cond-mat0103273/W00.TEX
1: \documentstyle[12pt,epsf]{article}
2: 
3: \textheight  230mm
4: \textwidth   150mm
5: \topmargin   -15mm
6: \oddsidemargin 10mm
7: \evensidemargin 10mm
8: 
9: 
10: \begin{document}
11: 
12: \begin{center}
13: 
14: {\Large\bf Energy price risk management}
15: 
16: \vspace*{.6cm}
17: 
18: {\large Rafal Weron\footnotemark}
19: 
20: \footnotetext{E-mail: rweron@im.pwr.wroc.pl}
21: 
22: \vspace*{.3cm}
23: 
24: {\small
25: Hugo Steinhaus Center for Stochastic Methods,\\
26: Wroc{\l}aw University of Technology, 50-370 Wroc{\l}aw, Poland
27: }
28: 
29: \end{center}
30: 
31: \vspace*{.5cm}
32: 
33: \noindent
34: {\bf Abstract:} The price of electricity is far more volatile than that of other commodities 
35: normally noted for extreme volatility. Demand and supply are balanced on a knife-edge 
36: because electric power cannot be economically stored, end user demand is largely weather 
37: dependent, and the reliability of the grid is paramount. The possibility of extreme 
38: price movements increases the risk of trading in electricity markets. However, a number
39: of standard financial tools cannot be readily applied to pricing and hedging electricity 
40: derivatives. In this paper we present arguments why this is the case.\\
41: 
42: \noindent
43: {\it PACS:} 05.45.Tp, 89.30.+f, 89.90.+n\\
44: {\it Keywords:} Econophysics, electricity price, risk management, mean-reversion\\
45: 
46: \vspace*{.5cm}
47: 
48: 
49: \section{Introduction}
50: 
51: Energy price risk management is still in its infancy compared to the more 
52: developed interest rate and foreign exchange markets \cite{fusaro,warrick,ww00}. 
53: However, we have to bear in mind that commodity markets are not anywhere near as
54: straightforward as financial markets. They have to deal with the added complexity 
55: of physical substance \cite{bll99}, which cannot simply be manufactured, transported 
56: and delivered, at the press of a button.
57: 
58: An innate energy industry conservatism coupled with highly profitable years caused stagnation 
59: despite the two oil price shocks of the 1970s. But the world oil price collapse of 1986
60: and the beginning of electric utility deregulation and privatization throughout the world 
61: are continuing to drive change in energy commodity markets \cite{fusaro,ICC,kurtz}.
62: 
63: In the wake of the recent price run-ups and defaults, managers have been forced to 
64: review their credit and counterparty risk policies. Traditional credit analysis has 
65: emphasized the financial risk associated with the failure of a buyer to pay for the 
66: goods purchased. Although this can be a concern in the power market, recent events 
67: have highlighted the substantial and perhaps less predictable market risk resulting 
68: from the failure to deliver by a seller \cite{calpx}. The defaults in late June 1998 
69: threw the US buyers into a superheated Midwest market, desperate for replacement power. 
70: This resulted in soaring prices that reportedly topped out at \$7,500 in real-time 
71: trading -- 300 times the average price of \$25/MWh!
72: 
73: The possibility of extreme price movements increases the risk of trading in electricity 
74: markets. Unfortunately a number of standard financial tools cannot be readily 
75: applied to pricing and hedging electricity derivatives. But before we explain why
76: let us briefly describe today's electricity markets.
77: 
78: 
79: \section{Electricity markets}
80: 
81: The deregulation of the electricity industry is giving way to a global trend toward 
82: the commoditization of electric energy \cite{green}. This trend has recently intensified 
83: in Europe and North America, where market forces have pushed legislators to begin removing 
84: artificial barriers that shielded electric utilities from competition. As a result,
85: during the last decade, we have witnessed a major explosion in the number of nontraditional 
86: power suppliers and financial engineers marketing electricity and electricity derivatives 
87: in the wholesale power markets. Only in the early four year period 1993-96 over 200 new 
88: marketers (qualified energy brokers) have appeared on the US electricity market \cite{ICC}.
89: Similarly, during the last six months more then 70 companies have obtained licenses to
90: trade electricity on the just liberalized Polish wholesale power market.
91: 
92: Organizations which have been used to long-term fixed price contracts are now 
93: becoming increasingly exposed to price volatility and, of necessity, are seeking
94: to hedge and speculatively trade to reduce their exposure to price risk.
95: The scenario in today's energy market is similar in many ways to the emergence of 
96: derivatives trading in the capital markets. From the modest beginnings 
97: in the late 1970s, financial markets have seen a massive explosion in the use of
98: derivative products. Starting with simple futures contracts and forward rate
99: agreements through swaps and on to increasingly ingenious and complex contracts.
