Professional Documents
Culture Documents
Abstract
This paper presents a valuation study of hydropower investment opportunities in the Norwegian context. According to NVE
(Norwegian Water Resources and Energy Directorate, the regulator), there is a potential of 39 TWh not yet developed (generation in a
normal year is approximately 120 TWh).
By using the conceptual real option framework of Dixit and Pindyck [1994. Investment Under Uncertainty. Princeton University
Press, Princeton, NJ] one can estimate the value of investment opportunities to NOK 11 million/GWh (EUR 1.4 million/GWh).
Furthermore, the optimal trigger price for initiating an investment based on electricity forward prices is calculated to NOK 0.32/kWh
(EUR 0.04/kWh). The analysis shows consistency between real option theory and aggregate investment behaviour in Norwegian
hydropower.
r 2007 Elsevier Ltd. All rights reserved.
0301-4215/$ - see front matter r 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.enpol.2007.07.021
ARTICLE IN PRESS
5902 F. Kjærland / Energy Policy 35 (2007) 5901–5908
2.00
1.50 work of Dixit and Pindyck (1994) and a discussion of the
1.00 relevance of this model on investment opportunities in
0.50
Norwegian hydropower. Section 3 describes the applica-
0.00
tion of this model, including a discussion of the input
1989 1991 1993 1995 1997 1999 2001 2003 parameters and the numerical analysis. Section 4 draws the
ear conclusions and implications.
Fig. 1. Hydropower plants under construction 1989–2004 (SSB (Statistics
Norway), 2006). 2. Methodology
electricity, one can learn that demand and generation must prices that are used in this analysis are the longest contracts
match each other continuously. If not in balance the that can provide reliable data, and thus are the best
transmission network will collapse. Current technology estimates of value of both generation assets-in-place and
gives no possibility of storing electricity.3 This non- investment opportunities.
storability aspect implies that electricity cannot be con- The described modelling implies that hydro generation
sidered a financial asset, eliminating the possibility for the basically behaves like a base load producer. This neglects
traditional arbitrage approach. The implication of this is the flexibility characteristics that are associated with hydro
that a megawatt hour of power cannot be held as an generation, and represents a shortcoming.
investment in a portfolio. Electric power can be neither
borrowed nor shortened, and then bought back and 3. Application of the Dixit and Pindyck model on
returned later. This violates the important spanning Norwegian hydropower investment opportunities
assumption in the Dixit and Pindyck framework, and the
traditional non-arbitrage approach for the valuation of 3.1. Data for applying the model
options. Finance-based asset pricing does therefore not
apply to power spot price dynamics. But efficient markets 3.1.1. Sigma, s
do apply to the pricing of derivatives on power. Therefore s is the standard deviation of the underlying asset. In
this study relates to forward contracts, which not are under this context,it implies that s is the standard deviation of the
the same restrictions. observed forward prices. The forward contract structure at
Because of these special properties of electricity, there Nord Pool has gone through a transition phase. The
are strong reasons for dealing with forward prices directly previous structure was based on the distinction of three
rather than endogenously through spot prices (Koekebak- seasons: winter 1 (1 January–30 April), summer (1 May–30
ker and Sødal, 2001). If one relates to the observed forward September) and winter 2 (1 October–31 December). There
and futures prices one deals with a tradable asset, and does were also year forward contracts. The new forward
not need to struggle with the complicated area of electricity contract structure is based on calendar month, quarter
spot price as a non-asset. While electricity is non-storable, (3 calendar months) and year contracts. This was
forward contracts are. Hence by relating to forward prices, introduced in 2004 (Nord Pool, 2005).
there should be no violation of the essential spanning To obtain an overview of the term structure and
assumption in the Dixit and Pindyck framework. volatility, it should be sufficient to concentrate on the
The model relies on a geometric random walk. This is a tertial/quarterly and yearly contracts. These are the longest
convenient assumption because it yields an analytical contracts that give the most relevant information when the
solution. The modelling of stochastic process is nonetheless purpose is to focus on the valuation of investment
controversial. The recommendation of Ronn (2002) is opportunities. The tertial (from 2004 quarterly) contracts
geometric Brownian motion (GBM) for forward priced- are tradable for the 2 following years after trading, while
based models of electricity prices. the yearly contracts are tradable for 3 years ahead. By
This model is in continuous time. In practice however, analysing these 9/11 contacts one can obtain a term
observed forward prices are restricted to discrete values. structure and a long-time volatility trend.
