Microsoft Word - 001.docx CHEMICAL ENGINEERING TRANSACTIONS VOL. 66, 2018 A publication of The Italian Association of Chemical Engineering Online at www.aidic.it/cet Guest Editors: Songying Zhao, Yougang Sun, Ye Zhou Copyright © 2018, AIDIC Servizi S.r.l. ISBN 978-88-95608-63-1; ISSN 2283-9216 Carbon Emission Performance in Chinese Extractive Industry Based on Stochastic Frontier Model and Stochastic Convergence Test Fengtao Hua School of Economics and Management, Anhui Normal University, Wuhu 241003, China hittile@ahnu.edu.cn The purpose of this study is to estimate the carbon emission performance and time trend of the extractive industry in China. In this paper, the stochastic frontier analysis was used to analyse the carbon emission intensity and estimate the potential of carbon emission reduction in this industry. Besides, the stochastic convergence model was applied to measure the difference of carbon emission intensity in this industry and test the time trend of carbon emission convergence. Finally, it’s concluded that at present, firstly, the carbon emission efficiency of the extractive industry takes a slowly rising trend, with greater potential for improvement; secondly, there exists the gap in carbon intensity between the sub-industries of the extractive industry, without any convergence trend. This reveals that the effective way to reduce carbon emissions of extractive industry is to narrow down the interindustry gap in carbon intensity. 1. Introduction 1.1 Research purposes With the energy and environment pressure increasing, the low-carbon road is an inevitable choice for China’s economic development in the future. China clearly puts forward the goal of carbon emission reduction by 2020. By then, the unit GDP will drop by 40-50% compared with 2005. In the Thirteenth Five-Year Plan for National Economic and Social Development, it’s proposed that CO2 emissions will reach its peak around 2030, and the carbon emissions per unit of GDP in 2030 will be 60%-65% lower than those in 2005. Non- petrochemical energy will account for about 20% of primary energy consumption (Lu et al., 2010). The extractive industry is a large carbon emitter in the energy consumption of all industries and is one of the traditional high-carbon industries. In order to fulfil the tasks of energy conservation and emission reduction and reduce the carbon emission intensity, it is necessary to effectively evaluate the carbon emission performance of the extractive industry. 1.2 An overview of carbon emission efficiency Carbon emission efficiency refers to the economic benefits from the carbon emissions caused by the activities of social economy entities. There are two types of indicators about the carbon emission efficiency: single factor efficiency indicator and total factor efficiency indicator (Herrala and Goel, 2012; Bretschger et al., 2011). The former means the ratio of total carbon emissions to one certain factor, such as CO2 emissions per unit of energy consumption or carbon emissions per unit of GDP (Hermeling et al., 2013; Fischer and Springborn, 2011). This indicator has a diversity of ratios, ignoring the relationship between other factors of economic activity and carbon emissions, with low precision. The latter refers to the economic effects based on the maximum expected output and minimum carbon emissions in the given conditions of input factors and technological endowments; actually, it means to calculate the technical efficiency of the decision unit under the constraints of carbon emissions. DOI: 10.3303/CET1866088 Please cite this article as: Hua F., 2018, Carbon emission performance in chinese extractive industry based on stochastic frontier model and stochastic convergence test, Chemical Engineering Transactions, 66, 523-528 DOI:10.3303/CET1866088 523 mailto:hittile@ahnu.edu.cn 2. Measurement method