100: The financial derivatives markets invented layer upon layer of new derivatives 
101: products using the basic building blocks to design tailor made hedges for 
102: customers \cite{coleman}.
103: 
104: Most derivatives markets begin with exchange traded futures. Global energy markets are 
105: no different. Heating oil futures appeared on the New York Mercantile Exchange (NYMEX) 
106: in November 1978 and futures on crude oil appeared there in March 1983. As in all other
107: markets options followed quickly -- on crude oil futures in 1986 and on heating oil a year
108: later. Commercial banks began providing commodity price risk management products in 1986,
109: when The Chase Manhattan Bank arranged the first oil swap \cite{ssw95}. 
110: Natural gas futures and OTC instruments began in 1990. And electricity futures contracts 
111: began trading on the world's most mature Scandinavian power market (Nord Pool -- 
112: the Nordic Power Exchange) in 1995 followed next year by the US (NYMEX) and 
113: Australian/New Zealand markets. 
114: First electricity options appeared on NYMEX in 1996. Last year Nord Pool introduced first 
115: exchange traded exotic options -- asian (average) options on electricity futures. 
116: Gas and electricity are now accelerating the change process of liquid trading and 
117: cross-energy commodity arbitrage. In effect, a conservative industry is continuing 
118: to be transformed through financial engineering. 
119: 
120: 
121: \section{Arbitrage pricing}
122: 
123: Demand and supply of electricity are balanced on a knife-edge because 
124: electric power cannot be economically stored, end user demand is largely weather 
125: dependent (see Fig. 1), and the reliability of the grid is paramount. 
126: Relatively small changes in load or generation can cause large changes in price and all 
127: in a matter of hours, if not minutes. In this respect there is no other market like it. 
128: 
129: \begin{figure}[htbp]
130: \centerline{\epsfxsize=10cm \epsfbox{fig1.eps}}
131: \caption{Nord Pool spot system price (top) and mean temperatures in Oslo, Norway (bottom) 
132: for the whole 1998 year. Clearly lower temperatures cause higher power consumption 
133: (heating) and thus raise electricity prices.
134: }
135: \end{figure}
136: 
137: Because of storage problems standard arbitrage type arguments cannot be used to price a number of
138: electricity derivatives \cite{ww00}. For example, when pricing a forward contract on crude oil, 
139: i.e. a contract for delivery of a specified amount of oil for a fixed price $K$ at a specified 
140: location and time in the future, we use the formula
141: $$
142: K=U (1+rT) + C,
143: $$
144: where $U$ is the current price of crude oil, $r$ is the risk-free 
145: interest rate, $T$ is the time to maturity of the contract, and $C$ is the so called 
146: {\it cost of carry} (a sum of storage, insurance, spoilage, and obsolescence costs). 
147: The formula can be derived by analyzing the following strategy \cite{ww00,ww98,hull97}:
148: \begin{itemize}
149: \item take a short position in the forward contract (i.e. agree to deliver oil for 
150: a fixed price $K$ at maturity) and take a loan from a bank to finance buying crude 
151: oil worth $U$ dollars,
152: \item store the oil until maturity (for time $T$) incurring the cost of carry $C$,
153: \item at maturity deliver the oil to the buyer of the forward contract for a fixed price $K$
154: and return the loan (with interest) to the bank.
155: \end{itemize}
156: The forward price $K$ should be such that from today's (the time we sign the contract) 
157: perspective no arbitrage is possible, i.e. we cannot make money without taking risk.
158: Thus the forward price should equal today's price ($U$) plus interest ($UrT$) 
159: paid to the bank for lending the money plus the cost of carry ($C$).
160: However, for contracts written on electricity the cost of carry is very large (or even
161: infinite) compared to the value of the delivered commodity and for this reason arbitrage 
162: type arguments cannot be used to price electricity derivatives. So what can we do? Well, 
163: we can either use other methods (eg. weather correlations, consumption prediction) for 
164: pricing such derivatives or use derivatives written not on electricity itself but on 
165: other derivatives.
166: The former method is used for pricing the first layer of derivatives -- forwards and
167: futures on electricity, whereas the latter for next layers -- options on futures 
168: (note that all exchange traded options are written on electricity futures and not 
169: electricity itself), swaptions, etc.
170: 
171: 
172: \section{Volatility}
173: 
174: In Figure 2 the average daily prices of the California Power Exchange 
175: (CalPX) spot market and their returns are plotted for the period April 1st, 1998 -- 
176: January 31st, 2000. Note that, unlike in the financial markets, electricity is traded 
177: every hour of the year -- including nights, weekends and holidays. Average daily price 
178: is a simple reference index constructed by adding up all 24 hourly prices during a day 
179: and dividing the sum by 24. One hour is the smallest time interval when prices can change, 
180: because in spot electricity trading prices are set constant for delivery of power during 
181: a certain hour. 
