by Jason Kaminsky and Richard Matsui, with kWh Analytics in Oakland
Investors in rooftop solar companies and portfolios, and lenders to the sector, are using big data to draw useful insights and improve their valuation techniques, creating an opportunity for thoughtful developers to differentiate their operations by skillfully demonstrating transparency into the performance of their assets.
Such data is becoming an essential element of underwriting and will be critical to attracting the volume of capital needed to scale.
Solar is entering a new era in which operating portfolios are changing hands.
Projects are being refinanced after only a few years of operation, and current investors are seeing that their investments have liquidity in a secondary market, whether by banks, insurance companies or yield cos.
As tax equity vests through the five-year holding period and securitization becomes more prevalent, the market will see more opportunities for refinancing portfolios and for more liquid debt products.
Residential solar portfolios are a unique financial product that straddles consumer credit risk and project finance. On the one hand, there is an individual customer; on the other, there is an operating asset. This has major implications for the types of data available as a tool for risk management and underwriting. With a few years of operating history and thoughtful analysis, an investor is better able to evaluate the risk of an existing portfolio vis a vis a new portfolio.
Unique characteristics of the solar industry, including the potential for “underwater” leases or power purchase agreements, make this a particularly important assessment.
In the same way that a high debt-to-income ratio may predict an impending consumer default on a credit card, or an underwater home mortgage may predict a homeowner default on the mortgage, there are important leading indicators that an investor in solar assets can assess.
Figure 1 shows electricity production of two portfolios of rooftop solar systems. Most people agree that portfolio 2 in the probability distribution is a riskier portfolio; although both portfolios have the same mean performance ratio, the variance in portfolio 2 is higher.
Going into a deal, all investors must make assumptions about system production. However, with even a couple years of operating data, a colorful picture of the actual portfolio performance emerges. Investors are analyzing more than just the average performance ratio of the portfolio, since the average conceals problems hidden in the tails. They want to look at the distribution of performance and understand what quantity of homeowners have grossly underperforming or over-performing systems — and why — since those customers are more at risk of having a upside-down value proposition. Those customers may not be realizing the promised savings, or may resent purchasing electricity and then giving it back to their utility for free.
Our data shows that it is not uncommon for solar portfolios to have a double-digit percentage of systems with performance materially different than what was underwritten.
The production data can then be combined with other data sets to quantify the risk of “underwater” leases or power purchase agreements. By analyzing asset-level information, an investor can determine the probability that a homeowner’s solar contract is or will become an out-of-the-money contract.
This risk is particularly acute in California, where impending utility rate reform spurred by AB 327 is guaranteed to change the economics of solar for both new and existing solar customers. To appropriately assess the embedded risk of “underwater” contracts in the portfolio, it is critical to run scenario analyses of contract terms against future utility rates. These customers are more likely to be upset and seek a renegotiation of their contracts (or worse), jeopardizing expected cash flows and putting increased pressure on the servicer of the contract to manage these situations.
Figure 2 shows an overlay of sample PPA rates (in yellow) as compared the proposed decision by the California Public Utilities Commission on rates for Southern California Edison. In this example, a significant number of customer contracts will be underwater by 2019 if the proposed decision is ultimately approved, with the majority of the portfolio upside down by the end of the contract term if utility rates increase at a nominal 2% inflation rate. (Note: The proposals were still under review by the CPUC as of the time of this writing.)
Underwater contracts, contract transfers and operating performances all place more burden on the servicer, and there is historical data on servicing, too. An analysis of default rates and servicing issues is a preview of other hidden risks of a portfolio.
The servicer is the first line of defense in seeing how concerned customers are being managed and the quality of the customer’s experience. In mortgages, the value of a mortgage security is a function of not only the underlying assets, but also the quality of the servicer and its ability to deal with upset or delinquent customers. It is commonly accepted in the timeshare industry that the value of a timeshare cash flow stream is partly a function of the originator of the timeshare contract; similarly for solar, there will be variances in default rates depending on the choice of servicer and sales partner. (Was the customer oversold?). In other asset classes, investors rely on industry data to benchmark servicers and managers against one another, and we observe the solar industry starting to do the same. In fact, our data shows that FICO scores are only one component of default risk, and that detailed portfolio analysis can be used to better scope the financial risk in a deal.
