Do riders with volatility controls cost VA investors potential gains?
This article originally appeared in the June/July 2013 issue of MorningstarAdvisor magazine. To subscribe, please call 1-800-384-4000.
Insurers are increasingly limiting investment options in their variable annuities that have riders to strategies with built-in volatility management. Volatility overlays mitigate the risks for insurance companies by embedding some hedging within the investment strategy. But are investors who buy the riders giving up a portion of potential gains in return? We investigate this question by comparing two contracts with similar benefits; one with a volatility control overlay and one without an overlay.
Most asset-allocation offerings use the percentage of stocks to serve as a proxy for risk. These are commonly billed as “target-risk funds.” A “conservative” target-risk fund typically holds between 20% and 50% of its assets in stocks; moderate, moderate growth, and growth options are analogously defined as portfolios with increasing amounts of equity exposure. This approach to investment- risk classification is reasonable because a portfolio gets more volatile as the percentage of stocks it holds increases. But risk here is understood in relative terms. Where this classification system falls short is in ignoring the absolute level of volatility. Exhibit 1 charts the rolling 30-day standard deviation of a “moderate” target-risk portfolio. It shows that the realized volatility was far from “moderate” during the difficult markets of 2008–2009, when standard deviation reached a high of 40%, a level of volatility that would have been unexpected even from an aggressive portfolio.
That experience was an unpleasant surprise to most investors, and it drove many investment firms to explore approaches that might provide a much smoother investor experience. Insurance companies had even more reason to explore alternative approaches to managing their underlying subaccounts: As portfolio volatility increased, so did the cost of hedging guarantees accompanying the investments. The insurers had effectively sold options and needed ways of managing the associated risk. The higher the unpredictability (or standard deviation) of the underlying investment, the higher the risk of an option sold on it.
The equity weight within a two-asset portfolio that meets a volatility target is the ratio of the volatility target to the forecasted volatility of the risky asset:
where WE is the weight in equities, is the target volatility, and is the forecasted equity volatility. Correlation between the two assets and volatility of the second asset is assumed to be zero. According to this equation, to maintain constant portfolio volatility level, the weight in equities should decrease when forecasted volatility goes up and increase when forecasted volatility goes down.
A variety of sophisticated techniques forecast volatility, but simpler approaches rely on either the Chicago Board of Options Exchange Market Volatility Index—otherwise known as the VIX, a measure of the market’s expectation of stock market volatility over the next 30 days— or recent realized volatility. Because insurance companies mark their exposures to market, their risk is mostly related to market-implied volatility. As a result, many of the volatility overlay approaches in practice use marketimplied volatility as a signal. Although VIX is the most well-known of all implied volatility measures, it may not be the most appropriate option for insurance companies because it is a short-term measure. A better measure that matches the duration of insurance companies’ liabilities is the Deutsche Bank Equity Volatility Index–US 12-Month, a proxy for stock-market implied volatility for the coming year.
Using the equation above and the Deutche Bank volatility index as our forecast of volatility, the equity weights for a 10% target volatility portfolio commencing on Dec. 6, 1999 (the volatility index’s inception date), are charted in Exhibit 2. When volatility goes up, the equity target goes down (and vice versa).
Target Volatility and Guarantees in Practice
Variable annuities come with a variety of optional features that an investor can buy for an additional annual fee. To illustrate the effect of a managed volatility strategy, we focus on just the Guaranteed Minimum Withdrawal Benefit. With a GMWB, the seller of the benefit may guarantee that the benefit base grows by at least a minimum percentage every year, irrespective of the performance of the underlying subaccount. For example, on each anniversary of the contract, the benefit base would increase by the greater of the annual return on the subaccount, or 5%. In addition, the riders may allow for a periodic step-up in the benefit base. For example, an annual step-up might reset the benefit base to the account value if the account value were greater than the benefit base on the anniversary date. This benefit base could then be used to calculate the withdrawal amounts at the time the investor decides to annuitize. Therefore, the higher the benefit base, the better it is for the investor.
