This is an interactive narrative map. Expand, drag and select.
This is an interactive narrative map. Expand, drag and select.
There’s a wonderful story I always make a point of telling whenever trustees are debating the issue of member choice in their retirement funds. The story involves Harry Markowitz, the esteemed academic who won the Nobel Prize in Finance for his Modern Portfolio Theory – the same theory that underpins the principle that members can change their risk profile on their portfolio by changing its asset mix towards less risky assets.
Anyway, as the press reported it, Professor Markowitz was asked what asset mix he would select for his own retirement fund. Without a blink, his pithy reply was….”How do I know what’s going to happen? I guess I’d invest 50% in equities and 50% in bonds!”
What makes his response particularly significant is that research shows that this is pretty much the answer that most of us mortals give …Nobel prize holder or not. When asked to make a decision about anything as complicated as determining a twenty-year investment strategy that can assure that the value of your growth in savings keeps pace with the cost of living, most people freeze up, and make decisions that better reflect a “path of least resistance” than a carefully calibrated decision. This means that, more often than not, they’ll simply divide their investment choice between all the options on offer. So….give a member two options to choose from for their retirement strategy, and they divide their allocations 50/50. Give them three options, and they’ll typically divide it 33/33/33, give them four options and …..you see the drift.
What’s amazing about this reflex reaction is that it seems to defy rational insight. For example, when Thaler and Benartzi were testing the phenomenon, they gave three groups of pension fund members the following options to choose from1:
True to form, the almost universal response was 50:50 in all three groups – no matter what the equity exposure translated into given these allocations. So, in the case of Group A, their 50/50 split left them with 50% in equities. In Group B, they had an exposure of 73% in equities, and Group C’s split decision netted them a mere 35% in equities! (Graph 1 below)
Nor does the decision seem to be predicated on members’ assessments of their risk profile. Thaler and Benartzi discovered that when they gave workers an opportunity to choose between their own age/risk-appropriate portfolio and the portfolio of the median participant, 8 out of 10 participants preferred the median portfolio to their own. The implication here is that members have little clear insight or commitment to the investment rationale of their own portfolio. As such, when confronted with the reality that the average member of the fund has a very different portfolio, the preference was clearly to go along with what the rest of the members did – whether this is an appropriate solution or not.2
But we are describing only one of a half dozen behavioural phenomena that are contributing to the most important reality our industry has yet to properly face. In spite of the extensive amount of intellectual capital being applied by investment professionals to identifying the “optimal investment strategy” for accommodating every possible attitude for risk, every possible funding requirement, over every imaginable time-frame, it appears as though the industry is still failing at the most basic step in the process: getting the member into the right solution in the first place – and then getting them to where they need to go over time to address a lifetime of savings requirements and investment concerns.
As Olivia Mitchell and Stephen Utkus summarised the problem in their excellent working paper “Lessons from Behavioral Finance for Retirement Plan Design”: “Behavioral research…challenges some of the most central assumptions of decision-making: that workers are rational, autonomous, microcalculators who exercise independent and unbiased judgment when it comes to their retirement plans.”3
We are only just beginning to appreciate how powerful this force is. Consider this conclusion of the authors, for example. In their estimation, the dominant factor dictating investor decision-making in retirement fund investing appears to be inertia – not a rational process of weighing up the options. Inertia is what keeps members from signing up for plans in the first place (unless plans are compulsory); it’s what keeps members from increasing their savings rate when failure to do so will clearly create a drag on achieving their replacement ratio targets; it’s what accounts for the fact that once investment programmes are set, only 10% of members make any further changes in their portfolio asset allocation (unless a Lifestage portfolio forces the appropriate change over time)4; and, most disturbingly, it’s what accounts for the fact that even when pension plans set up elaborate member education programmes, these too fail to get members any closer to their optimal investment strategy over the course of their membership.5
Why is inertia such a dominant force? To begin with, we probably grossly underestimate how difficult and complex a job retirement plan investing really is. As Mitchell and Utkus state the problem:
“Being good at retirement savings requires accurate estimates of uncertain future processes including lifetime earnings, asset returns, tax rates, family and health status and longevity. In order to solve this problem, the human brain as a calculating machine would need to have the capacity to solve many decades long-time value of money problems, with massive uncertainties as to stochastic cash flows and their timing.”6
So inertia is really just one way our brain tells us that it simply doesn’t know how to resolve all the complexity.
But, there are a number of other equally dysfunctional behaviours that our brains revert to during the complex member decision-making process and these also demand our attention.
