基于TGARCH-VineCopula的電價(jià)波動(dòng)分析及風(fēng)險(xiǎn)度量研究

謝 航，賴春羊，曾 宏，馬光文，陳仕軍,2，王建華

基于TGARCH-VineCopula的電價(jià)波動(dòng)分析及風(fēng)險(xiǎn)度量研究

謝 航1，賴春羊1，曾 宏1，馬光文1，陳仕軍1,2，王建華3

(1.四川大學(xué)水利水電學(xué)院/水力學(xué)與山區(qū)河流開發(fā)保護(hù)重點(diǎn)實(shí)驗(yàn)室，四川 成都 610065；2.四川大學(xué)商學(xué)院，四川 成都 610065；3.國(guó)家能源大渡河公司，四川 成都 610041)

在市場(chǎng)化交易中，計(jì)及電價(jià)波動(dòng)信息的風(fēng)險(xiǎn)度量可以幫助市場(chǎng)利益相關(guān)者規(guī)避風(fēng)險(xiǎn)。為此，結(jié)合TGARCH與VineCopula理論，提出一種電價(jià)波動(dòng)分析及風(fēng)險(xiǎn)度量的新方法。該方法用TGARCH建立日前、實(shí)時(shí)及輔助服務(wù)交易電價(jià)邊緣分布，通過VineCopula擬合各交易電價(jià)的多維相依結(jié)構(gòu)。基于得到的相關(guān)系數(shù)與尾部關(guān)系分析各交易電價(jià)之間的動(dòng)態(tài)波動(dòng)規(guī)律，并測(cè)度電價(jià)動(dòng)態(tài)波動(dòng)風(fēng)險(xiǎn)。實(shí)證分析證明，該方法不僅可以捕捉負(fù)荷容量比和可再生能源滲透率作用下價(jià)格波動(dòng)的變化，還可以較為準(zhǔn)確地描述各交易電價(jià)的非線性關(guān)聯(lián)結(jié)構(gòu)，進(jìn)而捕獲日前、實(shí)時(shí)、輔助服務(wù)交易電價(jià)之間逐時(shí)動(dòng)態(tài)波動(dòng)特征。此外，與其他方法相比還能更有效地降低組合波動(dòng)風(fēng)險(xiǎn)。

電價(jià)分析；TGARCH；VineCopula；風(fēng)險(xiǎn)度量

0 引言

電價(jià)是電力市場(chǎng)的支點(diǎn)，是電網(wǎng)公司、發(fā)電企業(yè)、市場(chǎng)監(jiān)管部門在評(píng)估、決策、監(jiān)督時(shí)參考的關(guān)鍵指標(biāo)[1-4]。隨著國(guó)內(nèi)新一輪電力體制改革工作的推進(jìn)，國(guó)內(nèi)已有8個(gè)區(qū)域開始現(xiàn)貨市場(chǎng)試點(diǎn)運(yùn)行[5]，電價(jià)風(fēng)險(xiǎn)管理對(duì)參與現(xiàn)貨交易的相關(guān)者顯得尤為重要?，F(xiàn)階段，國(guó)內(nèi)外針對(duì)電價(jià)風(fēng)險(xiǎn)管理的研究，按關(guān)鍵風(fēng)險(xiǎn)信息的角度可分為2類：1) 基于電價(jià)水平信息，設(shè)計(jì)風(fēng)險(xiǎn)管理機(jī)制和決策方法[6]；2) 基于電價(jià)波動(dòng)信息，進(jìn)行風(fēng)險(xiǎn)度量[7]、評(píng)估[8]與預(yù)警[9]，其中，風(fēng)險(xiǎn)度量是后兩者的基礎(chǔ)。對(duì)于風(fēng)險(xiǎn)度量，研究人員較常使用金融風(fēng)險(xiǎn)領(lǐng)域的參數(shù)法和半?yún)?shù)法計(jì)算風(fēng)險(xiǎn)的大小[10]，它們的重點(diǎn)在于選擇準(zhǔn)確的模型對(duì)電價(jià)波動(dòng)擬合，這突顯了價(jià)格波動(dòng)分析的重要性。