of carbon emission efficiency Determining the production frontier boundary is the key to the measurement of technical efficiency. In this paper, the stochastic frontier analysis (SFA) was adopted, which is the production function of cross-sectional data independently proposed by Aigner et al. (1977) and Meeusen and Broeck (1977). Based on this function, Battese and Coelli (1995) made some improvements to process panel data and expand the application scope of model. Therefore, this paper selects this model to measure the carbon emission efficiency: )exp();( itititit uvxfY   (1) Y is the output, x the input vector, and β the estimated parameter. The error term is the complex structure. Let vi and ui be the same as above, z is the variable that affects the technology inefficiency, and δ is the coefficient vector of the influencing factor. In order to reduce the risk of estimation bias by the production function error, the more flexible production function was selected in this paper. Taking the total industrial output (Y) as output and carbon dioxide (CO2), labour (L) and capital (K) as input, formula (1) can be further expressed as itititit itititit ititit itititit uvInLInK InLInCOInKInCO InLInKInCO InLInKInCOInY     )()( )()()()( )()()( 9 2827 2 6 2 5 2 24 32210     (2) Subtracting In CO2 from both sides of formula (2), it’s given as: itititit itititit ititit itititit it it uvInLInK InLInCOInKInCO InLInKInCO InLInKInCOIny CO Y In     )()( )()()()( )()()( )1( 9 2827 2 6 2 5 2 24 32210 2     (3) Let yit=Yit/CO2, the carbon emission efficiency is defined as the ratio of the expected output per unit of CO2 output to the expected value of the production frontier boundary, i.e.: )exp( )0|( )( it itit it it u uyE yE TE    (4) Formula (4) shows that the carbon emission efficiency value is between 0 and 1. The closer to 1, the higher the efficiency, and being equal to 1 means that it reaches the boundary of the production frontier and the existing technology is fully utilized (Ang, 2005; Ang et al., 1998). The parameters to be estimated in the formula were calculated using the simultaneous likelihood estimation method. The measurement software was Frontier4.1. 3. Data selection and empirical results According to the methods in the IPCC Guidelines for National Greenhouse Gas Inventories (2006), the carbon emissions of various sub-industries in extractive industry is calculated by the product of various energy consumption and the corresponding CO2 emission factor; the sample period is from 2007 to 2015; There are seven types of energy: coal, coke, gasoline, kerosene, diesel, fuel oil and natural gas. The CO2 emissions is calculated as:   )(2 jijtit EFECCO (5) where, CO2it is the total carbon emissions of i industry in t year; ECijt is the standard coal consumption of j industry in i year; EFj is the carbon emission factor of energy j. Standard coal and CO2 conversion standard coefficients for various energy sources are as follows Table 1. 524 In this paper, the input variables included CO2 emissions, capital investment (net fixed assets), and labour input (average number of all employees) in the extractive industry; the output indicators were the main business incomes of various industrial sectors. According to the industry classification of China Industrial Economic Statistics Yearbook (2007-2015), the extractive industry is divided into seven sub-industries. A total of 410 observed values/sub-industries were collected. Table 1: Standard Coal and CO2 Conversion Coefficient for Various Energy Sources Energy Types Standard Coal Conversion Coefficient (kg Standard Coal/kg) CO2 Emission Factor (kg/Standard Coal) Coal 0.714 2.791 Coke 0.974 3.134 Petrol 1.471 2.039 Kerosene 1.471 2.039 Diesel Oil 1.457 2.168 Fuel Oil 1.429 2.265 Gas 1.330（kg/Standard） 1.624 Note: The converted standard coefficient of coal was derived