Abstract
The problem of drawing down one’s savings in retirement is known as “decumulation” and is a topic of growing policy importance. This article is devoted to understanding the psychology inherent in decumulation decisions and to identifying policy options that may develop from those findings. We begin by describing key psychological inputs to these decisions, including self-control, temporal discounting, loss aversion, fairness, and psychological ownership. We then briefly review the current evidence testing those processes within particular decumulation domains. Finally, we offer five possible policy approaches for influencing decumulation decisions, including addressing financial literacy, providing defaults, encouraging precommitments, offering disclosures, and building customized decision support systems. These interventions differ according to individuals’ financial sophistication and allow policy makers flexibility in determining the level of support to provide.
Keywords
Tweet
Tough decisions in retirement? New article on Psychology of Decumulation by Prof. Suzanne Shu (@UCLAAnderson) & @SteveShu (@CassBusiness)
Key Points
Drawing down one’s savings in retirement—known as “decumulation”—is of growing policy importance.
Current research in both psychology and economics has begun to identify how psychological drivers can affect individuals’ decumulation decisions.
Individual differences in self-control, discounting, and loss aversion, plus affective reactions such as feelings of fairness and psychological ownership can drive financial decisions during decumulation.
We propose five main policy approaches to influencing decumulation decisions, depending on the decision maker’s level of psychological sophistication.
Financial industry regulators and social program administrators can implement low-cost solutions, such as disclosures and defaults, to improve the well-being of tens of millions of retirees.
Introduction
The field of judgment and decision making has long concerned itself with the everyday decisions that can directly affect individuals’ well-being. An early application of behavioral decision research was to the domain of increasing retirement savings; researchers realized that the same behavioral factors that may lead us to undersave (Laibson, 1997; Madrian & Shea, 2001; Thaler & Shefrin, 1981; Tversky & Kahneman, 1992) could instead be leveraged to create programs that increase savings, such as Save More Tomorrow (Thaler & Benartzi, 2004). Although these applications of psychology to the accumulation of wealth have been extremely successful at encouraging participation (Choi, Laibson, Madrian, & Metrick, 2004) and higher savings rates (Benartzi & Thaler, 2013), comparatively unexplored has been the complementary problem of how individuals make decisions that improve their well-being during retirement. In the financial decision-making literature, the problem of drawing down one’s savings in retirement is known as “decumulation,” and it is a topic of growing policy importance. This article is devoted to understanding the psychology inherent in decumulation decisions, based on existing research, and to identifying policy options that may develop from those findings.
To understand the importance of individuals’ decisions around decumulation, it is helpful to start with a description of the size of the problem. Baby boomers are now retiring at the rate of almost 10,000 per day, making the issue of decumulation important to both public policy experts and the financial services industry. The need to help retirees with decumulation decisions has led to the growth of a large industry in financial advisors, and more recently robo-advisors, who can offer guidance on self-management of assets. But not all retirees even have assets to manage; among retirees in the bottom half of the income distribution, social security benefits are the majority of income in retirement (Poterba, 2014). Optimizing social security claiming can, therefore, have a significant effect on a retiree’s well-being (Scott, 2012), but remains an understudied and poorly understood decision process. In addition to decisions around social security, retirees can choose from among other retirement income solutions such as annuities, reverse mortgages, pensions, and 401(k) draw down solutions.
The policy and economic issues around decumulation decisions, such as social security claiming and annuity choice, have been well known to policy makers and economists for some time. However, the psychological issues inherent in those decisions have only recently attracted the interest of researchers in the behavioral sciences. Consumer decisions about decumulation represent choice problems with large financial stakes and limited learning opportunities, difficult consumption trade-offs, multiple sources of uncertainty, issues of trust and branding, and long time periods. All of these attributes are topics on which behavioral science research can offer important insights. In this article, we first explore the research within psychology on broad topics relevant to decumulation decisions, then discuss behavioral science research that has looked directly at decumulation, before ending with policy implications.
Key Psychological Inputs to Decumulation Decisions
Decumulation decisions are, in general, decisions that individuals need to make about how to best draw down existing resources during the decades after they end their working careers. Those resources may be acquired wealth, such as retirement savings, housing, or other assets, or they may be access to entitlement income programs such as social security or a pension. Because these types of decisions are typically only made once, involve trade-offs between current and future income, and have long-run implications, both self-control and temporal discounting are key psychological inputs. Feelings of ownership toward the resources, loss aversion for saved resources, biases in predicting future events such as life expectancies, and general issues of fairness are also important (see Knoll, 2011). We review each of these psychological drivers in the section below.
