可再生能源发电份额对德国市场化电价的影响

doi:10.18402/resci.2021.08.05

摘要/Abstract

摘要：

近年来可再生能源研究表明,风能和太阳能发电份额的增加对市场化电价有显著的影响。德国可再生能源发电市场已形成较完善的市场化电价机制,研究德国电价的运作机理有利于推动中国可再生能源市场化电价机制改革。本文通过建立分位数回归模型,估算了不同电价分位数的优序效应,主要分析了风能和太阳能对德国电力市场现货价格的影响。研究表明：①风能和太阳能发电量的增加均能带来电价的降低。当考虑以电价中位数的四分位距衡量的电价波动时,对于中等负荷水平,风能发电对平均高峰价格的影响明显大于太阳能发电;在其他情况下,太阳能发电对平均高峰价格的影响更大。②用电需求水平较低时,风能发电份额的增加会增加电价波动性;用电需求水平较高时,风能发电份额的增加会降低电价的波动性;用电需求水平中等时,太阳能发电降低电力价格的波动性。③支持可再生能源开发整合的政策应在风能和太阳能之间寻求平衡,即形成合理的太阳能、风能的发电份额组合。本文研究了发展较成熟的德国可再生能源发电的市场化运行机理,对中国可在生能源市场化改革有借鉴意义。

关键词: 可再生能源, 电力价格, 市场化, 分位数回归, 优序效应, 价格波动, 德国

Abstract:

Renewable energy research in recent years has shown that the increase in the share of wind and solar power generation has a significant impact on market electricity prices. The German renewable energy power market has formed a relatively complete market electricity pricing mechanism. Examining the operating mechanism of German electricity prices is conducive to promoting the marketization reform of electricity pricing of China’s renewable energy power market. This study mainly analyzed the impact of wind and solar energy on spot prices in the German electricity market. It established a quantile regression model to estimate the priority effect of different electricity price quantiles. The research results show that: (1) The increase in wind energy and solar power generation can bring about a reduction in electricity prices, and when considering the fluctuation of electricity prices measured by the interquartile range of the median electricity price, for the medium load level, the impact of wind power generation on the average peak price significantly larger than solar power generation. In other cases, solar power has a greater impact on average peak prices. (2) Depending on the level of total demand, the increase or decrease in the share of wind power generation may increase or curb the volatility of electricity prices under different demand levels. The increase in the share of solar power generation stabilizes the difference in electricity prices under moderate demand levels. (3) Therefore, policies supporting the development and integration of renewable energy should seek a balance between different power generation methods, that is to form a reasonable combination of solar and wind power generation shares. This study examined the mechanism of operation of the relatively mature German renewable energy power market, which is of reference value for the marketization reform of China’s renewable energy.

Key words: renewable energy, electricity price, marketization, quantile regression, priority effect, price fluctuation, Germany

孟思琦, 孙仁金, 郭风. 可再生能源发电份额对德国市场化电价的影响[J]. 资源科学, 2021, 43(8): 1562-1573.

MENG Siqi, SUN Renjin, GUO Feng. Impact of renewable energy power generation share on Germany’s electricity prices[J]. Resources Science, 2021, 43(8): 1562-1573.

图/表 13

表1

表2

表3

图1

图2

图3

表4

非线性模型（3）在不同阈值下的参数估计

阈值	风能发电量			太阳能发电量
阈值	$β 1, τ W$	$β 2, τ W$	$β 3, τ W$	$β 1, τ S$	$β 2, τ S$	$β 3, τ S$
$τ A = 0.15, τ Z = 0.85$
0.1	-0.353***	-0.1602***	-0.1453***	-0.4477***	-0.0573**	-0.0505
0.2	-0.3167***	-0.1484***	-0.1846***	-0.3419***	-0.0844***	-0.2106**
0.3	-0.3094***	-0.1504***	-0.1841***	-0.2864***	-0.0999***	-0.2737***
0.4	-0.2608***	-0.1546***	-0.1888***	-0.2416***	-0.1068***	-0.284***
0.5	-0.2429***	-0.1474***	-0.1943***	-0.2419***	-0.1013***	-0.3111***
0.6	-0.2074***	-0.1415***	-0.1894***	-0.2581***	-0.1297***	-0.4179**
0.7	-0.2017***	-0.145***	-0.2041***	-0.2434***	-0.1404***	-0.5628**
0.8	-0.1871***	-0.1445***	-0.2309***	-0.2606***	-0.1761***	-0.5669**
0.9	-0.2097***	-0.158***	-0.2679***	-0.3547***	-0.2347***	-0.368***
$τ A = 0.20, τ Z = 0.80$
0.1	-0.3329**	-0.1514***	-0.1194***	-0.3708***	-0.0585**	-0.0816*
0.2	-0.3041***	-0.1479***	-0.1577***	-0.2961***	-0.0894***	-0.3264***
0.3	-0.2512***	-0.1507***	-0.1646***	-0.2534***	-0.0901***	-0.3548***
0.4	-0.2429***	-0.1505***	-0.1763***	-0.2253***	-0.1103***	-0.4244***
0.5	-0.2366***	-0.148***	-0.18***	-0.1931***	-0.1178***	-0.4078***
0.6	-0.2178***	-0.1415***	-0.1741***	-0.2365***	-0.139***	-0.3934***
0.7	-0.1994***	-0.1457***	-0.1985***	-0.2471***	-0.145***	-0.57***
0.8	-0.1912***	-0.1466***	-0.2175***	-0.2364**	-0.1747***	-0.5646***
0.9	-0.2008**	-0.1584***	-0.2414***	-0.3099**	-0.2244***	-0.5776***