Consider this example:
Can’t make a decision about the optimal path to retirement investing? Then the “path of least resistance” heuristic (short cut) leads members to either:
The examples go on and on. We will come back to a few more. Bottom line is that unless our industry gets serious about understanding and addressing these types of behavioural issues, and works collaboratively to develop ways to mitigate against these sub-optimal decisions, then the best laid investment plans will continue to fail the members – and not because the investments have gone sour.
What behavioural finance is beginning to teach us is that we need to re-examine a number of conventional practices if we are going to get it right. What follows is a discussion of where these practices are failing the process, why they’re failing the process, and what we, the broad group of fiduciaries responsible for these assets, can do to ensure a better outcome. We grant that these suggestions may prove to be quite provocative at first – but, in light of the findings of our authors – they may actually redress some of the problems we all grapple with.
Just how do we structure investment options to members of pension funds?
It is the rare investment tender or beauty parade of manager skills that doesn’t request that an asset manager provide trustees with their very best (house view) recommendation for an “aggressive”, “moderate” and “conservative” investment option. What comes across as a perfectly straightforward request actually masks a minefield of potential problems.
Here’s the crux of it. What exactly is meant by “aggressive”, “moderate”, “conservative”? Do the consultants, the trustees, the members, and the investment managers/multi-managers even begin to share the same understanding of these adjectives and, more importantly, do they have the same expectations as to how these different solutions should:
Even more problematic is whether members can meaningfully assess their own risk profile. Chances are that a 28-year-old mother who is the sole provider for two children is going to view her current situation as precarious and readily tick the “conservative” risk profile box. But is a cash or the low-risk portfolio the right long term retirement funding strategy for someone who will undoubtedly be required to work for most of her life to provide the required family support?
Similarly, is it the right strategy for a senior member of the fund, who has the bulk of their external wealth tied up in market-related assets, to compound their bet on the markets so close to retirement by selecting a high risk, high performing single manager portfolio for their “asset holding of last resort”?
These are complex questions. But two things are clear:
Risk profiling questionnaires are unlikely to get us any closer to the right answer. With such thought-provoking questions such as “do you like bungee-jumping?”, it’s little wonder that research is now suggesting that risk aversion questionnaires do little more that help members identify what their attitude towards risk is at that specific moment in time – an insight that may bear little relationship at all to what their attitude would be should their pension fund replacement ratio fail to hit its mark twenty years from now.
To begin with, we are up against an all-too-human tendency known as “hyperbolic discounting”.7 This is the mental sleight of hand that leads individuals to undervalue (under-prioritise) future long term benefits and overvalue the nearer dated opportunities. It’s the same behavioural phenomenon that results in many members of provident funds simply opting to take as large a lump sum payout as they possibly can at retirement, in spite of all the well-meaning advice to invest that cash in some form of annuity. And it’s the same phenomenon that results in risk profiles merely capturing the risk attitudes of individuals to the environment that surrounds them in their present circumstances, and not the environment that might meet them 20 years from now if they failed to meet their funding requirements. No questionnaire would be likely to elicit that emotional response.
For most trustees, consultants, and service providers such as ourselves, the primary focus is typically on the extremely complex problem of making sure that the investment strategy or range of investment strategies selected have the highest probability of delivering what is promised to members. The irony is that much of this good work may be completely undone by – of all innocuous-seeming things – how we present these options on the decision forms we give members!
To start with, “the path of least resistance” phenomenon stacks the odds heavily in favour of whatever gets listed first. Even when a default option is made available, if it is listed last, as they often are, the very members who should be opting for the default may not be reading far enough into the document to actually select it.
But the behavioural phenomenon known as “framing” also plays an important role. Here the issue refers to how the layout of the options may end up inadvertently leading the decision-makers to the wrong decision simply because of the way the decision was “framed”.
Consider that most option sheets typically list available choices in order of their riskiness. List the least risky portfolio first and this will become the dominant choice. List the most risky portfolio first and this too will get selected more times than our calculated distribution would suggest. For many individuals, listing something first subliminally suggests that it may well be best.
