Behavioral Finance Q&A with Shachar Kariv – Part 1

The University of California, Berkeley economics department chair discusses the role of “decision science” in improving participant decisionmaking and closing the retirement income gap.

Shachar Kariv is the Benjamin N. Ward Professor of Economics and the Economics Department Chair at U.C. Berkeley. Like other thought leaders, Kariv believes “decision science” is redefining the way people save and invest money, especially for retirement.

He admits retirement readiness and decision theory aren’t exactly the standard fare for economists in his position—but the trillions of dollars Americans have saved in the form of tax-qualified retirement assets comprise a critical piece of the U.S. investing landscape, he says. Beyond this, it is vital for a healthy economic future that Americans save enough to take financial responsibility for themselves and their families in retirement.

Finally—unlike economic challenges that so commonly break down by income quartile or political affiliation—everyone who hopes to retire one day, at any income level, must confront the difficult task of giving up resources today for the benefit of one’s future self.



Q: How does the chair of economics at U.C. Berkley get interested in personal saving issues and the thinking around retirement readiness and behavioral finance?

I am a game theorist and a decision theorist by training. I was drawn to economics because of the big unanswered questions surrounding financial decision making – such as, who accumulates wealth and why? And how do we make people better decisions makers?  Our tools are creativity, invention and the language of mathematics combined with exponential computing power. 

Experimental data and natural data if gathered in sufficiently large and rich data sets can begin to unlock and explain the actions of human beings and human behavior.  With this, we can help people with retirement, with savings and with the tradeoffs they face in their financial life.

We are not mathematicians, we’re not doing this for the sake of the theory—we are using mathematics and data analytics to help improve financial decision making in the real world. It takes a universal language to solve a universal problem such as retirement and mathematics is that language.  If you give me a better language, I will gladly adopt it.


Q: Can you expand on that? How do you think traditional models that are used in retirement planning have missed the mark?

More and more I became concerned about how many of the tools used to understand individuals are actually wrong, especially in the retirement space.  Overreliance on stated preferences and inconclusive psychology based tools invented by the industry and behavioral “experts” are geared toward entertaining us with our quirks and minimally meeting regulatory standards.  We must do better and reach for more robust and innovative techniques to understand and measure people’s preferences.  Once we understand peoples’ preferences, only then can we advise them.

Many times when I hear criticism about economics, I try to remind people that, economists and psychologists are trying to explain many of the same areas of human and group behavior, Economists, however, are willing to be strong and wrong – in short, we are willing to see our explanations’ of human behavior falsified.  In fact we welcome that.  Once proven wrong, then economists can revise their models and you can identify their mistakes.

We are in a new age and we need to rely on those methods that are scientifically rigorous and allow for continuous improvement and innovation.  I can argue that in an era of smart data, psychology, is not equipped to solve the big problems of retirement and financial decision making.    People cannot remember what they purchased yesterday, let alone accurately state to you their preferences toward risk, time or legacy issues.

The future retirement solutions must be based in science in order to scale across millions of individuals each with unique and differing risk, time and social preferences. Our research is focused on uncovering individual preferences using mathematical models to give us more precise measures of human behavior—to predict human behavior and in turn help retirement decision makers understand themselves and their options.

A big leap forward in game theory and decision theory has been the digital revolution—we have started working with data and collecting huge amounts of data that really illuminate what we were only able to speculate about before. We had some of these ideas earlier, but now we can really test them experimentally and in a scientifically rigorous way.

Working with data is so important in economics—today we are running experiments that do not have to rely on the way people say they will spend and save their money. Instead, we’re actually able to run analyses on real money and the real financial decisions that individuals have made in the past and over time. This is a huge leap forward for the quality of the results.


Q: And what is the data showing us?

Some of the modeling that is the most compelling and explains the most about the way people make decisions is what I like to call a ‘trade-off model.’

Let me explain this: When we look at the way human beings are making decisions, financial or otherwise, we can derive three fundamental trade-offs that they are considering and which really offer a strong framework which we can use to predict the way people will react to certain circumstances. I believe all the decisions we make in our lives, financial or otherwise, are determined by a mixture of these three trade-offs.

So the three big trade-offs are 1) risk potential versus return potential; 2) gratification today versus gratification tomorrow; and 3) the self versus the other—meaning both the self versus other people and the current self versus the future self. 

