Evidence-based economics

Why putting evidence before equations would make economics more like science

October 21, 2010

Economics needs to change. By this I do not mean that the structure of the economy itself needs to change, although reform is clearly needed on that front. Rather, I refer to the academic discipline of economics, which studies the economy and events that take place in it.

After more than two hundred years of theorising, academic economists have failed to settle any of the fundamental controversies in their subject. In microeconomics, the assumption of rational self-interest, core to the discipline at least since neoclassical economics coalesced in the 1870s, has always been open to question. Today this assumption is under fierce attack from other social scientists and from the new school of behavioural economics, which seeks to use psychology to explain economic behaviour.

Similarly, in macroeconomics, the only period of broad consensus was from the 1940s to the 1970s when Keynesian macroeconomics ruled the roost. Before that, myriad schools of thought competed, though non-interventionism dominated. And, since the 1970s, macroeconomics has fractured into various flavours of Keynesian and New Classical thinking that offer wildly different prescriptions for policy.

Not even the scope of the subject is well defined. Many economists today would agree with Lionel Robbins, the British economist who in the 1930s defined economics as ‘a science which studies human behaviour as a relationship between ends and scarce means which have alternative uses’. Using this definition, economists have produced theoretical models that establish what people driven by rational self-interest would do under various circumstances where scarcity prevails.

This view makes economics an ‘imperial’ subject with a claim to be the theoretical foundation of all other social science – but only if its assumption of rational self-interest holds true. The result has been controversy over whether the boundaries of economics are wide, encompassing all social science, or narrow, encompassing just commercial matters.

However, if the assumption of rational self-interest is not sufficient to explain human behaviour (as seems highly likely to be the case) then economics might not even be a very good explanation of commercial matters. Without capturing more of the behavioural vagaries of the real world, it would at best only partially answer the question which originally preoccupied Adam Smith and the other classical economists of the late 18th and early 19th centuries: an ‘inquiry in to the nature and causes of the wealth of nations’. Economics might just be a remote body of ‘rational choice theory’ that needs augmentation with other ideas to be applicable to the real world.

Either way, despite centuries of investigation, everything in economics remains open to debate.

This is worth contrasting with the natural sciences. No physicist seeks to revive the theory that the sun orbits the Earth. Neither do any seek to restore the theory that heat is transmitted by a fluid called ‘caloric’ that flows invisibly and weightlessly from hot bodies to cold ones. Likewise, no biologist seeks to replace Darwin’s theory of natural selection with the earlier theory of the French naturalist Jean-Baptiste Lamarck that evolution occurs mainly through slight acquired characteristics being transmitted to offspring and cumulating to large changes over time.

Scientific controversies do occur. The present controversy over just how severe man-made climate change might become is a well-known example. But scientists are committed to settling controversies through rigorous analysis of the evidence. This is the essence of the scientific method and, once the relevant evidence has been fully explored, controversies are decisively resolved.

Not so in economics. Economic theories are established primarily by abstract reasoning and are only cursorily checked against the evidence. When contradictory evidence does become impossible to ignore, ad hoc hypotheses are often generated to keep favoured ideas alive. In the early 1980s, for example, grafting assumptions that prices are ‘sticky’ and slow to adjust onto ‘rational expectations’ models of macroeconomics re-produced Keynesian results in which the economy can overshoot on either side of full-employment equilibrium – even though the idea of rational expectations had supposedly disproven this.

It could be argued that social science is fundamentally different from natural science and that it is impossible to validate social theories to the same extent as natural ones. Certainly, given the vagaries of society, there will always be more uncertainty in the conclusions of social science. But this does not mean that we should just accept completely contradictory viewpoints sitting side by side at the head of a major discipline. We all live in the same material universe and there can only be one underlying reality.

The solution is evidence-based economics. By sticking closely to the evidence of the real world, we can eliminate theories that do not fit and narrow-down to those which best explain the evidence. It may even turn out that seemingly contradictory theories are each valid in that they apply in different circumstances, but we must still work out what these circumstances are through empirical research.

Evidence-based methods in economics can be applied at any level. On a global scale it is fruitful to compare the periods 1873-1914, 1945-1971 and 1971-2010. These were each characterised by a distinct set of international economic institutions and we can see the effects of these institutions because the world was free from global wars to confuse matters.

The first and third of these periods were economically quite unstable, while the middle period was highly stable in financial and economic matters. It could be argued that the economic stability of the post-1945 era was due to strong growth arising from favourable demographics, post-war rebuilding and the adoption of a back-log of technologies delayed in Europe by the world wars. But the period 1873-1914 had also seen strong forces for growth arising from imperial expansion and adopting the innovations of the ‘second industrial revolution’ (electricity, chemicals, the automobile and so on). In any case, there are many historical examples where strong growth has led eventually to speculation and collapse. So the view that the stability of the 1950s and 60s was due to rapid growth does not hold water.

An alternative explanation is that economic stability from the late 1940s was due to the tethering of currencies to gold and the prevalence of fixed exchange rates. But, again, the periods 1873-1914 and 1945-1971 were alike in that they featured gold-backed currencies and fixed exchange rates. Indeed, these institutions were stronger in the first period than in the second. So the gold standard does not explain stability either.

The real explanation for the impressive stability of the 1945-1971 period seems to be that it was characterised by a set of institutions that included a large share for the state in the economy, controls on cross-border capital flows and tight regulation of the financial sector. These set the period apart form others and seem to be the root of its stability. They also mitigated inequality and probably fostered growth.

But we are still left with the question of whether some elements of this framework mattered more than others. One way to answer such questions is more detailed time-series analysis examining how economic trends changed after elements of a particular economic framework were introduced or removed at different points in time. Another is to examine more closely jurisdictions with regulatory frameworks divergent from the wider consensus prevailing in a given period.

In the years prior to 2007, Australia, Canada and Brazil, with their more intrusive financial supervision, provide such cases.

Further solutions are to trace the ways in which changes to regulatory regimes and other conditions have affected specific firms and sectors in the economy, to look at the behaviour of individual economic actors and to conduct economic experiments in artificial settings.

None of these tools are perfect, but they are much better than abstract equations disembodied from the real world.

Sean Harkin is a financial risk consultant based in London. He previously worked in the field of cell and molecular biology, and is author of The 21st-Century Case for a Managed Economy.