A reinsurance strategy, like all risky investments, does have the potential to incur significant losses. But, when those losses occur, insurers and reinsurers raise premiums to restore their capital. Thus, the highest expected return to insurers and reinsurers tends to be after catastrophic losses. Patience and disciplined rebalancing are keys to successfully harvesting the reinsurance premium over time.
The bottom line is that reinsurance risks help diversify the risks of a traditional stock and fixed income portfolio while also offering potentially higher returns. In addition, reinsurance offers the potential for equity-like returns with less volatility (about half) and less downside risk than equities. Furthermore, the returns to reinsurance are uncorrelated to the returns of all the other assets in a typical portfolio. We believe the most effective method for gaining exposure to the reinsurance premium is to invest in a fund that partners with large reinsurance firms through the purchase of quota shares, eliminating the market beta risk investors would incur if they instead owned reinsurance company stocks. Indeed, market beta risk would account for roughly 80 percent of the total risk of a reinsurance company’s stock. In other words, we want to isolate and hold only the risk specific to reinsurance. We also want to minimize the concentration risk associated with catastrophe bonds (which tend to have the vast majority of their risk profile connected to losses from hurricanes in the United States), another element of investing in reinsurance.
Quota Shares
In a quota share arrangement, a fund receives a specified percentage of the premium of a defined book of reinsurance business (say, 5 percent of a reinsurer’s global natural catastrophe business) and also pays the same percentage of the losses (5 percent in our example). Quota shares, which transfer risk from the reinsurance company to the fund, are typically one-year agreements. Thus, just as was the case with alternative lending, fund families generally use closed-end interval funds because they avoid the daily liquidity requirements of open-ended mutual funds.
Reinsurance companies incur high costs in originating business (underwriters, risk models, office overhead, etc.). However, the marginal cost of origination is less than the average cost. The investing fund typically agrees to share in the expenses at a rate attractive to both sides. Additionally, the ceding company usually agrees to cap the interval fund’s maximum loss at the amount of collateral posted. The interval fund puts up the required capital as collateral to demonstrate that it has the capacity to pay claims.
The extra capacity gained by ceding premiums to the interval fund allows reinsurers to underwrite more protection for their insurance company clients without straining their capital. In return, investors get access to an aligned, diversified pool of risk that has equity-like expected returns. Quota shares permit access to broader classes of reinsurance risk and offer an opportunity for higher returns (but with less liquidity) than catastrophe bonds, to which the fund also typically will have some limited exposure. The added diversification benefit is one reason we prefer funds that use quota share arrangements to access the reinsurance premium.
Catastrophe Bonds
Catastrophe (cat) bonds are risk-linked, high-yielding, non-investment-grade debt instruments that transfer a specified set of risks from the issuer to investors. Most typically, they are linked to losses from hurricanes in the United States. As of 2017, the cat bond market had grown to about $25 billion since the first issuance in 1994. Cat bonds are mainly used to provide liquidity for the reinsurance fund. Of course, these bonds are inherently risky and usually have maturities of less than three years. If no catastrophe occurs, the insurance company pays a coupon to investors. However, if a catastrophe does occur, the principal would be forgiven and the insurance company would use this money to pay their claimholders. The natural attraction of cat bonds is that there is no reason to believe events such as natural disasters are correlated with the risks of other financial assets.
Thus, cat bonds offer a source of potential return truly unrelated to other major asset classes (drops in the S&P 500 Index or actions by the Federal Reserve don’t cause earthquakes and hurricanes). And, in general, the reverse is also true. As a result, they provide diversification benefits.
Steven Clark, Mike Dickson Jr. and Faith Roberts Neale contribute to the literature on cat bonds with their July 2016 study “Portfolio Diversification Effects of Catastrophe Bonds,” which covers the 14-year period from 2002 through 2015. The authors observe that from January 2006 to December 2014, the Eurekahedge ILS Advisers Index showed negative returns in only seven out of 108 months. The largest negative monthly return, -3.94 percent, was in March 2011, the month the Tohoku earthquake occurred. The second largest monthly loss was just -0.74 percent, in September 2008, followed by the third largest loss, -0.57 percent, in October 2008.
