Saturday, December 10, 2016

Understanding Hayek's Knowledge Argument (2): Uncertainty and the Narrative Fallacy

(Previous Entry)

Last time we looked at Hayek’s knowledge argument. The argument that makes the case in favour of free markets and against central planning. The logic was simple enough: Society must solve its distributional problem (who makes what, when and for whom) and to do this it must solve a knowledge problem (figure who wants/needs what, whether it can be made, when and by whom). Hayek argued that central planners are bad at solving this knowledge problem. They cannot access all the relevant information and are prone to biases and distortions. Conversely, Hayek argued that free markets are good at solving this problem: the price mechanism amasses relevant information and communicates it to market actors in a simple, easily-decodable form (namely: the price of goods and services).

At the end of the last post, we noted some limitations to Hayek’s argument. He believed that the price mechanism worked best when it was undistorted by government interference (e.g. as when a government sets an artificial price ceiling or price floor). He also never argued that prices could contain all the information relevant to market decisions. Indeed, his whole argumentative framework presumed that some individuals would have a localised informational advantage over others, which they could exploit in the short term, but which would eventually be detected and reflected in the market price. This makes Hayek’s view quite distinct from modern, efficient market theories of price which, in their strongest form, claim that market prices reflect all relevant information.

There was another limitation to Hayek’s argument that was briefly mentioned: that market behaviour can itself distort price. This is a more serious criticism of Hayek’s position and warrants an extended treatment. I want to deal with one form of this criticism in the remainder of this post. It’s from Richard Bronk’s article “Hayek on the Wisdom of Prices: A Reassessment”. I think Bronk presents a very interesting argument about the role of storytelling in markets and how this can distort price signals.

1. How Markets Distort Prices
The storytelling argument is one part of Bronk’s critique of Hayek. Although it is by far the most interesting part, I think it is worth starting with the more general critique. In rough form, that critique works like this:

  • (1) The free market can only address the knowledge problem if free market prices are undistorted over the long-term.
  • (2) Free market prices can, on at least some occasions, be distorted over the long term.
  • (3) Therefore, the free market cannot, on at least some occasions, address the knowledge problem.

As you can see, the argument is hedged quite carefully. As I mentioned in part one, it is unlikely that any system can ‘solve’ the knowledge problem. Certainly, Hayek never claimed this. We can never have all the knowledge relevant to making market decisions. New information is being discovered all the time, and novel production processes are being implemented. This is a good thing. Nevertheless, some systems are better at amassing and communicating the relevant information than others. That’s all the knowledge argument claims. This criticism is pushing back by suggesting that the market doesn’t do a good job on at least some occasions.

Now let’s look at the premises of the argument. The first premise focuses on distortion over the long term. Why does it do this? The reason is that the Hayekian knowledge argument —like the famous ‘invisible hand’ argument of Adam Smith — was quick to highlight the self-correcting powers of the market. To Hayek, the free market was a distributed mess of actors, working off tacit and subjective knowledge, responding to localised variations in price, and discovering new information when possible. One would expect the distributed mess of actors to result in price signals that oscillated or varied relatively frequently. Sometimes this might result in price distortions (as when one market player spots an arbitrage opportunity that was missed by his peers), but one would expect these distortions to be ironed out over the long-term. It is only if the distortions persist that a problem really emerges. It was obvious to Hayek that certain forms of governmental interference (e.g. legally prescribed price floors) had a long-term distorting effect. The question here is whether the emergent behaviour of distributed market actors can result in similarly long-term distorting effects.

That’s what the second premise claims. Bronk raises several lines of argument in favour of this premise. They are:

  • (4) In some markets, prices are produced by narratives/stories about the future and these narratives/stories are prone to distortion.
  • (5) Wealth inequality means that a limited number of actors have an excessive amount of control over market prices: this can result in longer term price distortions.
  • (6) Market prices do not reflect all the relevant costs (or benefits) of market activity (the externalities problem).
  • (7) Information asymmetries between market actors can lead to price distortions.

Some of these reasons will be familiar to economists or, really, anyone who has taken an introductory economics course. The externalities problem highlighted by premise (6), for example, is widely-known. The classic illustration involves the environmental damage that results from some productive processes, e.g. the downstream pollution that is caused by a factory that dumps toxic materials into the local river. The costs that result from the pollution are often not reflected in the factory’s unit production cost. This means they charge a lower market price than they really should.* Legal rules are often used to correct for externalities of this sort. They might allow those downstream from the factory to sue for the pollution costs. The cost of litigation might then be factored into the unit production cost. There is an immense field on scholarship about this in law.

Likewise, premise (7) is already a feature of the Hayekian argument. Hayek liked the fact that there are information asymmetries between market actors. It is the fact that one party knows more than another that makes it (a) so hard to centrally plan economies and (b) makes the price signal so useful (because the asymmetries eventually get reflected in price). He thinks this is a feature, rather than a bug, of the free market. Bronk takes him to task for this. He thinks Hayek’s position ignores just how much of a distorting effect these asymmetries can have.

But I’m not going to say anymore about premises (6) and (7). I think what Bronk has to say in favour of (4) in particular, and to some extent (5) is far more interesting. In fact they end up being connected, as we shall see.

