In December 2017, in a commemoration of the Reformation 500 years earlier, some heterodox economists led by Steve Keen posted “33 theses” on the door of the London School of Economics. These were largely putting forward a positive vision of what economics should be, rather than directly attacking the mainstream. But the main point was that the neoclassical perspective has developed “an unhealthy intellectual monopoly … [that] dominates teaching, research, advice to policy, and public debate”.
The Guardian published an accompanying article by Larry Elliott, one of the signatories, as well as a number of letters. The article made some more direct criticisms of economics, e.g. its claims to be a science. This provoked a number of articles in reply, notably one in Prospect magazine by several important academic economists.
There are good reasons to challenge the current state of economics. But not all the criticisms that are sometimes made are justified. It is important to get the issues right, because misdirected attacks have a number of unfortunate consequences. First, they absolve the real culprit, as in a miscarriage of justice: not only is an innocent one wrongly accused, but the real one is left free to do more damage. Secondly, they encourage counter-attacks that both allow the status quo to be defended, and lead to a dispute about the wrong issues. This in turn provides a misleading basis for trying to construct a better economics.
The relationship of mainstream economists to neoclassical theory
There is a danger of generating more heat than light, on both sides. Heterodox economists tend to see mainstream economics as monolithic – but the reverse is true, it is highly diverse. Other charges are that all economics research is “neoclassical”, and that it is a priori and mathematical, which do not bear close scrutiny – many heterodox economists have an over-schematic view of what happens in academic economics.
Conversely, responses by mainstream economists to these attacks do not typically engage with valid criticisms of fundamental neoclassical theory. Rather, they are phrased in terms of what economists actually do in their personal research and other activities. The Prospect article is a case in point. The authors are doubtless correct in saying that their work makes a valuable contribution to policy on a diverse range of topics, and they defend mathematical modelling as a technique that has the aim of generating useful evidence.
But their position on traditional neoclassical theory is that they distance themselves from it: they do not believe that Homo economicus (that humans are selfish and rational) applies widely – except to corporations; or that markets are perfect and frictionless, citing asymmetric information, frictions and market power. And it is true that few academic economists believe that general equilibrium theory is a good description of the economy – largely because the theorists themselves have demonstrated that the conditions for it to be a good representation of the actual economy are completely unrealistic. Most modern economists are concerned not with abstractions and sophisticated mathematics, they are engaged in data analysis, in gathering evidence.
Thus, the mainstream contains many academics whose opinion of traditional theory is that much of it is wrong. They do not regard it as true, but rather as a starting point, a benchmark – to understand the real-life economy requires major deviations from the textbook view. Doubtless there are still some true believers in the “religious” aspects of neoclassical economics, especially in such areas as the nature and origin of money, where the reliable, evidence-based account has not yet displaced the discredited textbook view. The same is true of much of macroeconomics. But that does not mean that heterodox critics are right to tar the whole of economics with that brush.
And yet some default ways of thinking, derived from traditional theory, are still pervasive throughout the profession. It is no coincidence that the crisis was not predicted – and is still not well understood – by mainstream economists, given that the foundational concept of neoclassical theory is self-clearing markets.
What are the implications for economic theory? How to understand the basic causes that underlie how the economy works? The accumulation of evidence should enable researchers to generate new theory, that then replaces the defective parts of inherited “standard theory” – as it would in the natural sciences. But this has not yet happened systematically.
Even when useful, mainstream research typically does not replace defective theory
The focus on evidence does allow the work of mainstream economists to be socially useful, for example in policy advice. A standout example has been the demonstration that a minimum wage is not as dangerous to employment as was traditionally thought on the basis of standard theory. And that it has a positive impact on living standards which extends beyond its immediate recipients.
Labour economics in current practice is very empirical. But even there, the success of the policy has implications for “theory”, in the sense of the basic account of how the relevant section of the economy (in this instance, the labour market) works. After all, if standard theory predicts unemployment to rise sharply after implementation of minimum wage legislation, and this is not observed in practice, there must be something wrong with the theory.
The neglect of the task of creating new theory based on evidence means that the basic models are still there. In labour economics, this includes the competitive model which implies that non-frictional unemployment cannot occur. Another is the canonical search-and-matching model, which is scarcely more realistic: Pissarides, who was awarded the Nobel Memorial Prize in 2010, stated that “unemployment consists of workers who lose their jobs because it is not to their advantage (and to their employer’s advantage) to continue employed”. This is not a good description of e.g. the closure of a steel works! What is needed is a theoretical account of the economy that explains unemployment as it actually occurs in the real world.
And even if some policy-related work, as with the minimum wage, has managed to escape the shackles of neoclassical dogma, this is not universally so. Some policy advice is still distorted by received notions that derive from traditional theory. For example in the context of predicting the likely effects of “free trade” agreements, the model may assume full employment, even though the real-life adverse impacts of such agreements are likely to include loss of employment.