To test their diversification benefit hypothesis, Clark, Dickson and Neale analyzed cat bonds in several different ways. The following is a summary of their findings:
Because their correlations with other financial assets are low, strong support exists for the notion that cat bonds provide substantial diversification benefits when they are added to an investment opportunity set already consisting of traditional asset classes (U.S. equities, international equities, bonds, commodities and real estate).
Cat bonds demonstrate significant portfolio diversification benefits in various out-of-sample analyses, including naïve 1/N strategies (where N is the number of factors, asset classes or investments being considered) and when using a mean-variance model, a minimum-variance model and a volatility-timing model.
The payoffs from cat bonds cannot be replicated in the mean-variance space by portfolios holding only other asset classes.
The standard deviation of cat bonds is much smaller than the standard deviations shown by other indices, contributing to a significantly greater Sharpe ratio (which assumes normal distributions). However, the cat bond index does display extreme kurtosis and has the longest left tail (you would expect large losses when low-probability, but high-risk, events occur).
Cat bonds added substantial diversification benefits during the 2008 global financial crisis, reducing drawdown measures and conditional value-at-risk in times of market distress.
Cat bonds’ diversification benefits were confirmed by Peter Carayannopoulos and M. Fabricio Perez, authors of the 2015 study “Diversification through Catastrophe Bonds: Lessons from the Subprime Financial Crisis.” However, they also found that cat bonds were not zero-market-beta assets during the 2008 financial crisis. In fact, the dynamic correlation coefficients of cat bonds with the market and the corresponding hedge ratios were statistically and significantly positive during the crisis. They believe that the rising correlation between cat bonds and the market was caused by the “structure of CAT bond trust accounts and the composition of the assets used as collateral in the trust account.”
The authors go on to add: “Assets used as collateral in these trust accounts proved to be of lesser than expected quality and, furthermore, counterparties in swap agreements, put in place in an effort to immunise collateral asset returns from market fluctuations, were exposed to considerable credit risk or even defaulted during the crisis.”
The good news is that Carayannopoulos and Perez found the effects of the financial crisis on cat bonds had disappeared by 2011, as their correlation with the market returned to its statistically insignificant, pre-crisis levels. They noted: “These results may imply that the new and improved collateral structures created for CAT bonds issued after 2009 have been perceived as effective by market participants. These new structures attempt to enhance the credit quality of the collateral asset and include limits to the type of assets permitted in the collateral account, and constant monitoring and reporting of the collateral account balance.”
Despite their lack of an investment-grade rating, cat bonds typically do not entail credit risk because they are fully collateralized with proceeds, generally in U.S. Treasury money market products. They also are generally
floating-rate notes, with minimal duration risk. In terms of performance, while 2018’s bond yields tend to be lower than in the not-so-distant past, for the period from January 2002 through June 2016, the Swiss Re CAT Bond Index returned 8.0 percent versus 7.4 percent for high-yield bonds and 6.4 percent for the S&P 500 Index.
While past performance is no guarantee of future results, during this time period cat bonds not only produced better returns than high-yield bonds or the S&P 500 Index, but they did so with the lowest volatility (2.4 percent versus 6.6 percent for high-yield bonds and 17.3 percent for the S&P 500 Index). They also experienced a smaller maximum drawdown, just 4.3 percent versus 26.5 percent for high-yield bonds and 55.3 percent for the S&P 500 Index.
In short, cat bonds provide liquidity and diversification for a well-diversified interval fund whose risks consist mostly of quota shares.
Risk and Expected Return
As we have discussed, expected return is a statistical estimation for the compound return of an asset class over the long term. It is important to understand that expected returns are the mean of a very wide potential distribution of possible returns. Thus, they are not a guarantee of future results. Expected returns are forward-looking forecasts, and are subject to numerous assumptions, risks and uncertainties, which change over time. Actual results may differ materially from those anticipated in an expected return forecast.