已有研究表明電價(jià)波動(dòng)主要呈現(xiàn)“均值回復(fù)”、“極值跳躍”和“杠桿效應(yīng)”等規(guī)律[11-13]。文獻(xiàn)[14-16]發(fā)現(xiàn)結(jié)合廣義自回歸條件異方差模型(Generalized Auto-Regressive Conditional Heteroskedasticity, GARCH)[17]與自回歸移動(dòng)平均模型(Auto-Regressive Moving Average Model, ARMA)[18]或自回歸積分滑動(dòng)平均模型(Auto-Regressive Integral Moving Average Model, ARIMA)[19]，可以較好地描述電價(jià)波動(dòng)的均值回復(fù)及異方差性。文獻(xiàn)[20-21]指出EGARCH與TGARCH模型還適合刻畫電價(jià)波動(dòng)的杠桿效應(yīng)。文獻(xiàn)[22-23]構(gòu)建了考慮電價(jià)水平與負(fù)荷容量比、可再生能源滲透率等外生因素的TGARCH模型，其比EGARCH能更合理地解釋電價(jià)波動(dòng)的杠桿效應(yīng)。但上述研究對(duì)日前或?qū)崟r(shí)交易電價(jià)的波動(dòng)特性研究的較多，較少關(guān)注輔助服務(wù)以及多個(gè)交易品種電價(jià)間的關(guān)聯(lián)性。事實(shí)上，同國(guó)外PJM、Nord Pool等電力市場(chǎng)一樣，國(guó)內(nèi)廣東、四川等試點(diǎn)省份的市場(chǎng)交易規(guī)則均指出市場(chǎng)主體有參與輔助服務(wù)交易的義務(wù)[24-25]。文獻(xiàn)[26]研究了負(fù)荷與日前、實(shí)時(shí)和輔助服務(wù)交易電價(jià)之間的波動(dòng)關(guān)系，但缺乏定量的分析。VineCopula理論廣泛應(yīng)用于多變量間非線性相關(guān)結(jié)構(gòu)的研究，比如文獻(xiàn)[7]運(yùn)用VineCopula捕獲了多個(gè)發(fā)電商損益間的尾部關(guān)聯(lián)性。

關(guān)于電價(jià)波動(dòng)風(fēng)險(xiǎn)度量方面，通常應(yīng)用風(fēng)險(xiǎn)價(jià)值(Value at Risk,VaR)作為度量風(fēng)險(xiǎn)的指標(biāo)[27]。文獻(xiàn)[20]證明了Copula-VaR模型度量風(fēng)險(xiǎn)的效果較好，并且在度量組合風(fēng)險(xiǎn)上優(yōu)勢(shì)顯著。文獻(xiàn)[7]證明了適用于Copula-VaR與電力市場(chǎng)有關(guān)的動(dòng)態(tài)風(fēng)險(xiǎn)度量。

綜上，本研究在借鑒已有研究基礎(chǔ)上，提出一種電力交易價(jià)格波動(dòng)分析與VaR度量的新方法。結(jié)合TGARCH門限模型刻畫逐時(shí)刻日前、實(shí)時(shí)與輔助服務(wù)交易電價(jià)的波動(dòng)特征，并分析各交易電價(jià)波動(dòng)特征；引入VineCopula理論構(gòu)建日前、實(shí)時(shí)和輔助服務(wù)市場(chǎng)交易電價(jià)的0～23時(shí)(h)多維相依模型，進(jìn)而定量分析不同交易價(jià)格波動(dòng)之間的關(guān)聯(lián)；計(jì)算三種置信度(即0.90、0.95、0.98)下電價(jià)與組合電價(jià)的動(dòng)態(tài)波動(dòng)風(fēng)險(xiǎn)曲線；最后通過實(shí)例證明本研究所提方法的可行性。

1 研究模型

1.1 動(dòng)態(tài)波動(dòng)分析

1.1.1邊緣分布建模

1.1.2聯(lián)合分布建模

1.2 波動(dòng)風(fēng)險(xiǎn)度量

VaR是廣泛得到應(yīng)用的風(fēng)險(xiǎn)度量指標(biāo)[31]，能有效反映價(jià)格波動(dòng)劇烈時(shí)可能導(dǎo)致的最大損失，具體數(shù)學(xué)表達(dá)式為

