In a recent article, new World Bank chief economist Paul Romer writes that macroeconomics has become trapped in a similar stasis to fundamental physics, each field being dominated by a single failing theory. However, in fundamental physics, string theory is an elegant potential unifier of quantum mechanics, particle physics and general relativity, while in macroeconomics, neoclassical theory fails to account for the most striking observation: that capitalism is prone to crisis.
String theory imagines that there is one fundamental ingredient in the universe, the string, and that each fundamental particle is a different mode of vibration. Amazingly, it is capable of producing a universe with physics as rich as our own. The trouble is that string theory has a vast number of solutions, each corresponding to a possible universe. There is no way of knowing why we live in our particular universe and not one of the alternatives. Thus string theory has shifted the question “Why this particle spectrum, with these sets of properties?” to “Why this shape for the universe as opposed to some other shape?”. It is hard to currently imagine, but one day a new idea may move string theory forward. There are a few alternative fundamental physics theories, but they do not do any better than string theory when confronted with the unanswered questions.
In macroeconomics, on the other hand, there is striking empirical evidence against the mainstream neoclassical theory and its assumption of general equilibrium: the occurrence of crises such as those of 2008 and the Great Depression. As Romer points out, the standard response is that external “shocks” cause any behaviours not predicted by the theory. But if there is always the potential for unpredicted shocks to suddenly occur, then the theory isn’t very useful. A complete theory of macroeconomics must predict and explain crises, especially given their catastrophic impact. However, unlike in physics, and overlooked by Romer, there do exist alternative theories with promise to explain what the mainstream cannot.
A fundamental failing of neoclassical macroeconomics is that it ignores the effect of money and debt on the economy. In a recent paper Steve Keen plotted debt change against unemployment in the US from periods containing “normal times” and economic crisis (including the Great Depression and the 2008 crisis). The correlation is very high in all cases. The same was found for debt acceleration. While these data contradict neoclassical theory, they support alternative theories with endogenous money, i.e. theories that correctly posit that commercial banks create money out of “thin air” when they make loans. The real effect of endogenous money on the productive economy is a key pillar of the heterodox post-Keynesian school of economics.
Keen’s Minsky-inspired mathematical macroeconomic model captures, via just three concise differential equations, key features of the data in the run up to 2008, notably a period of decreasing volatility, followed by a sudden debt-induced crash. While aspects of the model can be critiqued, and it can’t actually make precise quantitative predictions, it could be taken as a useful starting point for new theory. Differential equations provide the natural mathematics for a truly dynamical system that does not merely fluctuate around equilibrium. It is striking that most economists have not been trained in differential equations.
Romer goes into detail on the “identification problem”, namely that of successfully estimating model parameters accurately enough for a model to be useful. He correctly concludes that there is a tendency to overestimate the accuracy of parameter estimates. This issue generalises to heterodox models and will always make reliable quantitative prediction hard. Nevertheless, if a model can reproduce and predict certain macroeconomic patterns then it can be deemed useful, especially in cases for which model behaviour is not sensitive to precise values of parameters. Held to this standard, there is good potential for progress in macroeconomic theory, as exemplified above.
It is interesting to compare string theory and macroeconomics because they refer to complete opposite ends of the spectrum of systems one can study. String theory is relevant to the most microscopic level, while arguably macroeconomics relates to the most macroscopic level, describing macro level properties of a system composed of possibly the most complex possible components - human beings. There is something understandable about the desire to concoct a theory with simple building blocks at the most fundamental level. On the other hand, it is hard to fathom how macroeconomists continue to be so obsessed with the notion of equilibrium, and mostly eschew insights from studies of other complex systems.
One could make an analogy between Newtonian physics and neoclassical economics (note here I drop the word macro and consider all levels). There is evidence against Newtonian physics in certain scenarios, notably for very heavy objects, and the mathematical framework is insufficient for describing complex systems. At the macro level, the fabric of spacetime is curved; Einstein’s theory accounts for this, and Newton’s theory fails. Neoclassical theory can trace its roots back to Smith and Ricardo in the 18th and early 19th century, arguably the “Newtons” of economics. The theory is successful in certain scenarios. At the micro level, considering a single product being sold in a large world with static background economic conditions, an equilibrium can be reached between supply and demand. At the macro level, however, the distribution of money matters, and economic agents operate in a climate of uncertainty, and these features arguably lead to a dynamical, complex system, typically operating away from equilibrium. Post-Keynesian theorists attempt to address all this.
To conclude, there is indeed trouble with both (fundamental) physics and macroeconomics. However, the nature of this trouble differs, and perhaps Romer could have been kinder about the accomplishments of string theory and the potential in (heterodox) macroeconomics. It is hard to see a way forward for physics given that it is currently impossible to experimentally probe a scenario in which the theorised breakdown of quantum mechanics and general relativity occurs. But none of the rival theories has accomplished anything string theory hasn’t. In macroeconomics there is empirical evidence against the mainstream theory, but there exist at least the seeds of alternative theories that can explain data that the mainstream theory cannot. There is thus much greater potential for progress, and it is time for neoclassical economists to engage more actively with alternative theories. The field can, and eventually will, do much better. Romer’s article is a positive step.
Adam Barrett is based at the University of Sussex and holds an EPSRC Complexity Science Fellowship. He has a PhD in string theory from the University of Oxford, and is currently working on a project on the complexity of financial and ecological stability.