Again with these caveats in mind, our expected return estimate for a well-diversified, quota shares-based fund is similar to the expected return for U.S. equities. Our estimate for a best-case year (meaning losses from catastrophic events are less than expected) is a return of about 11 percent to 12 percent. Investors should also expect that, in about 10 percent of years, losses will be about 10 percent. In a year like 2005, when three major hurricanes, including Hurricane Katrina, hit the United States, losses in such a fund would likely be in the 15 percent to 20 percent range. 2017, which also had three major hurricanes as well as two major wildfires in California, saw losses in the low double digits. Reinsurance losses larger than any ever experienced historically might cause a fund to lose somewhere in the range of 30 percent to 40 percent. Keep in mind that when losses occur, premiums are raised to restore capital and incorporate the new risk information.
Asset Location
Because a reinsurance strategy is tax inefficient (the income from it is ordinary), investors should strongly prefer to hold this investment in tax-advantaged accounts, unless they are in a low tax bracket. In certain situations, it may be appropriate for high-tax-bracket investors to use a reinsurance strategy in taxable accounts, particularly if it is being employed instead of municipal bonds.
Again, because interval funds do not provide daily liquidity, they cannot be held inside a 401(k) plan unless the participant has created a self-directed brokerage account.
Accessing Reinsurance Risk
There are currently two options investors can utilize to access reinsurance risk: the Stone Ridge Reinsurance Risk Premium Interval Fund (SRRIX) and the Pioneer ILS Interval Fund (XILSX). Our preferred vehicle, the one we recommend for our clients, is SRRIX.
The next three chapters, which cover the variance risk premium, the alternative style premium and time-series momentum, examine more complex strategies that, to many, may not be as intuitive as alternative lending and reinsurance strategies. We have done our best to provide clear explanations.
Chapter 7:
The Variance Risk Premium
Economists have long puzzled over the simultaneous demand from consumers for risk-reducing insurance and risk-increasing lottery tickets. Every year, people spend trillions of dollars on the two combined. These behaviors may seem unrelated, but they actually are symmetrical forms of risk transfer. The insurance policy is a means of risk transfer in which a buyer pays to eliminate the possibility of an extreme, rare, downside event—such as premature death, destruction from an earthquake or hurricane, or an equity market crash—while the lottery ticket is a means of risk transfer in which a buyer pays to create the possibility of an extreme, rare, upside event.
Again, the price of such a risk transfer decomposes into two parts: an expected payout and a risk premium to compensate the seller for the uncertain nature of any payout, which may be sudden and dramatic. Financial markets are full of strategies that resemble insurance or lotteries. For example, in options markets, the premium to compensate the seller for the uncertain nature of the payout is called the variance (or volatility) risk premium (VRP). This risk premium is rational for both the buyer and the seller. The buyer willingly pays it to create, or to eliminate, uncertainty. The seller charges it for taking the risk. The evidence indicates that the more remote the risk, the higher the ratio of risk premium to expected payout. Over a large sample size, an expected fair value is set by the probabilities of outcomes, and options sellers charge a premium to that fair value in compensation for providing a risk-transfer service.
The VRP refers to the fact that, over time, the option-implied volatility has tended to exceed the realized volatility of the same underlying asset. This has created a profit opportunity for volatility sellers—those willing to write volatility insurance options, collect the premiums and bear the risk that realized volatility will increase by more than implied volatility. Investors are willing to pay a premium because risky assets, such as stocks, tend to perform poorly when volatility increases. In other words, markets tend to crash down, not up. Thus, the VRP isn’t an anomaly we should expect to be arbitraged away. Because the VRP’s risks (specifically, the sale of options performs poorly) tend to show up in bad times (when risky assets perform poorly), we should expect a significant premium. Another way to think about this is that investors pay to hedge catastrophic outcomes; they want to transfer the risk of a terrible outcome, like their house burning down or the price of oil going to $200 per barrel. Thus, they knowingly and willingly pay above fair value to eliminate that risk. As a result, the VRP should be considered a unique risk premium that individuals with long investment horizons and stable finances can harvest because they have the ability to accept cyclical risks that show up in bad times.