from the China Energy Statistical Yearbook 2014, and the CO2 emission factor was calculated by IPCC (2006). Table 2: Variable-definition Variable Symbol Definition Industry Output Y Main Business Income ($100 million), 2007 as The Base Year Labour Input L Employees at the end of the year (tens of thousands of people) Capital Input K Net Annual Fixed Assets ($100 million), 2007 as The Base Year Carbon Dioxide CO2 Carbon Dioxide Emissions (Million Tons) Based on the stochastic frontier model, the simultaneous maximum likelihood estimation was made for the 7 sub-industries in the extractive industry from 2007 to 2015 to obtain the calculated results of the parameters to be estimated as shown in Table 3 (calculation results are retained as 3 decimal places). Table 3: Regression result of carbon emission result in China's industrial sector variable Coefficient (T value) variable Coefficient (T value) β0 3.058***(9.004) β5 -0.302***(5.401) β1-1 -1.947***(11.497) β6 -0.094**(2.074) β2 0.558***(5.221) β7 -0.127***(3.911) β3 0.771***(3.246) β8 -0.233***(4.990) β4 0.574***(12.875) β9 0.179***(2.215) σ2 0.703***(12.817) γ 0.842***(5.004) Log function value 998.024 Note: The t-statistics are indicated in parentheses, and ***, **, and * indicate rejection of the null hypothesis at the significant levels of 1%, 5%, and 10% respectively. According to Table 3, the coefficient (β1-1) is -1.947, indicating that as the carbon emission scale increases, the carbon emission efficiency gradually decreases. The ranges of the values (β1~β6) and the coefficient symbol are in accordance with the economic implication and statistical significance. In addition, the γ value was 0.842, which was significant at the 1% level, with the satisfactory fitting degree, indicating that the inefficiency of carbon emissions is the main reason that the extractive industry deviates from its production frontier. Table 4 shows that in the extractive industry of China, the carbon emissions from mining and washing, oil and gas extraction, and other mining industries are highly efficient, with carbon emission efficiency above 0.8; in the non-ferrous mining and non-metallic mining and dressing, the carbon emission efficiency is the lowest, which does not exceed 0.5, with the difference of about 0.3 from the highest three industries. This indicates that there are great differences in the carbon emission efficiency of various sub-industries in the Chinese extractive industry. Also, the overall carbon emission efficiency of the industry in Chinese extractive industry is approximately 0.61 to 0.74 every year, that is, the annual carbon emission reduction potential is approximately 24% to 39%. From Fig.1 (carbon emission efficiency of various sub-industries in the extractive industry), it can 525 https://www.baidu.com/link?url=ckk6RsHJg7MdYlJkUkfscNVskSm_I8XGCK_XA9pE01owJeLVxZd8ixZJrYFz4i9mliyOM53AHDpu80fVvOtm3gb7q3HmJsC93TM1Uq4P-jNYuVFMqxeYr3FoKM16Kh6Z&wd=&eqid=cc5010ee0000f242000000045b0042a2 https://www.baidu.com/link?url=45F2U_uQwY7y92Lq1We2v3sNSgkw7FC6Lrq6VkbuPjbT4ROUSYXNG5cgax7t6qLl1qRDUDR_JfU1pp9b4o3OGaOrMjwVU0TcyeDDIC62zcW&wd=&eqid=bf96772700014860000000025b0041df https://www.baidu.com/link?url=U5zfSL9auK6Enh0VC0FjeL4jy3vEc4exZGRAJi2dKgvK0n1XoqC8F4WRyAgc5qi5c5HnuqnsRbsu5CZ2Z4W17bhOP8ECi4ZQL0ZUMdCq2zoD3kWEz3EGM_Oq7BfD1GeA&wd=&eqid=8b97ec150000f2ba000000025b003d88 https://www.baidu.com/link?url=EfMJd36uDKv3DY868V1I9yKfYsCMHb8P4yuSXKHqqJbIb4QAsUIsKh4moxES9mRsC9UMVnpcr2Z7WMjk4QSSRQa5NAvsuFFCR4jtvTq9pLFNyRsZrxirUtkv67-9n4M0&wd=&eqid=cfbcb6890000dffa000000025b003db3 https://www.baidu.com/link?url=kaZjQ-9M8NzSL_6CfIHVLYz7x8AccRvbut9Td6ier7OecMkqyWyg0Fl_EbBkuKJoPi8MpgU7BTCsPQ7_LipgdGjjmp2UhYewk7shLBAh0rKCXXCMT9KyShGuf_UxZcPa&wd=&eqid=806677040000e60a000000025b003de3 