Why we save less for old age than we should

By Joanna Tyrowich & Krzysztof Makarski
Posted on June 28, 2019

How often do you think about the grocery shopping you’ll do after you retire? Or about the clothes you’ll buy when it’s time to retire? If your answer is “never”, do not feel confused. Our brains are not wired to think forward1, even about dinner tomorrow, let alone decades ahead.

Why is that a problem? Because the models that governments use to design pension systems assume that we optimize our consumption for our entire lifespan, and that stick to this plan vigorously. They also assume we have access to all available information, perfect foresight of the future and essentially unlimited capacity to process this data instantaneously.

Obviously, these assumptions are wishful thinking. In order for the models to deliver rigorous policy implications across various scenarios, they have to be tractable, just like life is. But with the development of new modelling techniques – and ever growing computing power – we can attempt to depart from rational expectations in the interest of capturing the average person’s personality. In our project, we relax the rationality assumption, turning the agents in our model into real people.

“People usually fail to stick to their pension planning and do not put money aside.”

Time inconsistency: real life

One thing people often do in real life is fail to stick to their own plans or follow through2,3. This phenomenon is called time inconsistency. In the context of retirement and pension planning, the plan is usually the idea to put money aside. But when it comes to actually putting money aside, we do not, for a bunch of reasons. For example, whatever we do not consume today actually makes us sad immediately, whereas the gratification of future consumption is delayed into the foggy future. Also, putting money aside is not nearly as exciting as spending it immediately.

The same applies to young people’s idea of retirement: it seems like a really good idea to continue to work into old age, until we actually reach that age. By the time we turn, say, sixty years old, staying professionally active for a few more years is no longer attractive. Time inconsistent agents save much less money than they would have wanted, making it impossible for them to consume as much as they would have preferred at retirement.

The element of surprise

There is also lot of empirical evidence that people assume the future will be pretty much like today4. Naturally, when they find out that tomorrow is quite different than today, they update their beliefs about the rest of their future – but still they do not expect the world to change much. That’s why we allow the agents in our model to be surprised by the economy. They are also surprised by how long they actually live – substantially longer than their own (grand)parents.

Once the economy and the population stop changing, people who are adaptive learners have nothing more to learn about, making them essentially rational. Except, of course, that the economy and the population always change. When we introduce adaptive learners, we learn that they save and work less than they should, and consume a bit too much. As a consequence, they cannot consume as much as they would have liked in their old age.

Mitigating old-age poverty

That said, many people are consistent enough, but with their current workload, family life and other activities, they simply don’t have the time – or the skills – to invest in financial markets. They would be quite skilled at putting money aside but their funds effectively earn meagre interest5. Agents who can store wealth, but cannot earn any serious interest, will not save nearly as much: giving up consumption today for hardly any return from postponing it to the future, makes them think about saving for retirement relatively late in life. For most of their working life, they simply consume their income.

Clearly, providing these agents with government-subsidized old-age saving instruments can help them achieve higher consumption in old age. We can mitigate old-age poverty and achieve greater social cohesion. In this project, we construct a model with four agent types, who behave more like people in real-life situations. We then study if these agents can be effectively convinced to save more, for instance by giving them a tax exempt on old-age savings.


  1. 1. https://think.ing.com/articles/daniel-gilbert-the-miracle-of-being-able-to-think-forward/
  2. 2. https://www.youtube.com/watch?v=dqxQ3E1bubI
  3. 3. https://slideplayer.com/slide/6023071/
  4. 4. Foster, G. & Frijters, P. (2014). The formation of expectations: Competing theories and new evidence. Journal of Behavioral and Experimental Economics, 53, 66-81.
  5. 5. freakonomics.com/podcast/everything-always-wanted-know-money-afraid-ask-rebroadcast/