182: 
183: \begin{figure}[htbp]
184: \centerline{\epsfxsize=10cm \epsfbox{fig2.eps}}
185: \caption{Average daily spot prices (top) and their returns (bottom) for the California
186: Power Exchange since the opening of the exchange (April 1st, 1998) till January 31st, 2000. 
187: }
188: \end{figure}
189: 
190: The price of electricity is far more volatile than that of other commodities normally 
191: noted for extreme volatility. Applying the classical notion of volatility -- the standard 
192: deviation of returns (i.e. logarithmic price changes: $r_t= \log x_{t+1} - \log x_t$), 
193: we obtain that measured on a daily scale for a series roughly one year in length: 
194: \begin{itemize}
195: \item treasury bills and notes have a volatility of less than 0.5\%,
196: \item stock indices have a moderate volatility of about 1-1.5\%,
197: \item commodities like crude oil or natural gas have higher volatilities (1.5-4\%),
198: \item very volatile stocks have volatilities not exceeding 4\%,
199: \item and electric energy has the highest volatility -- up to 30\%!
200: \end{itemize}
201: However, when measured on different time scales, electricity price volatility does not 
202: behave like that for financial instruments. For the data illustrated in Fig. 2
203: daily volatility is about 23\%, whereas monthly (30-day) volatility is about 33\%. 
204: This is much less then predicted by Brownian motion (the distance traveled 
205: by a particle is proportional to the square root of time) for which we would obtain
206: $23\% \times \sqrt{30} \approx 125\%$. Thus Black-Scholes type formulas \cite{ssw95,ww98,hull97} 
207: should in general overestimate premiums of long-term options written on electricity! 
208: 
209: Another feature of electricity price volatility is its seasonal character.
210: The daily and weekly seasonality of volatility can be illustrated \cite{o93,o97}
211: by the intra--weekly plot of mean absolute hourly price changes, see Fig. 3. 
212: The statistical week is divided into 168 hours from Monday 0:00-1:00 to Sunday 23:00-24:00. 
213: Each bar represents the mean absolute change in prices for every hour from the same 
214: week--days counted for energy prices from CalPX. The sampling period starts with the 
215: opening of the exchange (April 1st, 1998) and lasts until January 25th, 2000, so that 
216: we analyze 95 full weeks.
217: 
218: \begin{figure}[htbp]
219: \centerline{\epsfxsize=10cm \epsfbox{fig3.eps}}
220: \caption{Intra--weekly plot of mean absolute hourly price changes for the CalPX spot
221: market. The statistical week is divided into 168 hours from Monday 0:00-1:00 to Sunday 
222: 23:00-24:00.
223: }
224: \end{figure}
225: 
226: The patterns of volatility are clearly correlated to the on-peak/off-peak specification 
227: of the market. The lowest volatility is observed on the weekends and during night (off-peak) 
228: hours. However, unlike for the global interbank FX (currency) market \cite{o97}, 
229: the volatility during weekends is of the same order of magnitude as that for working days.
230: High volatility is observed during on-peak working day hours, with a maximum for hour 
231: 15:00-16:00. Saturday has a similar volatility pattern, but on Sunday the maximum is 
232: postponed till 17:00-18:00.
233: 
234: 
235: \section{Autocorrelation of returns}
236: 
237: Seasonality of a time series of returns $r_t$ can be demonstrated by plotting 
238: the autocorrelation function \cite{beran94}
239: $$
240: {\bf acf}(r,k) = \frac{\sum_{t=k+1}^{N}
241:   (r_t-\bar{r})(r_{t-k}-\bar{r})}{\sum_{t=1}^{N} (r_t-\bar{r})^2},
242: $$
243: where $N$ is the sample length and
244: $$
245: \bar{r} = \frac{1}{N} \sum_{t=1}^{N} r_t,
246: $$
247: for different time lags $k$ as in Fig. 4. For electricity spot price returns there is 
248: a strong 7-day dependence which, when we think about it, is not that surprising. However, 
249: what is surprising is the fact that this dependence structure lasts almost forever 
250: (or as long as the analyzed data set)! For most financial data autocorrelation of returns
251: dies out (or more precisely: falls into the confidence interval of Gaussian random walk)
252: after 10-20 days and long-term autocorrelations are found only for squared returns 
253: or absolute value of returns \cite{ww98,o93,o97,bp97,ms99}.  
254: 
255: \begin{figure}[htbp]
256: \centerline{\epsfxsize=10cm \epsfbox{fig4.eps}}
257: \caption{Lagged autocorrelation function for CalPX spot price returns (see Fig. 2).
258: Dashed horizontal lines represent the 95\% confidence interval of a Gaussian random walk.