1.3 TGARCH-VineCopula模型的構(gòu)建

基于TGARCH-VineCopula的動(dòng)態(tài)波動(dòng)分析及風(fēng)險(xiǎn)度量模型構(gòu)建流程如圖1所示。

圖1 TGARCH-VineCopula的建模流程

建模步驟如下。

步驟6 依據(jù)1.2節(jié)所述方法估計(jì)VaR值。

2 實(shí)證分析

2.1 實(shí)證數(shù)據(jù)統(tǒng)計(jì)分析

表1 主要描述性指標(biāo)統(tǒng)計(jì)

注：上表列出的是0～23 h的各序列的平均值。

2.2 動(dòng)態(tài)波動(dòng)特征分析

圖2 LJung-Box、ARCH-LM檢驗(yàn)結(jié)果

圖3 DMP等電價(jià)序列非線性相依關(guān)系

2.3 動(dòng)態(tài)波動(dòng)風(fēng)險(xiǎn)分析

比較圖4與圖 5可知，第2)種組合(1,2,3,45)波動(dòng)VaR整體低于第1)種(2,1)組合以及RMP、RC、RP、MR波動(dòng)VaR。根據(jù)現(xiàn)有交易規(guī)則，市場(chǎng)成員不能只參與日前現(xiàn)貨交易，所以各市場(chǎng)成員可以在考慮自身容忍風(fēng)險(xiǎn)水平上，分析日前、實(shí)時(shí)、輔助服務(wù)交易的關(guān)系，結(jié)合單個(gè)或組合交易品種一天的電價(jià)波動(dòng)風(fēng)險(xiǎn)，分散交易風(fēng)險(xiǎn)，使利益最大化。

圖4 三種置信度下動(dòng)態(tài)VaR的變化曲線

圖5 三種置信水平下動(dòng)態(tài)組合VaR的變化曲線

Fig.5 Performance of dynamic portfolio VaR curvesunder three confidence levels

圖6 三個(gè)置信度下組合VaR曲線

3 結(jié)論

附表1 0～23 h DMP序列邊緣分布模型系數(shù)估計(jì)

Attached Table 1 Estimation of marginal distribution model coefficients of 0～23 hrs DMP

01234567891011 偏態(tài)分布 0.310.120.100.090.080.100.080.080.100.160.060.05 0.090.780.900.900.920.830.920.920.850.260.940.95 (1.00)(1.00)(1.00)(0.99)(0.58)(1.00)(1.00)(1.00)(1.00)(1.00)(1.00)(1.00) 0.0000.000.000.000.000.000.000.000.000.000.000.00 121314151617181920212223 偏態(tài)分布 0.060.110.060.240.270.190.130.150.160.170.320.27 0.940.760.940.720.000.270.830.850.840.830.100.08 (0.99)(1.00)(1.00)(1.00)(0.56)(1.00)(1.00)(1.00)(1.00)(1.00)(1.00)(1.00) 0.000.000.000.000.000.000.000.000.000.000.000.00

附表2 0～23 h RMP序列邊緣分布模型系數(shù)估計(jì)

Attached Table 2 Estimation of marginal distribution model coefficients of 0～23 hrs RMP

01234567891011 學(xué)生t分布 0.350.650.300.250.270.100.490.200.080.280.970.68 0.030.060.700.710.570.880.510.770.860.000.000.32 (1.00)(1.00)(0.70)(0.76)(0.14)(0.85)(0.62)(1.00)(0.99)(1.00)(0.48)(0.07) 0.000.000.130.000.000.000.000.000.000.000.000.00 121314151617181920212223 學(xué)生t分布 0.230.030.040.030.480.410.470.500.300.160.640.34 0.770.970.960.850.000.420.120.500.700.810.150.66 (0.74)(1.00)(1.00)(0.94)(1.00)(1.00)(0.81)(1.00)(1.00)(1.00)(1.00)(1.00) 0.000.000.000.000.000.000.000.000.000.000.000.00