A better example of framing is perhaps illustrated in another fascinating study by Benartzi and Thaler. In this study they developed four portfolios ranging from portfolio A to D that each reflected a different risk profile ranging from most aggressive to least aggressive. Different groupings of these portfolios were presented to three different groups of members to determine their preferences. Astonishingly, although portfolio C reflected the same risk profile in each grouping, how it was positioned relative to other options being offered to members impacted their preference for the portfolio. For example, if the C portfolio was positioned last – indicating that it had the most extreme risk profile, it would garner the lowest number of preferences. If C was one of two options available (although still the last listed option) the preferences increased – but still didn’t exceed the first option. And If C was sandwiched in between two other options, this is when it would attract the highest number of votes (this is when the “select the median” heuristic clearly kicked in). See Graph 2 below.8
There is an accepted wisdom that members of pension funds who are well-informed about the nature of investments will make better decisions about long term retirement investments. Ironically, these are often the very members or trustees who lobby hard to have more aggressive or exotic options represented on the member choice platform. But again, behavioural research is suggesting that better education does not necessarily lead to better investment decisions.
In addition to the issues of hyperbolic discounting and framing that afflict all of us, we add to the list the “anchoring heuristic”. Anchoring simply tells us that the starting point of your investment experience can often have a greater influence on what you invest in and how, than where your end point should dictate.
Mitchell and Utkus investigated 2.3 million participants in the Vanguard Group’s pension plans:
This not only illustrates how sensitive participant investment decisions are to then-current market conditions but how these conditions can “colour” future decisions for some time.
Interestingly enough, the same phenomenon persists among investment professionals – who should theoretically know better. The rule of thumb is never to pick a fund manager who has never experienced a bear market. But, in fact, the issue might be more subtly complex. Select a manager who began their career when certain conditions persisted and those conditions have a more dominant impact on their thinking than lay people would like to imagine.
What Mitchell and Utkis conclude is that no matter how well educated the investor (remember our Harry Markowitz story), their investment decisions are typically formed by both the current market conditions and some element of their “anchoring” experience.
Again, trustees genuinely believe they are serving their members well if they provide for more member preferences. But Sethi-Lynegar et al. caution that the reverse is actually true.9 Offering more choice actually triggers the “inertia” shutdown response that members experience when decision-making becomes too complex. In fact, the optimal number of choices for effective decision-making is probably significantly lower than most trustees imagine – with five options possibly stretching the human decision-making mechanism to near sub-optimality.
A better heading here would probably be: “You actually can get members/trustees to stop using surveys and peer group rankings as crutches for their decision-making.”
But perhaps we first need to understand why it is that there is such an extraordinary weighting given to past performance and rankings.
Mitchell and Utkus argue that this specific behaviour owes much to two phenomena:
As cryptic as these two phenomena sound – they are really quite straightforward.
The representativeness heuristic simply refers to the tendency of human beings to see patterns in a series of numbers or results when, statistically speaking, there is really only randomness. Give a manager three straight, five straight, and even ten straight years of top performance and invariably they will be regarded as superstars – in spite of the fact that odds are still high that such an outcome is well within the range of a random distribution of outcomes.
By contrast, the “availability” heuristic kicks in when there is simply so much information to process to develop a meaningful insight, that any distillation of this complex data to provide a readily available assessment (through, say, a oversimplified ranking of manager performances) becomes a welcome crutch in the decision-making process.
So when trustees are presented with performance surveys, the double whammy of the “representativeness” heuristic and the “availability heuristic” kick in to ensure that trustees and members (and even consultants) mistakenly leap to the conclusion that top quartile performance must be a powerful indicator of manager skill – although nothing could be further from the truth.10
In summary, the value-destroying behaviour that’s triggered is that investors will typically abandon a perfectly viable long term investment strategy in the belief that short – or even medium term - underperformance is somehow indicative that the underlying manager responsible for performance delivery is somehow devoid of skill.
But, as the study by Watson Wyatt11 in the UK suggests, the real outcome is that trustees and their consultants simply end up firing managers just as their fortunes begin to turn. Investment philosophies and styles do have a nasty habit of working for certain parts of a market’s cycle and not in others – so when the cycle turns against the strategy, the mistaken view is that the manager’s skill has somehow vanished.
How much does this bad timing cost a fund? The timing problem alone suggests that fund managers who may have outperformed by as much as 4.4% p.a. before being hired go on to underperform the fired manager, whose performance has now recovered. In fact, the stronger the pre-hiring performance, the weaker the post-hiring delivery. Conversely the stronger the underperformance of the fired manager before being fired, the stronger their outperformance after being fired. But timing isn’t the only factor that destroys value. Changing managers on a fund incurs costs that research suggests aren’t necessarily recouped.
Clearly, then, finding the mechanism that can curtail this reflexive churning has significant implications for meeting, if not preserving, long term returns.
There is nothing new in these findings. So a rather provocative question comes to mind: Is our inability to curb these practices a function of the incredibly powerful behavioral influence at work with boards of trustees, or is it a function of the fact that it may not be in the industry’s interest to encourage such a change in behavior?
