Executive Summary

• This paper provides an objective framework for helping financial advisors assess the merits of variable annuities with living benefit riders. The first part introduces advisors to the current state of academic research on the subject. The second part provides a context for evaluating a variable annuity with a popular guaranteed lifetime withdrawal benefit (GLWB) rider if such a product had been available for purchase before two historical “worst case” market scenarios.
• The academic community is increasingly accepting of the concept of living benefit riders as a tool for helping retirees insure portfolios against sequence-of-returns risk and longevity risk.
• The irrevocability and low internal rates of return of immediate annuities have likely been barriers to their acceptance. Greater flexibility, higher potential returns, and revocability may make variable annuities with GLWB riders a superior alternative to immediate annuity contracts.
• Back-testing of a hypothetical variable annuity with a 5 percent GLWB rider indicates that the 1973–1974 bear market was not long enough or severe enough to force the insurer to meet its guaranteed income obligations through reserves. But the 2000–2002 bear market data suggest that insurance carriers do indeed take on a degree of investment risk in return for the rider premiums they collect.
• The high expenses of variable annuities significantly moderate sub-account performance relative to comparable lower-cost mutual funds, and may lead some readers to conclude that mutual funds are a superior planning option. Alternatively, other advisors may find value in the security they provide during bear market nadirs. This peace of mind may help investors avoid panic selling, remain appropriately weighted in equities, and remain disciplined in their allocation decisions.

John H. Robinson is managing director of a Honolulu, Hawaii-based independent wealth management firm.

Variable annuities are controversial products in the investment world. They have long been pilloried for their high internal expenses, opaque and excessive commissions, onerous surrender charges, and inadequate disclosure. In addition, scores of investor complaints and lawsuits alleging unsuitability and misrepresentation, particularly toward seniors, have made variable annuities a hot-button issue with regulators. The relatively recent introduction of so-called “living benefit” riders, however, has added an intriguing new dimension to the debate, and has dramatically changed the variable annuity marketplace. These riders, which have proliferated in myriad varieties, typically enable contract holders to invest for capital appreciation through equity-based sub-accounts or pre-constructed asset allocation models, while offering a guarantee of income even if the underlying investments perform poorly.
 
The popular appeal of living benefit guarantees is undeniable. Since the riders were introduced in the early part of this decade, sales of variable annuities have soared. Total assets in variable annuities now add up to well over $1 trillion, with guaranteed withdrawal benefit riders selected on approximately 70 percent of all new variable annuity contracts (National Underwriter 2005). While some longstanding media critics of variable annuities proclaim that the new features are merely superfluous, overly complex bells and whistles that only add fees to an already expensive and gimmicky product (Katt 2006, Quinn 2006, Orman 2006), the appeal of investing for growth with an income safety net has converted some former outspoken critics of variable annuities to advocates (Clements 2005, Cruz 2005a and 2005b, Milevsky 2007).
 
Academic research will be central to establishing (or debunking) the legitimacy of these products. Given their novelty and the speed with which living benefit riders have evolved, there is, to date, a relative paucity of published research on the subject. But the impact these benefits are having on the investment industry and, in particular, on the allocation tendencies of baby boomers, has attracted the attention of academia. At present, scores of working papers are wending their way toward journal publication.
 
It’s not surprising that product complexity, conflicting media perceptions, and sales rhetoric from the insurance companies also have led to confusion and division in the financial advisor community over the merits of variable annuities with living benefit riders (Opiela 2007). This paper provides an objective framework for helping financial advisors evaluate contemporary variable annuity products. The first part of the framework reviews the major lines of academic research and gives the reader a substantial reading list for more in-depth study. The second part adds to the existing body of research by examining how one popular contemporary living benefit rider would have performed under historical “worst case” scenarios. Specifically, the paper seeks to determine two key issues:

1. Whether the cost of a guaranteed lifetime withdrawal benefit rider might have been justified in previous down markets—that is, whether there has ever been a market scenario in which insurers might have had to pay claims from reserves in the past.
2. Whether such a rider would have provided any real or psychological benefit relative to simply investing in lower-cost equity mutual funds.