Self-Control
The psychology of self-control explores the tension between the long-term and short-term goals held by the individual (Wertenbroch, 1998). For example, individuals may appreciate the value of having higher income in the future if they wait to spend down their assets, but they also desire to use the income immediately for enjoyable consumption. Short-term consumption goals may win out over the long-term savings goals, resulting in myopic behavior and self-control failure (Hoch & Loewenstein, 1991; Kirby, 1997; O’Donoghue & Rabin, 1999; Strotz, 1955).
Solutions to self-control problems proposed by the psychology literature include mental budgets or strict preset allowances (Heath, 1995; Krishnamurthy & Prokopec, 2010; Thaler, 1985) and penalties for breaking self-imposed rules (Shefrin & Thaler, 1988). To be successful, rules must be clearly defined and ideally not able to be revised in the heat of the decision (Baumeister, 2002; Carver & Scheier, 1981). One danger is that rules that allow easy divergence may be redefined at the moment of the decision so that the individual can justify indulging in immediate wants (Cheema & Soman, 2006). On the contrary, overly strict rules may also end with failure, as individuals ultimately feel unable to stick to their strict consumption pattern and ultimately abandon the larger goal (Cochran & Tesser, 1996). Building consumption paths with small amounts of flexibility does increase individuals’ ability to persist with larger goals (Sharif & Shu, 2017).
Temporal Discounting
Individuals can differ in their temporal discount rates, that is, the amount by which they reduce the perceived value of future outcomes. The long time frames involved in decumulation decisions mean that even minor differences in discounting may, over the long term, explain large differences in how retirees manage their retirement income. The role of intertemporal patience in decumulation decisions also fits hyperbolic discounting, under which the value of near-future outcomes is heavily reduced but far-future outcomes are more weakly reduced (Ainslie & Herrnstein, 1981; Thaler, 1981). Time discounting measures predict individual differences for financial and health decisions, such as high credit card debt (Meier & Sprenger, 2010), gym attendance (DellaVigna & Malmendier, 2006), and smoking behavior (Bickel, Odum, & Madden, 1999). Because future outcomes are heavily discounted, temporal discounting is often one of the most apparent reasons (but not the only reason) for self-control failures.
In particular, hyperbolic discounting predicts a significant disconnect between immediate and near-future outcomes (Ainslie, 1975; Frederick, Loewenstein, & O’Donoghue, 2002; Hoch & Loewenstein, 1991; Kirby & Maraković, 1995; Loewenstein & Prelec, 1992). One impact of hyperbolic discounting is that plans made for future decisions (e.g., a plan to claim social security benefits at the age of 66 years rather than 62 years) can be overruled when the closer outcome date actually arrives (age = 62 years). A variety of scales and measures capture individual differences in temporal discounting (e.g., Hardisty, Appelt, & Weber, 2013; Schreiber & Weber, 2016), allowing researchers and policy makers to link this individual difference to decumulation decisions, a point we will revisit shortly.
Loss Aversion
Thinking about decumulation during retirement requires considering longevity risk: The need for assets and wealth to persist for as long as the individual lives, which is unknown. The primary risk is that income will run out while the individual is still alive. Risk aversion predicts that individuals will be more conservative with their spending, and will have a preference for options that reduce longevity risk by increasing future income. In other words, longevity risk and risk aversion should combine to make retirees prefer annuities and later social security claiming, both of which reduce the risk of running out of income in late retirement.
Although utility models of decumulation rely heavily on risk aversion to predict behavior, a more psychological model can use a different but related predictor: loss aversion. Loss aversion, as a psychological predictor of decumulation decisions, reflects the perspective that delays in utilizing resources may result in a loss of earned benefits. For social security, not claiming early may result in a perceived loss of income relative to a breakeven calculation or relative to the individual’s own working-years contributions into social security. Thus, loss aversion has different implications than risk aversion. Although risk aversion leads to later claiming and higher preference for products such as annuities, loss aversion leads to earlier claiming and dislike of annuities due to concerns that assets or entitlements will be “lost” or wasted. These concerns about potential loss (i.e., a prospective loss) may motivate individuals to utilize entitlements as soon as possible.