Historical Evidence
Data going back as far as 1873 shows the existence of a VRP. In the paper “Option Markets and Implied Volatility: Past Versus Present,” published in the November 2009 issue of The Journal of Financial Economics, Scott Mixon presented the results from his analysis of data hand-collected from newspapers published between 1873 and 1875. Mixon calculated implied and subsequently realized volatility, and found that options prices reflected a persistent, large and positive spread of 11.8 percentage points between implied and realized volatility for the most liquid options. That is a huge risk premium. Today, markets are more liquid, more transparent and more efficient. Thus, we should not expect to see such a large premium.
Graham Rennison and Niels Pedersen, authors of the September 2012 paper “The Volatility Risk Premium,” studied the period from June 1994 through June 2012 and 14 volatility markets, including markets associated with stocks, bonds, commodities and currencies. They found strong evidence of a variance risk premium, with its magnitude ranging from 0.9 percent in currencies to 2.2 percent in equity indices, 2.9 percent in 10-year interest rate swaptions and 4.4 percent in commodities futures. They concluded that the risk-return trade-off in volatility strategies compares favorably to traditional stock and bond strategies, and that these strategies exhibit low correlation to equities. They also found that Sharpe ratios ranged from 0.7 for currencies to 1.2 for U.S. interest rates and commodity futures. These compare favorably to the historical 0.4 Sharpe ratio for the market beta premium. Thus, the authors recommended that investors consider an allocation to VRP strategies.
Using equity index options, Roni Israelov, Lars Nielsen and Daniel Villalon, authors of the study “Embracing Downside Risk,” which appears in the Winter 2017 issue of The Journal of Alternative Investments, showed that the vast majority of the equity risk premium derives
from accepting downside risk versus seeking participation in the upside. They found that, over the period from 1986 through 2014, greater than 80 percent of the equity risk premium was explained by the willingness to accept downside risk. They also discovered that the ex-post VRP was positive 88 percent of the time and averaged 3.4 percent per year.
Israelov, Nielsen and Villalon found a similar result when they examined the upside and downside risk premium in Treasury bonds, with 62 percent of the excess return coming from the downside. The lower percentage they found in bonds should not be surprising because Treasuries have less downside risk than stocks. When they examined gold, the authors found 100 percent of the excess return was from downside risk. They also found strong results when they looked at credit default swaps, with the return for accepting downside risk accounting for more than 100 percent of the excess return.
In addition, Stone Ridge examined the VRP for the 10 largest stocks over the period from 1996 through 2012, breaking down that period into three sub-periods. The firm’s researchers found a persistent and stable premium. From 1996 through 1999, the VRP was 4.3 percent. From 2000 through 2009, the premium was 3.9 percent. And from 2010 through 2012, it was 4.1 percent. Stone Ridge also found strikingly similar patterns in implied volatility curves around the world. In international markets, as in the United States, more short-dated and more out-of-the-money options have higher expected VRP returns in both single-stock and index options.
The VRP has been well documented, and is best known in U.S. equities. For example, the implied volatility of S&P 500 Index options exceeded the realized index volatility 85 percent of the time from January 1990 to September 2014. What’s more, options historically have traded at about 4.4 percentage points above subsequent realized volatility. It is important to understand that this should not be interpreted to mean the options market tends to overestimate future volatility. Instead, the more likely explanation is that options prices incorporate a risk, or insurance, premium. Most investors are risk-averse, and so they are willing to pay a premium to hedge downside risk. Buying volatility insurance options provides that hedge or insurance. The large premium also exists because of an imbalance in supply and demand. This imbalance arises because there are likely far more natural buyers of volatility insurance options than there are sellers.
Reducing the Risk of Black Swans Page 7