附表3 0～23 h RC序列邊緣分布模型系數(shù)估計(jì)

Attached Table 3 Estimation of marginal distribution model coefficients of 0～23 hrs RC

01234567891011 偏正態(tài)分布 0.060.040.050.500.490.400.190.030.020.070.080.10 0.020.850.870.500.510.600.720.870.980.870.720.90 (1.00)1.000.61(0.78)(0.78)(0.46)(1.00)1.00(1.00)(1.00)(0.86)(0.80) 0.000.000.000.050.050.690.000.000.000.000.410.33 121314151617181920212223 偏正態(tài)分布 0.090.000.050.050.030.010.000.000.200.200.140.02 0.900.590.910.900.900.990.000.960.800.800.000.16 (1.00)0.250.020.271.00(1.00)(1.00)(1.00)0.790.79(1.00)(1.00) 1.540.000.000.000.000.000.001.880.000.000.000.00

附表4 0～23 h RP序列邊緣分布模型系數(shù)估計(jì)

Attached Table 4 Estimation of marginal distribution model coefficients of 0～23 hrs RP

01234567891011 學(xué)生t分布正態(tài)分布學(xué)生t分布 0.360.430.64(0.90)0.330.100.270.180.270.210.090.00 0.600.290.36(0.06)0.570.840.630.720.400.790.891.00 0.280.06(0.09)(0.37)0.410.810.25(0.25)(0.02)(0.21)0.481.00 0.000.030.000.000.000.060.000.020.400.000.000.00 121314151617181920212223 學(xué)生t分布正態(tài)分布學(xué)生t分布 0.150.250.030.040.020.070.030.910.901.000.380.40 0.810.410.940.870.890.900.870.080.100.190.500.35 (1.00)(0.49)(0.82)1.001.000.02(1.00)(0.35)(0.24)(0.15)(0.03)(0.23) 0.000.010.020.010.000.000.010.000.000.000.030.00

附表5 0～23 h MR序列邊緣分布模型系數(shù)估計(jì)

Attached Table 5 Estimation of marginal distribution model coefficients of 0～23 hrs MR

01234567891011 正態(tài)分布 0.050.050.000.010.560.000.050.050.020.330.000.00 0.950.950.970.960.120.990.950.950.980.151.000.95 (0.87)1.000.19(0.90)0.730.54(0.10)1.001.00(0.35)1.00(1.00) 0.000.000.000.000.000.000.000.000.000.000.000.00 121314151617181920212223 偏正態(tài)正態(tài)分布 0.150.000.100.021.000.000.020.060.010.000.040.07 0.811.000.920.970.830.960.980.940.990.990.950.93 (1.00)(0.15)0.86(1.00)1.000.64(1.00)(0.06)1.00(1.00)(1.00)(1.00) 0.000.000.000.000.000.000.000.000.000.000.040.00

注：括號(hào)中的數(shù)值為負(fù)數(shù)

[1] 謝開.美國(guó)電力市場(chǎng)運(yùn)行與監(jiān)管實(shí)例分析[M].北京: 中國(guó)電力出版社, 2017.

[2] 姚星安, 曾智健, 楊威,等.廣東電力市場(chǎng)結(jié)算機(jī)制設(shè)計(jì)與實(shí)踐[J].電力系統(tǒng)保護(hù)與控制, 2020, 48(2): 76-85.

YAO Xing'an, ZENG Zhijian, YANG Wei, et al.Electricity market settlement mechanism design and practice in Guangdong[J].Power System Protection and Control, 2020, 48(2): 76-85.

[3] 曾佳妮, 林曉凡, 鐘樺, 等.可再生能源消納背景下直接交易中的二次出清與定價(jià)[J].電力建設(shè), 2020, 41(10): 30-37.

ZENG Jiani, LIN Xiaofan, ZHONG Hua, et al.Secondary clearing rules and price determination in direct electricity trading with renewable energy consumption[J].Electric Power Construction, 2020, 41(10): 30-37.

[4] 杜威, 劉東, 李文云, 等.基于配網(wǎng)節(jié)點(diǎn)電價(jià)的綜合能源系統(tǒng)隨機(jī)優(yōu)化[J].電力建設(shè), 2020, 41(12): 141-151.