Using loss aversion, rather than risk aversion, to explain decumulation preferences has several implications. Behavioral manipulations moderate loss aversion in a variety of tasks (e.g., Ariely, Huber, & Wertenbroch, 2005; Gäechter, Johnson, & Herrmann, 2007; Novemsky & Kahneman, 2005). Thinking about decumulation decisions in a loss aversion context suggests applying loss aversion manipulations to interventions. Various methods can measure loss aversion at the individual level (e.g., Abdellaoui, Bleichrodt, & Paraschiv, 2007; Brooks & Zank, 2005; Payne, Shu, Webb, & Sagara, 2015; Seaman, Green, Shu, & Samanez-Larkin, 2018; Toubia, Johnson, Evgeniou, & Delquié, 2013). As with the temporal discounting measures noted above, knowledge of a given individual’s loss aversion level may assist policy makers in customizing interventions to better assist people’s decumulation decision making.
Fairness
Consumers of financial products directly consider perceived fairness, and that perception informs their attitude measures for such products. Fairness judgments depend on not just how outcomes are shared between consumers and firms, but also on the transparency and procedural aspects of the system that determines the outcomes (Bies, Tripp, & Neale, 1993). Thus, issues of perceived fairness for decumulation products such as annuities or pension payment options may be captured through both direct questions about benefit fairness (Kahneman, Knetsch, & Thaler, 1986) and questions about the process underlying the payouts. Ongoing research on annuities and other financial decisions (Shu & Morelli, 2012; Shu, Zeithammer, & Payne, 2016) suggests that individual perceptions of perceived fairness for financial products can significantly affect preferences for those products. In particular, for annuities, individual measures of perceived fairness predict overall annuity preference, consistent with behavioral explanations for the widespread underpurchasing of annuities, also known as the annuity puzzle (Benartzi, Previtero, & Thaler, 2011; Hu & Scott, 2007).
Psychological Ownership and Endowment
Related to perceived fairness are feelings of psychological ownership toward retirement assets and entitlements. Ownership is clearly an aspect of the loss aversion concept discussed earlier, as without ownership, there would not be loss. A key aspect of loss aversion is reference dependence (i.e., comparison with the starting point); the object must be integrated into the individual’s endowment so that not having it is seen as a loss. In other words, the individual must assess “is it a loss?” before loss aversion becomes relevant. Psychological ownership is the feeling that something is “mine” (Pierce, Kostova, & Dirks, 2001). Legal ownership of an object is not necessary; anticipatory possession or pseudo-endowment can have similar psychological effects to legal ownership, even without legal possession (Ariely & Simonson, 2003; Hoch & Loewenstein, 1991). For example, manipulations such as the amount of deliberation spent on the decision can affect “prefactual ownership” of an option (Carmon, Wertenbroch, & Zeelenberg, 2003).
One implication of both loss aversion and psychological ownership is the endowment effect, in which objects are more highly valued by owners than by nonowners (Kahneman, Knetsch, & Thaler, 1990, 1991). A direct measure of psychological ownership predicts the valuation disparities that underlie the endowment effect (Peck & Shu, 2009; Shu & Peck, 2011). Once an object, or even a financial benefit, becomes part of an individual’s psychological holdings, delaying the benefits of that entitlement becomes more difficult. Loss aversion and/or an endowment effect—and by extension feelings of ownership toward financial holdings—occur for various consumer financial decisions, including investments (Anagol, Balasubramaniam, & Ramadorai, 2016; Samuelson & Zeckhauser, 1988), bank accounts (Giné, Goldberg, Silverman, & Yang, 2018), mutual funds (Frazzini, 2006), and houses (Genesove & Mayer, 2001). Individuals are also more likely to pay for an extended warranty for new purchases when they feel higher psychological ownership toward the new purchase (Lessard-Bonaventure & Chebat, 2015). In each of these cases, loss aversion and psychological ownership drive individuals to make financial decisions meant to preserve the already endowed option.
Summary
Taken together, individual differences in self-control, discounting, and loss aversion, plus affective reactions such as feelings of fairness and psychological ownership (e.g., “I deserve the payments I contributed over my working life”), all can drive financial decisions during decumulation. These are certainly not the only psychological influences, however. For instance, framing affects longevity judgments, a key input to most decumulation decisions; individuals’ estimated mean life expectancies are nearly 10 years longer when they are asked how long they expect to “live-to” rather than “die-by” (Payne, Sagara, Shu, Appelt, & Johnson, 2013). Other psychological processes, such as goal generation (Bond, Carlson, & Keeney, 2008; Gregory & Keeney, 1994) and positive and negative affect (Isen, 2001), have yet to be directly explored within the decumulation space. Work on how decision making changes with differences in education and financial literacy (e.g., Fernandes, Lynch, & Netemeyer, 2014) and with increased age (Li, Baldassi, Johnson, & Weber, 2013) is also important to incorporate. We hope more behavioral science researchers will take up the call to study how these topics relate to financial decisions during retirement.
