What do retirement investors really want? Quantifying income preferences and trade-offs

Retirement investors’ preferences extend well beyond the basic metrics—expected returns and expected volatility—that often drive the design of investment products. Industry practitioners lack (1) a shared language to articulate those preferences and (2) methods for identifying the trade-offs necessary to meet desired retirement outcomes. T. Rowe Price’s five-dimensional (5D) framework seeks to fill those gaps.¹

Developed in 2024, our framework enables us to evaluate and quantify both the supply and demand sides of the retirement income landscape. We believe our 5D framework can help the industry better understand retiree needs, provide solutions more aligned with those needs, and, ultimately, help individuals feel more certain in their retirement.

This paper focuses on the demand side of the framework, which can be instructive when designing and offering income solutions to meet the diverse preferences of retirement investors. Using our 2024 research as a guide, we explore key findings and insights on the relative importance that retirement investors assign to each of the framework’s five attributes and the associated trade-offs between them.²

Research highlights

Among other key findings, our 2024 research indicated that retirement investors:

1. Are willing to give up a meaningful percentage of their retirement income to secure a guaranteed stream of income for life.
2. May require less liquidity than anticipated—i.e., they most value access to a portion of their savings.
3. May be less risk averse than is often assumed.
4. Can tolerate some volatility in the recurring income their investments can generate.

We believe these findings collectively make a strong case for retirement income solutions that help investors draw down their savings in a relatively consistent and predictable manner, along with a guarantee component to help protect against longevity risk.

See content below the “Exploring the relative importance of attributes” section for detailed research findings and insights.

5D framework overview

Before diving further into our research, it’s helpful to review the fundamentals of the 5D framework.

The framework centers around five mutually exclusive attributes that, in our view, fully capture an individual’s income experience in retirement (Fig. 1). A defining characteristic of the 5D framework is that a performance gain on any one of the five attributes requires a compromise on at least one of the remaining four. This dynamic speaks to the crucial reality of investing—trade-offs are required.

As an example, Fig. 2 illustrates a preference profile for a hypothetical retirement investor who is primarily concerned about longevity risk and wants guaranteed lifetime income. This investor prefers low payment volatility (i.e., a stable income stream) and a higher payment level in retirement, perhaps to compensate for past undersaving. Given these priorities, the investor is willing to forego some liquidity and accept a moderate level of balance depletion risk.

With the trade-off dynamic established, the next question is: What are the “exchange rates” for trade-offs?

There are two ways those rates can be determined:

• Supply side: letting the financial market efficient frontier tell us what rate the market can provide. This is rational and objective.
• Demand side: letting retirees tell us what rates they are willing to accept. This is subjective, and behavioral aspects play a major role.

When the two sides are brought together, a potentially desirable retirement income solution can be created. 

Exploring the relative importance of attributes

Our 2024 research allowed us to generate relative importance scores for each of the five attributes—offering a quantitative method for understanding what retirement investors value and, crucially, how they make trade-off decisions. The following data yield salient insights for defined contribution plan sponsors exploring the possibility of retirement income solutions.

As shown in Fig. 3, survey respondents assigned the greatest relative importance to longevity risk hedge, followed by unexpected balance depletion. Together, the two attributes summed to greater than half of the total score—an unsurprising result to us, as survey respondents in the qualitative component of our research identified “not running out of money before I die” as a top concern (Fig. 4).

“...many investors value solutions that strike a balance between income and liquidity.”

Notably, survey respondents assigned equal importance to payment level and liquidity. This indicates that while the industry is focused on solving the retirement income puzzle, many investors value solutions that strike a balance between income and liquidity.

Payment volatility was assigned the least relative importance, at only 9%. One possible explanation for the low score is that the respondents, being U.S.-based, may have assumed that Social Security benefits would be foundational to their total retirement income and were therefore willing to tolerate some fluctuation in their investment income. That explanation appears consistent with the range of volatility (+/- 24%) we explored in the survey; Social Security benefits compose roughly half of the typical U.S. retiree’s total income,³ and spending fluctuates meaningfully in retirement.⁴ In other words, a 24% change in investment income is likely to preserve enough desired stability for retirees at a total income level.

From a plan sponsor perspective, the relative importance scores suggest a need for retirement income solutions that help investors draw down their savings in a relatively consistent and predictable manner, paired with a guarantee component to help protect against longevity risk.

Trade-offs between attributes

Further analysis allowed us to explore how individuals make trade-off decisions between attributes. The relative importance scores revealed that longevity risk hedge was the most important attribute for our survey population in aggregate, but this did not reveal how much of a longevity risk hedge individuals believed they needed or what value they placed on it.

