تصمیمات تخصیص دارایی های اطلاعات غلط و اطلاع از بازنشستگی اعضای طرح خود راهبری

Most defined contribution pension plan members misunderstand asset allocation, but those with higher levels of wealth managing their own money are less likely to be confused. Younger, more-educated, higher-earning advice-receiving males with a planner mindset hold more equity. Notably, an understanding of asset allocation accentuates the impact of the key factors age, income and a planner mindset.

On February 6th, 2005, during his State of the Union speech, President Bush renewed his call for social security reform. One of his key proposals was to create personal retirement accounts whose returns would be a function of worker decisions. Unfortunately, the extant evidence suggests that future retirees often make potentially costly mistakes in managing their self-directed retirement accounts. Many do not begin saving till very late and do not save enough once they start (Mitchell & Utkus, 2004). Investment decisions are made though perceived to be suboptimal by the decision-makers (Benartzi & Thaler, 2002). Risk is not well understood: it is common to believe that an individual security is less risky than a market index (Benartzi, 2001). People become increasingly hesitant and even paralyzed when offered additional asset choices (Iyengar, Jiang, & Huberman, 2004). Ignoring the most basic lessons of diversification, future retirees put far too much money into company stock, and fall prey to recency and representativeness in chasing winners (Benartzi, 2001). And, despite all these shortcomings, plan members are all too sure that they are doing the right thing (Bhandari & Deaves, 2006). One finding that has been particularly troubling is that many future retirees do not understand asset allocation. One way this has been demonstrated is when survey respondents are given several fund menus to choose from, they are often egregiously inconsistent in their equity exposure or risk-taking (Benartzi & Thaler, 2001). More specifically, they sometimes employ a “diversification” or “1/n heuristic,” whereby all n funds on offer are given roughly equal allocations without regard for underlying risk. What makes this confusion potentially damaging is that it is often observed that the asset allocation decision is the most important one for an investor’s long-term portfolio performance (e.g., Brinson, Hood, & Beebower, 1986 and Brinson, Singer, & Beebower, 1991). 1 The question of an investor’s optimal allocation is complex and as yet unresolved. Nevertheless, most would agree that it is in large part driven by one’s only noisily observed risk tolerance. Additionally, under reasonable conditions theory agrees with industry practice that risk-taking should decline with age and proximity to retirement ( Bodie, Merton, & Samuelson, 1992). Financial planners typically recommend an equity share that not only conforms to an investor’s risk attitudes, but also declines one for one (1%/year) as people approach retirement. The purpose of the current paper is to conduct an analysis of the asset allocation knowledge and decisions made by individuals in self-directed retirement accounts. In doing so, we combine two strands in the asset allocation literature, that on confusion and that on demographic determinants. We make use of a survey of approximately 2000 Canadian defined contribution pension plan members. While surveys are commonplace, this one is unique in being consciously designed to investigate the extent to which plan members fall prey to certain investment knowledge gaps and behavioral pitfalls. One of the pitfalls investigated by the survey instrument was the extent of asset allocation confusion among participants.2 While this is old ground, unique to our paper, we then turn to the demographics of the problem. There has been some research on the demographics of investment knowledge inadequacy and behavioral biases. For example, gender and overconfidence appear to be associated, with men being more susceptible (Lundeberg, Fox, & Punccohar, 1994 and Barber & Odean, 2001). Further advances on who is most at risk might make it possible to direct educational offerings appropriately (MacFarland, Marconi, & Utkus, 2004). We go on to explore how the stock-bond mix is impacted by demographic factors. Our dataset provides distinct advantages relative to previous work which looks at actual 401(k) or 403(b) behavior along four dimensions: ignorance, framing, inertia, and sample breadth.3 First, and most importantly, others treat all individuals equally, though it is clear that some individuals are ignorant of asset allocation, while others understand it. Our paper is unique in being able to differentiate, and to observe how behavior differs between the groups. Second, the 1/n heuristic indicates that for the ignorant the stock-bond mix will be a function of the frame, which in this context is the menu of choices available. This is particularly a problem for papers examining the revealed behavior of 401(k) or 403(b) investors, as member choices will be influenced by the frame. We obviate this problem as we are able to examine individuals who are consistent regardless of the frame. Third, member choices are sometimes not really choices at all, but rather inertia-induced defaults. In our sample, inertia is not an issue, as all must express their preferred allocations. Fourth, in most 401(k) or 403(b) studies members either come from a single company or a single pension provider, which raises questions about the extent to which findings can be generalized. On the other hand, our sample makes use of a broad range of plans. 4 It should be noted that the use of surveys as opposed to large sample studies of actual investor decisions offers both advantages and disadvantages. On the negative side of the ledger is the suspicion that respondents, not being incentivized in any obvious fashion, are unlikely to respond with any care: that is, salience is lacking. While it is likely that the respondents are mature individuals who take their pensions quite seriously, beliefs are not the same as actions, and survey questions can be misunderstood. If the problem is noise creation, this is a weak criticism, since it just makes signal extraction all the more challenging. The problem of bias is not so easily countered, so any results must be scrutinized with this possibility in mind. Still Kühberger, Schulte-Mecklenbeck, and Perner (2002, p. 1164) argue that “the general consensus among psychologists seems to be that hypothetical choices give a reasonable, qualitatively correct picture of real choices.” On the positive side of the ledger, a survey can allow for great flexibility. Graham and Harvey (2002, p. 189) point out in their study of a survey of CFOs that “large sample studies often have weaknesses related to variable specification and the inability to ask qualitative questions.” Indeed, since one of the major goals of the present study is to observe differences in behavior between those who signal an understanding of asset allocation and those who do not, it would be difficult to broach this issue with naturally-occurring data. In the next section we provide a brief literature review of asset allocation confusion and decision-making. Section 3 describes the survey instrument and presents results on the extent of asset allocation confusion. The next section turns to the demographics of the issue. Section 5 moves on to exploring the determinants of equity exposure, with a focus on differentiating between the behavior of those who understand asset allocation and those who do not. The final section concludes.

