Firms, agency, and evolution

Abstract

A recent (though controversial) trend in economics has been to appeal to evolutionary theory when addressing various open questions in the subject. I here further investigate one particular such appeal to evolutionary biology: the argument that, since markets select (in a standard biological sense) firms as coherent units, firms should be seen to be genuine economic agents. To assess this argument, I present a model of firm/office selection in a competitive market, and show that there are cases where markets can select for firms/offices as collective units – and thus, as agents of their own – but also that there are cases where they do not. In this way, I try to make the evolutionary argument for the agency-based view of the firm more precise.

Keywords:

• evolutionary economics
• evolution
• biology
• group agency
• methodological individualism

Acknowledgements

I would like to thank Jason Alexander, Geoffrey Hodgson, Kelly Schulz, and three anonymous referees for this journal for useful comments on previous versions of this paper.

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

^1. Of course, the general idea here is old – see e.g. Veblen (1898) and Schumpeter (1942).

^2. Note that there are also many other questions surrounding the nature and workings of firms (Alchian, 1950; Coase, 1937; Enke, 1951; Nelson & Winter, 1982; Nickerson & Zenger, 2004; Radner, 2006; Satz & Ferejohn, 1994; Schmalensee & Willig, 1989). These, though, are not so important here.

^3. Note also that Nelson and Winter (1982, chap. 4 and pp. 124–128) argue that human agents can be subsumed under a similar model as well: they are embodiments of skills. If that is so, then this view would have the implication that there are no genuine agents at all that it is worth tracking in an economy.

^4. Note that this question is related to the plausibility of ‘methodological individualism’: should we make room in the social sciences for genuine higher level (group) agents, or should we only permit agency at the level of the individual person (Watkins, 1952)? However, since even the formulation of the latter thesis is controversial (Hodgson, 2007), I prefer to phrase the issues here without reference to the latter term.

^5. For example, Hodgson and Knudsen (2010, pp. 170–171, footnote 9) write: ‘Organizations as here defined have the capacity for goal-directed behavior, irrespective of whether goals are actually declared. In this sense, an organization has the capacity to be a “collective actor” (Knight 1992, 3).’ They then note that ‘[t]he importance of […] firm-specific capabilities and learning effects mean that the firm often has the necessary cohesion to qualify as an interactor’ (Hodgson & Knudsen, 2010, p. 173) – where an interactor is (following Hull, 1988) defined as an entity that ‘interacts as a cohesive whole with its environment in such a way that this interaction causes replication to be differential’ (Hodgson & Knudsen, 2010, p. 165, see also 167). Turning this around, this seems to imply that if firms qualify as interactors, then that is (probably) due to the fact that they have the kind of cohesion that gives them the capacity to be a collective actor. However, the textual situation here is not entirely clear, in that Hodgson and Knudsen (2010) also seem to want to embrace Nelson and Winter's routine-based picture of the firm (see e.g. Hodgson & Knudsen, 2010, pp. 173–179). Fortunately, assessing this is not so important here. The goal of the present paper is to discuss an interesting evolutionary biological argument for the agency-based view of the firm; whether this argument is also the one suggested by Hodgson and Knudsen (2010) can be left open here.

^6. At times, some evolutionary economists also suggest that their aim is more heuristic in outlook (Hodgson & Knudsen, 2010; Vromen, 2001; Witt, 1999). In particular, they note that they are merely interested in providing an evolutionarily grounded ‘how possibly’ story about how it is possible for firms to be agents of their own (for more on ‘how possibly’ stories, see Brandon, 1990). However, since it was already known that it was possible (in some sense at least) for firms to be economic agents of their own (Bacharach, 1999; Foss & Klein, 2008; Richardson, 1972; see also Hodgson, 1999, chap. 11), this does not appear to be a particularly interesting argument to make (Schulz, 2013). Hence, I will not consider this further.

^7. It is important to note that MLS2 does not require that the fitness of a group is completely unconnected to any of the features of its parts. The point is just that group fitness is measured in terms of the expected reproductive success of groups – not of the individuals making up the groups. See also Okasha (2006).

^8. There is an alternative route to the same conclusion. In particular, one can follow Godfrey-Smith (2009, pp. 39–40), and, in all cases, define evolutionary biological individuals as units of selection. This leads to the same conclusion as the one in the text, since, on this picture, cases of MLS1 do not (in general) constitute cases where groups really are units of selection – for example they are often to be seen merely as instances of frequency- or (social) environment-dependent selection of individual organisms. Either way, the point is that one needs to distinguish group-level influences on evolution in the MLS1 sense from those in the MLS2 sense. See also Okasha (2006).

^9. Note that much the same worry has been raised about the possibility for ecosystems to evolve (Godfrey-Smith, 2009; Sterelny, 2001; Sterelny & Griffiths, 1999, chap. 8).

^10. In this way, the model sits between a purely instrumentalist account of economic modelling and a fully realist one: the claim is not that the model ‘proves’ that firms sometimes are economic agents, but nor is the idea just to show that it is sometimes possible to model firms as economic agents. The claim is that the fact that it is sometimes possible to model firms as economic agents is evidence for them sometimes being economic agents – and that we can say in more detail when, exactly, this is the case.

^11. Alternatively, in terms of the routine-based approach, there is a chance for the behavioural routines supported by the expectation-exceeders to fail to be maintained within a given firm.

^12. For more on decentralised organisational structures, see Radner (1993), Carley (1996), Witt (2003), and Gindis (2009).

^13. This thus mitigates the concerns raised in Hirshleifer (1977) and Frank (2003): the present model presents a case where it is at least possible to make sense of the idea of office reproduction, rather than just firm growth. For a good recent overview of some of the work on firm spin-offs, see also Klepper (2009). See also below.

^14. This was done using NetLogo; the code is available at http://tinyurl.com/ofurolb

^15. These results have also been confirmed with a midlevel analysis that simultaneously varied r between 1, 4, and 7, E₁ between 0.01 and 0.41 in 0.1 increments, and E₂ and E₃ between 0.4 and 0.8 in 0.1 increments. The details of the results are available at http://tinyurl.com/pp4ec62

^16. This last fact should not be overemphasised – it is a straightforward consequence of the fact that this is a selection-based model with a monotonically increasing fitness function. Note also that there can be long periods where several firms coexist in the market.

^17. The number of expectation-exceeders there are in the initial pool of applicants (i.e. p_E) also matters, but it seems primarily to concern the length of time it takes before a firm establishes itself in the market.

^18. Here it is also important to recall (see also note 7) that MLS2 does not require group fitness to be completely independent of any property of the lower level (here, it is dependent on the number of expectation-exceeders in an office).

^19. In this way, the present model partly contradicts and partly supports claims made in Enke (1951), Hirshleifer (1977), and Frank (2003): there are cases where talking of firm reproduction does seem plausible. See also Klepper (2009).

^20. It is worthwhile noting that the model displays, in the first instance, office selection. Firm selection is merely a by-product of office selection. This, though, does not alter anything of importance for the present discussion.

^21. Of course, the model can also be used – perhaps in conjunction with other models – to investigate various other questions, for example concerning economic growth or innovation (see e.g. Nelson & Winter, 1982). As noted earlier, this is not the aim in the present context, but of course, there is nothing speaking against using the model for these other purposes as well.