Literature Review

A survey of the existing body of academic literature finds two separate lines of research pertaining to variable annuities with living benefit riders. The first line has evolved from research on retirement income sustainability. Numerous studies have found that (1) 30-year inflation-adjusted sustainable withdrawal rates have historically varied widely and (2) achieving sustainability for withdrawal rates much above 3–4 percent depends not just on one’s asset allocation, but also on whether one happens to retire at a good or bad time in the investment markets (Bengen 2001; Ameriks, Veres, and Warshawsky 2001; and Chen and Milevsky 2003).
 
Specifically, research has established that one of the biggest threats to the financial security of retirees is the chance that their portfolios will suffer losses early in retirement after they have begun income withdrawals. This threat has been coined “sequence of returns risk” (Milevsky and Kyrychenko 2007). In addition to this risk, the tendency of many retirees to require initial withdrawal rates of greater than 3–4 percent, combined with increasing life expectancies, has led to concern over “longevity risk.” The notion that sound asset allocation alone, based upon the principles of modern portfolio theory, cannot guarantee lifetime income sustainability suggests a need for products and strategies that can ameliorate these two retirement-specific risks.
 
The potential for annuities to increase the likelihood of income sustainability has been recognized for some time, and has even led many researchers to proclaim that retirees should be using annuities to a far greater extent than they traditionally have (Ameriks, Veres, and Warshawsky 2001; Chen and Milevsky 2003; Babbel and Merrill 2006; Hu and Scott 2007). Much of the early research focused on immediate annuities as a solution. But the problem, as noted in several papers, is that retirees’ concerns about sequence-of-returns risk and longevity risk were outweighed by certain structural disadvantages of immediate annuities. Specifically, irrevocability, illiquidity, low internal rates of return, and loss of the asset upon death collectively explain the reluctance of retirees to turn to annuities as a solution to the problem of lifetime income sustainability.
 
The dynamics of the value proposition of annuitization changed with the advent of living benefit riders attached to variable annuity contracts. In contrast to immediate annuities, living benefit riders—through delayed annuitization—allow investors to retain control of the asset, terminate the contract, and potentially pass on a remaining account balance to heirs. Thus, a recent focus of academic research has been on determining whether variable annuities are, in fact, a superior alternative to traditional immediate annuities, and whether retirees are rational in their widespread adoption of variable annuities with living benefit riders. A review of this research suggests that there is growing academic support for the legitimacy of variable annuities with living benefit riders (Babbel and Merrill 2006; Milevsky and Kyrychenko 2007; Hu and Scott 2007; Scott, Watson, and Hu 2007).
 
A second line of research has focused on whether variable annuity products are fairly priced for consumers. To do this, some researchers have sought to examine how the cost of variable annuity riders and other internal expenses compares with the insurance companies’ actual hedging cost. Interestingly, two separate studies that modeled risk in different ways each concluded that the insurance cost of a simple 7 percent return of principal guaranteed withdrawal benefit rider was .73–1.6 percent versus a typical actual rider cost of .3–.5 percent (Milevsky and Salsbury 2006, and Dai, Kwok, and Zong 2007). While this research suggests that the cost of living benefit riders in variable annuity contracts may be justified or even underpriced, as Milevsky and Salsbury freely concede in their paper, research in this area is limited by the fact that the real actuarial data and pricing methodologies are proprietary to the insurance companies.
 
In fact, other academics have expressed skepticism over the validity of these models. Hardy (2005) notes that modeling of hedging costs versus the actual price of living benefit riders is extraordinarily complex and suggests that published modeling attempts to date fall short because (1) there is little good data on policyholder behavior, (2) there is no published data on actual lapse rates, and (3) the products are new and evolving. Other researchers have approached the pricing issue from a different perspective. In addressing consumer concerns over variable annuity expenses, Mitchell, Poterba, Warshawsky, and Brown (1999), and Babbel and Merrill (2006) examined annuitization pricing on immediate annuities and found that insurance company mark-ups relative to the mortality tables of those who actually bought annuities were low and that such mark-ups have been declining over the past decade.
 