Absent a cost consideration, the typical answer to this type of question is “the more, the better.” By putting cost into the equation, our method revealed how much survey respondents were willing to “pay”—i.e., give up in the form of retirement income—to get additional longevity risk protection at different protection levels. Consequently, we could determine the true desire for longevity risk protection. 

For simplicity’s sake, we used payment level as the cost, or reference point, which enabled us to easily calculate trade-off rates between attributes.

Longevity risk hedge

Beginning with the longevity risk hedge attribute, we were able to quantify how individuals assigned value to various portfolio horizons. Using payment level as the reference point allowed us to estimate the “price tag” that individuals were willing to pay to achieve a desired level for the longevity risk hedge—e.g., extending their portfolio horizon to 10, 15, 20 years, etc.

• To make our research findings more tangible, we cite the example of a hypothetical retirement investor who has saved enough to retire at age 65 and intends to use those savings to pay themselves $100,000 each year in retirement until age 75.⁵
• Each of the price tags shown in Fig. 5 represent how much annual income our hypothetical individual would be willing to give up to extend their portfolio horizon by various increments beyond age 75. The dollar amounts shown in the price tags are cumulative, moving from left to right (Fig. 5).⁶

As a point of reference, the Social Security Administration’s life expectancy calculator⁷ indicates that a woman currently age 65 (i.e., born in 1960) can expect to live until age 87 and therefore will need her retirement savings to sustain nearly 22 years in retirement. This is notably short of the 35 years of investment income that our hypothetical retiree has “paid for” (see the pink price tag in Fig. 5), but it suggests that individuals nonetheless value the longevity risk hedge and are willing to pay for it.

Once we estimated how much annual income our survey respondents would be willing to sacrifice in exchange for the potential to receive a regular paycheck until age 100, we sought to understand how much more they would be willing to pay if that income stream were guaranteed for life.

Key finding

Our research suggests that these investors would be willing to reduce their annual incomes by an additional 5.9% in exchange for guaranteed lifetime income. This means that the hypothetical investor in Fig. 5, who already has stated a preference for receiving an annual income of nearly $91,200 until age 100, would be willing to further reduce that annual paycheck to approximately $85,800 if the amount were guaranteed for life.

There are two possible explanations for that 5.9% price tag.

• One could argue that since most Americans won’t live until age 100, a 35-year portfolio horizon is long enough to fully hedge longevity risk. Thus, our hypothetical retiree would be giving up an additional 5.9% of annual income simply for the word “guaranteed.”
  • This speaks to the behavioral aspect of designing retirement income solutions. While sacrificing an additional 5.9% of annual income may not be optimal from a quantitative perspective, it could still be worthwhile to the investor, who might feel an added sense of stability and peace of mind about the risk of outliving their savings.
• Alternatively, we could assume that the guarantee represents additional years of income being purchased, in which case we want to determine how many years that 5.9% price tag represents.

We’ve already established that our hypothetical retiree is willing to reduce their annual income by 4.5% to extend their expected portfolio horizon from age 85 to age 100—equivalent to 0.3% of annual income for every additional year of income. If we extend that per-year price beyond age 100, the 5.9% cost of the guarantee buys roughly 20 more years of income. This could imply that our hypothetical retiree has a subjective perception that longevity risk protection until age 120 is necessary and is rationally willing to obtain that additional protection by purchasing a lifetime guarantee. 

Unexpected balance depletion

Unexpected balance depletion, the attribute with the second-highest relative importance score in our survey, speaks to a retirement investor’s tolerance for market risk. Again, the concept of what could be considered rational versus irrational is relevant here, as retirement income preferences are not purely objective.

Continuing to use payment level as our reference point, we were able to estimate how much a retirement investor would be willing to give up to achieve varying levels of market risk in their retirement portfolios (Fig. 6).

Key finding: Similar to the example in Fig. 5, in which our hypothetical retiree was highly motivated to purchase a guarantee, we found that retirement investors in our survey were most motivated to reduce the odds of premature balance depletion to 1 in 300. This is a notable finding, as it provides a clear risk budget for designing retirement income solutions.

In the savings or accumulation phase of retirement investing, there are established norms—such as the traditional balanced portfolio (60% equity/40% fixed income) or a glide path—that serve as touchstones that solutions providers can modify to meet the risk preferences of different investor cohorts. However, in the spending or decumulation phase, there is no generally accepted starting place that can be adjusted for retirees’ unique preferences and risk tolerances.

One reason for this missing touchstone is the fact that the appropriate level of portfolio risk depends not only on the retiree’s tolerance for risk, but also on the desired level of future income. Based on different balance depletion probabilities, our study tells us that the additional benefit from reducing portfolio depletion risk to less than 1 in 300 is not considered highly valuable by retirement investors. Hence, the 1 in 300 number could serve as a baseline for part of the risk profile for retirement income solutions in the mass market.

