Covariance risk: What it is and how to manage it
Project sponsors, banks and tax equity investors in transactions with hedges may be overlooking some risks that wind projects are bearing. Each risk should be borne by the party best able to manage it. In some deals, this may not be happening.
One such risk is covariance risk.
There has been a fundamental shift in how electricity is sold by independent generators. As utilities cut back on the amount of electricity they are buying under long-term contracts from independent generators, financial parties, like banks and commodities firms, entered the market to buy power. Utilities have tended to buy “as-generated power,” meaning they pay a fixed price regardless of how much power is generated and — critically — when it is generated. In contrast, financial parties typically buy power in fixed blocks: with a set volume of power every hour over the life of the contract.
Financial parties buy power this way either so that they can match up with a predictable load or, more commonly, so that they have a known volume of power to sell to the physical consumers of electricity.
Selling fixed volumes of power in every hour of a contract creates challenges for an electricity supplier like a wind farm. The owner of the project does not know, and has no control over, how much electricity it will produce in any given hour, and even though there are seasonal and diurnal averages, what actually happens in any hour is highly variable.
”Covariance risk” is the risk that a project will have a strong (typically negative) relationship between generation and price — so an hour of abnormally high generation will correspond to a low power price, and vice versa. While this condition can limit the value of power from a merchant wind farm relative to baseload energy, it is particularly challenging when the project has made fixed hourly delivery commitments (physically or financially) as the project not only misses out on revenues during high price hours, but is in fact a buyer of energy during those hours due to a need to purchase any shortfall between its hourly commitment and its hourly generation.
Chances are the reason the project came up short is a shortage of fuel: the wind died. And if the wind dies at a single project, it likely died at all of the neighboring projects — which means overall energy supply to the region has fallen, driving up energy prices — so the cost to cover the generation shortfall will
To put this condition in financial terms, the current cap on energy price in ERCOT during a supply scarcity event is $9,000. Suppose a large wind farm in ERCOT has committed to sell 50 megawatt hours of electricity during a certain hour for a fixed price of $20 a MWh. It is a hot day in August. The wind dies and power prices spike to $9,000 a MWh in that hour. The project is at risk of having to buy 50 MWh at $9,000 each just to sell them under the existing contract for $20 each — a net cost to the project of $449,000 for a single hour.
The insurance markets are typically better positioned to absorb that kind of risk than are independent generators because insurers have a much greater capacity to absorb and diversify that risk.
An insurer can hold wind risk in Texas, solar risk in Australia, hydro risk in Uruguay, and so on, with the idea that extreme weather patterns are unlikely to hit every area simultaneously. The ability to diversify the risk makes the insurer the party best able to manage location-specific, weather-driven risks.
Balance of Hedge
Renewable generation projects can manage covariance risk through a hedging product called a “balance of hedge.”
The balance-of-hedge product is designed for projects that will sign or have already signed a hedge with a bank or commodity trader to swap floating market prices for fixed prices on fixed volumes of power. It transfers the risk of being short during high prices and long during low prices. The insurer will assign an expected value to all of the residual excess short and long positions. Because of the enormous amount of potential volatility, the insurer will price the risk below the expected value so that it should make money for the insurer during an “average” weather year, but will eliminate the project’s exposure to extreme weather conditions.
For example, July 2018 was very hot in Texas. Power prices spiked during a period when wind speeds were low. Anyone with a bank hedge that month probably had a rough month. A balance of hedge smooths out the pattern of cash flow for a project with a fixed-quantity price hedge. The underlying hedge converts the floating revenue for a project selling its electricity into ERCOT into a fixed revenue stream, but if it is a fixed-volume hedge, it does not protect a project from coming up short on the fixed volume that the project has promised and having to cover the shortfall in floating revenue owed under the hedge. The balance of hedge covers this risk.
There may be only a limited appetite for a balance of hedge at the project level for an existing tax equity deal. The tax equity investor and lender have already underwritten the transaction based on their evaluations of the power contract and hedge. Most sponsors would do better to have the project company sign the balance-of-hedge contract with the insurance company when the tax equity is first put in place. Doing it later requires consent from the tax equity investor, who may be reluctant to reopen a transaction, especially as it may require a re-marking of the position.
From a credit perspective, a letter of credit is typically used as collateral for the balance of hedge. This is often posted at the sponsor level. However, if the sponsor lacks access to an LC facility or wants to offer a lien instead, then the lien must be harmonized with the lien-based collateral that has probably been provided to the bank that is the counterparty to the main bank hedge. Anyone entering into a bank hedge without putting the balance of hedge in place at the same time should negotiate for the ability to use incremental liens as collateral for the insurance company that is the counterparty under any balance of hedge put in place later.
REsurety is not an insurer. We support balance of hedge transactions by providing analyses to insurers who use those analyses to offer and set the price of balance-of-hedge products. While other insurers are working to enter the market, the vast majority of balance-of-hedge contracts — and other related products — have been offered through a partnership between Allianz Risk Transfer and Nephila Climate.