For our analysis, we compared two portfolios with different GMWB riders:
1. 5% annual credit with a fee of 150 basis points and annual step-up with the underlying portfolio invested in a static mix of 50% stocks and 50% cash.
2. 5% annual credit with a fee of 150 basis points and annual step-up with the underlying portfolio managed to a 10% volatility target. The equity allocation here changes based on our equation by moving assets in and out of cash.
A forward-looking Monte Carlo analysis of these products usually involves generating thousands of market paths using future expected returns and covariance of the asset classes that are the building blocks of the underlying portfolios. However, we wanted to capture the relationship between implied volatility and subsequent market returns, so we relied on historical daily data for this analysis. We used the S&P 500 Index as a proxy for stocks and the three-month Treasury bill as a proxy for cash.
Our approach was to track the benefit bases and values for accounts that start on each business day since the inception of the Deutsche Bank volatility index. In other words, we had a contract for every day since the index’s inception through the end of 2011. We tracked the last contract entered on the last business day of 2011 until the end of 2012, which meant we were using daily market data between the years 2000 and 2012. For each start date, we then compared the growth of each portfolio while accounting for the annual credits and step-ups. We assumed that there were no withdrawals before the end of the analysis period (December 2012).
Exhibit 3 repeats our rolling 30-day volatility chart, but this time with the rolling standard deviation of the portfolio including the volatility overlay. As we expected, the volatility of the portfolio with the overlay is much more muted.
For the period since inception of the volatility index through December 2011, or 3,019 business days—with each business day representing an investor that entered the contract on that day—the following table compares the success rate of the managed volatility overlay versus no overlay.
These results show that, using historical volatilities and returns, the volatility overlay results in a better deal for 35% of GMWB contracts, an equally good deal for 34% of the investors, and a bad deal for 31% of the investors. The reason 34% of the cases end up being the same is because the annual credits dominate the actual returns on the portfolio. In other words, the annual credit of 5% was better than the annual returns of both the portfolios. However, if we split the entire time period into two periods—pre-2009 and 2009 and later—an interesting pattern emerges.
For contracts entered before 2009, the volatility overlay trails only 14% of the time. For contracts entered in and after 2009, it trails 81% of the time. Not surprisingly, the GMWB with the volatility overlay is most effective when the investor goes through a market crash like that of 2008. It is not that effective in a rising equity market like the period between 2009 and 2011.
We must emphasize that these results are specific to the assumptions, models, and indexes we used. It is possible to improve upon the volatility forecasting approach, which would make a more compelling case for the volatility overlay.
It might seem that making managed volatility an investment approach is an overreaction by the makers of variable-annuity products. In practice, however, the pricing on the guarantees with traditional asset-allocation portfolios did not account for severe market conditions such as those in 2008. Insurers realized that providing compelling guarantees in this low-interest-rate environment while charging similar fees requires them to change the way underlying investments are managed, as well as to enhance their hedging programs. Managed volatility is one way insurers can control their costs, because just when the cost of hedging is high (which is directly related to market volatility), they have to engage in less hedging because of the relatively small allocation to risky assets.
As a further illustration, imagine you are short a put option and you experience profits and losses on a daily basis based on the changes in price of the put option. As the volatility of the underlying security increases, the price of the put option increases. Because you are short the option, you have to register a loss, and as volatility fluctuates, the volatility of your profits and losses also goes up. However, if the volatility of the underlying holding is kept constant by changing your position size, you experience less volatility in your profits and losses.
Market volatility is only one of the parameters that goes into the valuation of guarantees. As insurers try to reduce uncertainties around their profits and losses, we expect more changes to their offerings. Some of the changes that are already in practice include using passive products. This eliminates the guesswork of understanding the composition of the underlying subaccount when it uses actively management funds. Another practice is using additional overlays like option-based strategies that reduce gap risk—the risk that a quick drop in equities leaves the insurance firm with a huge mismatch between the underlying subaccount and liabilities.
We hope that the industry’s cost reductions make these volatility overlays even better deals for investors who want a smoother ride during the next market crash. Improved investment and hedging approaches would seal the deal for investors.