https://www.baidu.com/link?url=w_9FV5mbaGDQ3rfKE9drwwNlXL0-pVaYmsUgDNk-I5aPohuBAjurBFvYFkCMmknm3nYT0s0_FvKgytM06NHFLTBvka2I81R4YqvQ9mqxMiS&wd=&eqid=a6f42255000107cd000000025afd6ea3 https://www.baidu.com/link?url=w_9FV5mbaGDQ3rfKE9drwwNlXL0-pVaYmsUgDNk-I5aPohuBAjurBFvYFkCMmknm3nYT0s0_FvKgytM06NHFLTBvka2I81R4YqvQ9mqxMiS&wd=&eqid=a6f42255000107cd000000025afd6ea3 https://www.baidu.com/link?url=P3dN81ArB9ZfnFACZkUnuoUZH5pgb9c2ePHgjGvhpyRqWNqKGQNDjN5uXuFakacPewR0wXCmRkMQoBccyHSk6xt7fr3LJzqCiP9iFdlvU2L5iJsequiiLtu1ZAY3Kar4&wd=&eqid=cd78764800013ca5000000045afd7458 be seen that, in addition to the auxiliary industries for extraction, the carbon emission efficiency of other industries shows a gradual upward trend from 2007 to 2015. The entire extractive industry rose from 0.617 in 2007 to 0.737 in 2015, a slight decrease from 2011 to 2013. According to the above data analysis, the overall efficiency of the industry is not high and the increase is small and unstable(Springmann et al., 2015). There is a huge space for carbon emission reduction. Therefore, the carbon emission efficiency of the extractive industry, especially non-ferrous mining and non-metallic mining and dressing should be improved. It is of great significance for the extractive industry to take the “green” sustainable path of energy conservation and emission reduction. Table 4: Carbon Emission Efficiency in Various Industries 2006 2008 2009 2010 2011 2012 2013 2014 2015 Mining and Washing of Coal 0.801 0.827 0.892 0.801 0.881 0.877 0.865 0.880 0.887 Petroleum and Natural Gas 0.809 0.804 0.811 0.869 0.842 0.884 0.854 0.865 0.870 Ferrous Metal Ores 0.414 0.447 0.481 0.524 0.568 0.506 0.533 0.593 0.590 Non-Ferrous Metal Ores 0.515 0.491 0.528 0.516 0.581 0.549 0.552 0.564 0.591 Nonmetal Ores 0.817 0.804 0.821 0.807 0.857 0.864 0.860 0.871 0.902 Support Activities Mining 0.413 0.424 0.416 0.485 0.491 0.410 0.496 0.451 0.424 Mining of Other Ores 0.801 0.817 0.827 0.834 0.858 0.905 0.911 0.922 0.928 All Industry 0.617 0.621 0.610 0.635 0.665 0.670 0.658 0.710 0.737 Figure 1: Carbon emission efficiency of various sub-industries in the extractive industry 4. Analysis for stochastic convergence of carbon emission efficiency in various sub- industries of extractive industry Stochastic convergence is to test whether one variable has a persistent impact on another variable. It effectively solves the problem whether there exists the convergence in the short term (Shrestha & Timilsina, 1996; Torvanger, 1990). Assuming that the relative carbon intensity of each industry tends to its respective compensating-differentials-equilibrium level over the long term, and it does not change over time, then, the relative carbon intensity RCI (relative carbon intensity) of each industry at time t can be written as the sum of RCIe and ut: t e t uRCIRCI  (6) tt vtvu   0 (7) where, RCIe is the equilibrium level that does not change with time; ut is the deviation degree of the relative carbon emission from the equilibrium level, which is decomposed into a definite linear trend and stochastic process; v0 is the initial deviation of the carbon emission intensity from the equilibrium level, and β is the deterministic convergence rate. Then, the formula (6) is re-written as: tt vtRCI   (8) 526 where, α=RCIe+v0. If RCIt doesn’t have a unit root, then the impact on RCIt is only temporary, and will still return to its compensating-differentials-equilibrium level in the long term, indicating that the carbon emission intensity of the industry is randomly converged. In this paper, based on the method of Evans & Krass (1996) and Carlino & Mills (1996), it’s assumed that for each industry, at n=1, 2, ∙∙∙∙∙∙; if and only if the difference between the carbon emissions degree ynt of n industry in year t and the average y*t of the carbon emissions of all industries during the year t is the stable series, the carbon emission intensity of these n industries shows the