Month: March 2019

Doughnut Economics (5): Understand emergent behaviour. Oh, and what externalities are

/ rolfgroeneveld / 2 Comments

The fourth of Kate Raworth’s “Seven Ways to Think Like a 21st-Century Economist” is to replace the old, rather Newtonian model of economics by one that treats the economy as a complex adaptive system. There is a lot in this chapter I can agree with, but my enthusiasm is rather dampened by her embrace of the Suzuki Fallacy that in economics, “externality” means “irrelevant”.

Economists’ role model: nineteenth-century physics

In this chapter Raworth makes a rather convincing case that the efforts of nineteenth-century economists like Walras and Jevons to make economics the physics of the social sciences still linger in ways that are less than pretty. Before I read Doughnut Economics it had already struck me how articles in top economics journals seem to try to emulate theoretical physics with their complicated lemmas and formal proofs, and I often cite Jagdish Bhagwati’s joke that nice economists reincarnate as physicists whereas nasty economists reincarnate as sociologists. Rather ironically, Raworth cites Friedrich Hayek, the original neoliberal, as a prominent critic of the emulation of physics. Hayek features more than once in this chapter, and I can’t help wondering whether Raworth might have credited him more if only she had overcome her aversion to his neoliberal legacy to seriously consider his ideas. But I’m getting ahead of the story here.

In a sense, her criticism of this Newtonian economics falls in the same trap as the Newtonians themselves. General equilibrium is the wrong model, she says. Its assumptions are too stringent, too unrealistic, and in any case the Sonnenschein-Mantel-Debreu theorem has shown that in a system with multiple agents and multiple interdependent markets there is no such thing as a single equilibrium. I had to google for the theorem (hey, I never said I’m a expert) but I’m not so worried by it. Both the Newtonians and Raworth appear to be looking for the right model of an economy, like phycisists are still looking for the right model to explain such things as mass and gravity. But in economics, as in any science dealing with very complicated systems, all models are wrong, and so is the general equilibrium model. So “wrong or right?” is the wrong question.

Unlike physicists and, apparently, some economists, scientists in such fields as ecology, meteorology, and climate science aren’t bothered with finding the ‘right’ model: they know their models are gross simplifications of reality. Few ecologists would waste much time proving theorems of their models, or pretend that those models are accurate representations of reality. But like the computable general equilibrium models used by applied economists, their models help understand complex systems, and evaluate policy. As George Box said, all models are wrong but some are useful.

Externalities – oh puh-lease!

There it is, on page 143:

At the same time, drop the economist’s beloved notion of ‘externalities’, those incidental effects felt by people who were not involved in the transactions that produced them – like toxic effluent that affects communities living downstream of a river-polluting factory, or the exhaust fumes inhaled by cyclists biking through city traffic. Such negative externalities, remarks the ecological economist Herman Daly, are those things that ‘we classify as “external” costs for no better reason than because we have made no provision for them in our economic theories’. […] Far from remaining a peripheral concern ‘outside’ of economic activity, addressing these effects is of critical concern for creating an economy that enables us all to thrive.

Her characterisation of externalities as “effects felt by people who were not involved in the transactions that produced them” is fairly accurate, but the suggestion that economists ignore them because there is “no provision for them in our economic theories” or that “externality” means “not in the economy” is nonsense of a degree that makes one understand why so many economists dismiss the entire book. How can Raworth get this so spectacularly wrong?

And yes, I’m being defensive here. That is because my field, environmental economics, studies hardly anything else than externalities. Externalities are central to environmental economics, because they are market failures that can do a lot of real harm. There is a lot you can say about the behavioural model underlying the concept of market failure, including externalities. It treats people as incentive robots who cannot change their attitude and just blindly follow whatever impulses the market gives them. It has become a common excuse (among economics majors, especially) to act selfishly: sorry, you gave me the wrong incentives! But the idea that economists call environmental pollution an externality because they prefer to ignore it is the number one nonsensical canard popular among lazy thinkers who prefer to attack a caricature of economics rather than the actual thing. And it’s a shame she is perpetuating this myth, because it puts off a lot of people who otherwise may have learned about her more sensible ideas.

Complexity, emergence, and two liberal thinkers

In a sense, the chapter’s title (Get savvy with systems) is misleading: any economic model is a representation of a system. Raworth’s alternative is no different in this respect, and as I argue above, she falls into the same old trap of presenting her model as the final accurate representation of economic systems.

But there is an interesting difference between the two approaches. Mainstream economic models usually have only one equilibrium and one optimum (the two of which might not be the same – that is one way of understanding market failure). This property is based on how their costs and benefits depend on production levels. Marginal costs increase: in other words, the sacrifice needed to attain the next thing gets bigger with how many things we already have. Meanwhile, marginal benefits decrease: the more things we already have, the less we appreciate the next thing. Naturally, there comes a point where we have so many things that the costs of the next thing equal its benefits – voila, nous avons a market equilibrium.

Economists do acknowledge that things do not always work this way. Some technologies have network externalities: the more people are on Facebook, the more friends a new Facebook user can connect with through his newly opened account. This is probably one of the reasons why the Dutch company Hyves eventually stopped its network and switched to online games. Paul Krugman won his Nobel Memorial Prize for his contribution to a theory of international trade based largely on economies of scale: by specialising in a particular product, countries can make it at lower costs than would be possible if they had broadened their portfolio. Both network externalities and scale economies have the property that even if external conditions are the same, different starting conditions or even sheer coincidence can give very different outcomes (although I doubt Hyves would ever have beaten Facebook).

Ecologists have known for a long time that ecosystems can have such path-dependency. Ecosystems can also suddenly flip from one state to another. Shallow lakes, for example, usually have either clear water with much submerged vegetation and high densities of predatory species such as pike, or very turbid water with limited vegetation and a high density of bottom-feeding species such as bream. Like with scale economies, positive feedbacks drive this property: turbid water favours bream, which by its feeding behaviour maintains turbidity.

Ecologists have taken this idea much further, however. Rather than systems with a handful of equilibria, which can be described by a handful of equations, many ecosystems are best described as assemblages of a large number of individual agents, each with simple decision rules, but whose joint behaviour can produce highly complex outcomes. Such outcomes are difficult to predict from the properties of the individuals and emerge from the system when we just let it run its course.

The models that describe such complex adaptive systems, usually called agent-based or individual-based models, are not used as often in economics as in ecology. Ironically, to find emergent behaviour in economic thinking we need to look at two economic thinkers who have been most criticised by Kate Raworth and others. Adam Smith’s invisible hand was a metaphor to describe how a form of order emerges from a seemingly chaotic process. Friedrich Hayek described similar processes in his writings. Interestingly, however, I see few of these insights translated into quantitative models. Every now and again you come across agent-based models in the economics literature, but they are certainly no mainstream economic tools. I suspect the problem with such models is that it is difficult to distil generalisable insights from them, while that is precisely what most academic economists are looking for. And to understand why they do so we need to go back to economists’  tendency to take physics as a role model.

Previous posts

Next post