Ironically, the suggestion that variable annuity contracts may be underpriced has led some observers to question the insurance companies’ claims-paying ability (Powell 2007). Babel and Merrill note that investor perception of even a remote possibility of insolvency has been shown to deter investors from considering insurance products. But the authors also note that the odds are extremely slim that investors would be harmed by insurer defaults resulting from underpricing of riders. If an insurer were to become insolvent, the contract holder would become a creditor of the insurance company and would receive distributions as the failed company’s assets are liquidated by the receiver. In the event that there are insufficient assets available to cover the remaining payments to the contract holder, the National Organization of Life and Health Guaranty Associations (NOHLGA) would work with state guaranty associations to provide continued support and protection. Since 1988, NOLGHA has stepped in to successfully facilitate and continue annuity coverage for 87 failed insurance companies, including many of the dramatic insurance company failures of the early 1990s (Babbel and Merrill 2006).
 
In summary, the body of published academic research pertaining to living benefit riders and variable annuities is expanding rapidly, and much of this research seems to suggest that there is increasing acceptance of the product as a potential tool for helping retirees achieve lifetime income sustainability. Readers are encouraged to review the papers referenced in this article to formulate their own opinions.

Back Testing Under Historical ‘Worst-Case’ Market Scenarios

While a review of the academic research provides a useful background for understanding the role that variable annuities with living benefit riders may play in helping retirees attain income sustainability, it provides little guidance to the financial advisor about how the products will function in practice and in what contexts they may be most suitable. As a starting point, one useful way to examine whether the cost of a variable annuity with a living benefit rider is justified is to examine how the contract would have performed during stressful market periods in the past. This paper examines the value investors might have received from a variable annuity with a popular living benefit rider if this contract feature had been available to investors before the two most stressful market environments of the past half century: the 1973–1974 and 2000–2002 bear markets.
 
One obvious challenge in conducting this back-testing exercise is that most variable annuity sub-account choices do not have sufficient track records. This hurdle is crossed by using the S&P 500 Index as a proxy for the annuity sub-accounts, with the performance reduced each year by the expenses of the annuity. This approach seems reasonable, since (1) many variable annuity contracts offer S&P 500 sub-accounts or ETFs as investment options and (2) critics of variable annuities often argue that investors’ interests would be better served with low-cost index funds than by investing in mutual funds within an annuity contract. Examining how investors might have fared with an index fund in a variable annuity contract, relative to investors who bought the same index fund on its own, should provide good apples-to-apples fodder for the debate.
 
The rider that has been chosen for this analysis is a 5 percent guaranteed lifetime withdrawal benefit (GLWB) rider that is available in various iterations in several top-selling variable annuity contracts. Under this rider, the insurance company typically guarantees the investor a lifetime (non-annuitized) systematic withdrawal rate of up to 5 percent of the initial investment value, even if the contract’s investment value has been exhausted due to a combination of withdrawals and poor investment performance (so long as withdrawals have not exceeded 5 percent in any given year).
 
In addition, insurers offering this rider usually give investors the ability to step up the lifetime guaranteed withdrawal amount to 5 percent of the value on future contract anniversary dates, assuming the value on the anniversary dates is higher than the initial investment amount. For example, if an investor deposited $100,000 into an annuity contract and, 10 years later, the highest contract anniversary value is $200,000, the investor’s maximum guaranteed lifetime withdrawal rate would be 5 percent of $200,000, or $10,000 per year. In this analysis, the total expenses of the variable annuity, including the rider cost, have been set at 2.05 percent a year, which seems to be representative of the insurance cost of a moderately priced GLWB variable annuity contract.¹ 
 
To make the output from this exercise as realistic and meaningful as possible, two different scenarios are explored for each bear market period. The first scenario considers a hypothetical 55-year-old investor who makes an initial investment of $100,000 ten years before retirement. The second scenario considers a hypothetical 65-year-old investor who invests his $1 million retirement nest egg and requires annual distributions immediately. In both scenarios it is assumed that the investors retire in the year before the bear market and that they receive their first annual withdrawal distributions after the first year of the market decline (that is, the worst possible time to retire and begin withdrawing from one’s portfolio).²

Bear Market Back-Testing Results

Tables 1 and 2 present the investment results for the two aforementioned investors who retired immediately before the 1973–1974 bear market. Tables 3 and 4 present the results for the investors who had the misfortune of retiring immediately before the 2000–2002 bear market. Examining this data leads to a number of interesting observations. First, as presented in Tables 1 and 2, the bear market environment of 1973–1974 was not severe enough to lead to depletion of either the mutual funds or the annuity contracts. Indeed, the historical record shows that the stock market eventually rebounded and both the fund and annuity values would have grown significantly over time, even after withdrawals.