Balance liquidity

Balance liquidity and payment level were assigned equal relative importance by the retirement investors we surveyed, although both were valued below the longevity risk hedge and unexpected balance depletion attributes.

Consistent with its perceived lower level of importance, we saw smaller price tags across all the levels tested for the balance liquidity attribute. In this regard, an important finding from our research is the far left-hand price tag in Fig. 7. This shows that the retirement investors in our survey were willing to sacrifice the greatest percentage of retirement income to ensure that at least a quarter of their retirement savings were liquid and accessible.

In this instance, our hypothetical individual was willing to pay 3.4% of their annual income to be able to access a quarter of their total asset balance whenever they wanted. We then observed a diminishing willingness to pay across the next two balance quartiles.

• The retirement investors in our survey were only willing to pay 1.4% of annual income for an additional quartile (from 25% to 50%) of balance liquidity, a huge step down from the 3.4% they were willing to pay for the first quartile.
• That number fell even further—to a minimal 0.5% of annual income—for a third quartile (50% to 75%) of balance liquidity. This declining willingness to pay implied that these investors saw a diminishing benefit in adding liquidity, if we assume their decisions were rational.
• Interestingly, the retirement investors we surveyed showed extra motivation to have 100% balance liquidity, as suggested by the higher price tag (1.3% of income) for that last quartile of liquidity. 

We can glean three key takeaways from this analysis

1. The value assigned to different levels of balance liquidity is not linear.
2. Retirement investors may need less liquidity than we expected.
3. The combined results for the balance liquidity and longevity risk hedge attributes make the case for partial annuitization.

Payment volatility

Finally, as mentioned earlier, payment volatility was assigned the lowest relative importance among the five attributes in our survey, as demonstrated by the comparatively small dollar amounts in the price tags in Fig. 8. We interpret these results as underscoring the conclusion that retirement investors are willing to tolerate some volatility in their annual income because they have the flexibility to adjust their spending, with the support of some stable income from Social Security.

Concluding thoughts

With accumulation-oriented investment products, quantifying trade-offs seems less important. Typically, the investor determines the portfolio’s risk level through some basic risk profiling methods, and the market decides the reward for the risk taken.

However, retirement income preferences are complicated and highly individualistic. What one investor values may be completely different from another. The designers of retirement income products must therefore consider how different risks are prioritized, how rewards are allocated, and the trade-offs that result from that prioritization and allocation. In this process, quantifying the values that retirement investors place on competing objectives is a critical first step

Through our comprehensive survey, we have added empirical depth to our understanding of the preferences of retirement investors and how they may vary across investor populations. These findings offer important insights for plan sponsors as they strive to meet the diverse needs of plan participants. And they enable us to further refine the design of our retirement solutions, giving investors more support in achieving the retirement they want.

 

Appendix: Study methodologies

The study used conjoint analysis, a statistical technique designed to understand how individuals value different attributes of a product or service. By presenting respondents with a set of specific, hypothetical attributes, conjoint analysis helps identify the trade-offs people are willing to make. This method reveals the relative importance of each attribute in decision-making and is often used to inform product design.

Basic assumptions of conjoint analysis:

• A product or service is a collection of attributes
• A product’s utility is a function of the utilities of the attributes
• Utility is a predictor of behavior, and estimating utilities involves conducting a linear regression.

Additionally, the qualitative component of the study featured a best-worst scaling survey methodology known as MaxDiff analysis. This survey technique is used to determine preferences or priorities among a set of items. It asks respondents to select the most and least important or appealing items from a list, providing a clear ranking of options.

Key evaluation metrics

Relative importance scores: the proportional impact that each attribute had on a respondent’s choices. The importance score is a relative measurement, so the sum of the impacts from all five attributes is normalized to 100% and the results are expressed as percentages.

The dollar amounts shown in Figs. 5–8 were calculated using estimates from a probabilistic model fitted to the conjoint analysis. In the model, payment levels were characterized as percentages. However, we express the payment levels in dollar terms by applying the percentages to a base annual income of $100,000.

IMPORTANT: Estimates are subject to numerous limitations, are based on assumptions, are hypothetical in nature, and do not represent actual investment results. Altering assumptions used could significantly alter the results shown, and results could vary over time and with each use of the model. Actual results and outcomes may differ significantly and outcomes are not assured. Costs and other fees associated with an actual investment are not considered, which could alter conclusions made by readers when taken into account. Not representative of an actual investor’s situation or product. Not to be construed as a recommendation to buy or sell any security.

Risks: Guarantees are subject to the claims-paying ability of the insurer. There is no assurance that any objective will be achieved. Investments involve risks including possible loss of principal.