convergence trend. tn i ittnttnnnttn uyyyyyy , 1 * 1, * 11, * , )()()(       (9) In formula (9), the convergence should be determined by whether the autoregressive parameter ρn is zero: if the carbon emission intensity between industries is convergent, then ρn is negative; if it’s divergent, then ρn is zero. The calculation formula for the relative carbon intensity RCI of various industries is: )()()( * ,* ,, t t y y InyInyInRCI tn tntn  (10) According to the meaning of the above model, the test was made for the stochastic convergence of relative carbon emissions, i.e., whether there exists the unit root in formula (11). The unit root test is divided into two types: variable unit root test and panel data unit root test. The former includes IPS, ADF-Fisher, and PP-Fisher method; the latter includes ADF, PP, KPSS, DF-GLS, and MZ test method. Considering the short span of the sample periods in this paper, three test methods of IPS, ADF-Fisher and PP-Fisher, as well as the three lag- periods were selected. The test results are as follows: Table 5: The Effect of Random Convergence Test test tatistic ADF-Fisher PP-Fisher IPS lag phase 1 2 3 1 2 3 Mining and Washing of Coal statistics (p value) 0.370 (0.814) 0.119 (1.104) -1.372 (0.955) 0.694 (0.776) -1.112 (0.557) 0.924 (0.411) 0.275 Petroleum and Natural Gas 0.472 (0.377) 0.527*** (0.078) 0.825 (0.772) 0.831 (0.901) 0.533 (0.718) -1.608 (1.399) 0.927 Ferrous Metal Ores 0.583 (0.411) 0.213 (0.381) 0.517 (0.472) 0.383*** (0.085) 0.661 (0.418) 0.211 (0.191) 1.247 Non-Ferrous Metal Ores 0.745 (0.660) -1.005 (0.972) 2.040 (1.954) -1.044 (0.992) 1.005 (0.976) 1.124 (1.104) -0.559 Nonmetal Ores 0.910 (0.739) 0.601 (1.087) -1.546 (1.322) 0.492 (0.329) 0.339 (0.514) -1.025 (1.271) 0.887 Support Activities Mining 1.244 (1.071) 0.183 (0.204) -0.677 (0.572) 0.956 (1.584) -0.947 (1.224) 0.223 (0.190) 0.351 Mining of Other Ores 1.314 (1.124) 1.280 (1.214) -0.557 (0.417) 0.670 (0.721) 1.447 (1.270) -0.417 (0.625) -1.263 All Industry 0.743 (0.698) -1.290 (0.884) 0.883 (0.625) 0.334 (0.419) 0.497 (0.372) 1.247 (0.899) 0.674 Note: The lag order of the IPS indicator is determined by the AIC criterion. From the test results in Table 5, it can be seen that, except for the two-order lagging ADF-Fisher test in coal mining and washing industry and the 1-order lag PP-Fisher test in the non-ferrous mining industry, all test estimates at the significant level of 10% are consistent with the original assumption that there exists the unit root. Therefore, basically it can be judged that the seven sub-industries of the extractive industry in China do not have the time trend of stochastic convergence. This means that the carbon emission efficiency gaps in the sub-industries of the extractive industries will not be automatically eliminated in terms of overall or internal relations (Leontief & Ford, 1971). 527 https://www.baidu.com/link?url=XHhvZh9iui4fQ4ZCBy0-Vm06_4y-gR6xXhU-1UuxKPyTobzMIZ5PNmu7EW7Gmusnt38toqghTgCrkjEhAixpvP74frB6F2osItEXVoW6ceSqmOZBWqY9gclqXjaKwfuQ&wd=&eqid=ae24731b00006a5a000000025afe5406 https://www.baidu.com/link?url=j-jrnr-ZX7N_goDkPXOsAaWFW8JFxyQ7uHFoXBvG-UH60ijr5smOn5e4sIZ8jpB0nwXI8W1XxBOtcML6n5K6pECmPD05er0zbKfrIqIUQb7&wd=&eqid=b4d71faa0000ce59000000025afe5429 5. Conclusions In this paper, the stochastic frontier analysis and stochastic convergence analysis were conducted to empirically test the carbon emission efficiency of the extractive industry in China and the differences between sub-industries. The following conclusions are made: Firstly, as a whole, the carbon emission efficiency of extractive industry is slowly rising, and there is room for further improvement. This means that in the Thirteenth Five-Year Plan in China, the extractive industry can achieve emission reduction targets under the condition that the industry output is further increased. The