表4

表5

不同阈值下基本变量对分位数范围IQR的估计

变量	系数	基本变量对IQR的估计
阈值	$τ A$	0.15	0.20
	$τ Z$	0.85	0.80
负荷	$β 1 L$	0.016	0.034
	$β 2 L$	0.058	0.067
	$β 3 L$	0.099**	0.109*
风能发电量	$β 1 W$	0.144***	0.142***
	$β 2 W$	0.002	-0.007
	$β 3 W$	-0.124**	-0.132***
太阳能发电量	$β 1 S$	0.094	0.061
	$β 2 S$	0.177***	-0.166***
	$β 3 S$	-0.317*	-0.496**

表5

图4

表6

线性模型（1）和非线性模型（3）的风能与太阳能发电量的参数估计

$τ$	风能发电量				太阳能发电量
$τ$	$β τ W$	$β 1, τ W$	$β 2, τ W$	$β 3, τ W$	$β tτ S$	$β 1, τ S$	$β 2, τ S$	$β 3, τ S$
日指标
0.1	-0.178***	-0.400***	-0.177***	-0.161***	-0.067***	-0.474***	-0.054**	-0.028
0.2	-0.166***	-0.311***	-0.158***	-0.215***	-0.107***	-0.416***	-0.080***	-0.419**
0.3	-0.164***	-0.359***	-0.154***	-0.198***	-0.118***	-0.359***	-0.106***	-0.424**
0.4	-0.168***	-0.253***	-0.159***	-0.219***	-0.140***	-0.281***	-0.109***	-0.450**
0.5	-0.166***	-0.240***	-0.154***	-0.241***	-0.146***	-0.278***	-0.105***	-0.486*
0.6	-0.160***	-0.191***	-0.150***	-0.244***	-0.170***	-0.275***	-0.140***	-0.525*
0.7	-0.170***	-0.195***	-0.155***	-0.255***	-0.196***	-0.265***	-0.159***	-0.597*
0.8	-0.172***	-0.197***	-0.150***	-0.285***	-0.241***	-0.292***	-0.194***	-0.061
0.9	-0.184***	-0.221***	-0.166***	-0.345***	-0.267***	-0.369***	-0.245***	-0.141
高峰指标
0.1	-0.158***	-0.182**	-0.160***	-0.189***	-0.030*	-0.114	-0.035*	-0.074
0.2	-0.175***	-0.175***	-0.175***	-0.251***	-0.066***	-0.076	-0.065**	-0.078
0.3	-0.172***	-0.173***	-0.170***	-0.266***	-0.087***	-0.101*	-0.074***	-0.319*
0.4	-0.177***	-0.177***	-0.173***	-0.301***	-0.096***	-0.144**	-0.077***	-0.376*
0.5	-0.181***	-0.181***	-0.172***	-0.304***	-0.103***	-0.143***	-0.093***	-0.514*
0.6	-0.186***	-0.186***	-0.182***	-0.304***	-0.115***	-0.121***	-0.06***	-0.500*
0.7	-0.189***	-0.189***	-0.186***	-0.290***	-0.147***	-0.123***	-0.145***	-0.379
0.8	-0.203***	-0.203***	-0.197***	-0.334***	-0.156***	-0.144**	-0.158***	-0.141
0.9	-0.214***	-0.214**	-0.189***	-0.484***	-0.211***	-0.104**	-0.211***	-0.188
非高峰指标
0.1	-0.169***	-0.211***	-0.173***	-0.149***	-0.014	-0.769***	0.012	0.816
0.2	-0.166***	-0.197***	-0.164***	-0.178***	-0.094*	-0.640***	-0.054	-0.160
0.3	-0.161***	-0.209***	-0.150***	-0.164***	-0.155**	-0.528***	-0.075	-0.454
0.4	-0.157***	-0.194***	-0.157***	-0.165***	-0.146***	-0.407***	-0.095	-1.032
0.5	-0.156***	-0.192***	-0.154***	-0.162***	-0.149**	-0.508***	-0.089	-1.047
0.6	-0.153***	-0.214***	-0.147***	-0.164***	-0.146**	-0.471***	-0.070	-2.049*
0.7	-0.155***	-0.199***	-0.152***	-0.175***	-0.147***	-0.544***	-0.094**	-2.520*
0.8	-0.161***	-0.198***	-0.148***	-0.183***	-0.201***	-0.543***	-0.148***	-3.431*
0.9	-0.178***	-0.240***	-0.159***	-0.214***	-0.314***	-0.441**	-0.259***	-4.806