Let’s think about how this applies to a big financial decision—a question like, how much money should I save for retirement and where should I keep this money? Of course we can see how the risk versus return question plays out—do I want to keep my money in a risk product that could grow quickly and fall quickly? Or do I want to store this under my mattress and try to protect it until tomorrow?

But we also have to consider whether we want to trade off income today for the potential of collecting the income in the future—there is a strong psychological bias to favor the present self, and the ability to overcome this bias is a huge indicator of whether someone will be successful in saving for retirement.  Quantifying what we call “present bias” is not as simple as asking someone “are you present biased” – you must recover if this is true from their decisions.  This is what we do.

The final determinant is whether the individual cares deeply about things like leaving money to the next generation, or saving enough so that the younger generation—family or society in general—will not be burdened in taking care of the older generation.

If you know how a person will come down on these three trade-offs, you’ll have a pretty good idea about how to make them successful in the retirement savings effort.  Generic education of retirees without a good understanding of what we call their Economic Fingerprint is too imprecise—it’s why we lose people.  We need to target educational content or training to help them overcome psychological biases, such as favoring current consumption, fearing losses to the detriment of future outcomes.



Q: What have you learned about how individuals come down on these trade-offs?

Our biggest finding is that people are very heterogeneous and do not fit into the neat buckets that the industry has forced them into.  The existing investor profiling methods are crude and statistically unjustifiable. 

You cannot explain preferences by standard observables such as age, gender, occupation, income or even IQ.  [In the U.S. work force,] if you take a population of any kind you will find as much or even more diversity within that group than across the groups. The industry is missing the chance to better understand investors. 

We’ve learned quite a lot about how people solve the trade-offs from major, academically verified research panels and studies that we have conducted across the U.S., Netherlands and forthcoming in China and South Korea.


Q: How does this apply practically for plan sponsors and advisers? Should they ask these three questions?

Being that I’m a game theorist, it won’t surprise you that I’m a big fan of gamification. That’s why I have gotten involved in a company called Capital Preferences.

Together with the firm, we have created a series of ‘risk and ambiguity games,’ through which individuals are asked to make a series of theoretical decisions, which are structured like retirement investments.

 In each decision you are asked to choose between two investment opportunities, Investment A and Investment B. In the risk game, there is a 50% probability that one of the investments will provide a positive profit while the other returns $0. In the ambiguity game, the probabilities change from a known 50% to an unknown probability that ranges between 20% and 80%. The potential combinations of allocations to Investments A and B are represented on a randomly generated downward sloping budget line where you are asked to pick a point that represents your ‘portfolio.’ The profit from these portfolios is intended to represent a meaningful percentage of one’s overall net worth.

Using this type of a gaming environment helps us get around psychological biases—it helps individuals look at their own situation more objectively. We encourage them just to react to the individual questions and to be as honest as possible—and this is easier because they are not talking about their own wealth, but instead just a theoretical and abstract choice.

By learning peoples’ risk preferences this way, we should be able to build the proper portfolio for their risk needs. 


Q: What other insights have emerged through this psychological and behavioral approach?

I think the key thing for sponsors and advisers to learn is that there is often a really big and really important gap between an individuals’ stated risk preferences and their true risk preferences—what we like to call their revealed risk preferences, which only come to light through looking at historical data and through the gamification approach I’ve already described.

We all know how this happens—we like to share our good intentions when we are asked about financial decisionmaking.

Today far too many plan sponsors and advisers are driving their decisions based on plan participants’ stated preferences, which they gather through a short questionnaire in many cases. We’ve all seen these—they ask the client to report their own preferences.

Now, there are a couple problems here. There are situations in which the client is attempting to manipulate the results—perhaps they have lost money in investments before and they have decided they don’t want to take any more risk in the markets again, so they purposefully answer the questions to make them seem really risk averse, even knowing they could probably take on some risk and be fine.

In other cases, even when the client has no incentive to deceive, they’re just not in touch with their own true risk preferences. They don’t really know how much risk they can take, usually because they don’t know what they’ll need in retirement, so they try to answer out of good intentions. The result is the same—the person ends up with the wrong portfolio allocation.

Especially in the retirement space, thinking about what you may want or need 40 years down the road is next to impossible. One thing we have learned from the data is that people’s fundamental risk outlook doesn’t tend to change that much over their lifetime—while other characteristics about them can change substantially. This is encouraging because if we can get people saving the right way early on, we can hopefully serve them very well and keep them on track over time.