Given both the severity and rarity of events such as the 1973–1974 market crisis, one might be tempted to conclude that the annuity expense is unnecessary since, to date, it appears that there has never been a bear market scenario in which an insurance company would have had to pay claims from its reserves under a GLWB rider. But Tables 3 and 4 suggest that such a conclusion might be premature. Although there is obviously no way to predict whether the stock market will rebound as well over the next few decades as it did following the 1973–1974 crisis, both hypothetical investors in all scenarios (index fund and annuity) would have portfolio values through 2006 far below what they had upon retirement at the end of 1999. In fact, as presented, in the purple highlighted row, a down year of 20 percent or more in 2007 would likely doom both the index fund and the variable annuity portfolios to premature depletion.
 
A second important observation concerns the difference in wealth accumulation between the index fund on its own and the index fund in the variable annuity through both bear markets. At first blush, Tables 1 and 2 appear to offer textbook examples of the drag expenses have on performance over time and of the importance of keeping investment expenses low. In Table 1, the index fund investor would have withdrawn $438,668 from his retirement portfolio from 1973 to 2006 and the index fund would have grown from $100,000 in 1963 to $3,513,241 through 2006. The variable annuity over this time would have produced just $379,950 of income and would have grown to just $1,366,618. Similarly, in Table 2, the index fund value would have grown from $1 million at the start of 1973 to more than $13 million through 2006. In contrast, the annuity contract would have grown to just over $5 million over this period.

But a problem with using this data to proclaim the superiority of the index fund over the annuity is that this conclusion is based entirely on hindsight and fails to acknowledge the fact that future returns at any particular point in time are entirely unknown. The significance of this concept can be demonstrated by examining the “low water marks” for each portfolio (highlighted in red in each table). While it is true that the index fund value is higher than the variable annuity value for each corresponding year, at the low points in both bear markets both the index fund and the variable annuity portfolios in all four tables were down 45–60 percent a mere two to three years into retirement. At that point, neither the index fund nor the variable annuity investors would have been able to predict if or when the stock market would recover. Consequently, it seems reasonable to surmise that index fund investors would have been far more nervous about the possibility of depleting their retirement savings early into retirement than the variable annuity investors, who would at least be able to take comfort in knowing that their initial 5 percent income withdrawal streams were guaranteed for life.
 
A third observation concerns the comparison of the results for the investors who used the index fund and annuities for ten years of accumulation before retirement. As presented in Tables 1 and 2, the annuity investors had significantly smaller retirement nest eggs upon retirement than the index fund investors. Given the 2.05 percent expense differential and ten years of compounding, such differences are not particularly surprising. Wide differences would, of course, be expected for every ten-year sequence of returns, regardless of whether it was a good, average, or below-average accumulation period. Even greater differences in wealth accumulation would be expected for accumulation periods longer than ten years.
 
From these rather simple observations about the power of compounding, it seems reasonable to conclude that investors whose primary goal is accumulating wealth for retirement would be better served by keeping expenses low, and that the decision to buy a variable annuity with a GLWB rider should be made only when the investor is retired and ready to begin withdrawals (namely, when the investor’s goal has shifted from wealth accumulation to retirement income sustainability).
 
Yet it should be noted that the GLWB riders in a number of popular variable annuity contracts offer investors certain guarantees during the accumulation period that are designed to offset the compounding disadvantage from higher expenses. Common guarantees include marking the accumulation benefit for lifetime withdrawals to the highest anniversary date during the accumulation period or guaranteeing that the accumulation benefit will grow at no worse than a fixed rate (typically 5 to 6 percent) for a certain number of years (usually ten).
 
