Learnings from the lab: Average and maximum bias in marginal reasoning

Background

In addition to partnering with credit unions, fin-tech companies, and non-profits, we also strive to be a thought leader in the space by conducting studies in the lab. One such study provides a strong theoretical framework for we make budgeting decisions between categories.

Study on Effective Budgeting

There is evidence that, when people budget, they do not do it rationally (Kourilsky & Murray, 1981). That is, they don’t work towards a budget that gets them as much happiness as possible for each dollar they spend. In this study, we investigated whether people budget in a rational way, and how we can help them to budget more judiciously.

Rational budgeting requires that people think about the benefits and costs of small changes. Suppose I have a weekly budget, and I have $300 for food and $100 for entertainment. To determine whether this is a good budget, I should consider the benefits and costs of small changes in the budget. If I take one dollar from my food budget and put it in my entertainment budget, what is the cost of eating less well, and what is the benefit of having more entertainment? If the benefit of having more entertainment outweighs the cost of eating less well, I should transfer the dollar from my food budget to my entertainment budget. And I should continue to transfer dollars until it is no longer true that the benefit of having more entertainment outweighs the cost of eating less well. Economists call this marginal thinking, thinking about small changes.

We hypothesize people make errors in marginal reasoning, and we look for solutions. We hypothesize that people don’t think about the impact of small changes, but that they just think about how happy consumption in either category makes them on average. Moreover, we hypothesize that consumption episodes that make people extremely happy have too much influence on their budgeting decision. To test this hypothesis, we ran a series of experiments on budgeting across a number of categories.

Experiment

In a series of budgeting games, subjects allocated money to two budget categories. Some subjects focused on how small differences in budget allocations would impact their happiness (“marginal utility”), which is the correct way to do it. However, we tempted some subjects to focus on how happy consumption in either of these categories makes them on average (“average utility”). And we tempted others to focus on consumption episodes that made them extremely happy (“maximal utility”).

In the budgeting games, subjects receive information on the (diminishing) return to buying items in 2 categories (gems and precious stones), they make a binding decision on the allocation, and earn money depending on their items.

Results

In the Control condition, most people made a rational budgeting decision (the error was $0). In contrast, when there was a decoy (high average or high maximum utility of items in the suboptimal budget category), many people made a fully irrational budgeting decision (the error was $3).

In the Control treatment, there is a clear mode for $0 error in budgeting. In contrast, in all the other conditions, the distribution is bimodal: While $0 error is still a peak, there is a second peak for fully irrational behavior (an error of $3).

This provides evidence for an average bias and a maximum bias in marginal reasoning. This, according to our knowledge, is a novel finding in behavioral economics. What it means is that when someone is considering to transfer some money from his food budget to his entertainment budget, he is probably not thinking about small changes: about how much happiness he would lose from consuming a bit less food, and how much happiness he would gain from a bit more entertainment. Instead, he might be thinking something like: “Oh, I really like food, because I love my Friday night restaurant visits with my friends! I want more money in my food budget!”

These Friday night restaurant visits, however, should not determine his budgeting though, because it is not something he would want to cut (it is too nice to cut), and it is also not something he can add more of (because there are only so many Friday nights on a month). He should think about things like his sporadic afternoon snacks: this is something he might cut, or add more of. But it doesn’t easily come to mind.

The results of this study are groundbreaking and we plan to partner with budgeting companies like MoneyComb, to create better budgeting tools for consumers. MoneyComb is an app that helps consumers reduce their spending, while increasing their happiness by focusing consumers spending on the things that matter to them. Now we know which errors people make, and we have a platform to study these errors, we can test solutions.