The key to enabling these analyses is historical data, and the ability to look beyond a simple average to see the strengths and weaknesses of a diverse, distributed portfolio. Investors are increasingly building in-house capabilities or working with third party risk management firms to keep tabs on their portfolio performance. Not only is this data management prudent from an asset management and portfolio surveillance perspective, but they also know that this data will increasingly be required for the secondary market. The ratings agencies have continued to push for greater data disclosures about the historical performance of distributed solar portfolios, which has resulted in delayed or cancelled transactions for would-be issuers who are unprepared for the sudden need for quality data management.
Institutional investors have lots of options for their investments. In the consumer fixed income market, they can buy into investment products backed by residential mortgages, autos, student loans, credit cards, among others — and now solar.
Solar investments are long-dated deals. Most consumer products that are securitized are short-term consumer agreements; for example, the average term for auto deals is 5.5 years and for credit card receivables is usually eight to 10 months. The average term of a solar loan/PPA/lease, by contrast, is 15 to 30 years, which is really long for an unsecured loan and makes some investors uncomfortable. The head of securitized products at Janus Capital Group recently shared with Bloomberg that Janus is unsure of the solar asset class due to the long weighted-average life of the investment. The notable exception is mortgages, which also have a 20 to 30 year term and lead to a robust set of loan-level data sharing requirements.
A solar panel is a great product because its value proposition to the homeowner is clear: as long as the solar system is generating electricity, it is worth at least as much as the electricity being delivered, less the cost of upkeep. It provides a cash flow benefit to the homeowner every single month, in contrast to an auto that may sit in the garage unused. A home with a purchased solar system on the roof is worth more than one without solar.
However, there is an open question regarding the value of a system that has to be removed from the roof. With hardware costs reaching record lows while the costs of customer acquisition, construction labor, permitting and other expenses remain relatively high, most industry observers agree that the salvage value of the underlying collateral is minimal and that, therefore, the most economical solution in the event of a default is a negotiated settlement with the homeowner.
The value on the home is, therefore, a function of system production (kWh) and utility rates ($/kWh). This adds a heavy dose of project finance to the deal: bond repayment requires the system to work, is dependent on utility rates and servicing costs, and has latent warranty risk if manufacturers go bankrupt. The data about these risks are essential to underwriting.
There are no other consumer securitizations that rely on the functioning of an asset for repayment. While there is a lemon law for autos and people do expect their houses to provide shelter, repayment is not directly tied to the miles per gallon of the car or the amount of rain the house shields.
Additionally, the construct of the solar agreement is such that the contract usually stays with the home. People go through divorces, deaths, work transitions, or any of a number of other life events where they need to leave their homes.
We have seen portfolios with double-digit percentages of reassignments in fewer than five years.
A few interesting things may happen when a customer wants to leave his or her home before the end of the contract term. First, the customer will scrutinize the current contract and may choose to renegotiate the contract, particularly if the customer is in an underwater contract. Second, the customer may prepay to buy the system, or pay to move it to a new home. Alternatively, the contract could transfer to the new homeowner, in which case the credit characteristics of the portfolio have just changed.
Although the solar issuer has the authority to reject transfers for poor credits, we see that in practice this is damaging to its reputation and that credit migration does occur over time.
Two, five or 10 years after a deal initially closed, the characteristics of the borrowers in the portfolio may be dramatically different than at inception.
In just a few years, the solar rooftop industry has grown from using “kilowatts” to “megawatts” to “gigawatts” as the unit of measure to quantify growth. Owing to the tremendous volume of projects already deployed, the industry now has valuable information about important risk factors, such as system performance and customer payment performance. Investors are recognizing the importance of mining this data before making investments.