表6

表7

模型（1）和（3）中风能和太阳能发电量对电价不同区间的影响差异

$τ$	$β τ W - β τ S$ Ⅰ	$β 1, τ W - β 1, τ S$ Ⅱ	$β 2, τ W - β 2, τ S$ Ⅲ	$β 3, τ W - β 3, τ S$ Ⅳ
日指标
0.1	-0.111***	0.074	-0.132***	-0.142
0.2	-0.059***	0.104	-0.078***	0.204
0.3	-0.045**	0.100	-0.047***	0.235
0.4	-0.27	0.027	-0.050***	0.241
0.5	-0.029	0.048	-0.048	0.255
0.6	0.010	0.085	-0.010	0.281
0.7	0.027	0.070**	0.004	0.442
0.8	0.058**	0.095**	0.043	-0.234
0.9	0.084*	0.149	0.068	-0.204
高峰指标
0.1	-0.127***	-0.069	-0.135***	-0.116
0.2	-0.109***	-0.098	-0.111***	-0.174
0.3	-0.085***	-0.023	-0.096***	0.054
0.4	-0.081***	0.015	-0.095***	0.075
0.5	-0.078***	0.002	-0.080***	0.210
0.6	-0.070***	-0.027	-0.075***	0.197
0.7	-0.042***	-0.025	-0.041***	0.090
0.8	-0.046**	-0.004	-0.038***	-0.201
0.9	-0.002	-0.082	0.022	-0.297
非高峰指标
0.1	-0.156**	0.348**	-0.245***	-0.254
0.2	-0.074*	0.342**	-0.195***	0.620
0.3	-0.007	0.386**	-0.117*	0.685
0.4	-0.013	0.347**	-0.064*	2.161
0.5	-0.017	0.360**	-0.057	2.660
0.6	-0.016	0.353**	-0.031	2.796
0.7	-0.019	0.491**	-0.035	2.364*
0.8	-0.040	0.341**	-0.070	3.185*
0.9	-0.145	0.386**	0.159***	5.166*

表7

表8

线性模型（6）和非线性模型（7）基本变量参数估计

变量	系数	指数
变量	系数	日指标	高峰指标	非高峰指标
线性模型（6）
负荷	$β L$	0.081	0.102**	0.117***
风能发电量	$β W$	-0.006	-0.055**	-0.009
太阳能发电量	$β S$	-0.200***	-0.181***	-0.299***
非线性模型（7）
负荷	$β 1 L$	-0.002	0.030	-0.012
	$β 2 L$	0.048	0.066*	0.020
	$β 3 L$	0.105*	0.163*	0.049*
风能发电量	$β 1 W$	0.179**	-0.004	0.104
	$β 2 W$	0.010	-0.029*	0.019*
	$β 3 W$	-0.184***	-0.295*	-0.058**
太阳能发电量	$β 1 S$	0.104	0.009	0.057
	$β 2 S$	-0.181***	-0.176***	-0.396***
	$β 3 S$	-0.114	-0.0115	-5.478

表8

表9

线性模型（8）和非线性模型（9）基本变量参数估计

变量	系数	指数
变量	系数	日指标	高峰指标	非高峰指标
线性模型（8）
负荷	$β τ L$	-0.030	0.116**	0.077**
风能发电量	$β τ W$	0.001	-0.033**	-0.004
太阳能发电量	$β τ S$	-0.184***	-0.080***	-0.282***
非线性模型（9）
负荷	$β 1, τ L$	-0.175*	0.048	0.079
	$β 2, τ L$	-0.041	0.095	0.070
	$β 3, τ L$	-0.034	0.032	0.079*
风能发电量	$β 1, τ W$	0.214**	0.013	0.001
	$β 2, τ W$	0.014	-0.039*	0.012
	$β 3, τ W$	-0.150***	-0.209**	-0.061**
太阳能发电量	$β 1, τ S$	0.146	0.002	0.244
	$β 2, τ S$	-0.200***	-0.119***	-0.308***
	$β 3, τ S$	-0.150	0.388	-1.798