DU Wei, LIU Dong, LI Wenyun, et al.Stochastic optimization of integrated energy system considering distribution locational marginal price[J].Electric Power Construction, 2020, 41(12): 141-151.

[5] 孫可, 蘭洲, 林振智, 等.國(guó)際典型電力市場(chǎng)阻塞管理機(jī)制及其對(duì)中國(guó)的啟示研究[J].電力系統(tǒng)保護(hù)與控制, 2020, 48(12): 170-178.

SUN Ke, LAN Zhou, LIN Zhenzhi, et al.Transmission congestion management mechanism of typical international power markets and possible guidance for China’s power market[J].Power System Protection and Control, 2020, 48(12): 170-178.

[6] 路嘉琦, 李強(qiáng), 宋少群, 等.基于概率邊際電價(jià)的價(jià)格風(fēng)險(xiǎn)管理機(jī)制設(shè)計(jì)[J].電網(wǎng)技術(shù), 2020, 44(12): 65-74.

LU Jiaqi, LI Qiang, SONG Shaoqun, et al.Price risk management mechanism based on probabilistic[J].Power System Technology, 2020, 44(12): 65-74.

[7] 謝敏, 胡昕彤, 柯少佳, 等.基于vine-Copula的發(fā)電商運(yùn)營(yíng)損益動(dòng)態(tài)風(fēng)險(xiǎn)VaR評(píng)估方法[J].電力系統(tǒng)自動(dòng)化, 2019, 43(5): 39-45, 52.

XIE Min, HU Xintong, KE Shaojia, et al.vine-Copula based dynamic risk VaR assessment method for operation profit and loss of generation companies[J].Automation of Electric Power Systems, 2019, 43(5): 39-45, 52.

[8] 謝敬東, 陸池鑫, 魯思薇, 等.基于序關(guān)系-熵權(quán)法的電力市場(chǎng)風(fēng)險(xiǎn)評(píng)估[J].中國(guó)電力, 2021, 54(6): 71-78.

XIE Jingdong, LU Chixin, LU Siwei, et al.Evaluation of status of electricity market risk based on order relation-entropy weight method[J].Electric Power, 2021, 54(6): 71-78.

[9] 黃仁輝, 張集, 張粒子, 等.整合GARCH和VaR的電力市場(chǎng)價(jià)格風(fēng)險(xiǎn)預(yù)警模型[J].中國(guó)電機(jī)工程學(xué)報(bào), 2009, 29(19): 85-91.

HUANG Renhui, ZHANG Ji, ZHANG Lizi, et al.Price risk forewarning of electricity market based on GARCH and VaR theory[J].Proceedings of the CSEE, 2009, 29(19): 85-91.

[10] 熊尚飛, 鄒小燕.電力市場(chǎng)價(jià)格風(fēng)險(xiǎn)價(jià)值與波動(dòng)預(yù)測(cè)研究綜述[J].電力系統(tǒng)保護(hù)與控制, 2014, 42(2): 146-153.

XIONG Shangfei, ZOU Xiaoyan.Value at risk and price volatility forecasting in electricity market: a literature review[J].Power System Protection and Control, 2014, 42(2): 146-153.

[11] SIMONSEN I.Volatility of power markets[J].Physica A: Statistical Mechanics and its Applications, 2005, 355(1): 10-20.

[12]CHIU H Y, CHEN T F.Impact of volatility jumps in a mean-reverting model: derivative pricing and empirical evidence[J].The North American Journal of Economics and Finance, 2019, 52: 101-112.

[13] OTTESEN S ?, TOMASGARD A, FLETEN S E.Multi market bidding strategies for demand side flexibility aggregators in electricity markets[J].Energy, 2018, 149: 120-134.

[14] GARCIA R C, CONTRERAS J, AKKEREN M V, et al.A GARCH forecasting model to predict day-ahead electricity prices[J].IEEE Transactions on Power Systems, 2005, 20(2): 867-874.

