Risk premia in electricity spot markets - New empirical evidence for Germany and Austria

Transcription

1 Risk premia in electricity spot markets - New empirical evidence for Germany and Austria Niyaz Valitov Schumpeter School of Business and Economics University of Wuppertal, Germany valitov@wiwi.uni-wuppertal.de April 8, 2016 Niyaz Valitov ENERDAY / 19

2 Motivation Overview This paper analyzes whether risk premia are paid in the German/Austrian day-ahead market for electricity. It investigates the impact of left-skewed spot prices on the risk premium. It investigates the impact of negative spot prices on the risk premium. It tests whether risk premia can be explained by price risks in the long-term. Niyaz Valitov ENERDAY / 19

3 Equilibrium model for risk premia Bessembinder and Lemmon (2002, JF) develop an equilibrium model for risk premia that takes into account the physical properties of electricity and the convexity of the power production curve. Identical producers and identical retailers participate in a forward and in a spot market. The optimal forward position contains risks faced by the industry: variability in retail revenues and in power aquiring/production costs. Niyaz Valitov ENERDAY / 19

4 Equilibrium model for risk premia The equilibrium forward price P F can be expressed as: P F = E(P W ) + αvar(p W ) + γskew(p W ) (1) where α and γ depend on: spot price P W, the number of participants, the retail price, and the production cost function. If spot prices are positive and right-skewed, α < 0 and γ > 0. Positive skewness (median < mean) in spot prices reflects the possibility of large price spikes in marginal production costs. Niyaz Valitov ENERDAY / 19

5 Literature review Several empirical studies are based on the model by Bessembinder and Lemmon (2002): Longstaff and Wang (2004, JF) provide evidence for risk premia at the wholesale market in Pennsylvania, New Jersey, and Maryland (PJM). Haugom and Ullrich (2012, EE) extend upper study and find that risk premia decrease over time. However, they cannot confirm the model results by Bessembinder and Lemmon (2002). Results of Viehmann (2011, EP) suggest that traders in the German/Austrian day-ahead market are willing to pay risk premia during the period from Niyaz Valitov ENERDAY / 19

6 Day-ahead market in Germany and Austria Day-ahead trading in Germany and Austria takes place at two power exchanges and via bilateral contracts: Energy Exchange Austria (EXAA): Auction results published at a.m. Market area: Germany/Austria. European Energy Exchange (EPEX): Auction results published at p.m. Highest liquidity for the market area Germany/Austria. Continuous over-the-counter (OTC): Most trading activities are between 8 a.m. and p.m. Prices at a.m. coincide with EXAA prices. Niyaz Valitov ENERDAY / 19

7 Data Hourly price data from EXAA (forward) and EPEX (spot). Time period from 2005 to 2015 with subsamples: Oct. 01, Sep. 30, 2008 Oct. 01, Nov. 17, 2015 Viehmann (2011) New observation period Subsamples with negative electricity prices: Oct. 01, Oct. 14, 2013 Oct. 15, Nov. 17, 2015 Neg. EPEX prices Neg. EPEX and neg. EXAA prices Niyaz Valitov ENERDAY / 19

8 EPEX prices from Oct Nov Skewness Hour Mean Min Max S.D. All prices Pos. prices Neg. prices All Niyaz Valitov ENERDAY / 19

9 Calculation of risk premia Calculation of average realized risk premia for each hour: rrp i = 1 T T (EXAA i,t EPEX i,t+1 ) (2) t=1 t-tests with Newey-West standard errors. Subsamples for positive electricity prices. Niyaz Valitov ENERDAY / 19

10 Tests for risk premia from Oct Nov All days without neg. prices with neg. prices Difference Hour Mean t-statistic Mean t-statistic Mean t-statistic * ** ** * * * ** ** ** * ** ** ** *** *** * * * ** ** * *** *** *** *** t-statistics are based on Newey-West standard errors. Significance levels: *** p<0.01, ** p<0.05, * p<0.1. Niyaz Valitov ENERDAY / 19

11 Adjustments of the model by BL (2002) The model by Bessembinder and Lemmon (2002) needs to be adjusted for left-skewed (positive) and left-skewed (negative) spot prices. The market-clearing spot price in the model can be expressed as: ( ) Q D c 1 P W = a (3) where Q D is the total retail demand, N P is the number of producers, a is a variable cost parameter, and c is a constant. N P Positive spot prices become left-skewed if c ]1; 2[. Spot prices become negative if a < 0 and left-skewed if c ]0; 1[. Niyaz Valitov ENERDAY / 19

12 Adjustments of the model by BL (2002) Risk premia are determined by: P F E(P W ) = αvar(p W ) + γskew(p W ) (4) If spot prices are positive and left-skewed, α < 0 and γ > 0. Hence, risk premia are strictly negative. If spot prices are negative and left-skewed, α < 0 and γ < 0. Hence, risk premia can be either positive or negative. Niyaz Valitov ENERDAY / 19

13 Tests for long-term consistency of model parameters Regression of the ex post risk premium on the variance (divided by 100) and skewness of the spot price (Longstaff and Wang, 2004): rrp i = a + bvar i + cskew i + ɛ i (5) Long-term consistency is tested with a moving window of fixed length (365 days) for the variance and skewness of positive spot prices. Niyaz Valitov ENERDAY / 19

14 Results from rolling regressions (window: 365 days) Rolling estimation of variance parameter = b Oct Oct Oct Oct Oct Oct Oct Oct Oct Oct-15 b +/-2 s.e. Niyaz Valitov ENERDAY / 19

15 Results from rolling regressions (window: 365 days) Rolling estimation of skewness parameter = c Oct Oct Oct Oct Oct Oct Oct Oct Oct Oct-15 c +/-2 s.e. Niyaz Valitov ENERDAY / 19

16 Conclusion Risk premia are still paid in the German/Austrian day-ahead market, but their magnitude decreased remarkably. Model of Bessembinder and Lemmon (2002) can be adjusted for left-skewed and negative spot prices. Positive and left-skewed spot prices lead to negative premia in the affected hours. Negative spot prices lead to negative or positive risk premia in the affected hours. However, empirical results reveal unstable parameters for the variance and skewness of spot prices. Niyaz Valitov ENERDAY / 19

17 References Bessembinder, H., Lemmon, M.L., Equilibrium pricing and optimal hedging in electricity forward markets. Journal of Finance 57 (3), Haugom, E., Ullrich, C., Market efficiency and risk premia in short-term forward prices. Energy Economics 34, Longstaff, F.A., Wang, A.W., Electricity Forward Prices: A High-Frequency Empirical Analysis. Journal of Finance 59 (4), Viehmann, J., Risk premiums in the German day-ahead Electricity Market. Energy Policy 39, Niyaz Valitov ENERDAY / 19

18 Appendix α and γ are calculated as follows: α N P(x + 1) Nca x ([E(P W ] x P R [E(P W ] x 1 ) (6) γ N P(x + 1) 2Nca x (x[e(p W ] x 1 (x 1)P R [E(P W ] x 2 ) (7) where x is 1/(c 1), N the number of firms in the industry, and P R the fixed retail price. Each producer P i has the following cost function: TC i = F + a c (Q Pi) c (8) where F are fixed costs, a is a variable cost parameter, Q is the output, and c is a constant. Niyaz Valitov ENERDAY / 19

19 Appendix 4.0 Hourly mean risk premium ( /MWh) Viehmann (2011) Oct Nov Niyaz Valitov ENERDAY / 19