There is no such thing as a sunk cost fallacy

The advocates of the sunk cost fallacy state that, since an agent ponders in his decisions marginal costs against marginal incomes, any consideration upon sunk costs would be irrational. Notwithstanding, as soon as we accept the arguments of the said sunk cost fallacy and try to put its recommendations into practice, we discover that we have just become an easy prey of a more severe kind of irrationality: the one that concerns with intransitive preferences.

Jon Elster exemplifies the sunk cost fallacy with the case of a huge snowfall that pours onto the city the very same day we were planning to attend a theatre play whose tickets we had bought the previous days and are not refundable. Elster points out that, since our attendance to the play will not bring the money we had paid for the tickets back, there is no reason to make the decision on whether or not to attend the play on the basis of the sunk costs of the tickets. The correct reasoning should take into account only the cost of enduring the heavy snowfall in order to reach to the theatre where the play would be performed. Nevertheless, the same Jon Elster makes the disclaimer that a zealous observance of avoiding the sunk cost fallacy could lead to make choices following non transitive preferences.

If we change our mind every day, discarding previous decisions and assuming a new direction just because a new opportunity has arisen, we risk to end up in the ruin. Transitive preferences tend to assure the agent of a certain profit and non transitive ones exposes him to losses. In evolutionary games simulations, agents who act according to transitive preferences outshine agents who do not. It seems, then, that the rational agents walks on the edge of the razor, between sunk cost fallacies and non transitive preferences.

That is why there is not such a thing as a sunk cost fallacy. The rational agent, to be such, must ponder a whole plan against an alternative plan in a whole as well, which in some cases, both of them last several periods of time. It is true that in the “very short term” all past costs are sunk and that it only matters the opportunity costs, but most decisions are made in the short term, which lasts more than just a moment. Otherwise, the very concept of transitive preferences would lack any meaning.

Of course certain costs are sunk: if the flux of earnings that a good of capital produces just covers the variable costs of putting it to work (for example, a truck whose earnings just pay for the gas and the salary of the driver), the more rational choice is to use it until it becomes full obsolete and do not replace it with a brand new unit.

But the sunk cost fallacy does not provide a criterion to distinguish sunk costs from just mere costs of a single plan. What a rational agent with transitive preferences discards in his considerations will be named sunk costs, and what he does not, will not. A pure tautology.

Even the snowfall case does not explain satisfactorily the said fallacy: when the agent bought the tickets, their cost were inferior to the income of watching the play, but a heavy snowfall adds not a marginal cost but increases the marginal cost of the plan composed by the cost of the tickets plus the cost of enduring the snowfall.

Notwithstanding, the sunk cost fallacy derives into a philosophical puzzle: what is the subject? How are relations between time and being and between being and becoming. It seems that our permanence as rational agents depends mostly upon not to put into practice the opportunistic approach of the sunk cost fallacy ad libidum.

Moreover, the matter has a political strand: constitutional constraints demand from the authorities to take into account the weight of certain principles in their decisions and those principles could be disregarded if the decisions are purely made on the basis of expediency. If it is the same authority the one who decides whether certain constitutional principle should be followed or not, then all the citizens would be left exposed to arbitrariness.

The considerations about the length of the period a plan should last, the responsibility upon the consequences of our past choices, and the weight of the constitutional principles on the legitimacy of political decisions, become rational if they are not pondered by an isolated agent but in the framework of the interplay among several agents.

This framework of human interaction upon which the agent’s choices take place had been characterised by Friedrich A.Hayek as a spontaneous, or abstract or extended order. He proposed to leave the term “economics” to the explanation of the choices made by an isolated agent and to establish the science of “catallaxy” as the study of the complex phenomena involved in the said structure of interactions. In the same line, James M.Buchanan labelled the interplay of individual agents as “symbiosis” and proposed to redefine the task of the political economy to its study.  More recently, in 2009, Douglass C. North, John Joseph Wallis and Barry R. Weingast, in Violence and Social Orders: A Conceptual Framework for Interpreting Recorded Human History, coined the term “open access orders” to analyse the same set of events. To this stream of thought, it also belongs Vernon L. Smith’s own account of the concept of ecological rationality.