[15]王賀.系統(tǒng)邊際電價(jià)預(yù)測(cè)的GARCH建模[D].北京: 華北電力大學(xué), 2007.

WANG He.GARCH modeling system marginal price forecast[D].Beijing: North China Electric Power University, 2007.

[16]王瑞慶, 王宏福.基于t分布GARCH模型的電價(jià)波動(dòng)時(shí)變性研究[J].電力系統(tǒng)保護(hù)與控制, 2011, 39(23): 49-53, 59.

WANG Ruiqing, WANG Hongfu.Investigation on time-varying volatility of electricity price based on GARCH model with student-t distribution[J].Power System Protection and Control, 2011, 39(23): 49-53, 59.

[17]CHEN H, ZHANG J, TAO Y, et al.Asymmetric GARCH type models for asymmetric volatility characteristics analysis and wind power forecasting[J].Protection and Control of Modern Power Systems, 2019, 4(4): 356-366.

[18]HUANG L, YANG Y, ZHAO H, et al.Time series modeling and filtering method of electric power load stochastic noise[J].Protection and Control of Modern Power Systems, 2017, 2(3): 269-275.

[19]BOWDEN N, PAYNE J E.Short term forecasting of electricity prices for MISO hubs: evidence from ARIMA-EGARCH models[J].Energy Economics, 2008, 30(6): 3186-3197.

[20]劉麗燕, 鄒小燕.GARCH族模型在電力市場(chǎng)電價(jià)預(yù)測(cè)中的比較研究[J].電力系統(tǒng)保護(hù)與控制, 2016, 44(4): 57-63.

LIU Liyan, ZOU Xiaoyan.Comparison of GARCH models in electricity price forecasting[J].Power System Protection and Control, 2016, 44(4): 57-63.

[21]李停.基于ARMA-GARCH模型的電價(jià)預(yù)測(cè)與研究[D].重慶: 重慶師范大學(xué), 2015.

LI Ting.Electricity price forecasting and research based on ARMA-GARCH model[D].Chongqing: Chongqing Normal University, 2009.

[22]熊尚飛.考慮負(fù)荷容量比的電力市場(chǎng)電價(jià)波動(dòng)與風(fēng)險(xiǎn)價(jià)值—基于門限四歸模型TGARCH-I[D].重慶: 重慶師范大學(xué), 2014.

XIONG Shangfei.Price volatility and value-at-risk in electricity market considering load/capacity ratio—based on threshold GARCH-I model[D].Chongqing: Chongqing Normal University, 2014.

[23]王永茹.基于TGARCH模型的電價(jià)波動(dòng)性分析[D].濟(jì)南: 山東大學(xué), 2020.

WANG Yongru.Analysis of wlectricity price volatility based on TGARCH model[D].Jinan: Shandong University, 2020.

[24]廣東調(diào)頻輔助服務(wù)市場(chǎng)交易規(guī)則(試行)[EB/OL].https:// shoudian.bjx.com.cn/news/20180808/919432.shtml.

Guangdong FM auxiliary service market trading rules (trial)[EB/OL].https://shoudian.bjx.com.cn/news/20180808/ 919432.shtml.

[25]項(xiàng)中明, 石博隆, 吳昌, 等.考慮無功輔助服務(wù)分段報(bào)價(jià)的多目標(biāo)無功優(yōu)化方法[J].電力系統(tǒng)保護(hù)與控制, 2021, 49(6): 40-47.

XIANG Zhongming, SHI Bolong, WU Chang, et al.Research on AVC considering quotation rules of reactive power ancillary services in the electricity market[J].Power System Protection and Control, 2021, 49(6): 40-47.

[26]劉德旭, 馬光文, 陶春華, 等.PJM電力市場(chǎng)交易價(jià)格分布特征及波動(dòng)風(fēng)險(xiǎn)度量[J].電網(wǎng)與清潔能源, 2020, 36(9): 15-21.

LIU Dexu, MA Guangwen, TAO Chunhua, et al.Electricity prices distribution characteristics and volatility risk measurement in the PJM power market[J].Power System and Clean Energy, 2020, 36(9): 15-21.

