Any of us may be risk-taking in some situations and risk-averse in others, according to experimental research by Graham Loomes and Ganna Pogrebna, published in the May 2014 issue of the Economic Journal. Their study finds that measures of ''risk attitude'' for the same individuals vary considerably between different testing procedures and even within the same procedures when seemingly unimportant things are different.
Imagine that you went to a doctor to have a health condition diagnosed. The doctor offers you a choice of three different procedures to do this. Naturally, you would expect all three procedures to have similar outcomes – that is, they should tell you whether or not you have the suspected condition. And by choosing any one of these three diagnostic procedures, you would expect to get an accurate reading of your health state.
But what if this reading varies depending on your choice of diagnostic procedure and on the context in which this procedure is applied? For example, one procedure could say that you are ill while another may diagnose you as being perfectly healthy. Or perhaps the same test applied at the start of the day gives you a different reading than if it is done at the end of the day. Why might different procedures produce different results? And given this scenario, how do you choose the procedure that would give you the most accurate diagnosis?
In many areas of economics, researchers may use measures of ''risk attitude'' to help explain behaviour. There are several different procedures that could be used to measure risk attitude and these procedures are often assumed to provide similar results. But is this really the case?
In their research experiment, Loomes and Pogrebna test three procedures used in economics to measure risk attitude within the same group of individuals. They find that the readings from the same individuals vary considerably between procedures and even within the same procedures when seemingly unimportant things are different.
This suggests that most of us have imprecise (''noisy'') preferences towards risk. In other words, an individual may be risk-taking in some circumstances or contexts, and risk-averse in others. So care needs to be taken when using different procedures to measure risk attitude, as they are likely to be very context-specific and there is no ''one-size-fits-all'' risk score.
''Measuring Individual Risk Attitudes when Preference are Imprecise'' by Graham Loomes and Ganna Pogrebna is published in the Conference issue of the Economic Journal. Graham Loomes and Ganna Pogrebna are at the University of Warwick.