What Do We Need Homo Oeconomicus For?

A few days ago, the celebrity among German economists, Hans Werner Sinn, published a short piece in the Süddeutsche Zeitung in which he defends economics against common and, according to him, mistaken criticisms. I won’t take issue with all 6 points he raised (invisible hand, ecology vs economics, Keynesianism vs neoclassical economics, competition, neoliberalism and homo oeconomicus), instead focusing on the last one. I had started writing a post on this anyway, so Sinn’s commentary comes just-in-time. The main question behind my criticism of Sinn’s presentation of the problem is: what do we need homo oeconomicus for?

Let’s start by reminding ourselves what homo oeconomicus is: it is a behavioural model, a fictitious persona who possesses perfect information or can at least acquire as much of it as he needs; always acts so as to maximise self-interest based utility, does not let society influence her preferences, which are given and complete; her choices are always consistent and can be used to identify her “preference orderings”; her choices are based on prices only, as prices reflect all relevant information about goods and services. She is perfectly rational.

I once wrote a post here in which I criticised the behavioural model of homo oeconomicus as reflecting economist’s implicitly normative ideas about how people should think and act, rather than being an accurate (positive) description of how we do behave. Our preferences are not consistent, they are not stable, they depend on the social environment we are in, which does not allow us to be 100% selfish, even in case we are inclined to this. Our knowledge about the world around us is inherently limited and we perceive it in a systematically biased way. Prices are poor information providers for many goods and services. Some things do not have any. And so on.

Prisoners’ Dilemma.

Now, economists like this model a lot. At times, their love for it leads them to very strange comments on real-world human beings:

Evidently the run-of-the-mill players are not strategically sophisticated enough to have figured out that strategy DD is the only rationally defensible strategy, and this intellectual short-coming saves them from losing.

This is a quote from a landmark book on the game theoretical problem known as Prisoners’ Dilemma. It is a reaction of the authors on results from economic experiments in repeated prisoners’ dilemma games, in which actual people obviously didn’t arrive at the socially irrational, Pareto inferior Nash equilibrium predicted by economic theory (strategy DD).

What has Mr Sinn to tell us about homo oeconomicus? He argues that the idea behind using homo oeconomicus is not to make predictions about actual behaviour, but to show a counterfactual – how would a policy work if people were rational? Such an approach, it is claimed, helps to avoid paternalism. However, the paternalism argument only holds for those aspects of the homo oeconomicus that are considered “bad” – mainly, selfishness (and even then only under the assumption that the model still holds when people are not selfish). Meanwhile, many of its aspects are “good” things that people equally lack, as e.g. a rational treatment of uncertainties or non-conformity. The design of policies on the basis of such assumptions is paternalistic, as we know that people are not as rational (in the positive sense of the word), but implicitly ask them to be so if the so-designed policy is to work out.

Nonetheless, Sinn is right in pointing out that homo oeconomicus is just a model. Like any model, it constitutes a simplification of reality. The point of a model is not to be “realistic”, but to be able to provide explanations of reality that match our actual experiences and stand up to empirical tests. There are areas in which homo oeconomicus is useful as behavioural model (especially in relation to markets for private goods – much less so, however, when public goods are the object of analysis). Thus, I don’t think – at least not any more – that it is a good idea to discard it altogether.

The problem with homo oeconomicus, however, is that its use is extremely undifferentiated. It has been applied virtually everywhere, from the analysis of financial markets to “markets for marriage”. However, in areas in which knowledge limitations are important; in which social conventions play an important role; in which uncertainty and ignorance are prevalent; in which markets are imperfect or absent; in which power is distributed unequally; etc. – the “basic” homo oeconomicus is useless, or even harmful, as when analyses based on it are used to inform public policy. Indeed, one of the “founders” of homo oeconomicus, Francis Edgeworth, explicitly restricted its applicability to war and contract, deeming it inappropriate in other areas. In cases going beyond the “domain” of homo oeconomicus, there is a need to use a broader perspective. As one of my colleagues said: you may start with homo oeconomicus but then think about which extensions and modifications of this basic model are sensible in the given context. In some contexts, minimal modifications might be sufficient – in others, they may be so far-reaching that the end-result would have very little in common with the “economic man”.

