# Market Risk and its Implications for PPA Pricing

Power vs Other Asset Classes

Until recently, most renewable assets have been state-subsidized under Feed-in-Tariff (FiT) schemes, which shielded them from the fluctuating prices seen in wholesale traded power markets.

Utilities and commodity trading houses have long developed tools and skills to manage market risk, and indirectly offer such risk management services to project owners via a power purchase agreement (PPA). These contracts provide long-term price stability required to obtain project financing. While they are widely adopted in countries with no FiT schemes, the risk management techniques underlying their structuring are rarely understood outside utilities and trading houses.

The below graph shows annualized historical volatilities of various asset classes, i.e. a measure for the standard deviation of historically observed annual returns.

Fig.1: Historical volatilities for various asset classes

Exchange rates (EUR vs USD rates) and interest rates (futures on German Bundesanleihen) exhibit relatively low volatilities of typically below 10%, equities (Standard & Poors 500 index) are typically in the 10-15% range (current extreme levels are due to the massive equity crash).

In contrast, the volatility of power has been in excess of 20% for most of the last 4 years, experiencing periods of well over 30%. Given most financial firms have risk departments for the unique purpose of monitoring risk of assets less volatile than power, at least the same degree of care should be exercised when dealing with power exposures.

Measuring Market Risk with the Historical Simulation Approach

In the downloadable Excel file, we demonstrate the historical simulation approach as an easy-to-use and model-free approach which not only allows to quantify the risk of individual positions, but also enables the measurement of portfolio risk. Using historically observed prices, we demonstrate several basic concepts:

Measurement of (log)returns: For purposes of risk measurement we use so-called log-returns (i.e. the natural logarithms of price ratios between dates). A period commonly used for the calculation of VaR is 5 days, based on assuming that 5 days suffice to liquidate a position. Depending on the specific case, different choices may be taken.

Standard deviation of returns: We use the standard deviation of logreturns to characterize the breadth of the return distribution. The larger it is, the larger the financial risk of a position.

Cumulative Distribution Function (CDF) and Inverse CDF: By using these functions as shown in the spreadsheet, standard deviations of returns can be turned into P95 , P90 etc values. These are loss amounts at different confidence levels, i.e. levels of loss which are not exceeded at defined levels of probability (95%, 90% etc). Congratulations – at this point you have calculated VaR ! What we have derived...