The merits of guarantees during the accumulation period are not particularly obvious for the 1963–1972 or 1990–1999 illustrated periods because in both cases the ten-year period before the bear markets was strong and the high water marks (1972 and 1999) for both illustrations happened to be the years immediately before the start of the bear markets. The value of accumulation benefit guarantees might be more obvious if the accumulation periods extended to the low point of the bear markets. Under these scenarios, the 5 percent guaranteed lifetime withdrawal amount (based off the 1972 and 1999 high water marks) would likely be higher with the variable annuity than 5 percent of the market value of the index fund at the 1974 and 2002 market lows. Further study is needed to determine whether accumulation benefit features offer enough potential benefit to investors to offset the considerable compounding disadvantage caused by higher expenses.

Implications of This Analysis

One of the primary goals of this exercise has been to determine whether the cost of variable annuity contracts with living benefit riders is justified. Although back testing of the 1973–1974 bear market suggested that there has yet to be a scenario in which insurance companies would have to pay claims, analysis of the 2000–2002 bear market suggests that premature depletion scenarios are certainly within the realm of possibility, and that insurers do have legitimate risk exposure on the living benefit rider premiums they collect.³ Thus, although sequence-of-returns risk appears quite low, for those retirees who have the misfortune of retiring just before a prolonged bear market it is possible or perhaps even likely that GLWB riders will provide tangible value relative to a lower-cost mutual fund alternative. Assuming that these riders are priced fairly and competitively, if one accepts the general risk management principle that the most important risks to insure are those with high loss potential and low probabilities of occurrences, then the application of living benefit riders as “portfolio insurance” seems plausible. 
 
In addition to assessing the legitimacy of the insurance risk, a second goal of this paper has been to discern any real or perceived benefits investors might gain from variable annuity contracts with GLWB riders. The data in the tables clearly show that investors who buy variable annuity contracts sacrifice significant long-term performance than if they invested in lower-cost mutual funds, due to the impact of expenses over time. But as noted in the “low water mark” analysis, variable annuity investors might benefit from greater psychological comfort and peace of mind during severe market downturns than mutual fund investors who do not have a lifetime income guarantee.
 
Although the value of such benefits is obviously ephemeral, in taking this presumption one step further one might also surmise that the comfort afforded by the GLWB rider may make it more likely that annuity investors will remain invested in equities, whereas mutual fund investors may be more inclined to sell in panic. Anecdotal support for this assertion can be found in a 2002 AARP survey titled “Impact of Stock Market Decline on 50- to 70-Year-Old Investors.” The survey found that 36 percent of investors age 50–75 lost 25 percent or more of their retirement savings during the 2000–2002 bear market. The survey also found that 29 percent of retirees had returned to work as a result of investment losses and that a full 73 percent indicated that the market decline had forced them to modify their lifestyles or financial expectations.
 
Although there appears to be little published data specific to mutual fund flow patterns of retirees during this period, the fact that flows in general into equity funds peaked in 1999 (the market top) and turned negative in 2002 (the market nadir) is well documented (Investment Company Institute 2002). The propensity of investors to make bad timing decisions (buy high, sell low) is also well documented in behavioral economics (Goetzmann, Massa, and Rouwenhorst 2007; Friesen and Sapp 2007; Benartzi and Thaler 2007).⁴ Because living benefit features came into existence after the 2000–2002 bear market, it is impossible to determine whether variable annuity investors would, in fact, have been more inclined to ride out the storm in equities. But annuity asset allocation data presented by Milevsky and Kyrychenko (2007) demonstrates conclusively that the existence of a living benefit rider makes retirees much more comfortable investing in equities than retirees who buy annuities without one. Given that retirement income sustainability research has generally concluded that retirees seeking to maximize withdrawal rates should have as much as 75 percent or more of their portfolios invested in equities (Spitzer and Singh 2007), variable annuities may serve a valuable purpose in helping investors resist the temptation to become overweighted in stable fixed income investments that might increase their long-term probabilities of portfolio depletion.
 