Often, critique of homo oeconomicus – including my own – tends to be too undifferentiated. Critics make the same error as those economists and other social scientists who stick to the basic homo oeconomicus model, in that they implicitly assume that there is another, better, but equally one-size-fits-all behavioural model. Alas, there is none. Depending on the specific exercise, very different behavioural models can be useful, ranging from the basic all-rational homo oeconomicus to other, completely different “homini“.

7 thoughts on “What Do We Need Homo Oeconomicus For?

  1. The homo oeconomicus is a strange tool indeed. I think in terms of complexity/predictive value-ratio it is probably the most powerful behavioral model out there…

    I mostly agree with the paternalism-accusation. The HO-model is unproblematic only in a ‘negative’ context where policy is designed to prevent people from doing something (e.g. theft). Then applying HO is in effect to reckon with the worst case: The penalty must be strict enough to prevent a perfectly informed (sentence, benefit of theft), purely self-interested (maximal ‘income’) etc. being from undertaking this illegal action. It would obviously be careless to assume that everyone refrains from stealing simply due to moral reasons.
    By contrast, in a ‘positive’ setting where the question is not what action is to be prevented but what should be done, recommendations based on the HO-model quickly become problematic / paternalistic. A popular case in point is the use of depletable ressources, where the model leads to recommendations such as using up all resources during our lifespan or to exploit resources until marginal damages and benefits are equated regardless of minimum requirements for the stability of ecosystems. There is indeed the danger to shield such policy-recommendations against criticism from the environmental movement by categorizing it as irrational or based on ‘immature’ preferences.

    By looking closer, however, we see, just as you wrote, that this is not the model’s fault but rather the fault of its inappropriate application. Because with minimal adjustments the HO yields totally different, fully sustainable recommendations: If we include the benefits of future generations from using depletable resources the model tells us not to use any of them at all.

    It is interesting that such results are typically disregarded as “artefacts” or “not applicable in the real world” and that even ecological economists then resort to undoubtly paternalistic concepts such as merit goods. To me the homo oeconomicus model provides exactly the correct answer: If we really care for future generations it is most rational not to use any depletable resources at all. This is of course not feasible if we don’t intend to live like miserable peasants before the 18th century. Nevertheless, it puts our nose to the fact that we leech off future generations with virtually every step we take. Or that a rational resource use is impossible. Not bad for such a simple model I would say.

    • It’s interesting that your interpretation of my paternalism argument is different from my own, while our above mentioned colleague proposed another one. I am not entirely sure whether yours and mine are really essentially different, so I will write down the three and invite you to brainstorm with me;-) Here they are:

      1. My own version: It’s not paternalism when the policy analysis is based on the selfishness assumption behind HO (since we are assuming that it works even if people are “bad” – if they are “good”, i.e., unselfish, even better). It is paternalism when policy analysis is based on, e.g., the assumption of rational treatment of uncertainties or of non-conformism (e.g., no herding behaviour), as then the policy only works if we are behaving in such a “desirable” way – if we don’t, it will fail. So we are in a sense asked to behave rationally, HO gets a normative component.
      2. Your version (as I understood it): The distinction paternalism vs. no paternalism is dependent of the goal of the policy that is being analysed, i.e., whether it is meant to prevent people from doing something (HO “OK”) or whether it is meant to push us towards some desirable behaviour (HO not OK). This seems quite similar to my perspective, I am not sure, however, whether they are identical.
      3. Our colleague’s version: When the main assumption is the selfishness component of HO, this might create a tendency on the side of the government to engage in paternalistic actions to overcome this supposed selfishness. On the other hand, if the assumption is some of the “positive” components of HO, the State might feel no need to intervene and thus let us more freedom.

      My intuitive impression is that all three versions might be compatible with one another because they emphasise different aspects of the same problem. E.g., the seeming contrariness of 1 and 3 may be due to the fact that in 1 it remains clear that HO is just a model and the results of the policy analysis are the source of paternalism (which policy is chosen), while in 3 the delineation between model and reality seems to be blurred and it is the model itself (taken as a true representation of reality) is the source of paternalism. What do you think?

