Every year around this time, the Royal Swedish Academy of Sciences awards the Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel—aka the Nobel Prize in Economics.
This year, the Prize went to Abhijit Banerjee, Esther Duflo and Michael Kremer “for their experimental approach to alleviating global poverty.” The three economists pioneered the use of randomized controlled trials to improve the living standards of poor people in developing countries.
A good example of this experimental approach can be found in Duflo et al (2012). In the early 2000s, Duflo and her two coauthors, Rema Hanna and Stephen P. Ryan, carried out a randomized experiment to test the effects of financial incentives on teacher absenteeism in India. The experiment was a success: introducing incentives in teachers’ salaries resulted in absenteeism falling by 21 percent in the treatment group.
It should be noted that the use of controlled experiments in economics isn’t very common. As stated by Nobel-Laurates Paul Samuelson and William Nordhaus in the 1985 edition of their Principles of Economics manual,
“economists cannot perform controlled experiments like chemists or biologists because they can’t easily control other important factors. Just like astronomists or meteorologists, they usually have to solely use their observation”
In effect, experiments are difficult to implement in the field of economics due to the impossibility to control and isolate all factors influencing the variable of interest. As a result, economists need to resort to other empirical methodologies to gain a better understanding of economic phenomena.
Most economists use complex statistical models to try to detect causal relationships between variables using real-word data. For example, we can develop a model that estimates the impact of inflation on unemployment (or vice versa). Nonetheless, we must bear in mind that these models need to have a solid theoretical framework to support them. Otherwise, results can be misleading. Furthermore, statistical models tend to be very sensitive to the initial assumptions. Small changes or omissions in the model can lead to different or even contradictory results.
Despite the prevalence of such techniques in the profession, the last decades have witnessed the emergence of important research based on the use of controlled experiments as well other alternative methodologies. One of the pioneers was Vernon Smit, who was awarded the Nobel Prize in Economics in 2002 for the use of controlled experiments in the study of market mechanisms.
Nobel laurates Daniel Kahneman’s and Richard Thaler’s research on behavioral economics is also based on controlled experiments. One of their most popular experiments is the dictator game. A person is given $20. He can either give the other player $2 and keep the rest, or split the $20 evenly. The experiment yielded surprising results: most participants chose the second option, undermining the idea that economic agents always seek to maximize their own utility.
Elinor Ostrom, the first woman to receive the Nobel in Economics, didn’t use experiments but case studies as a research method. Ostrom examined hundreds of communities to understand the governance of natural resources in these communities. She found that communities tend to develop their own rules in order to make an efficient use of their resources and avoid the so-called Tragedy of the Commons.
The use of controlled experiments and case studies isn’t without problem. Case studies are difficult to generalize due to the particularities of each case. Similarly, experiments may work in the controlled environment where they are performed, but they aren’t always transferable to other settings.
Whether it be through statistical models, controlled experiments, case studies or any other methodology, empirical evidence is essential in the study of economic phenomena. In this sense, social sciences aren’t very different from natural sciences: theories must be tested against the real world.