2. The Narrative Fallacy in Market Prices
The term ‘narrative fallacy’ has a known meaning (I believe it comes from Nassim Taleb’s work). It refers to the fallacy of assuming that events are causally connected into neat explanations or ’stories’ when they may be chaotic and disconnected. People commit the narrative fallacy all the time. They look at a successful individual, note some obvious or identifying features of their lives, and assume these events are part of their success story. They forget that others may share those features and not be successful, and that an individual life is made up of thousands of events, all of which might be somehow relevant to their success. They favour the neat, attractive success story, over the complexity of real-world causation.

Bronk doesn’t use the term ‘narrative fallacy’ in his article, but his worry about price distortion seems to be premised on it. The difference is that where most people talk about the narrative fallacy retrospectively — i.e. about how we spot narrative patterns in historical data that are spurious or misleading — Bronk talks about it prospectively — i.e. about how our storytelling often projects into the uncertain future.

He argues that Hayek’s claims about the knowledge-amassing powers of the market underplay the role of radical uncertainty in human life in general and in some markets in particular. The reality is that we don’t know the future very well. Sure, there are some events that we can predict with certainty. If you get an astronomer in the room, she will be able to predict the occurrence of the next solar eclipse with remarkable accuracy. But other aspects of human life are much more uncertain.
This is particularly true of social events. Look at the recent political turmoil in the UK and the US. If you asked people two years ago whether Donald Trump would be president or whether the UK would leave the EU, you would have received lots of different answers, all phrased in varying degrees of certainty and uncertainty. Some people tried, hard, to put figures on the probability of these events, and some of those estimates were more accurate than others, but many were a long way off. And if you asked the majority of people what they based their predictions on, it probably wouldn’t have been some careful parsing of data or some probability-infused model, it would have been some simple ‘story’ or ‘narrative’ they spun about the current political climate and how it would project into the future.

Uncertainty plays a big role in markets too, particularly in markets that deal with expectations about the future. Debt and equity markets are the obvious examples. When you buy stocks and shares, you are usually making predictions about future prices of those stocks and shares. When you loan someone money, you are making a prediction about their likely ability to repay the loan in the future. People operating in debt and equity markets often build sophisticated statistical models to account for that uncertainty (they try to convert the uncertainty into risk), but those models often create an illusion of knowledge where none resides. In reality, players in those markets are dealing with what Bronk calls ‘ontological uncertainty’ and they replace knowledge with hopes and imaginings:

When faced with ontological uncertainty, everyone is feeling his way forward with little firm indication of what the future will bring. With no single correct vision of the future out there to anchor expectations, and a vast space of possible outcomes, prices reflect not so much our decentralised knowledge as the way we imagine or hope the future to be… 
(Bronk 2013, 95)

These hopes and imaginings in turn become stories about the future (“the market is going to go up and up because…”) and these stories end up guiding market behaviour. But stories are not knowledge.

This narrative fallacy might not be too much of a problem if all the players on the markets told their own stories about the future and the diversity of the stories would thereby prevent any large, systematic long-term distortion in prices. But this isn’t what happens. Some stories become conventional wisdom and they start to dominate the minds of the individuals that determine market prices.

Bronk argues that some markets that involve lots of technological innovation and change are particular susceptible to this problem. The innovation changes the conditions under which the market operates, and creates further ontological uncertainty about the future. Nowhere was this more true than in the financial market in the lead-up to the crisis of 2008:

The decade leading up to the crisis was characterised by massive financial and policy innovation, which led to high levels of ontological uncertainty [reference omitted]. In such conditions, investors duly relied on convention and new era stories, which in turn engendered widespread confidence. 
(Bronk 2013, 96)

We did not have the distributed cognition that Hayek wanted from the market. Instead, we had groupthink and ‘groupfeel’. This distorted the price mechanism in a significant way.

This is where the link with wealth inequality comes in (premise 5). Wealth inequalities often translate into power inequalities. Those with more money have greater access to and influence over the market storytellers. A good example would be the power that large financial corporations had over ratings agencies at the time of the 2008 crash. The ratings produced by the ratings agencies functioned as storytelling devices. They helped to support and confirm the conventional narrative about the future value of mortgage-backed securities. But the agencies were paid for these ratings by the corporations who created the instruments being rated. They were also advised by these corporations as to how the instruments worked. As a result, the corporations had a significant influence over the narrative.

3. Conclusion

To briefly recap, Hayek extolls the messy, distributed, knowledge-amassing powers of the price mechanism. Lots of individuals, acting for profit-seeking reasons, discover information that is relevant to the production and supply of goods and services. They use this information to set prices, and others respond to those prices in ways that allow the price to incorporate more information. The market price thus becomes an information communication device par excellence, addressing the knowledge problem that plagues central planning.

But this is only true if the price mechanism is undistorted. It can obviously be distorted by government interference. Can it also be distorted by the emergent behaviour of actors on the free market? That’s what Bronk has just argued. In particular, he has argued that in markets involving uncertainty as to future prices, storytelling tends to replace knowledge. What’s more, certain stories become conventional wisdom. These stories can have a significant distorting effect on the market. This reduces the scope and effectiveness of the Hayekian argument.

*Note: the reverse can happen too, i.e. downstream benefits do not accrue to the manufacturer when they probably should. Intellectual property laws are, arguably, attempts to address this positive externality problem.

No comments:

Post a Comment