The existence of asset allocation confusion among self-directed retirement account members is documented using a dataset of Canadian DC plan members. Consistent with Benartzi and Thaler (2001), confusion is widespread. Further, those with higher levels of wealth who actively manage their own portfolios have a clearer sense of asset allocation and are less likely to err. Consistent with previous work, the stock-bond mix is impacted by gender, age and income: younger, higher-earning males hold more equities. A reasonable inference seems to be that no strong bias arises from the use of hypothetical survey data rather than actual portfolios. Moreover, we produce some new insights. Those with higher levels of education, receiving advice and having a planner mindset take on more risk. Additionally, those who understand asset allocation are more likely to be risk-takers. This comes through in the enhanced sensitivity of several key factors for those who are AAC. For example, while those who are ignorant do decrease their equity share a little with age, the decrease is greater for the AAC group. Some care is called for. Can we definitively conclude that those who understand asset allocation are closer to optimal behavior than those who do not? It is impossible to be certain. The reason is that most theoretical models are not precise enough. For example, while models do call for a positive relationship between income and equity share, they do not specify what the partial derivative should be. So it is always possible that those who are consistent overshoot. This issue awaits further research. This caveat notwithstanding, the evidence here seems to reinforce the importance of education to increase the likelihood that individuals will conform to normative precepts. Bernheim and Garrett (2003) and Lusardi (2004) document that workplace financial education increases saving. Admittedly though some are not likely to access or be swayed by education. In such cases, judicious defaults are arguably best. In this regard, it has been demonstrated that automatic enrolment is helpful in getting people to save (Madrian & Shea, 2001), and that a program whereby people lock themselves into future scheduled deferral increases is effective in inducing people to save more (Thaler & Benartzi, 2004). In the context of asset allocation, defaults into age-based asset allocation funds, or, even better, the increasingly popular lifecycle-type funds designed to dynamically adjust asset allocation as individuals approach retirement (Holden & VanDerhei, 2005), are sensible.