U.S.-based foundations control assets of approximately $900 billion and give $60 billion annually in support of social causes. This support is pivotal to the success of so many worthy efforts. Yet we find ourselves in an investment environment most believe is overvalued, thus making future returns challenging. How can foundations not only continue their level of support but possibly even grow it? One word: guarantees.

The balance sheet is one of the most underutilized assets foundations have to support their missions. Why can a balance sheet be so powerful? Because most foundations, if rated by a credit rating agency, would receive the highest possible rating of AAA. This is a unique advantage; there are only two U.S. corporations still rated AAA. Foundations can be considered AAA because most have unencumbered endowments and very few liabilities beyond their required 5 percent annual spend. This often leaves 95 percent of the endowment available to support a guarantee. If foundations used just 3 percent of the $900 billion of assets in their collective endowments for guarantees, the amount of support could grow by $27 billion. This could provide a real boost in a time when increased philanthropic support is truly needed.

Since Kresge began issuing guarantees in 2011, the foundation has made $70 million in guarantee commitments through 23 transactions. We have seen our guarantees effectively encourage investment from those shying away from known or perceived risks, help launch new organizations or business models, and support investments where repayment is based on the proof of positive social outcomes. Our typical transaction duration is 10 years. On average, our guarantees have been outstanding for five years, and we haven’t paid out anything on these commitments to date. We have successfully catalyzed impact on the ground while unlocking $780 million of additional investor capital without spending a dollar from our endowment.

But few things in life are free, so how should we think about the “cost” of a guarantee? Even the best-intentioned and structured guarantee has the potential to be called on for payment. So how do we model this when there is no obvious or easy comparable? Often guarantees that catalyze investment in low-income communities have unique considerations and loss patterns that don’t follow traditional markets. But by using an understood index in the capital markets, we can at least look at a familiar point of comparison. The best place to start is the corporate high-yield market, a proxy to measure “risky” borrowers. Typical issuers of high-yield debt tend to have limited operating history, be heavily indebted or experiencing financial distress. The highest-rated corporate high-yield bond is BB. The chart below shows the historical annual and cumulative default rates for a BB bond.

The probability of default on these bonds is just above 10 percent.  The default rate in the first year is relatively high and each subsequent year the probability drops and remains low. ¹

Borrower default rates ultimately determine the cost of any guarantee program (we assume no recovery). Let’s assume we make a 10-year, $10 million guarantee, and we have various levels of defaults. Below is a table with the expected loss based on various annual default rates.

The calculations for this table are straightforward, since we assume we lose a set amount each year for a fixed period (annual default rate x 10 years). Fortunately, we won’t experience this loss on day one so, while we are waiting for the possibility for our guarantee to be called, it remains invested in our endowment, earning a return of 7.5 percent. This is the time value of money. We could perfectly calculate our expected loss if we knew exactly when the guarantee would be called. Unfortunately, we don’t. Again, we turn to the high-yield market for help. The market has robust historical data on both the timing and the size of expected losses (like the BB bond default chart above). We can use the probability curves of similarly rated bonds to our borrowers’ creditworthiness as a proxy to estimate the cost of each guarantee. While this is the most accurate method, a shortcut, without having to model every bond curve, is to use a fixed annual default rate. Below is a table that models various annual default rates and the cost in today’s dollars of our same hypothetical $10 million guarantee.

The cost in today’s dollars, assuming a 1 percent annual default rate for our $10 million hypothetical guarantee, is $686,408. We think of this cost as the “grant” we could give today to provide the same social impact as the guarantee we are providing. This can be a useful comparison as we think about the social impact expected from each form of capital.

The cumulative loss of our hypothetical guarantee is 7 percent, slightly below the 10 percent loss rate for a high-yield BB bond. We’ve included higher annual loss rates in the table to give an idea of even “riskier” guarantees. These costs range up to 27 percent for a portfolio that has a 4 percent annual default rate, which means 40 percent of the borrower’s default. This rate of default seems extreme but helps frame the upper end of the costs for a very risky guarantee program.

In 2015, Kresge committed $150 million in guarantees to support the work of our programs. We believed these guarantees could attract another $750 million of capital into work to expand opportunity in America’s cities. We assumed default rates in the range of 7 percent to 10 percent for our borrower group. Using the concept above, we estimated the value of our expected losses for the program would be $11 million to $15 million. So, the question we asked ourselves was, “Would we spend $15 million today for the chance to leverage $750 million of capital to the causes we care about?” That’s a 50x multiplier! The answer was a resounding, “YES!” So far, Kresge’s $150 million commitment has exceeded expectations; with $70 million invested so far, it has attracted $780 million of private capital, again, without requiring any payouts.

I’d encourage other foundations to expand their toolkit to include guarantees. We have shown that they catalyze other capital, the cost can be estimated, and the market is receptive. To help you take your first step, our Social Investment Practice is working on a multi-funder guarantee bank to lower the operational burdens to get you started. We’ll share more on that soon, and I hope you join us.

Rob Manilla serves as The Kresge Foundation’s chief investment officer. Follow his team on Twitter @kresgeinvests.

^{1 Because of the nature of the guarantees we tend to make, we would not expect to see such significant losses in year one. In many cases, we benefit from reserves in the transactions we guarantee, sit behind the lender who must take the first loss or have negotiated guarantee payment only upon the wind-down of a fund. Each of these structures works to mitigate our early exposure.}