表9

参考文献 26

[1]	Shahnazi R, Shabani Z D. Do renewable energy production spillovers matter in the EU?[J]. Renewable Energy, 2020, 150:786-796. doi: 10.1016/j.renene.2019.12.123
[2]	Goodarzi S, Perera N, Bunn B. The impact of renewable energy forecast errors on imbalance volumes and electricity spot prices[J]. Energy Policy, 2020, DOI: 10.1016/j.enpol.2019.06.035. doi: 10.1016/j.enpol.2019.06.035
[3]	赵新泉, 王闪闪, 李庆. 市场交易补偿可再生能源的正外部性研究[J]. 中国人口·资源与环境, 2020, 30(8):42-50.
	[ Zhao X Q, Wang S S, Li Q. Research on the compensation of positive externality of renewable energy through market trading[J]. China Population, Resources and Environment, 2020, 30(8):42-50.]
[4]	李力, 朱磊, 范英. 可再生能源配额机制下电力投资最优序贯决策模型[J]. 管理评论, 2019, 31(9):37-46.
	[ Li L, Zhu L, Fan Y. Optimum sequential investments model on renewable energy power investments under renewable portfolio standards[J]. Management Review, 2019, 31(9):37-46.]
[5]	毕清华, 范英, 蔡圣华, 等. 基于CDECGE模型的中国能源需求情景分析[J]. 中国人口·资源与环境, 2013, 23(1):41-48.
	[ Bi Q H, Fan Y, Cai S H, et al. Analysis of China’s primary energy demand scenarios based on the CDECGE model[J]. China Population, Resources and Environment, 2013, 23(1):41-48.]
[6]	李杨. 政府政策和市场竞争对欧盟国家可再生能源技术创新的影响[J]. 资源科学, 2019, 41(7):1306-1316.
	[ Li Y. Impact of government policy and market competition on renewable energy innovation in EU countries[J]. Resources Science, 2019, 41(7):1306-1316.]
[7]	Yan H, Tian C S, Wu C Y. Modelling of carbon price in two real carbon trading markets[J]. Journal of Cleaner Production, 2020, DOI: 10.1016/j.jclepro.2019.118556. doi: 10.1016/j.jclepro.2019.118556
[8]	Lin B Q, Zhu J P. Determinants of renewable energy technological innovation in China under CO₂ emissions constraint[J]. Journal of Environmental Management, 2019, 247:662-671. doi: 10.1016/j.jenvman.2019.06.121
[9]	Nicolli F, Vona F. Heterogeneous policies, heterogeneous technologies: The case of renewable energy[J]. Energy Economics, 2016, 56:190-204. doi: 10.1016/j.eneco.2016.03.007
[10]	Frauendorfer K, Paraschiv F, Schürle M. Cross-border effects on Swiss electricity prices in the light of the energy transition[J]. Energies, 2018, 11(9):2188. doi: 10.3390/en11092188
[11]	Ketterer J C. The impact of wind power generation on the electricity price in Germany[J]. Energy Economics, 2014, 44:270-280. doi: 10.1016/j.eneco.2014.04.003
[12]	Woo C K, Horowitz I, Moore J, et al. The impact of wind generation on the electricity spot-market price level and variance: The Texas experience[J]. Energy Policy, 2011, 39(7):3939-3944. doi: 10.1016/j.enpol.2011.03.084
[13]	Rintamäki T, Siddiqui A S, Salo A. Does renewable energy generation decrease the volatility of electricity prices? An analysis of Denmark and Germany[J]. Energy Economics, 2017, 62:270-282. doi: 10.1016/j.eneco.2016.12.019
[14]	Hagfors L I, Bunn D, Kristoffersen E, et al. Modeling the UK electricity price distributions using quantile regression[J]. Energy, 2016, 102:231-243. doi: 10.1016/j.energy.2016.02.025
[15]	许明强. 高级计量经济学教学中分位数回归的经济解释[J]. 大众投资指南, 2019, (12):293-294.
	[ Xu M Q. The economic meanings of quantile regression in the teach of senior econometrics[J]. Public Investment Guide, 2019, (12):293-294.]
[16]	Hagfors L I, Paraschiv F, Molnar P, et al. Using quantile regression to analyze the effect of renewables on EEX price formation[J]. Renewable Energy and Environmental Sustainability, 2016, DOI: 10.1051/rees/2016036. doi: 10.1051/rees/2016036
[17]	Bunn D, Andresen A, Chen D, et al. Analysis and forecasting of electricty price risks with quantile factor models[J]. The Energy Journal, 2016, 37(1):101-122.
[18]	Gianfreda A, Bunn D. A stochastic latent moment model for electricity price formation[J]. Operations Research, 2018, 66(5):1189-1203. doi: 10.1287/opre.2018.1733
[19]	Yi B W, Xu J H, Fan Y. Coordination of policy goals between renewable portfolio standards and carbon caps: A quantitative assessment in China[J]. Applied Energy, 2019, 237:25-35. doi: 10.1016/j.apenergy.2018.12.015
[20]	赵新刚, 任领志, 万冠. 可再生能源配额制、发电厂商的策略行为与演化[J]. 中国管理科学, 2019, 27(3):168-179.
	[ Zhao X G, Ren L Z, Wan G. Renewable portfolio standards, the strategic behavior of power producers and evolution[J]. Chinese Journal of Management Science, 2019, 27(3):168-179.]
[21]	高雷, 苏辛一, 刘世宇. 可再生能源消纳责任权重下的新能源合理弃电率研究[J]. 中国电力, 2020, 53(12):136-142.
	[ Gao L, Su X Y, Liu S Y. Research on reasonable curtailment rate of new energy under renewable energy consumption responsibility weight[J]. China Electric Power, 2020, 53(12):136-142.]
[22]	Bayer C, Hanck C. Combining non-cointegration tests[J]. Journal of Time series analysis, 2013, 34(1):83-95. doi: 10.1111/jtsa.2012.34.issue-1
[23]	康宁. 分位数回归模型及在金融经济中的应用[D]. 合肥: 合肥工业大学, 2016.
	[ Kang N. Quantile Regression Model and Its Application in Financial and Economy[D]. Hefei: Hefei University of Technology, 2016.]
[24]	王焱, 汪震, 黄民翔, 等. 基于OS-ELM和Bootstrap方法的超短期风电功率预测[J]. 电力系统自动化, 2014, 38(6):14-19.
	[ Wang Y, Wang Z, Huang M X, et al. Ultra-short-term wind power prediction based on OS-ELM and bootstrap methods[J]. Automation of Electric Power Systems, 2014, 38(6):14-19.]
[25]	Shaffer B. Misunderstanding nonlinear prices: Evidence from a natural experiment on residential electricity demand[J]. American Economic Journal: Economic Policy, 2020, 12(3):433-461. doi: 10.1257/pol.20180061
[26]	王秀花. 基于混频已实现GARCH模型的波动预测与VaR度量[D]. 广州: 华南理工大学, 2018.
	[ Wang X H. Mix Frequency Realized GARCH Models: The Forecast of Volatility and Measure of VaR[D]. Guangzhou: South China University of Technology, 2018.]