Assessing the Value
A white paper on our website called “The P99 Hedge that Wasn’t” looks at the hourly performance of every operating wind farm in Texas. We were able to use this data to analyze how a wind farm that purchased a bank hedge would have performed historically, including through the 2014 polar vortex, the 2011 heat wave and other major weather-driven events.
That said, a perfect view of the past cannot guarantee future performance. A good example occurred when coal plants dropped out of the ERCOT generation fleet in 2017 and the ERCOT reserve margin shrank, increasing the likelihood of high price events during low wind periods. Predicting how pricing will be affected in a market with less thermal generation and much more wind and solar is hard. You are predicting how various weather and commodity conditions will interact with a generation stack that has never existed before. Every month there are more wind farms in Texas than ever before. We spend a lot of time looking at how projects and markets performed over the last five or six years under high and low gas prices, high and low temperatures and high and low wind speeds, and analyzing how this is likely to change over time.
That depth of analysis is critical to insurers’ ability to underwrite balance-of-hedge and related products. Fundamentally, our job is to build a distribution of risk. We provide information around that distribution and identify sources of uncertainty and insurers like Allianz and Nephila use that information to offer and price products.
On average, the market has underestimated covariance risk in bank hedges and, in particular, in the Texas market. Utilities have taken this risk historically under long-term contracts where they commit to take whatever electricity is generated.
The covariance issue is weather-driven. High heat or extreme cold during low weather events is what causes significant changes in the as-generated versus fixed quantities of power. The year 2018 saw some unusually cold temperatures in January and some unusually high ones in July. This led to a significant amount of dislocation, and the market woke up to the exposures that projects are bearing.
Now we are in 2019, and we see pretty diverging views across the market about what happened last year and how it might change. If we re-live a 2011 heat wave with the current generation supply stack, nobody knows how that will play out, but it is clear that it would be a significantly bad event for almost any wind project with a standard fixed-quantity hedge.
Solar v. Wind
Solar developers should think about covariance risk the same way as wind developers.
A lot of solar is being built in Texas, in part because power prices are high in the summer when wind speeds are low, so there is an attractive pricing dynamic for a solar operator. At the same time, the whole solar industry is aware of what happened in California with the duck curve. More solar electricity is generated during the middle the day than the grid requires. If the grid sheds the excess electricity, it can depress power prices in the same way that happens during an especially windy hour in the winter in Texas.
The prevailing view currently is that the extremely rapid growth of wind in Texas compared to solar creates a great opportunity, but the solar industry in Texas has the potential to become a victim of its own success. The question is where is the equilibrium reached, and how big of a role storage will play.
The focus on Texas has been driven by the fact that most of the financial hedging for wind projects to date has been in ERCOT. However, interest in hedges is expanding into other regions like SPP and MISO where the same relationship can be seen between wind speeds and power prices. There is less wind in PJM, so there is less of the causal issue of high wind speeds pushing down power prices, but there is still the same general correlation of lower prices during high wind periods. The severity of the issue varies from one market to the next, but it affects every power plant whose output is intermittent.
The longest balance of hedge being offered today is 10 years. Pricing gets more expensive the longer the term in some markets, but not in all markets.
The concern about price spikes during low wind events in the summer is most acute for the next three to four years in Texas. That is partly due to a belief that solar capacity additions will help to moderate price spikes during the summer months when extreme covariance risk is most acute.
The issue of covariance is not unique to the seller of electricity. If a project enters into an as-generated power purchase agreement with a corporate buyer, it will have transferred the covariance risk away from the project and to the buyer of electricity.
Microsoft has been the most active in thinking about and managing this risk, and it was the first to embrace a solution through use of a “volume firming agreement.”
A “volume firming agreement” works in much of the same way as a balance of hedge: it locks in a fixed value to the sum of the hourly short and long positions held by a corporate buyer who is meeting a fixed-shape load with an as-generated PPA.
Suppose a data center requires 50 megawatts of power every hour to run its operations. If it has signed an as-generated PPA with a wind farm to manage the risks of its electricity costs, how well that PPA performs as a hedge on energy costs depends on the correlation between the wind project’s output and power prices. For example, if the wind dies and power prices spike and the data center still must buy 50 megawatts of power each hour, the data center is buying very expensive power despite the fact that it signed a PPA to mitigate energy price risk.
Microsoft decided it would like to shed that risk to an insurer in the same way that a project does.
Usually, the underlying PPA has already been signed and the volume firming agreement is added after the fact as a way to convert the PPA into something that is significantly more effective in managing the energy costs of a corporate buyer.
However, we are starting to see more corporate offtakers look at putting a volume firming agreement in place at the same time the PPA is signed. That gives them certainty about how their PPAs will perform as expected from the start.
In some cases, the project selling power under the PPA may or may not be aware of the volume firming agreement as the corporate buyer has a view that it should be free to manage its risk however it chooses without having to involve the project. In other cases, it becomes a three-party discussion among the project, the corporate buyer and the insurer. In either case, the PPA and the volume firming agreement are separate contracts.
Overall, projects and their investors should expect offtake arrangements to be much more dynamic in the future. Whereas traditional 20+-year busbar PPAs managed nearly all of a project’s risks for a long period, today offtake contracts are typically shorter term and have various flavors of risk management.