conclusion is in accordance with the view of Zhang et al. (2013). Second, within the extractive industry, there is a huge interindustry gap in carbon emission efficiency, and this gap has continued to widen. This indicates that in the process of carbon emission reduction, the costs paid by various industries are not the same, and then reducing the carbon intensity gap between industries is an effective measure to achieve the overall emission reduction target of the extractive industry. Reference Aigner D.J., Lovell C.A.K., Schmidt P., 1977, Formulation and Estimation of Stochastic Frontier Production Functions Models, Journal of Econometric, 6(1), 21-37, DOI: 10.1016/0304-4076(77)90052-5 Ang B.W., 2005, The LMDI approach to decomposition analysis: a practical guide, Energy Policy, 33(7), 867- 871, DOI: 10.1016/j.enpol.2003.10.010 Ang B.W., Zhang F.Q., Choi K.H., 1998, Factorizing changes in energy and environmental indicators through decomposition, Energy, 23(6), 489-495, DOI: 10.1016/S0360-5442(98)00016-4 Bretschger L., Ramer R., Schwark F., 2011, Growth effects of carbon policies: Applying a fully dynamic CGE model with heterogeneous capital, Resource & Energy Economics, 33(4), 963-980, DOI: 10.1016/j.reseneeco.2011.06.004 Carlino G., Mills L., 1996, Testing Neoclassical Convergence in Regional Incomes and Earnings. Regional Science and Urban Economics, 26(6), 565-590, DOI: 10.1016/S0166-0462(96)02137-0 Evans P., Karras G., 1996, Convergence Revisited, Journal of Monetary Economics, 37(2), 249-265, DOI: 10.1016/S0304-3932(96)90036-7 Fischer, C., Springborn M., 2011, Emission`s targets and the real business cycle: intensity targets versus caps taxes, Journal of Environmental Economics & Management, 62(3), 352-366. Hermeling C., Loschel A., Mennel T., 2013, A new robustness analysis for climate policy evaluations: A CGE application for the EU 2020 targets, Energy Policy, 55(4), 27-35, DOI: 10.1016/j.enpol.2012.08.007 Herrala R., Goel R.K., 2012, Global CO2, efficiency: Country-wise estimates using a stochastic cost frontier. Energy Policy, 45(6), 762-770, DOI: 10.1016/j.enpol.2012.03.007 IPCC., 2006 IPCC guidelines for national greenhouse gas inventories: volume two, Japan: The Institute for Global Environmental Strategies, 2008. Leontief W., Ford D., 1971, Air pollution and the economic structure: empirical results of input-output computation, Paper presented at Fifth International Conference on Input-output Technique, January, Geneva, Switzerland. Lu C.Y., Tong Q., Liu X.M., 2010, The impacts of carbon tax and complementary policies on Chinese economy, 38(11), 7278-7285, DOI: 10.1016/j.enpol.2010.07.055 Mecusen W., Broeck J., 1977, Efficiency Estimation from Cobb-Douglas Production Functions with Composed Error, International Economic Review, 18(2), 435-444, DOI: 10.2307/2525757 Mohtadi H., 1996, Environment, Growth and Optimal Policy Design, Journal of Public Economics, 63(1), 119- 140, DOI: 10.1016/0047-2727(95)01562-0 Shrestha R.M., Timilsina G.R., 1996, Factors Affecting CO2 Intensities of Power Sector in Asia: A Divisia Decomposition Analysis, Energy Economics, 18(4), 283-293, DOI: 10.1016/S0140-9883(96)00019-9 Springmann M., Zhang D., Karplus V.J., 2015, Consumption-based adjustment of emissions-intensity targets: an economic analysis for China`s provinces, Environmental & resource economics, 61(4), 615-640． Torvanger A., 1990, Manufacturing Sector Carbon Dioxide Emission in Nine OECD Countries, 1973-1987: A Divisia Decomposition to Changes in fuel mix, emission coefficients, industry, energy intensities and international structure, Energy Economics, 13(3), 168-186, DOI: 10.1016/0140-9883(91)90018-U Zhang D., Rausch S., Karplus V.J., Zhang, X., 2013, Quantifying regional economic impacts of CO2 intensity targets in China, Energy Economics, 40(2), 687-701, DOI: 10.1016/j.eneco.2013.08.018 528