Traditional theory as a “benchmark”
A widespread view of the place of standard theory is that it provides a benchmark, a starting point. The authors of the Prospect article endorse this view, both in relation to what they call ‘the infamous “homo economicus” theory’, and to the assumption of a perfectly competitive market.
They are undoubtedly correct that models need to be simple – “as simple as possible but no simpler” – this is true in any modelling context, not just in economics. (Their example is the London tube map, which “abstracts from unnecessary details”.) This is a standard view that tends to feature prominently in any defence of mainstream economics. But the issue is not just that textbook theory is over-simple, which in principle can readily be rectified by adding the necessary extensions.
The other problem with economic modelling, as currently done, is that if one starts off with the wrong causal assumptions, without allowing the evidence to challenge them, the whole analysis and its interpretation can be distorted. For example, since the turn of the century, copious quantities of capital have flowed from China to the US. The actual causal relationships can readily be discerned from the evidence, and are widely understood, e.g. by high-quality economic journalists. But the academic literature is rooted in the puzzle set forth by Lucas in 1990: why does all capital not flow from rich to poor countries? This has resulted in the situation where the assumed direction of causation is the reverse of the one that operates in real life. The literature is concerned not with the actual cause(s) of the observed phenomenon, but with explaining why the observations of what happens in the real world deviate from the predictions of standard theory. Lucas’ reasoning was erroneous – although admirably clear – and this means that the search for a correction factor, to bring theory in line with evidence, is doomed from the start. If the first component is wrong, then the second is necessarily wrong too: it seeks to explain a deviation from something that does not exist. This is a double error.
An often-ignored problem with simplified models
Even simplification itself is not blameless. A core component of all modelling is to be explicit not only about what is included, but also what has been omitted. If the omissions are clearly spelt out, the simplification is transparent, and harm is unlikely to result. But in economics, this is not standard practice; researchers typically go straight for a model, with an accompanying story to justify it, so that the broader causal context is obscured.
The classic case, which is widely acknowledged, was the widespread pre-crisis use of DSGE models that omitted the financial sector. It meant that the assumption of frictionless financial transactions was invisible. Economics could learn good practice from natural sciences such as biology and geology, which first develop a broader, empirically-based causal theory. In economics, such an approach could provide a richer, multi-causal view that can be adapted as necessary, for different countries and periods. Models would then be embedded in this broader theory. The contrast is between taking a causal theory as the starting point – a theory that is based on the interlocking of different types of evidence, and that may be quite complicated; or starting with an admittedly over-simple model that can then be extended. The first of these, as used in the natural sciences, has an enviable record of generating secure knowledge, whereas the second has produced patchy results in economics.
Observers outside economics are often surprised that its dismal record in relation to the crisis is not leading to a transformation of the discipline. Among economists, despite lip-service being paid to the need for change, one finds denial and an inability to change – or even to see clearly what needs to be changed. This is coupled with an appeal to the inevitable specialization within academic disciplines: everyone happens to be working on another topic. The authors of the Prospect article sidestep all the issues, merely saying “Like most economists, we do not try to forecast the date of the next financial crisis” – but the issue is not forecasting the date, it is understanding the causes.
The major criterion for accepting theory should be that it accords with the evidence. Everyone wishing to explain the real world should acknowledge this, whether they regard themselves as mainstream economists or not. It does not mean that they need to change what they are doing – as I already said, much of this is socially useful. But it would require a change in awareness.
Sometimes such theory already exists, outside the mainstream, and merely needs to be brought in from the cold, as has been the case with money. This is the argument for pluralism, both in research and teaching. Many heterodox economists have expressed concern at the extent to which viewpoints contrary to the mainstream are systematically excluded, e.g. in discrimination in recruitment and promotion, and in the “leading” journals. That could only be defended as necessary to maintain standards if conventional economics were performing well in its task of explaining how the economy works, which is not the case.
But pluralism is not enough, because it only encompasses already-existing theories. It needs to be extended explicitly to encourage the development of empirically-based new hypotheses. They could be developed using the abundant evidence already available in economics – but this would require prominent journals to become open to publishing empirically-based causal hypotheses. Currently it is only possible to publish papers in a standard format. Ironically (from the viewpoint of evidence-based economics) the status quo favours papers that provide new evidence – but it excludes contributions that attempt to synthesize existing evidence to generate new and better theory.
Admittedly, generating new theory from evidence is not easy – especially if you are used to working in the ways that have become established tradition in economics. But it is not as hard as it may sound. It would transform economics from being a discipline in which abundant good evidence coexists with largely inadequate theory into a truly empirically-based discipline.