The Economist as outsider

Economics needs to change. Many academic and professional economists have been re-evaluating the subject since 2008, reflecting a widespread view within the economics profession that the financial crisis and its aftermath underlined its serious shortcomings. Employers say they can hire economics graduates who are technically able and can manipulate models, but that these graduates are wholly unable to apply what they have learnt in any real world context, have no practical data skills and are unaware of context or recent economic history. Changes must therefore be made to the way economics is taught, applied and developed.

Economics often cares more about logical rigour than reality, despite the increasingly desperate attempts by reality to get economists to pay attention. We put models, attempts to make sense of the world by including only relevant detail, at the heart of our methodology. A good model is a powerful tool for analysis and prediction. However, economists who accommodate reality into their models, for example by using rules of thumb with no micro-foundations, will often have their models dismissed by their peers. Models which are built on supposedly rigorous micro-foundations are flawed too since such models are lacking in sufficient real-world evidence. Nonetheless, these models are the ones used for practical applications.

So economists need to rely less on models alone. We need to supplement the analysis far more with narrative approaches, both from economic history and from other social sciences such as anthropology and sociology. A combination of models, field experiments, narrative accounts of the cultural context and history relevant to the experiment could add up to a genuinely powerful approach to economics. Economists must realize that evidence can consist of much more than data or statistics. Evidence does not even have to be quantifiable. Some things of value are not only non-monetary but wholly unquantifiable, such as freedom or citizenship. We economists should concentrate on our comparative advantage in analysis and empirical measurement, but we should also be prepared to supplement it with narrative and an acknowledgement of the unquantifiable.

The problem is that too many economists are just not very good at the kind of empirical work we should be doing. One frequent mistake made by economists is to test hypotheses against a specific counterfactual or alternative hypothesis. Even when they do explicitly consider them, many economists will fall into the trap of comparing a policy with a completely unrealistic alternative. There are, embarrassingly, many other examples of poor empirical work in economics, such as omitting variables and ignoring simultaneity and non-stationarity. Many economists are also remarkably uncurious about statistics – how they are constructed and adjusted, and how that might limit the conclusions they can support – and remarkably cavalier about making strong claims on the basis of weak econometrics.

To complicate matters further, interpreting economic evidence is not a simple business. In economics we are typically trying to test hypotheses about a small number of variables in a complex world, with a huge amount of feedback and simultaneity. This must be done using a relatively small amount of data, of dubious quality. Neither the current enthusiasm for randomised control trials, nor more realistic assumptions in economic models makes a difference to the sheer difficulty of the empirical challenge. There is a danger of over-fitting economic models and of preferring inaccurate precision to the accurate imprecision that would more properly characterise noisy data. Therefore trials and experiments, like other empirical methods, need to be undertaken with due care and humility in the face of our epistemological uncertainty about the world.

All this is not to despair of the project of turning economics as a whole into a more soundly-based empirical science. We don’t have nearly enough data. We don’t interpret it with sufficient care and we don’t use evidence to inform theory enough either. Economists do need to take both data and methodology far more seriously as a practical day-to-day matter. But there is still hope for economics. A new economics curriculum has been developed which several universities could start piloting from autumn 2014; economists are steadily embracing the behavioural findings of psychologists and cognitive scientists about people’s preferences; economists and economic policymakers are beginning to acknowledge the enthusiasm for randomised control trials and field experiments. Therefore, although the field of economics needs much reform and regeneration, economists can learn and grow from the mistakes which culminated in 2008 to develop economics into a balanced, modern study.