Scientific Evidence
Having now reviewed some of the key psychological processes involved in decumulation decisions, we turn to a brief review of the current evidence testing those processes within particular decumulation domains. Research in this area is relatively new and still sparse within the behavioral sciences. Work done within the economics literature has identified decumulation behaviors that are inconsistent with standard economic theory, leading those scholars to propose that behavioral explanations may be necessary. We provide insights from both of these streams of literature in the section below.
As noted above, a key decumulation decision for most Americans is the timing of their claim to social security benefits. Social Security Administration (SSA) benefits can be chosen to begin between the ages of 62 years and 70 years, provide an inflation-adjusted hedge against the risk of living long enough to outlast one’s savings, and are the single largest source of retirement income for retirees in the United States (Scott, 2012). Delaying is actuarially advantageous for a large subset of people (Shoven & Slavov, 2012), suggesting that “primary earners with average life expectancy should delay benefits to age 70 to maximize expected present value” (p. 4). Unfortunately, many people may not be making optimal decisions about when to start claiming their social security benefits. Contrary to advice, the majority of Americans start collecting at the age of 62 years or within 2 months of leaving the labor force, 80% or more claim their social security benefits before the normal retirement age of 66 years, and less than 5% delay to the maximum age of 70 years (Shoven & Slavov, 2012). Given the substantial risks of longevity and the financial benefits from delayed claiming, later claiming should be more popular and these percentages should likely be reversed (Shoven & Slavov, 2012; Tacchino, Littell, & Schobel, 2012).
Many of psychological drivers of decumulation reviewed earlier are at play in the SSA claiming decision. Because individuals may perceive benefits as already “owned,” deciding when to claim can resemble an endowment effect: That is, existing entitlements are highly valued, and the decision to delay looks like a potential loss, relative to either breakeven value or death before receiving benefits. Thus, individuals who are high in loss aversion should be less likely to delay claiming. Indeed, loss aversion, measured at the individual level, does predict preference for early versus late claiming, as do individually measured levels of temporal discounting and psychological ownership (Shu & Payne, 2013). These empirical findings support theories proposed by economic researchers on this topic. Individuals who delay claiming social security benefits beyond the age of 62 years are essentially purchasing an inflation-indexed annuity for the future (Brown, 2007); however, consumers sometimes view the purchase of an immediate annuity as gambling on their lives. Extending that idea entails examining annuity choice from the perspective of more behavioral approaches such as cumulative prospect theory (CPT; Tversky & Kahneman, 1992) because CPT is a better behavioral model than the classic expected utility model (Hu & Scott, 2007). Incorporating psychological measures, such as loss aversion, into models of claiming decisions may improve those models’ predictive power.
Another relevant example of possibly suboptimal decision making within the decumulation space is the question of whether retirees should purchase annuities. Life annuities allow retirees to exchange assets for a guaranteed lifetime stream of income payments, regardless of how long they live. Like delayed claiming, economists suggest that annuities are a good choice for risk-averse individuals to reduce longevity risk (Davidoff, Brown, & Diamond, 2005; Yaari, 1965), and yet few individuals purchase life annuities, leading to an “annuitization puzzle.” Economists seeking to understand the behavioral reasons for the lack of annuitization have identified loss aversion, framing, temporal myopia, and fairness as possible explanations for the annuity puzzle (Benartzi et al., 2011; Beshears, Choi, Laibson, Madrian, & Zeldes, 2014; Brown, 2007; Hu & Scott, 2007; Mitchell, Poterba, Warshawsky, & Brown, 1999) and some have tested interventions that reframe annuity income to see whether such presentations can change the decision (Agnew, Anderson, Gerlach, & Szykman, 2008; Brown, Kling, Mullainathan, & Wrobel, 2008).
Although much of this work has only proposed these behavioral explanations, some researchers have directly tested whether particular psychological drivers are involved in the annuity decision. For example, individual measures of temporal discounting can predict preference for annuities, integrated within a comprehensive utility model (Schreiber & Weber, 2016). Individual measures of perceived fairness, life expectancy, and numeracy affect life annuity attribute preferences when individual utility is measured and modeled in choice-based stated-preference surveys (Shu et al., 2016). This research tests multiple ways of presenting annuity information (cf. Brown et al., 2008), and considers the implications of such preferences for marketers and policy makers interested in promoting annuitization.