Nevertheless, the underlying factors in the industry, like the From the FTP server of Nord Pool, one can obtain a
market mechanisms in an industry with many participants, simple descriptive analysis of the tertial, quarterly and
the sensitivity of national reservoir level information and yearly contracts. Since the volatility parameter in GBM is
downpour and temperature forecasts as well as the an annualised volatility, the annualised standard deviations
regulators’ constant monitoring of the industry, are for the price return of the relevant contracts are shown in
continuous in nature. Therefore, it should not be con- Table 1 (from 2004 the prices are in euro. The change into
troversial to apply a continuous time model for the the Norwegian currency has been based on the average
stochastic process of forward electricity prices. currency in the actual year). These numbers reveal the
The value of investment opportunities in hydropower seasonal pattern and also the term structure of the
generation is for obvious reasons strongly related to future volatility (Lucia and Schwartz, 2002; Ronn, 2002). The
expected prices of electricity. Hydropower plants are numbers also reveal an increase in volatility after the shock
normally assumed to last for several decades. Investors winter of 2002–2003.
are thus looking way ahead of 3–4 years’ forward prices The results are consistent with earlier studies (Bjerksund
that can be observed at Nord Pool. However, there is no et al., 2000; Koekebakker and Ollmar, 2005). There is also
efficient market for longer forward prices. The observed consistency with the assumption of Ronn (2002). The
volatility is high for contracts with short time to maturity
3
and is convex decreasing for contracts with longer time to
There are some possibilities of storing. Advanced, expensive technol- maturity. An overall analysis of the development of the
ogies like pumped/storage hydro, high-pressure air facilities and batteries
can convert electricity into potential energy in other forms and convert it forward prices (Fig. 3) shows, however, an increase in both
back. This is, however, a rare opportunity and involves significant loss in price level and volatility level after the winter of 2002–2003.
conversion. This is ignored in the analysis. Overall this gives support to allow the level of s be about
ARTICLE IN PRESS
5904 F. Kjærland / Energy Policy 35 (2007) 5901–5908
20–25% as a base case volatility input parameter in the The P variable should be a measure of the level of
model. In the analysis later in this paper, the starting point forward prices. The choice of this paper is to let P be the
for the calculations is set with a s of 25%. average forward price of the six tertial (eight quarterly) and
3-year ahead forward prices. If one relates to these 9/11
3.1.2. Price, P contracts, one can calculate an average forward price at
s is derived from an analysis of the structure of a each trading day. The development of this representative
portfolio of forward contracts. Hence the P for forward forward price is shown in Fig. 3.
prices, which has to be the X-axes concerning valuation
of the investment opportunities and the trigger for
investments in the Dixit and Pindyck model frame- 3.1.3. Delta, d
work, has to be obtained from the same portfolio of In the theoretical introduction of the Dixit and Pindyck
forward contracts. model, d represented the net marginal convenience yield
from storage. The assumption was that the output was a
storable commodity. As pointed out, this is a complex issue
Table 1 concerning electricity. But in the hydropower-dominated
Average annualised standard deviation on yearly and tertial/quarterly
system of Norway, the water reservoir levels can serve as a
forward contracts 1999–2006
kind of inventory, leading to high reservoir levels
Year Average annualised std. deviation (%) (inventory) in the summer and low levels in the winter
(Fig. 4).
1999 15.0
2000 8.5
The definition of convenience yield is ‘‘the flow of
2001 18.7 services which accrues to the owner of a physical inventory
2002 25.5 but not to the owner of a contract for future delivery’’
2003 38.1 (Brennan, 1991). As already pointed out, electricity has
2004 16.9 some peculiar properties. Because of this some careful
2005 21.1
2006 26.6 considerations should be made. Botterud et al. (2002) point
out that there are asymmetry aspects between the supply
Fig. 3. Development of average forward price (øre/kWh/NOK/MWh), 7 September 1998–27 December 2006.