Catallaxy, Symbiosis, Complex Phenomena, and Open Access Order or Ecological Rationality are some of the aspects of what Karl Popper once called “critical rationalism” and supersedes old problems such as those of the instrumental or subjective reason. An authentic “toolbox,” ready to be used.

What makes robust political economy different?

17760989_10211701794931604_8034787751519057151_o

I encountered what would later become important elements of Mark Pennington’s book Robust Political Economy in two articles that he wrote on the limits of deliberative democracy, and the relative merits of market processes, for social and ethical discovery, as well as a short book Mark wrote with John Meadowcroft, Rescuing Social Capital from Social Democracy. This research program inspired me to start my doctorate and pursue an academic career.  Why did I find robust political economy so compelling? I think it is because it chimed with my experience of encountering the limits of neo-classical formal models that I recount in my chapter, ‘Why be robust?’, of a new book, Interdisciplinary Studies of the Market Order.

While doing my master’s degree in 2009, I took a methodology course in rational choice theory at Nuffield College’s Center for Experimental Social Science. As part of our first class we were taken to a brand new, gleaming behavioural economics laboratory to play a repeated prisoners’ dilemma game. The system randomly paired anonymous members of the class to play against each other. We were told the objective of the game was to maximise our individual scores.

Thinking that there were clear gains to make from co-operation and plenty of opportunities to punish a defector over the course of repeated interactions, I attempted to co-operate on the first round. My partner defected. I defected a couple of times subsequently to show I was not a sucker. Then I tried co-operating once more. My partner defected every single time in the repeated series.

At the end of the game, we were de-anonymised and it turned out, unsurprisingly, that I had the lowest score in the class. My partner had the second lowest. I asked her why she engaged in an evidently sub-optimal strategy. She explained: ‘I didn’t think we were playing to get the most points. I was just trying to beat you!’

The lesson I took away from this was not that formal models were wrong. Game theoretic models, like the prisoners’ dilemma, are compelling and productive analytical tools in social science, clarifying the core of many challenges to collective action. The prisoners’ dilemma illustrates how given certain situations, or rules of the game, self-interested agents will be stymied from reaching optimal or mutually beneficial outcomes. But this experience suggested something more complex and embedded was going on even in relatively simple social interactions.

The laboratory situation replicated the formal prisoners’ dilemma model as closely as possible with explicit rules, quantified ‘objective’ (though admittedly, in this case, low-value) payoffs, and a situation designed to isolate players as if they were prisoners in different cells. Yet even in these carefully controlled circumstances, it turns out that the situation is subject to multiple interpretations and understandings.

Whatever the textual explanation accompanying the game, the score on the screen could mean something different to the various players. The payoffs for the representative agents in the game were not the same as the payoffs in the minds of the human players. In a sense, my partner and I were unwittingly playing different games (although I lost within either rules of the game!).

When we engage with the social world, it is not only the case that our interests may not align with other people. Social interaction is open-ended. We do not know all the possible moves in the game, and we do not know much about the preference set of everyone else who is playing. Indeed, neither they nor we know what a ‘complete’ set of preferences and payoffs would look like, even of our own. We can map out a few options and likely outcomes through reflection and experience but even then we may face outcomes we do not anticipate. As Peter Boettke explains: ‘we strive not only to pursue our ends with a judicious selection of the means, but also to discover what ends that we hope to pursue.’

In addition, the rules of the game themselves are not merely exogenous impositions on us as agents. They are constituted inter-subjectively by the practices, beliefs and values of the actors that are also participants in the social game. As agents, we do not merely participate in the social world. We also engage in its creation through personal lifestyle experimentation, cultural innovation, and establishing shared rules and structures. The social world thus presents inherent uncertainty and change that cannot be captured in a formal model that assumes fixed rules of the game and the given knowledge of the players.

It is these two ideas, both borrowed from the Austrian notion of catallaxy, that makes robust political economy distinct. First, neither our individual ends, nor means of attaining them, are given prior to participation in a collective process of trial and error. Second, the rules that structure how we interact are themselves not given but subject to a spontaneous, evolutionary process of trial and error.

I try to set out these ideas in a recent symposium in Critical Review on Mark Pennington’s book, and in ‘Why be robust?’ in Interdisciplinary Studies of the Market Order edited by Peter Boettke, Chris Coyne and Virgil Storr. The symposium article is available on open access and there is a working paper version of my chapter is available at the Classical Liberal Institute website.