      • Thanks for the synopsis. I agree with your rationing under all three points, just not with your definition of paternalism in 1:
        (a) Agreed that a policy to discourage destructive behavior which is developed under the selfishness assumption should “work”. Even the most sophisticated destructive acts are anticipated by the rule. Mr. 305 also points this out in his example of theft.
        (b) Agreed that a policy to encourage constructive behavior which is developed while overestimating positive aspects of our rationality should “fail”. Imagine there is public pressure to adopt a policy for a mandatory insurance because research revealed that people underestimate certain risks. Assume this research is correct. If the details of the policy are now drafted under the assumption that peoples risk assessment is (nearly) perfect, the policy will achieve (nearly) nothing to address the issue and “fail”.

        However, a “working” policy does not necessarily indicate the absence of paternalism and a “failing” policy does not have to be paternalistic. In (b), a well-functioning policy would be based on a realistic assessment of people’s average abilities. Let’s assume this lies between the extremes of a perfect and a non-existent ability to assess risks correctly. This well-functioning policy would be appropriate for a large share of the population. To the smaller share of people rational enoughto buy insurance anyways, it would however be unnecessarily coercive. While this policy would clearly “work”, it is also clearly more paternalistic than the “failing” policy described in (b).

        The policy described in (a), on the other hand, “works”. However, a government that, for example, assumes all internet users to be psychopathically egoistic super-criminals (the negative interpretation of the homo oeconomicus) might decide monitor their behavior comprehensively. While this policy fulfills its purpose, most would likely consider it “invasive”. I would not call it “paternalistic” as it does not try to protect people from themselves, but the effect is very similar.

        So I am wondering, where our interpretations of paternalism or of the employment of the homo oeconomicus concept in policy making differ. I would interpret Mr. Sinn’s argument about showing a counterfactual as using the homo oeconomicus concept to identify one extremum of the continuum of positive descriptions of human rationality. This interpretation also seems to be present in Mr. 305’s argument to use the homo oeconomicus as a worst-case scenario: “The penalty must be strict enough to prevent a perfectly informed (sentence, benefit of theft), purely self-interested (maximal ‘income’) etc. being from undertaking this illegal action.”

        Is this positive interpretation the reason you characterize my argument, which you summarized in number 3 of your synopsis, as blurring the line between the model and reality? To me, the argument in 3 is based precisely on the contrast between a government, which bases its policy decisions on the assumption that people to behave according to a homo oeconomicus model, and a reality, in which they don’t. This hypothetical government is, of course, also a type of counterfactual.

  2. However, a “working” policy does not necessarily indicate the absence of paternalism and a “failing” policy does not have to be paternalistic.

    Of course not. That wasn’t my point. My point was that when a policy is designed on the basis of, say, the assumption that people can deal with uncertainty perfectly (or at least better than they actually do), its failure will be due to our “imperfectness” – from which the conclusion may be derived that we should do better, then the policy is going to work. For me, this is (implicit) paternalism. On the other hand, when a policy is designed on the basis of the assumption that we are perfectly selfish, there is no room for paternalism in the interpretation of its success/failure (because there is no implicit “be better if you want our policy to work” – if it doesn’t, the HO model was just not appropriate, it cannot be easily argued that it is “our fault”, because no sane person would argue that we should behave selfishly*). But you are right in pointing out that such a policy might well be invasive – which, however, is a different matter.

    Thus, I think that you identify paternalism in policy – I identify it in (tacit) interpretation of policy (failure). I think that these interpretations are not exclusive because they focus on slightly different things.

    * Although, when you look at the quote from Rapoport and Chammah (see above)…

    • I am glad we solved this. Your account of the difference in our interpretation of paternalism makes perfect sense to me. I guess when I think of paternalism, I am more worried of the state actually intruding into decisions I am capable of making, than of government’s (tacit) disappoinment that I am not perfect ;)


Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s