100
90
80
% of reservoir capacity
2006
70 2005
60 2004
2003
50
2002
40 2001
30 2000
20 1999
10
0
1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52
Week
where Pt is the spot price at time t, Ft,T is the future price 3.1.5. Investment, I
for delivery at time t+T, rT is the risk-free T-period The general picture when examining this sector is the
interest rate and kT is the per unit cost of physical storage. high entry barrier of high investment costs and the
This equation can be proved by normal non-arbitrage relatively low level of variable cost. Concerning invest-
reasoning (Pindyck, 2001). Storage cost is assumed to be ments, the reports from NVE seem to level 3 and
zero in this context (storage costs are vital in a normal 5 NOK/kWh as standard intervals concerning classification
discussion of convenience yield, but storage of forward of the potential new hydropower plants.
contracts seems reasonable to set as zero). The relative According to NVE (2004) reports, there is an overall
convenience yield becomes estimation that the economical limit for investment is
considered as 3 NOK/kWh. This limit has been increased
ct;T since then. An NVE report also shows the increased
¼ ½ð1 þ rT ÞPt F t;T . (2)
Pt knowledge of small-scale hydropower plants leading to a
significant upside change in potential (NVE, 2005). In this
To obtain a current real option value, the focus is on the
model framework, it therefore seems appropriate to use
latest quarterly and yearly contracts, which determine d. By
these numbers as the investment expenditure in the analysis.
using Eq. (2) one obtains the results shown in Table 2.
4
This parameter is under GBM related to the expected growth in the 3.1.6. Variable cost, c
current estimate of the forward curve P. If one e.g. calls the growth a, we There is low cost in production in this industry. The level
can set that a ¼ rd. of c is dependent on aspects like age, size and complexity
ARTICLE IN PRESS
5906 F. Kjærland / Energy Policy 35 (2007) 5901–5908
V(P)
an argument for letting this part of the costs be included in 200
the investments cost (which NVE refers to as ‘‘conserva-
150
tive’’) and letting the c parameter be closer to zero.
100
3.2. Analysis of option value and trigger price
50
The above discussion of parameter values gives a way of σ = 0.25 δ = 0.025 r = 0.045 I = 300 c = 3
0
calculating the option value in this sector according to the 0 2 4 6 8 10 12
following equations in the model framework of Dixit and P
Pindyck (1994):
Fig. 5. Value of investment opportunity as a function of average forward
V ðPÞ ¼ A1 Pb1 if Poc, (3a) price.
b2 ¼ 1=2 ðr dÞ=s2
Table 4
f½ðr dÞ=s2 1=22 þ 2r=s2 g1=2 . ð4bÞ Sensitivity analysis of the option value when the d parameter is changed
The constants A1 and A2 are expressed as Average Value of option (øre/kWh) (s ¼ 0.25, r ¼ 0.045,
forward price I ¼ 300, c ¼ 3)
r b2 ðr dÞ ð1b Þ (øre/kWh)
A1 ¼ c 1 , (5a) d ¼ 0.01 d ¼ 0.02 d ¼ 0.03 d ¼ 0.04
rdðb1 b2 Þ
20 1937 940 612 454
r b1 ðr dÞ ð1b Þ 25 2436 1189 776 575
A2 ¼ c 2 . (5b)
rdðb1 b2 Þ 30 2935 1438 941 698
35 3435 1687 1107 821
Fig. 5 shows the results on the basis of the previously
discussed volatility (s), convenience yield (d), risk-free rate
(r) and investment level of 300 (øre/kWh), and hence (Table 5). A lower d would result in a considerably higher
variable cost 3 (øre/kWh). option value.
P represents the average forward price of the included If a local company then possesses a potential of
forward contracts. The value V(P) is the value of an 100 GWh in its area, this gives a value of NOK 1.1 billion
investment opportunity in øre (0.01 NOK) per kWh. The according to this approach. This can be a qualified estimate
figure shows the significant value this option has depending for the value of the company beyond assets-in-place, and
on the forward price level. A change in input parameter I hence can be an estimate for a bid premium if such a
and hence c, to 500 and 5 as discussed earlier, reveals no company is involved in a merger or an acquisition.5
big difference in the results. What about the optimal timing of investing? The model
The option value of investment opportunities is mainly framework of Dixit and Pindyck (1994) has developed
the intrinsic in-the-money value, close to max{(P/d) the following equation that reveals P* when solved
(c/r);0}. The option value to stop producing when prices numerically:
are decreasing (‘‘time value’’) is low because of the low A2 ðb1 b2 Þ b2 ðb1 1Þ c
variable cost (c). ðP Þ þ P I ¼ 0, (6)
b1 db1 r
With the first set of parameters, at a present (late 2006)
level of forward prices of about 30, this gives an option 5
There are a number of other aspects involved in such an assessment
value of about 11 NOK/kWh (Table 3). This option value such as quota regulations, the relationship with private fall rights owners
is,however, sensitive to changes in both d (Table 4) and s (land owners) and the uncertainty of obtaining permission from NVE.