[27]馮玉明.基于VaR的電力市場(chǎng)價(jià)格風(fēng)險(xiǎn)度量[D].重慶:重慶師范大學(xué), 2012.

FENG Yuming.Electricity market price risk measurement based on VaR[D].Chongqing: Chongqing Normal University, 2012.

[28]賀學(xué)強(qiáng), 易丹輝.基于動(dòng)態(tài)Copula方法的股票組合VaR估計(jì)[J].統(tǒng)計(jì)與決策, 2010(17): 11-14.

HE Xueqiang, YI Danhui.Stock portfolio VaR estimation based on dynamic Copula method[J].Statistics & Decision, 2010(17): 11-14.

[29]AAS K, CZADO C, FRIGESSI A, et al.Pair-copula constructions of multiple dependence[J].Insurance Mathematics and Economics, 2007, 44(2).

[30]高江.藤Copula模型與多資產(chǎn)投資組合VaR預(yù)測(cè)[J].數(shù)理統(tǒng)計(jì)與管理, 2013, 32(2): 247-258.

GAO Jiang.Vine Copula model and VaR for multi-asset portfolio[J].Journal of Applied Statistics and Management, 2013, 32(2): 247-258.

[31]燕飛.基于VaR和ES的分位數(shù)損失QS的估計(jì)[D].武漢: 武漢大學(xué), 2017.

YAN Fei.Estimation of the quantile shortfall based on VaR and ES[D].Wuhan: Wuhan University, 2017.

[32]陶薈名.基于Copula-VaR模型在金融風(fēng)險(xiǎn)管理中的應(yīng)用研究[D].昆明: 云南財(cái)經(jīng)大學(xué), 2019.

TAO Huiming.Application of Copula-VaR model in financial risk management[D].Kunming: Yunnan University of Finance and Economics, 2019.

[33]PJM.Markets & operations data directory[EB/OL].(2019-10-18)[2021-03-21].http://dataminer2.pjm.com/feed/ ancillary_services/definition.

Fluctuation analysis and risk measurement of electricity pricing using TGARCH and VineCopula

XIE Hang1, LAI Chunyang1, ZENG Hong1, MA Guangwen1, CHEN Shijun1, 2,WANG Jianhua3

(1.College of Water Resources and Hydropower/State Key Laboratory of Hydraulics and Mountain River Engineering, Sichuan University, Chengdu 610065, China; 2.Business School, Sichuan University, Chengdu 610065, China; 3.Dadu River Company, China Energy Corporation, Chengdu 610041, China)

In a market-oriented transaction, risk measurement of electricity price fluctuation contributes to conduct risk management for market stakeholders.This paper proposes a new method for analyzing electricity price fluctuation and measure risk.It combines TGARCH and Vinecopula.This method applies TGARCH to establish the margin distribution of day-ahead, real-time and ancillary service transaction electricity prices, and uses Vinecopula to fit the multi-dimensional dependent structure of each transaction electricity price.Based on the Kendall rank correlation and tail correlation calculated from the method, the dynamic fluctuation characteristic between each transaction price is analyzed, and its risk is measured.Empirical analysis shows that this method can not only capture the change of price fluctuation under the combined action of load/capacity ratio and renewable energy penetration rate, but can also accurately describe the nonlinear correlation structure of each transaction price.This can capture the dynamic fluctuation characteristics of day-ahead, real-time and ancillary service transaction price.Also, it can more effectively reduce portfolio volatility risk in comparison to other methods.This work is supported by the National Key Research and Development Program of China (No.2018YFB0905204).

analysis of electricity price; TGARCH; VineCopula; risk measurement

10.19783/j.cnki.pspc.210504

2021-04-29；

2021-07-02

謝 航(1996—)，女，碩士研究生，研究方向?yàn)樗娺\(yùn)行管理及電力市場(chǎng)；E-mail: 409367753@qq.com

王建華(1973—)，男，通信作者，高級(jí)工程師，研究方向?yàn)殡娏κ袌?chǎng)。E-mail: 1294331990@qq.com

國(guó)家重點(diǎn)研發(fā)計(jì)劃項(xiàng)目資助(2018YFB0905204)

(編輯 葛艷娜)