There’s a new paper up at NBER — the National Bureau of Economic Research — that isn’t directly about retirement but still holds an important lesson for planning retirement.
The authors have 9 years of longitudinal data. Every year they would ask the same set of women questions like:
- How many children do you want to have in your life?
- Do you think you’ll ever change your mind about how many?
- Do remember how many you wanted 5 years ago?
They were testing three main things:
- How stable do people think their preferences are? (“I will never change my mind about wanting more than two children.”)
- How good are people at predicting changes to their preferences? (“Even if my first two children were both boys, I still wouldn’t want a third child.”)
- How good at people at remembering their past preferences? (“When you asked me five years ago, I definitely said that I wanted three children, not two.”)
Their results — that we over-estimate how stable our preferences are, that we are bad at predicting how our preferences will change, and that we conveniently rewrite history to pretend we never changed our mind — all seem directly applicable to retirement planning.
In retirement planning, the default assumption is that our preferences are 100% stable for the next 1/3rd of a century. Whatever we are spending today is what we will be spending in the year 2050. The default assumption is that we are amazing accurate at predicting how our preferences will shift. So good at it, that some people think that can “match liabilities” three decades out. And, of course, when we talk to other retirees for advice about their choices & decisions we only hear the “rewritten” version of things.
Some of the numbers from the study:
- Only 27% of women said they’d want another child if the first ones all turned out girls. In reality, 70% of women in that situation ended up saying they wanted another child. 50% of women mis-predicted how they would act & feel.
- 28% of women who said they would not have a child in the next 2 years ended up having one within that time frame.
- Only 31% of women could accurately recall their past preferences.
- Most women claimed to desire the exact same number of children three years ago as they desired today, a clear case of anchoring & post-hoc rationalisation.
This illusion of stability has potentially important implications. Incorrect expectations and beliefs might result in sub-optimal decision-making for some.
There’s no reason to believe these results are limited to women and their preferences about children. It seems more likely (and the authors point to a few other, similar, research papers) that it is a pervasive bias among humans.
Just as one example, a 2017 survey found that divorce among the 50+ crowd has doubled up to 10%.
Yet how many retirees actually consider that as a possibility in their planning? You have a bigger chance of divorcing than of running out of money due to a financial crash, yet retirees focus much more on one than the other. We assume that our partner preferences are 100% stable.
It seems to me like further evidence that our retirement planning should be based around retaining optionality, about keeping things flexible, and being humble about how well we can predict a future that is several decades out.
“The Illusion of Stable Preferences Over Major Life Decisions” by Mueller et al.