ARTICLE IN PRESS
F. Kjærland / Energy Policy 35 (2007) 5901–5908 5907
Table 5 2000
Option value as a function of average forward price including different
volatilities as input parameters
1500
Average Value of option (øre/kWh) (d ¼ 0.025, r ¼ 0.045,
forward price I ¼ 300, c ¼ 3) P* = 32.22
(øre/kWh)
F(P), V(P)-I
s¼0 s ¼ 0.20 s ¼ 0.225 s ¼ 0.30 1000
300 3 0.25 0.025 0.045 32.22 Fig. 6. Graphical solution of P* with basic parameters.
400 4 0.25 0.025 0.045 42.96
300 3 0.20 0.025 0.045 27.43
300 3 0.225 0.025 0.045 29.74
300 3 0.25 0.03 0.045 33.25 leading to a possible investment decision at some lower
300 3 0.25 0.02 0.045 31.24 trigger price.
300 3 0.25 0.025 0.03 29.52
300 3 0.30 0.02 0.06 35.46 4. Conclusions and implications
been under construction plants with generation capacity Botterud, A., Bhattacharya, A.K., Ilic, M., 2002. Futures and spot
less than 1.2 TWh (Fig. 2), lower than granted licences and prices—an analysis of the Scandinavian electricity market. Paper
presented at the 34th Annual American Power Symposium (NAPS),
much lower than applied licences. This reality gives support
Tempe, AZ, USA.
to the results of this real option approach. Brennan, M.J., 1991. The price of convenience and the valuation of
The analysis also shows that the s in the model, the commodity contingent claims. In: Lund, D., Øksendal, B. (Eds.),
volatility of the forward price, has a profound impact on Stochastic Models and Option Values. North-Holland, Amsterdam.
the option value and hence the hesitation in making Brennan, M.J., Schwartz, E.S., 1986. A new approach to evaluating
irreversible investments. There have always existed un- natural resources. In: Stern, J.M., Chew, D.H. (Eds.), The Revolution
in Corporate Finance. Blackwell, Oxford.
certainty factors for investor’s in this sector. The un- Bye, T., Hope, E., 2006. Electricity market reform—the Norwegian
certainty has been linked to aspects like demand, experience. In: Sørgard, L. (Ed.), Competition and Welfare—The
international fuel prices, transmission constraints and Norwegian Experience. The Norwegian Competition Authority,
climate. But there seems to be an increase in uncertainty pp. 21–50.
and hence volatility in electricity price development. This Bye, T., von der Fehr, N.-H.M., Riis, C., Sørgard, L., 2003. Kraft og
Makt—en analyse av konkurranseforholdene i kraftmarkedet. Minis-
significant increase in volatility can possibly be linked to a try of Labour and Administration.
number of controversial and partly unsettled political Clewlow, L., Strickland, C., 2000. Energy Derivatives: Pricing and Risk
issues as the possible introduction of green certificates,6 the Management. Lacima Publications, London, UK.
introduction of CO2 allowances, the home fall institute7 Deng, S.-J., Johnson, B., Sogomonian, A., 2001. Exotic electricity options
and the valuation of electricity generation and transmission assets.
and the emerging awareness of low levels of new
Decision Support Systems 30, 383–392.
investments, higher demand and transmission constraints. Dixit, A., Pindyck, R.S., 1994. Investment Under Uncertainty. Princeton
The increase in oil, gas and coal prices is also interfering. University Press, Princeton, NJ.