	平均数	中间值	最小值	最大值	标准差	峰度
日指标
价格	31.60	31.61	-47.46	101.83	11.37	-0.03
负荷	218.23	223.44	139.54	265.60	23.94	-0.54
风能发电量	39.68	31.74	3.87	136.69	29.01	1.14
太阳能发电量	16.30	15.76	0.95	39.65	10.49	0.24
高峰指标
价格	34.99	34.39	-36.76	126.50	13.74	0.84
负荷	244.40	250.40	167.60	306.40	29.99	-0.51
风能发电量	38.98	29.92	1.93	152.50	31.30	1.16
太阳能发电量	28.10	28.00	1.66	67.41	17.25	0.20
非高峰指标
价格	28.20	29.00	-58.17	77.11	9.76	-1.49
负荷	198.40	202.30	143.50	245.60	21.86	-0.41
风能发电量	38.95	30.87	3.24	131.86	27.41	1.14
太阳能发电量	3.85	3.04	0.01	13.11	3.60	0.57

	J-B测试				ADF测试
	价格	负荷	风能发电量	太阳能发电量	价格	负荷	风能发电量	太阳能发电量
日指标
测试	1848.8	58.21	258.0	75.68	1848.8	58.21	258.0	75.68
p值	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01
高峰指标
测试	2479.4	68.78	272.1	68.79	-2.864	-4.581	-8.204	-3.341
p值	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01	<0.05
非高峰指标
测试	4577.7	47.45	256.5	98.89	-8.107	-4.961	-7.483	-2.336
p值	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01	0.16