The science of choice – a new model for policy makers?

A little known branch of economics can help governments to set budgets and make policy decisions that are more in line with the will of the people.

When San Francisco was building a new rapid transport system in the 1970s, a young economist from the University of California at Berkeley set out to predict how many people would use it.

Using data from commuters and a mathematical model he had developed himself, Daniel McFadden worked out that just over six per cent would travel on the new train. His figure was well below the official forecast of 15 per cent, but it proved to be remarkably accurate.

McFadden’s work addressed something that had puzzled scientists for years – how to understand human decision making and predict consumer choices – and it won him the Nobel prize. His technique was soon adopted by transport authorities around the world and so a new branch of economics was born.

More than 40 years later, choice modelling remains an obscure and little understood discipline, even amongst skilled mathematicians and economists. However, it is widely used in market research to help companies decide what features to incorporate in a new product and how to price it. As the science reaches ever greater levels of sophistication, it is also finding new uses in public life - helping governments decide how best to allocate budgets and make policy decisions and even hold politicians to account.

Australian academics are at the forefront of choice modelling. Associate Professor Len Coote, Acting Dean of UQ Business School, together with Professor Jordan Louviere and Dr Campbell Rungie, has developed a new, more general and flexible model which offers deeper insights.

“Choice modelling tells us what features are important to the user and what they are willing to pay for them,” Associate Professor Coote explains. “It can tell manufacturers how best to configure products, optimise pricing and forecast demand.

“It can also help guide policy and strategy as we can ask what people are willing to pay for a change in the system. As it allows us to place an economic value on a policy change, even when there is no market, we can discover, for example, what patients value in a free health service or how much voters would be pay to protect the Great Barrier Reef.”

Choice modelling has its origins in the work of 1920s psychologists such as Louis Thurstone who tried to understand individuals’ choices and develop models to take account of apparently random decisions – why would someone who prefers vanilla ice-cream occasionally choose strawberry? Then in the 1950s, economists took an interest but focused on the impact of broader economic choices of populations of decision makers.

McFadden drew on this body of work to develop the first tractable choice model. Associate Professor Coote explains: “McFadden found a way to extend the psychologists’ work beyond individuals and build models to represent the preferences of samples of decision makers. His original model is widely used by policy makers and provides an overall picture decision makers’ preferences. However, it assumes that everyone has the same preferences. Since his early work, McFadden and other economists have put much effort into understanding individual differences. Our work closes the loop because, in addition to the aggregate preferences, it also gives you a deep understanding of the individual differences. Policy making can be improved if the nature of these differences is known.”

Their mathematical model extends McFadden’s model and also incorporates a model developed by Swedish mathematician Karl Jöreskog which psychologists use to study individual differences. It also allows different data sources to be combined which can help reveal subtle links between decisions made at different times and in different contexts.

“In addition to studying what decision makers want, we can study how decision makers differ and why,” adds Associate Professor Coote. He is currently working on a project with cystic fibrosis sufferers. These patients who would once have been expected to die in childhood can now live to 40 or 50 years old and their lifestyle has a major impact on their longevity.

The work attempts to understand what influences their willingness to follow treatment plans. Associate Professor Coote says: “Our model allows us to separate out the various sources of influence — the patient themselves, their doctors or friends and family — revealing in simple terms the otherwise complex interplay of the influences. If we know what compels people to follow a treatment plan, we can better educate them and develop more thoughtful policy responses.”

Governments are increasingly using choice modelling to inform policy decisions, in particular on environmental issues. Typically people are asked how much they would be willing to pay to fund improvements or enforce legislation. In one study in Tobago, where seawater pollution is a major problem, researchers found snorkelers were willing to pay more than non-snorkelers for cleaner water, although both groups felt strongly about the risk of ear infections, development on the coast and the amount of plastic waste. They were able to make clear policy recommendations — that the island should focus on reducing the health risks, ensure proper planning controls and improve waste collections.

Meanwhile public health services have discovered its potential to reflect people’s preferences in their budget decisions. A study on spending on health innovation in the UK found people preferred the money to be used to help people with disability, cancer and asthma but not drug addicts or the obese.

Choice modelling can offer a far more comprehensive picture of public opinion than a traditional ‘yes or no’ referendum. Following the ‘Brexit’ vote in the UK, a study by King’s College London found that people wanted to reach a deal with the EU – in contrast to Prime Minister Theresa May’s comment that ‘no deal for Britain is better than a bad deal’. It showed the perceived benefits of the UK being able to make its own laws and not pay into the EU budget would be outweighed by the need to obtain a visa and work permit in an EU country and the costs of trading outside the single market.

Associate Professor Coote says: “Wherever there is a decision to be made, choice modelling can help make us understand people’s preferences, including their political preferences. It can make the process of allocating resources far more in line with the true will of the people and hold politicians to a much higher standard of accountability.”

Last updated:
25 March 2019