259: }
260: \end{figure}
261: 
262: This 7-day cyclic correlation can be removed by differentiation, i.e. by constructing
263: the data series $z_t = r_{t+7} - r_t$. The lagged autocorrelation of $z_t$ is shown in 
264: Fig. 5. We can observe two evident outliers: for lag=1 day and for lag=7 days. Both
265: have negative correlations. This implies a strong mean-reverting property \cite{hull97} 
266: of the returns as was already suggested by the results of Section 3. 
267: 
268: \begin{figure}[htbp]
269: \centerline{\epsfxsize=10cm \epsfbox{fig5.eps}}
270: \caption{Lagged autocorrelation function for CalPX spot price returns after 
271: differentiation by 7 days. A strong mean-reverting property is revealed.
272: }
273: \end{figure}
274: 
275: 
276: \section{Final remarks}
277: 
278: As we have shown the price of electricity is far more volatile than that of other 
279: commodities normally noted for extreme volatility. However, the term structure of
280: volatility distinguishes electricity from most financial assets and forces us to use 
281: Black-Scholes type models with great care and a doze of skepticism. 
282: On the other hand, the mean-reverting property puts electricity in the same box as 
283: interest rates and suggests that the search for models of electricity price dynamics 
284: should be started with examining and calibrating certain interest rate models 
285: \cite{rb80,hw90}.
286: 
287: 
288: 
289: \begin{thebibliography}{99}
290: {\small
291: \vspace{-.3cm} \bibitem{fusaro} P.C. Fusaro (ed.), Energy Risk Management, McGraw-Hill, New York, 1998.
292: \vspace{-.3cm} \bibitem{warrick} J.B. Warrick, Wisconsin Energy Forum, Sept. 28th, 1999.
293: \vspace{-.3cm} \bibitem{ww00} A. Weron, R. Weron, Power Exchange: Risk Management Strategies, (in Polish), CIRE, Wroc{\l}aw, 2000.
294: \vspace{-.3cm} \bibitem{bll99} R. Bjorgan, C.-C. Liu, J. Lawarree, IEEE Trans. Power Systems 14 (1999) 1285.
295: \vspace{-.3cm} \bibitem{ICC} International Chamber of Commerce, Liberalization and privatization of the Energy Sector, Paris, July 1998.
296: \vspace{-.3cm} \bibitem{kurtz} D. Kurtz, Power Engineering Int. 7(4) (1999) 33.
297: \vspace{-.3cm} \bibitem{calpx} CalPX, A. Petursson, CalPX White Papers,\\ http://www.calpx.com/news/publications/index.htm, 1998.
298: \vspace{-.3cm} \bibitem{green} S. Green, Power Engineering Int. 7(4) (1999) 45.
299: \vspace{-.3cm} \bibitem{coleman} M. Coleman, FSD Int. White Papers, http://www.fsd.co.uk/www/news/papers.htm, 2000.
300: \vspace{-.3cm} \bibitem{ssw95} Ch.W. Smithson, C.W. Smith, Jr., D.S. Wilford, Managing Financial Risk, IRWIN, Chicago, 1995.
301: \vspace{-.3cm} \bibitem{ww98} A. Weron, R. Weron, Financial Engineering: Derivatives Pricing, Computer Simulations, Market Statistics, (in Polish), WNT, Warsaw, 1998.
302: \vspace{-.3cm} \bibitem{hull97} J.C. Hull, Options, Futures, and Other Derivatives, 3rd ed., Prentice Hall, 1997.
303: \vspace{-.3cm} \bibitem{o93} M.M. Dacorogna, U.A. M\"uller, R.J. Nagler, R.B. Olsen, O.V. Pictet, J. Intern. Money Finance 12 (1993) 413.
304: \vspace{-.3cm} \bibitem{o97} D.M. Guillaume, M.M. Dacorogna, R.R. Dave, U.A. M\"uller, R.B. Olsen, O.V. Pictet, Finance Stochast. 1 (1997) 95.
305: \vspace{-.3cm} \bibitem{beran94} J. Beran, Statistics for Long-Memory Processes, Chapman \& Hall, New York, 1997.
306: \vspace{-.3cm} \bibitem{bp97} J.P. Bouchaud, M. Potters, Theory of Financial Risk, (in French), Alea-Saclay, Eyrolles, Paris, 1997.
307: \vspace{-.3cm} \bibitem{ms99} R.N. Mantegna, H.E. Stanley, An Introduction to Econophysics: Correlations and Complexity in Finance, Cambridge University Press, Cambridge, 1999.
308: \vspace{-.3cm} \bibitem{rb80} R. Rendleman, B. Bartter, J. Financial and Quantitative Analysis 15 (1980) 11.
309: \vspace{-.3cm} \bibitem{hw90} J.C. Hull, A. White, Rev. Financial Studies 3 (1990) 573.
310: }
311: \end{thebibliography}
312: 
313: \end{document}
314: