12 Aug 2019

Behavioural economists argue that a person’s decision making depends on a reference point which they use to compare other options. This reference point can be determined by a person’s expectations about future outcomes. For example, if you had expected to go to your favourite Indian restaurant, you will be disappointed if you can’t get a table—but only because you had expected to go there.

While models incorporating these “expectations-based reference points” have been successful in explaining people’s behaviours, such as in the taxi driving industry or in sales, they have generated debate between neoclassical and behavioural economists. Neither “reference points” or “expectations” can be directly observed in standard economic datasets. Therefore, economists cannot conclusively test whether reference-dependence is an important determinant of behaviour in the real world.

In a study published in the August 2019 issue of The Economic Journal, David J Freeman provides a test which shows observable behavioural patterns that are inconsistent with the neoclassical model and indicate the presence of reference-dependence relative to expectations.

To explain this point, consider Anne who is planning a dinner out. She likes an Indian restaurant, Curry House, slightly more than a Thai restaurant, Thailicious, – but if she had been expecting Thai food, she would strictly prefer Thailicious over Curry House.

Anne is planning her dinner and knows it will be an unusually busy day at Curry House restaurant. Which will she choose? There is a 10% chance of being able to get a table at Curry House, but a 100% chance of getting one at Thailicious. Given the improbability of eating at Curry House, she expects to end up eating Thai, which leads her to prefer Thailicious over Curry House. Anne ends up going to Thailicious even if Curry House has space.

Standard economics of rational decision-making assume that if Anne prefers Curry House over Thailicious, she should prefer to make the risky plan with a higher chance of Curry House. The above example shows, through expectations-based reference-dependence, Anne violating the “axiom” of rational choice that lies at the basis of most economic models of decision-making involving risks.

Consider a related example: suppose that there is a second, excellent but consistently busy Indian restaurant called Vij’s next to the other two restaurants. Whenever Anne is expecting Indian food, she would go to Vij’s over Curry House and if she had expected there was some likelihood of getting a table at Vij’s, then the Curry House would feel like a disappointment.

After accounting for this, Anne decides that planning to go to Thailicious is her best option even if she knows that Curry House will not be busy. This is because she would follow through on an expectation of eating Thai, while she would have faced uncertainty and likely disappointment if she had instead expected to eat Indian food.

Therefore, when Thailicious and Curry House were both available but Vij’s was not, Anne would have planned for Indian. When there was some likelihood of the added, excellent option however, her plan would shift to Thai. While this may seem reasonable if the decision-making depends on expectation, it cannot be explained by a single rational preference. Instead, the choice of restaurant is affected by the potential of going to Vij’s when it might be available.

The research shows that these types of behaviour allow economists to uncover a person’s reference-dependent preferences. It pinpoints exactly what evidence economists need to reject standard models in favour of expectations-based reference dependence. Equally it shows what potentially observable behaviour is inconsistent with the model and can be used to refute it.

Expectations-Based Reference-Dependence and Choice Under Risk by David J Freeman is published in the August 2019 issue of The Economic Journal.

David J Freeman

Professor of Economics | Simon Fraser University