Policy Implications and Options
Drawing upon both the general psychological drivers of decumulation decisions, as well as the scientific evidence specific to such decisions, we now identify some policy implications of the findings. Note that we are not specifically advocating any particular approach, nor are we suggesting that some policies (i.e., claiming social security later, encouraging purchase of annuities) are preferable to others, even when traditional economic theory has recommended them. We leave it up to policy makers to determine which decisions can or should be improved, and in which ways. Instead, we outline five main approaches to influencing decumulation decisions. These approaches vary in psychological sophistication, as sorted in the list below, with the simplest interventions being educational, and the most sophisticated using individual psychological differences to customize decision support systems that guide the individual through the decision process.
Financial Literacy: Some researchers have advocated for financial training in advance of large decisions (Fernandes et al., 2014), similar to a “drivers’ license” for complicated or risky investment decisions (Alesina & Lusardi, 2006). Education programs for certain subgroups, such as retirees in advance of making decumulation decisions, could also be useful (Lusardi, Mitchell, & Curto, 2012; Meier & Sprenger, 2013). Given the impact of temporal discounting on these decisions, education for individuals with high discount rates, who may be most impatient for use of their assets, could have the highest return on investment (Schreiber & Weber, 2016).
Safe Automatic Options: Automatic options (i.e., defaults) have been a major success story in many policy environments (Thaler & Sunstein, 1999). Financial decumulation programs could include default options for asset drawdown built around guaranteed income options such as annuities (Agarwal, Driscoll, Gabaix, & Laibson, 2009). These low-cost “safe harbor” options would protect a proportion of individuals’ total wealth, while also still allowing discretion by the retiree for the remainder; such auto decumulation features could be most easily implemented in employee benefit programs such as defined contribution plans (Benartzi et al., 2011; Brown & Finkelstein, 2011).
Precommitment: Precommitment during an individual’s younger working years, including financial advanced directive contracts, is especially useful if people expect future cognitive decline (Agarwal et al., 2009). To encourage precommitment, and also compete against psychological ownership and loss aversion biases, guaranteed income programs could be adapted to allow for a small “silver lining” payouts in advance of full benefits. For example, if the SSA wished to encourage individuals to delay claiming their benefits, small lump sum payouts at the age of 62 years (in exchange for a precommitment to claim later) may help reduce the desire to claim immediately. Precommitment can also be a powerful tool in the annuity market, because younger people are more open to annuities (Schreiber & Weber, 2016; Shu et al., 2016), possibly because they have not yet built up strong ownership for the retirement assets that would be exchanged to purchase one. Options that allow purchase over time through smaller regular payments, rather than a large lump sum payment at retirement, also avoid problems of loss aversion and psychological ownership.
Disclosures and Framing: As noted above, many researchers have already tested framing interventions for annuities and social security framing. Some of these interventions focus on describing payouts as either losses or gains (Brown et al., 2008) whereas others test how providing different information formats affect decisions (Shu & Payne, 2013; Shu et al., 2016). Relatively untested is whether changes in terminology and communications, such as around ages (e.g., language that refers to the age 65 or 66 years as being at “full retirement age” despite benefits continuing to grow to 70 years) and earnings tests for social security claiming (e.g., reframing continued employment as an investment in future social security income), can shift choices (Benartzi et al., 2011). Although most information provision options remain voluntary and under the discretion of the financial providers, policy makers also have the ability to make specific information required via mandatory disclosures. Investigations underway in the area of financial disclosures also provide a useful guide to behavioral researchers investigating information options (H. Johnson & Leary, 2017).
Customized Interventions: Given that psychological research has provided well-tested, useful measures of many of the individual differences that can affect decumulation decisions, customized interventions and disclosures deployed to help individuals with these decisions are within reach. Imagine, for example, a financial robo-advisor that begins with an assessment of loss aversion, temporal discounting, and other key measures, and then leads the retiree through information designed to appeal to particular needs and preferences. Such decision support systems allow policy makers to work in concert with the individuals’ heuristics, biases, and preferences, rather than in opposition to them.
The action implications of these five approaches are highly relevant to both financial service industry regulators and social program administrators. Regulators have the ability to define appropriate default options, require particular disclosures or information displays, and demand that individuals complete individual assessments and training prior to receiving financial advice. Program administrators, such as those at the SSA, can work with policy makers to reframe benefits, change program descriptions, and even allow precommitment. Clearly, some of these interventions are lower cost than others; default settings, information displays, and disclosures are all relatively inexpensive to change, while more customized interventions such as decision support systems are more expensive to build and operate but yield more optimized outcomes for the retiree. Census projections for 2035 are that there will be 78 million people in the United States above the age of 65 years. These millions of retirees, and the families and providers who look out for their financial well-being, are counting on a smarter approach to decumulation.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.