The retirement problem

by vonNudge

The modern day retirement problem is very new. Higher living standards and life expectancies has made long retirements the norm, rather than the exception. Furthermore, people would often work at one employer all their life before retiring on a company pension. Pensions, otherwise known as defined-benefit retirement plans, provide a safe stream of benefits to their recipients. This naturally places a high risk and administrative burden on the plan provider. General Motor’s massive pension obligations was one of the main causes behind it requiring a bailout from the US government during the financial crisis. Defined-contribution plans are much more common nowadays. With these plans the sponsor’s obligation is made up-front, while the individual has to make all the financial decisions.

This is a curious state of affairs. We don’t expect people to defend themselves in court (and this even applies to professionals, with the old adage that a lawyer who defends himself has a fool for a client), and people who place their symptoms into Google to diagnose themselves over the internet are labelled as hypochondriacs. So why is investing any different?

Well, many people do employ financial advisers. But while doctors are obliged under the Hippocratic oath to help you, the financial industry lacks any true equivalent. Yes, there are codes of good practice in the industry, but they fail to overcome the incentives on offer. Simply put, the financial industry makes money by siphoning off a portion of your money. Take a full-service broker who charges 1% of assets on your £100,000 portfolio – or £1,000 a year. Increasing your rate of return by a further 1% will increase their compensation by the same amount, but that’s only £10. While a 1% increase in returns is very profitable for you – increasing your wealth by £1,000 in year 1 and further in future due to compounding – it’s barely worth the effort for your adviser. Therefore, your adviser is better off generating fee income, for example by buying and selling stocks at an unnecessary frequency (otherwise known as churning).

The government is aware of many of these issues, and has recently passed the Retail Distribution Review to try and deal with some of them. For example, many advisers used to receive secret kickbacks from the providers of certain mutual funds – which were often the highest-cost funds and the worst for the consumer. Although this is a good piece of legislation, there’s no way it can solve all the problems around adviser incentivisation. Advisers still have little incentive to make optimal decisions for you, and the fundamental information asymmetry underlying the relationships (ie that they know more than you do), means they are unlikely to get called up on any errors.

There is another way.

Nudge theory” proposes that small changes in representation, default options, or problem complexity can have a large impace on behaviour. For example, when applying for a driving license you’ll have to answer a question about organ donation. Countries vary in terms of the default option on this form. Some countries require non-donors to actively opt-out, while others require donors to voluntarily opt-in. Perhaps amazingly, the default option on this form has a huge effect on behaviour. Only 12% of the population choose to donate their organs in Germany, which has an opt-in system, compared to almost 100% in Austria, where an opt-out system is in place. The publication of Thaler and Sunstein’s book Nudge in 2008 has lead to a lot of political interest in the field, for example with the UK government setting up a dedicated “Nudge Unit” in 2010.

Some nudges have already been applied towards the retirement problem by some companies in the US. Automatic enrolment is one example. Just as with organ donation, requiring non-investors to opt-out of a defined-contribution retirement scheme leads to much higher levels of investment. Another example, called “Save More Tomorrow”, encourages investors to commit to increasing their investment contributions as their pay rises.

These examples are a good start to the problem, but unfortunately there is much more that could be done. The list of common investor mistakes, called biases in the behavioural finance literature, is huge. A few of the most costly examples are overconfidence, overtrading, overpaying for financial products, recency bias, failing to diversify sufficiently, and selling winners while holding losers (the disposition effect). This list is highly abbreviated, and it’s clear that automatic enrolment and escalation does nothing to solve them (although defaulting investors into an appropriate low-cost fund can help).

My personal view is that the impact of many of these biases could be greatly reduced with a helpful nudge or two. Much of the literature in the field of judgement and decision making in pyschology is concerned with noting participants’ poor performance on a range of abstraact reasoning tasks, and then finding experimental manipulations to improve their responses. Errors are often not innate, but due to a poor representation of the relevant information.

Take the mammography problem for example, a problem in Bayesian updating.

Suppose a woman tests positive for breast cancer.

The base rate of breast cancer in the population is 1%

If the patient has breast cancer there is an 80% probability that the patient will test positive

there is also a 10% chance of those without breast cancer testing positive.

What is the probability that this woman has cancer?

The correct answer is 7.5%, although the majority of participants fail to find this answer (they tend to greatly overestimate the probability of breast cancer, which can be slightly concerning as even trained doctors perform poorly on this task).

The problem is easier if we change the way the information is presented:

10 out of every 1,000 women will have breast cancer

of these 10 women, 8 will test positive

out of the remaining 990 women 99 will also test positive

therefore the chances of the woman having breast cancer is 8/(99+8)=7.5%

In this presentation of the problem most participants get roughly the correct answer. (Althoughwhy this might be the case is an issue up for debate.) The important point is that a simple reframing of the problem has lead to participants performing much closer to the predictions of a normative theory (a theory of how we should make decisions). Now, luckily there are many theories of how investors should make decisions in finance.

My hypothesis is that, just as in the mammography problem, individual investors can be made to act much more like these models with a few simple changes in how the information is presented. Take overtrading for example: a classic study showed that the more often you trade the less likely you are to get good returns on your portfolio. Now, there is mainly one simple reason for this: the cost of trading stocks is very high, often amounting to 5% for a round-trip trade. This is very costly given that the stock market generally increases by around 6% a year. One answer might be that one of the main costs of trading, the bid-ask spread, is paid out of an investor’s total return, and so might never be noticed after conducting a single trade. This is quite unlike other walks of life. We are used to paying costs with an up-front exchange of cash. I hypothesise that many investors would make fewer trades if this cost was made more salient, perhaps by requiring them to make a debit from their cash account for an amount equal to the bid-ask spread. If investors had to make a seperate payment of £50 for every £1,000 round-trip trade they might make a lot fewer trades. It quickly becomes obvious how the costs are adding up. Although paying the bid-ask spread out of their total return, or out of a seperate account, are logically equivalent, it might be that just as in the mammography problem, one representation helps by increasing the salience of the relevant information.

Changing the way that investors experience investment costs is one example of a nudge. Nobody is forced to change their choices, just as in opt-in versus opt-out in organ donation. But it might well be that investors will make better decisions with the new information format.