Economics should rediscover reason

The following comment by me appeared in the latest Fund Strategy (20 April).

The two chapters of the International Monetary Fund’s twice-yearly World Economic Outlook (WEO), published last week, provide interesting pointers to the character of the current downturn. They also exemplify the limitations of contemporary economics.

One chapter looks at the likely strength of the recovery while the other examines the spread of the crisis from developed to emerging economies via financial linkages. Both are based on divining patterns from numerous sets of historical data.

The first chapter shows that recessions associated with financial crises tend to be severe. Globally synchronised downturns, such as the present one, tend to last longer and are typically followed by weaker recoveries.

The other published chapter shows that financial linkages are a key channel of transmission through which economic downturns in advanced economies hit the developing world. Typically, capital flows to the developing world fall for a protracted period during such recessions.

However, there is a fundamental problem with the approach pursued in the WEO. Correlation does not prove causation. Just because two things tend to happen at the same time it does not prove that one causes the other. Although this principle is central to courses on statistics, it tends to be forgotten in the real world of economics.

Take a simple example. Most people with blonde hair probably have blue eyes. But this does not mean that being blonde causes people to have blue eyes. To prove there is a relationship, it would be necessary to do studies on the genetic level.

Confusing correlation and causation can easily lead to incorrect conclusions. For example, British children of families of Bangladeshi or Pakistani origin tend to do relatively poorly at school.

It would be easy, although wrong, to conclude that being Bangladeshi or Pakistani explained poor educational attainment. In fact, the explanation probably lies more in class. Bangladeshi and Pakistani migrants to Britain tend to come from poor rural areas. Migrants from India, in contrast, tend to be more middle class and do better at school.

Such confusions abound in economics and many other areas. Correlations are frequently, and often wrongly, assumed to suggest causal relationships.

The obsession with statistics in contemporary economics suggests a fear of rationality. Rather than logically examining relationships between different developments, many economists seem intent on using ever more muscular statistical tests.

Statistics are useful but they should only be the starting point of economic analysis. To go further demands the application of reason to unpick the true character of economic relationships.