What is the Pexapark calculation methodology?

Pexapark’s pricing is based on a layered approach, whereby a theoretical baseload price is adjusted by a number of structural, risk- and cost-based terms in order to determine PPA prices. Careful calibration of all price components against observed prices, both in liquid traded markets and PPA markets ensure a fully market-consistent methodology. Pexapark’s comprehensive model therefore allows for an unbiased comparison of structurally different PPA deals.

The breakdown of the PPA price calculation is as follows:

Components labeled “Price” or “Cost” are expected values as determined from static price curves, seasonality patterns etc. whereas “Risk” components derive from dynamic simulations and reflect price adjustments against potential losses. In the following, the methodology for determining all components will be outlined.

Pexapark’s forward curve is derived from both liquid and illiquid price inputs and shaped to monthly granularity. Liquid price inputs are typically settlement prices obtained from exchanges or other price providers (Please consult table “Liquid Price Inputs” in the Appendix). Illiquid prices are obtained from PPA transactions reported by Pexapark. Using this data, the first step is to determine the initial shape of the price curve. In a second step, this initial shape is superimposed with a monthly seasonality.

Where available in liquid price data, monthly prices are used. Beyond the short-dated liquid tenor range, for which monthly quotes are available, quarterly and calendar year quotes are used. In the illiquid tenor range, the shape of the forward price curve is modelled to be linear, fitting illiquid price quotes in a least-square sense. Below figure shows how the curve looks at this stage.

After initial construction of the price curve (as described above), monthly shaping is applied, in consistency with historically observed price ratios, resulting in a curve as shown below.

The price of a hypothetical constant volume baseload deal of same tenor as the PPA, valued on Pexapark’s monthly shaped price forward curve.

Wind and PV deals have a distinct seasonal/daily shaping of volumes, which results in a price difference to the theoretical baseload price. This is accounted for by the hourly profile cost. The sign of hourly profile cost can be negative (typically the case for wind) or positive (typically the case for PV).

Liquidity cost is the cost of hedging that the purchaser of a PPA expects to incur over the life of the deal. Given the illiquidity of longer-term deals, the assumption underlying the calculation of liquidity cost is that the deal will be stack-and-roll hedged. Stack-and-roll hedging means that a PPA is hedged by entering into an offsetting position in the (shorter) liquid tenor range. Maintaining this hedge means that during the life of the deal multiple roll-overs of the hedge will be required, causing bid-ask cost, as old hedge trades are closed out and new hedge trades are entered into. With a neutral view on future price evolution and in absence of portfolio effects, a PPA purchaser expects to incur said hedging cost, therefore Pexapark adjusts prices by this amount in order to arrive at a fair bid level.

The ability to simulate the dynamics of the price forward curve is a cornerstone of Pexapark’s quantitative framework. Monte-Carlo simulation techniques bring two essential capabilities to the Pexapark framework:

Based on Monte-Carlo simulation, it is possible to generate ensembles realistic price scenarios which allow for a wealth of risk analytics on PPA deals (Profit-at-Risk, potential future credit exposure, analysis of hedge efficiency etc).

Volumes which are unhedged or hedged with a tenor mismatch (such as is the case in stack-and-roll) result in outright price risk or risk with respect to the shape of the forward curve. Generating multiple forward price scenarios with Monte-Carlo, it is possible to assess the potential financial impact of curve risk at a defined confidence level.

Fixed hourly power structures can be priced on a power forward curve, as described above (theoretical baseload price ± hourly profile cost). However, the intermittent nature of generation of renewable energy sources results in two sources of risk:

The Pexapark simulation model amalgamates both sources of financial risk into a single risk figure at a defined confidence level.

The setup of a simulation framework comprises three main elements: i) the choice of a forward curve model, ii) the calibration of the curve model and iii) the simulation.

Pexapark uses a two-factor model for forward prices to correctly account for the tenor-dependency of forward curve dynamics. The model distinguishes two regimes: i) a short-tenor regime (typically 1-3 years) and ii) a long-tenor regime (typically beyond 3 years). While volatility and correlations for long tenors are essentially constant, the short-tenor regime typically exhibits higher volatility and low correlation to the long end of the curve. The model features two stochastic driving terms, for the long and the short end of the curve and allows for specifying correlation between the long and the short end of the curve.

Realistic calibration of the model requires a choice of parameters which is consistent with historically observed volatilities and covariances. Pexapark has devised a robust calibration approach in order to ensure that outputs of the forward curve model match historical market observations. The calibration procedure delivers all required model parameters that are required as inputs to the Monte-Carlo simulation.

The forward curve, constructed as described above, is used as starting point of curve dynamics simulations. The shape and level of the forward curve are stepped forward in time via Monte-Carlo simulation of the stochastic differential equation of the forward curve model. Repeating this procedure multiple times creates an ensemble of scenarios of forward curve evolution. Statistical evaluation of the simulation data generated in this way allows for computing various risk metrics.