Conclusion 

The goal of this paper has been to provide a framework that will help financial advisors look beyond biased marketing materials and negatively stilted media portrayals in order to come to their own conclusions regarding the merits of variable annuity contracts with living benefit riders.
 
The first part of this framework involved bringing readers up to speed on the current state of academic research. The primary take-away from this is that there seems to be growing academic consensus that variable annuities may be an effective tool for helping retirees protect against sequence-of-returns risk and longevity risk.
 
The second part of the framework was intended to provide advisors with a simple, practical way to evaluate living benefit riders by examining how annuity investors might have fared relative to mutual fund investors through the two worst bear markets in the modern history of the stock market. A primary finding is that, while the conditions required to cause insurance companies to pay claims under lifetime income benefit riders are rare, carriers do appear to carry real claims risk (that is, the riders do not appear to be a pure profit gimmick for the insurers). Further, while the probability of failure appears low, the effect of premature depletion of one’s retirement savings can be catastrophic, suggesting that retirees may be rational in electing to use living benefit riders as “portfolio insurance.” Placing variable annuity contracts with GLWB riders in this historical context provides a basis for advisors to judge for themselves whether the risk of portfolio failure is sufficient to warrant the lower compounding rate that results from the annuity’s higher expenses.
 
In terms of broad suitability standards, variable annuity contracts with living benefit riders seem best suited for investors who are retired and are ready to shift from wealth accumulation to the withdrawal/decumulation phase of their financial lives. The value that these contracts provide to investors who are five or more years away from retirement is less clear. This analysis has shown that the expense burden of variable annuities will likely be a significant drag on performance for investors who are still in the wealth accumulation phase. Further study is needed to determine whether the high water mark or step-up accumulation features included in many contemporary living benefit riders provide enough potential value to make up for the compounding lag caused by higher expenses. As a starting point to this investigation, the data provided by Milevsky and Kyrychenko suggest living benefit guarantees may have some value during the accumulation phase by helping investors feel more comfortable allocating a larger percentage of their portfolios to equities.
 
In closing, it is worth noting that the growing academic acceptance of the product comes at a time when the suitability of variable annuities for seniors is increasingly being challenged by state and federal regulatory authorities. In fact, virtually every regulatory authority, including the Financial Industry Regulatory Authority, the Securities and Exchange Commission, state securities administrators, and state insurance commissioners, has posted information on their Web sites cautioning investors about investing in variable annuities. Yet the dichotomy between the academic community and the regulators’ perceptions of variable annuities may be less striking than it appears. A closer examination of the regulatory position suggests that many of the suitability concerns arise not from the structure of the annuity contracts, but from the manner in which they are sold to the public. According to the North American Securities Administrators Association, seniors account for 44 percent of all investor complaints received by state securities regulators, and as much as 60–65 percent of individual state securities administrators’ case loads involve complaints of fraud or misrepresentation of annuity products (including both variable and equity index annuities). As referenced earlier, the high incidence of fraud and misrepresentation in variable annuity sales can be attributed to (1) the complexity of the product, (2) the fact that fees and commissions are largely opaque to investors, and (3) the fact that many annuities pay commissions that are considerably higher than the fees or commissions on other investment products. Given the potential of variable annuity contracts with living benefit riders to address the single biggest concern of most retirees—lifetime income sustainability—the importance of improving disclosure, eliminating conflicts of interest, and banishing unethical sales practices cannot be understated.

Endnotes

1. Of the 2.05 percent total expenses, .40 percent represents the GLWB rider cost, and 1.65 percent represents “mortality and expense” charges. It is from the mortality and expense charges that revenue is paid to the insurance company and to the selling agent.
2. To avoid complicating the issue with a discussion of the tax implications of investing in variable annuities versus mutual funds, it is assumed that the variable annuities and the traditional S&P 500 Index fund are owned in individual retirement accounts.
3. Although beyond the scope of this paper, future analysis using bootstrapping or Monte Carlo simulations might be useful to further quantify the probability of portfolio failure. It is worth noting that some academic research on this subject has already begun. One study (Milevsky and Salsbury 2006) estimated the likelihood of portfolio failure for a slightly different form of living benefit rider to be 11.7 percent under normal market volatility assumptions and as high as 39.9 percent using more extreme volatility assumptions. 
4. As a matter of interest, a review of behavioral economics literature finds little support for the notion that investors who employ a financial advisor perform better than investors who self-direct. 

