# [UIS, Tu 10:45] Uncertainty and Investment Strategy

## 1.Evaluation of Compressed Air Energy Storage Plant Based on Stochastic Electricity Price Simulations (Keles, Möst, Fichtner)

**Introduction**

Liberalisation of energy markets, climate policy and the promotion of renewable energy change the framework conditions of the formerly strictly regulated energy markets. Generation companies are mainly affected from these changing framework conditions, as they are exposed to the different risks from liberalised energy markets in combination with huge and to a large extent irreversible investments. Uncertainties, which generation companies are facing, are especially the development of electricity prices as well as fuel prices. On the other side the increase of volatile feed-in of renewable electricity makes new investments in energy storages necessary. These investments have to be evaluated with the help of new approaches which can capture the uncertain parameters.

**Methodology**

In this analysis a compressed air energy storage (CAES) plant is evaluated considering the volatility of electricity prices. Therefore price paths for electricity are simulated with the help of stochastic processes, such as a mean-reversion process. Thereby a regime-switching approach is used for two different price regimes, one for the base price development, one for price peaks. The price simulation considers also deterministic components as trend and seasonality of the electricity price series. Then the simulated price paths are used within an optimization model, which maximizes the annual contribution margin of the plant during its economical lifetime. The contribution margin corresponds to the income from electricity sales on the spot market and from the power sales on the reserve market minus the total variable costs of the plant dispatch. Based on the optimum contribution margin, the net present value (NPV) of the CAES-plant is calculated to support the investment decision for this plant type. A Monte-Carlo simulation based on 1000 simulations is carried out to strengthen the evaluation results. At last the results based on the stochastic price simulation are compared with a deterministic calculation, to point out the impact of stochastic parameters.

**Results**

The evaluation of the CAES-plant with the above described methodology results in a positive NPV, if the interest yield of the investor equals to 5% per year. However, the NPV of the plant changes to a negative value for higher interest yields (i.e. 10%). For the deterministic case, the NPV is already negative for a interest rate of 5%.

**Conclusion**

At last it is worth mentioning that the NPV and therefore the investment decision is strongly dependent on the expected interest yield of the investor. Besides, the comparison between the stochastic and deterministic approach shows how the stochastic one, considering a larger price data series and its volatility, can avoid an underestimation of the CAES-plant value. It has to be annotated that transmission fees for the consumption of electricity have to be paid additionally, which has not been considered in the calculation. Thus, the CAES plant is still less profitable and it is not a lucrative investment option at the estimated market prices.

## 2.Investment under Uncertainty or the Recovery of Strategy in Investment Models (Kolmsee)

**Starting Point**

Over the next years there is a strong need for investments in new generation capacity. This is true not only for Europe but also for the USA and many of the emerging countries. While in the past investment conditions have seemed to be stable and easy to forecast even long term today’s decision makers face strong price volatility, a changing political framework and also a changing demand structure as even big industrial consumers might quit production or rapidly build up new production lines. Under these conditions the instruments to support investment decisions are revised.

**Not only on Methodology**

While the Discounted Cash Flow (DCF) -methodology is still the most common instrument there is a broad literature on case studies using real option theory for decision making on investments in new power generation plants. While calculation for financial options is straight forward based on Black-Scholes-Merton-Model or a Montecarlo simulation, real options have led to tools like the event tree analysis based on decision theory modeling. Event trees represent in a very simple form hysteresis in decision theory. I will argue that the simple event tree model is better than a simple calculus but neither a satisfying guide nor representation for real world investment decision. I will mainly stay in the analytical framework given by formal decision theory. As excessive mathematical formalism is one of the hurdles for using real options models I will keep my arguments as close to a non-formal language as possible and avoid elaborated mathematical representations of decision theory models.

**Main Arguments**

The decision tree model has been praised as good representation of options as treated by rational actors in real world decision situations. It has therefore also been applied to larger investment decisions like power plants. First I will argue that although a real improvement compared to a simple calculus the model – as applied in its standard form – does not deal with the difference between risk or uncertainty and the lack of information or insecurity. Secondly I will point at a blind spot of the model itself if applied to long term investment decision like the one in power plants. These decisions are necessarily set in a market environment which involves other players. Therefore the decision tree model should be embedded in a framework as given by negotiation theory which deals with n-steps, n-participants situations.

**Conclusions**

If these arguments are correct long term decisions should not be based on a more or less simple mathematical representation of the future only. In the long run a company faces many options which can be valued in a market environment only. Therefore investment models should necessarily consider possible competitor`s (re)action. This leads to a mixture of scenario analysis as the most simple tool to evaluating complex situations and more complex negotiation or n-players game theory models. In any case it brings strategy back into the evaluation of investments. It strongly depends on the market positioning of a company whether future investments will pay back – because in a competitive environment competitors do have a strong impact on the value of these investments.

## 3.Strengthening Investment Decisions for Power Generation Assets Using Portfolio Analysis – E.ON's Experience (Frueh, Soria-Lopez, Bodewig)

The use of modern portfolio theory in strategic investment decisions for generation assets provides insights on different generation mixes while considering the impacts on the risk/return trade-offs of new assets entering into an existing portfolio of generation assets. By concentrating on the overall portfolio of assets instead of each individual project's financial and strategic merits, a corporation determines an efficient set of portfolios called the efficient frontier. From the efficient frontier logical inferences can be made concerning the intensity of the different types of fuel, technologies, regions and other factors which tend to populate the efficient frontier. These inferences are not only valuable for individual projects; they can be utilized in the conceptualization of long-term strategic plans and initiatives.

In this study we used a set of market scenarios and analytical models (DCF) to evaluate E.ON's existing generation systems and projects under development or investigation. The outputs of these analytical models are the characteristics for each project or system by scenario that are the required inputs into the newly developed portfolio tool. These inputs together with the probability assumptions for the scenarios and predefined constraints (e.g. CO2 emissions, CAPEX, market share etc.) are used by the portfolio tool to determine the efficient frontier. The portfolio algorithm uses an optimization procedure (linear programming) that locates the discrete points on the efficient frontier.

In order to create the efficient frontier, the metrics for risk and return needed to be identified. The metric for the return is the expected net present value of the portfolio. A major issue during our investigation was the appropriate risk measure to employ. We used three different risk metrics in order to compare and understand the impacts of the different risk measures on the choice of projects populating the efficient frontier.

A closer examination of the results leads to a better understanding of the selection of some projects over other projects. The results underscore the effect of any correlation of projects with the existing system or with other projects. These correlations lead to the portfolio effect in the determination of the efficient frontier. It is this effect that creates value in the risk/reward trade-offs and is crucial to decision makers and strategic planners.