In truth, the ultimate performance of the solution owes far less to individual manager performance and far more to the integrity of the long term structure of the asset management solution than trustees remotely imagine.12
The good news is that the pattern of reliance on performance histories and performance surveys can be broken. What’s required, however, are three critical factors that may not always be forthcoming.
What these sub-optimal decisions of members make clear is that the choice of the default portfolio for any member choice fund is absolutely critical. This will be the portfolio that inertia forces most members into selecting. The problem though, is that for many trustees who are genuinely interested in ensuring there is no potential come-back from members should the portfolio be deemed “unsuitable”, the default portfolio typically turns out to be the “least risk” portfolio. Unless more than 60% of the fund’s members are within five years of retirement this would clearly be the wrong portfolio choice for the bulk of the fund’s membership. It’s a widespread phenomenon that reflects the most prevalent abuse in retirement fund investing: reckless conservativism.
For many trustees, arriving at a meaningful insight for the default option demands the kind of “management” information that trustees rarely have access to. What’s really required is:
We believe the answer lies in a management decision-making tool that is readily accessible to trustees and should be reviewed at the start of every trustee meeting. These are the meaningful dialogues trustees should be engaging in.
We believe that our current developments in the alpha-lab tool provides trustees with just these insights. (Again, more on alpha-lab in the accompanying paper)
Bottom line, though, the default portfolio must reflect the optimal investment strategy for the most members – not the least risk option
So what are the most critical lessons that Behavioral Finance has to teach us about member choice decision-making? Possibly the most important lesson is that education alone is not going to resolve the bulk of these problems. Addressing the challenge demands that the industry itself provides the much needed structures and tools. If inertia is such a prevalent problem then we need to have programmes that automatically move the bulk of members to where they need to go, when they need to go, to meet their long term funding requirements. A Lifestage structure is certainly an excellent starting point. But with the increasing awareness with trustees that their responsibilities actually extend to ensuring that members get into the right post-retrement vehicles as well, we believe that the Lifestage construct needs to expand right through to post-retirement.
We have also learned that members “opt out” for the default option when decision-making becomes far too complex – and that it doesn’t take much to make members feel overwhelmed by the decision. As such, we need to provide trustees with meaningful management information on the membership profile of their funds and the decisions those members are making to help them determine an optimal default portfolio that will service the maximum numbers of members.
And finally, we need to change the whole evaluation process of the plan’s success from one that simply assesses how well the performance of the fund is doing when compared to the peer group to one where members assess how well they are doing at meeting their own long term goals and funding requirements. This shift in focus demands a tool that can be readily and regularly available to all members whether in paper form or via a web-based means. Only by changing this mindset will we be able to keep members and trustees focused on staying the course of their optimal investment plan and curtail the constant erosion to long term performance that comes from chopping and changing managers and strategies.
1 Benartzi and Thaler, 2001, “Naïve Diversification Strategies in Retirement Savings Plans,” American Economic Review, March 91(1): 79-98
2 Benartzi,Shlomo and Richard Thaler 2002 : “How Much Is Investor Autonomy Worth?” Journal of Finance ((57) 4: 1593-1616.
3 Mitchell, Olivia and Utkus, Stephen, 2003, “Lessons from Behavioral Finance for Retirement Plan Design” Wharton Business School Working Paper, p.30.
4 Madrian, Brigitte C. and Dennis F. Shea, 2001 “The Power of Suggestion: Inertia in 401(k) Participation and Savings Behavior.” Quarterly Journal of Economics. 116 (4) 1149-1187.
5 In a study by Choi, Laibson, Madrian and Metrick, they discovered that immediately following one of their educational seminars, they could elicit a 100% commitment to “action” from members attending their seminars. “In fact, however, over the next 6 months, only 14% did so”. Disheartening news to say the least.
6 Mitchell and Utkis, ibid, p.3.
7 Mitchell and Utkus, ibid, p.5.
8 Benartzi and Thaler 2001 ibid p 79 -98
9 Sethi-Iyenjar, Sheena, Gur Huberman, and Wei Jang. “How much choice is too much? Contributions to 401(k) retirement plans
10 This last point is a massively complex issue in itself, but for readers interested in exploring the point a bit further, please feel free to contact the author for further background information on the subject.
11 Watson Wyatt, “Hiring and Firing Managers” What if? April 2006
12 Again – another long discussion, covered in other equally long papers - but a meaningful one.
Please upgrade your internet browser to Edge or use Chrome for a better experience.
(NB: Your current browser is no longer supported.)