1112.
5. Brown, R.L., Durbin, J. and Evans, J.M. (1975) 'Techniques for testing the constancy of regression relationships over time', Journal of the Royal Statistical Society: Series B (Methodological), 37(2), pp. 149-163.
6. Dickey, D. and Fuller, W. (1981). Likelihood ratio statistics for autoregressive time series with a unit root. Journal of Econometric analysis, 49(4), pp. 1057-1072.
7. Faheem, J.B. and Mir, A.T. (2016) 'Energy crisis in Pakistan', IRA-International Journal of Technology and Engineering, 3(1), pp. 2015-16.
8. Godfrey, L. (1978) 'Testing for higher order serial correlation in regression equations when the regressors include lagged dependent variables', Econometrica, 46(6), pp. 1303-10.
9. Granger, C.W.J. and Newbold, P. (1974) 'Spurious regressions in econometrics', Journal of Economics, 1(2), pp.111-120.
10. Hali, S.M., Yong, S.I.D.W. and Kamran, S.M. (2017) 'Impact of energy sources and the electricity crisis on the economic growth: Policy implications for Pakistan', Journal of Energy Tech. and Policy, 7(2).
11. Halpern, J., Kenny, C. and Dickson, E. (2008) Deterring corruption and improving governance in the electricity sector: a sourcebook. No. 46708, The World Bank, pp. 1-245.
12. Sumera Iqbal, Jianxun Chu and Shafei Moiz Hali (2019) 'Projecting impact of CPEC on Pakistan's electric power crisis', Chinese Journal of Population Resources and Environment, DOI: 10.1080/10042857.2019.1681879
13. Imoro Braimah, O.A. (2012) 'Causes and effects of frequent and unannounced electricity blackouts on the operations of micro and small-scale industries in Kumasi, Journal of Sustainable Development, 5(2), pp. 17-36. D01:10.5539/jsd.v5n2p17.
14. Imran Nasee and Jawad Khan (2015) 'Impact of energy crisis on economic growth of Pakistan', International Journal of African and Asian Studies, 7, pp. 33-42. Available at: https:// core.ac.uk/download/pdf/234689844.pdf (accessed 02 February 2023).
15. Iqbal, S., Chu, J. and Hali, S.M. (2019) 'Projecting impact of CPEC on Pakistan's electric power crisis', Chinese Journal of Population Resources and Environment, 17(4), pp. 310-321.
16. Jamil, F. and Ahmed, E. (2013) An economic investigation of corruption and electricity theft. No. 2013. Pakistan Institute of Development Economics.
17. Jarque, C.M. and Bera, A.K. (1987) 'Regression Residuals', International Statistical Review/ Revue Internationale de Statistique, 55(2), pp. 163-172.
18. Jones, D. (1991) 'How urbanization affects energy-use in developing countries, Energy Policy, 19(7), pp. 621-630.
19. Junejo, I. and Khoso, J.R. (2018) 'Impact of electricity crisis on industrial growth in Pakistan: A multiple regression analysis approach', International Journal of Academic Research in Business and Social Sciences, 8(9).
20. Khan, D. and Ullah, A. (2019) 'Testing the relationship between globalization and carbon dioxide emissions in Pakistan: Does environmental Kuznets curve exist?', Environmental Science and Pollution Research, 26(15), pp. 15194-15208.
21. Kiran, A. and Kiran, F. (2016) 'Impact of electricity crisis on Pakistan textile Industry', International Journal of Economic and Business Review, 4(1), pp. 15-23. Available at: https:// eprajournals.com/IJES/article/8793/abstract (accessed 02 February 2023).
22. MRS NWANKWO, O.C. (2013) 'The effect of electricity supply on industrial production within the Nigerian economy (1970-2010), Journal of Energy Technologies and Policy, 3(4), pp. 34-42.
23. Shahbaz, M. and Feridun, M. (2012) 'Electricity consumption and economic growth empirical evidence from Pakistan', Quality and Quantity, 46 (5), p. 1583-1599.
24. Nawaz, S., Iqbal, N. and Anwar, S. (2013) 'Electricity demand in Pakistan: A nonlinear estimation', The Pakistan Development Review, 52(4), pp. 479-491.
25. Pesaran, M.H., Shin, Y. and Smith, R.J. (2001) 'Bounds testing approaches to the analysis of level relationships', Journal of applied econometrics, 16(3), 289-326.
26. Phillips, P.C.B. and Perron, P. (1988) 'Testing for a unit root in time series regression', Biom-etrika,75(2), pp. 335-346.
27. Rehman, A. and Deyuan, Z. (2018) 'Investigating the linkage between economic growth, electricity access, energy use, and population growth in Pakistan', Applied sciences, 8(12), 2442.
28. Qasim, M. and Kotani, K. (2014) 'An empirical analysis of energy shortage in Pakistan', Asia-Pacific Development Journal, 21(1), pp. 137-166.
29. Qazi, U., Iqbal, S., Zaheer, H. and Ur, R.T. (2020) 'Prepaid metering: A way forward for sustainable power sector in Pakistan', Energy Strategy Reviews, 31, 100546.
30. Sekrafi, H. and Sghaier, A. (2018) 'Examining the relationship between corruption, economic growth, environmental degradation, and energy consumption: A panel analysis in MENA region', Journal of the Knowledge Economy, 9(3), pp. 963-979.
31. Sekrafi, H. and Sghaier, A. (2018) 'The effect of corruption on carbon dioxide emissions and energy consumption in Tunisia', PSU Research Review, 2(1), pp. 81-95.
32. Shiu, A. (2004) 'Energy consumption and economic growth in China', Energy Policy, 32(1), pp. 47-54.
33. White, H. (1980) 'A heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity, Econometrica, 48(4), pp. 817-838.
34. World Bank, (2009) Reducing technical and non-technical losses in the power sector. Energy sector strategy background paper. Washington, DC: World Bank.
35. World Bank, (2017) World Development Indicators. Available at: https://databank.worldbank.org/ (accessed 02 February 2023).
OFFICIAL DOCUMENTS
1. HDIP. Energy Yearbook. Government Annual Report, Karachi: Hydrocarbon Development Institute of Pakistan, 1997-98.
2. HDIP. Energy Yearbook. Government Annual Report, Karachi: Hydrocarbon Development Institute of Pakistan, 1999-200.
3. HDIP. Energy Yearbook. Government Annual Report, Karachi: Hydrocarbon Development Institute of Pakistan, 2003-04.
The article was submitted: 04.09.2021; approved after reviewing: 07.02.2022; accepted for publication: 25.03.2023.