References

AARP. “Impact of Stock Market Decline on 50–70 Year Old Investors.” December 2002.

Ameriks, J., R. Veres, and M .J. Warshawsky. “Making Retirement Income Last a Lifetime.” Journal of Financial Planning December 2001.

Babbel, D. and C. Merrill. “Rational Decumulation.” Wharton Financial Institutions Center Working Paper No. 06-14, July 2006.

Bengen, W. “Conserving Client Portfolios During Retirement.” Journal of Financial Planning May 2001.

Bernartzi, S. and R. Thaler. “Heuristics and Biases in Retirement Savings Behavior.” Journal of Economic Perspectives February 2008.

Cary, Rick, “Variable Annuity Sales Hit Record $128.4 Billion in 2004,” National Underwriter—Life & Health March 7, 2005. 

Chen, P. and M. Milevsky. “Merging Asset Allocation and Longevity Insurance: An Optimal Perspective on Payout Annuities.” Journal of Financial Planning June 2003.

Clements, J. “For a Conservative Investment, Variable Annuities Are Too Costly.” Wall Street Journal C1, January 21, 2004.

Clements, J. “An Annuity That’s Worth a Second Look: Retirement Security—But at a Price.” Wall Street Journal C1, September 28, 2005.

Cruz, H. “New Benefit Could Be the Next Big Thing in Annuities.” Los Angeles Times June 22, 2005a.

Cruz, H. “Is a Variable Annuity Lifetime-Income Rider Right for You?” Los Angeles Times October 2, 2005b.

Dai, M., Y. K. Kwok, and J. Zong. “Guaranteed Minimum Withdrawal Benefit in Variable Annuities.” Working paper to appear in Mathematical Finance 2007.

Friesen, G. and T. Sapp. “Mutual Fund Flows and Investor Returns: An Empirical Examination of Fund Investor Timing Ability.” Journal of Banking and Finance September 2007.

Goetzmann, W., M. Massa, and K. G. Rouwenhorst. “Behavioral Factors in Mutual Fund Flows.” Yale School of Management working paper 2007.

Hardy, M. “Not-So-Easy Rider: The Guaranteed Minimum Withdrawal Benefit.” Financial Engineering News 2005.

Hu, W. and J. Scott. “Behavioral Obstacles to the Annuity Market.” Pension Research Council working paper 2007.

Katt, P. “The Good, Bad, and Ugly of Annuities.” Journal of Financial Planning November 2006.

Milevsky, M. and V. Kyrychenko. “Asset Allocation Within Variable Annuities: The Impact of Guarantees.” Pension Research Council working paper 2007.

Milevsky, M. and T. Salisbury. “Financial Valuation of Guaranteed Minimum Withdrawal Benefits.” Insurance: Mathematics & Economics 38 (2006).

Milevsky, M. “Confessions of a VA Critic.” Research January 2007.

Mitchell, O., J. Poterba, M. Warshawsky, and J. Brown. “New Evidence on the Money’s Worth of Individual Annuities.” American Economic Review 89, 5 (December 1999).

Opiela, N. “Variable Annuities: Emerging from the Dark Side?” Journal of Financial Planning March 2007.

Orman, S. “Here’s One Investment You Should Avoid.” The Costco Connection September 2006.

Powell, R. “Bargain-Buyers: Beware—New Research Suggests Variable Annuities May Be Dangerously Underpriced.” Marketwatch.com January 4, 2007.

Quinn, J. “A Guarantee? It’ll Cost You.” Newsweek May 8, 2006.

Scott, J., J. Watson, and W. Hu. “Efficient Annuitization: Optimal Strategies for Hedging Mortality Risk.” Pension Research Council working paper 2007.

Spitzer, J. and S. Singh. “Is Rebalancing a Portfolio During Retirement Necessary?” Journal of Financial Planning, June 2007.