Anyway, the model shows that by reducing the uncertainty Gjølberg, O., Johnsen, T., 2001. Electricity Futures: Inventories and Price
one can lower the option value (alternative cost) and hence Relationships at Nord Pool. Department of Finance and Management
trigger the investments at a lower price. More stability and Science, Norwegian School of Economics and Business Administra-
tion, Norway.
more predictable framework conditions will thus decrease Koekebakker, S., Ollmar, F., 2005. Forward curve dynamics in the Nordic
the uncertainty and possibly make electricity prices less electricity market. Managerial Finance 31 (6), 72–93.
volatile. This encourages more investments in small-scale Koekebakker, S., Sødal, S., 2001. The Value of an Operating Electricity
hydropower plants and probably other electricity genera- Production Unit. Agder University College.
Lucia, J., Schwartz, E.S., 2002. Electricity prices and power derivatives:
tion facilities.
evidence from the Nordic power exchange. Review of Derivatives
This real option analysis gives deeper insight into a Research 5, 5–50.
controversial issue. An option approach explains invest- McDonald, R., Siegel, D., 1986. The value of waiting to invest. Quarterly
ment behaviour in a way that is not captured by a Journal of Economics 101 (4), 707–727.
neoclassical NPV approach. Even if there are factors in the Milne, A., Whalley, A.E., 2000. ‘Time to build, option value and
application of the Dixit and Pindyck framework that are investment decisions’: a comment. Journal of Financial Economics 56
(2), 325–332.
disputable, the analysis shows that real option theory gives Ministry of Petroleum and Energy, 2006. Fakta 2006 (in Norwegian).
insight into the value of investment opportunities and Myers, S.C., 1987. Finance theory and financial strategy. Midland
aggregate investment behaviour in this industry. Corporate Finance Journal 5 (1), 6–13.
Nord Pool, 2005. Trade at Nord Pool’s Financial Market.
Norges Bank, 2007. NIBOR. From /http://www.norges-bank.no/stat/
References valutakurser/S.
NVE, 2002. Kostnader ved produksjon av kraft og varme.
Al-Sunaidy, A., Green, R., 2006. Electricity deregulation in OECD NVE, 2004. Beregning av potensial for små kraftverk i Norge (No. 19/
(Organization for Economic Cooperation and Development) coun- 2004).
tries. Energy 31 (6–7), 769–787. NVE, 2005. Kraftbalansen i Norge mot 2020 (No. 20/2005).
Berkovitch, E., Israel, R., 2004. Why the NPV criterion does not maximize NVE, 2006. Vannkraftpotensial.
NPV. Review of Financial Studies 17 (1), 239–255. NVE, 2007. Water reservoir statistics. From /http://www.nve.no/
Bjerksund, P., Rasmussen, H., Stensland, G., 2000. Valuation and Risk vannmagtall/vannmagtall98_06.aspS.
Management in the Norwegian Electricity Market. NHH. Pindyck, R.S., 1991. Irreversibility, uncertainty, and investment. Journal
Black, F., Scholes, M., 1973. The pricing of options and corporate of Economic Literature 29 (3), 1110–1148.
liabilities. Journal of Political Economy 81, 637–659. Pindyck, R.S., 2001. The dynamics of commodity spot and futures
markets: a primer. The Energy Journal 22 (3), 1–29.
6
For several years there were expectations of a common Norwegian Ronn, E.I., 2002. Real Options and Energy Management. Risk Books.
Swedish market of green certificates which would make investments in Ross, S.A., 1995. Uses, abuses, and alternatives to the net-present-value
renewable electricity generation like hydropower plants more profitable. rule. Financial Management 24 (3, Autumn), 96–102.
This was abandoned by the Norwegian Government in 2006. Schwartz, E.S., 1997. The stochastic behavior of commodity prices:
7
The home fall institute is founded in the Norwegian legislation of implications for valuation and hedging. Journal of Finance July, 923–973.
possessing and operating hydropower plants. Home fall essentially means Schwartz, E.S., 1998. Valuing long-term commodity assets. Financial
that private owned plants will without compensation be passed over to the Management 27 (Spring), 57–66.
authorities when the licence expires. The length of a licence is 60 years, but SSB (Statistics Norway), 2006. Nyttbar, utbygd og ikkje utbygd vasskraft
will possibly be 75 years. There has been considerable dispute on this issue etter fylke. 2003 og 2004. GWh.
since 2001, especially whether publicly owned plants should also be under Trigeorgis, L., 1996. Real Options: Managerial Flexibility and Strategy in
the same legislation. Resource Allocation. MIT Press, Cambridge, MA.