The Mirage of Microfoundations
批判了将微观基础引入战略管理和组织分析的项目，指出其忽视了1970年代宏观经济学微观基础项目的失败，存在模糊性和内在不一致，强调社会关系与个体同样重要。
If the project to bring ‘microfoundations’ into strategic management and organizational analysis (Abel et al., 2008; Felin and Foss, 2005, 2006, 2009) simply means that we ‘we stress the need to build microfoundations rooted in individual action and interaction’ (Felin and Foss, 2009, p. 162), then we should applaud it.1 But this simple and valid proposition is complicated by added ambiguities. Strangely, this current project: (a) ignores the failure in the 1970s of the project to build macroeconomics on secure microfoundations; (b) retains damaging ambiguities; and (c) is ultimately inconsistent in its own terms. I deal with these three issues in turn and focus on the insufficiency and rhetorical bias in their core claims and propositions. Mainstream economics has attempted to place economics on individualistic microfoundations. But this project faced insurmountable difficulties and essentially collapsed under the weight of its own internal problems. This episode has lessons for the current proposal. Its chief implication … is that the hypothesis of individual rationality, and other assumptions made at the micro level, gives no guidance to an analysis of macro-level phenomena: the assumption of rationality or utility maximisation is not enough to talk about social regularities. This is a significant conclusion and brings the microfoundations project to an end. A devastating consequence is the breakdown of the types of analysis based on individualistic ontologies where relations or structures are downplayed (Ingrao and Israel, 1990; Kirman, 1989). The microfoundations project in economics was based on the utility-maximizing individual. The work of James Coleman (1990) – which Teppo Felin and Nicolai Foss support enthusiastically – and of their co-author Peter Abell, also involves rational, utility-maximizing agents. Even if we overlook criticism of utility-maximization, there are further lessons from this former microfoundations episode in economics. While it is important and valuable to consider individuals and their psychology, we cannot get far by considering individuals alone. We have to consider relations between individuals as well. All social analysis requires some consideration of social structures, as well as individuals and their motivations. Given this, much ‘microfoundations’ rhetoric is one-sided and misleading. While social relations are always present (even if covert) in social science discourse, they are still not given enough emphasis in economics. If they were there would be much more dialogue with sociology. Organizations are more than individuals: without social relations and social positions they would not be organizations. They would simply be aggregates of isolated individuals. So statement (i) has to be supplemented by the observation (v) that organizations involve social relations, and there is no organization without social relations. Statement (ii) is also valid but inadequate. Ignored is the equally valid proposition (vi) that there are no conceivable causal mechanisms in the social world that operate solely at the individual level. The individual too is made up of component parts with biological, neurological, and psychological causes. The supra-individual and sub-individual levels have also to be taken into account. Although the difference may appear slight, it has enormous consequences. Emphatically (x) has never been achieved in practice: all viable explanations of social phenomena so far encountered involve relations between individuals as well as individuals themselves (Arrow, 1994; Nozick, 1977). By contrast, the content of version (y) is acceptable, because of its inclusion of both individuals and social relations in the explanantia. But consider what it means: social relations as well as individuals are essential to explain social phenomena. Many social theorists define structures as social relations. Accordingly (y) is equivalent to the idea that social phenomena should be explained in terms of individuals plus social structures. Both versions of (y) invoke dual explanantia. So its description as ‘methodological individualism’ is misleadingly biased towards one side of the story. It would be equally biased and erroneous to describe (y) as ‘methodological structuralism’. Neither is the term ‘methodological collectivism’ acceptable. The problem with (y) is of labels rather than content. We always have to start from relations/structures and individuals. There is no other viable explanatory strategy. Biasing the label in one direction or the other is highly misleading (Hodgson, 2007). The overall order of actions in a group is … more than the totality of regularities observable in the actions of the individuals and cannot be wholly reduced to them. … [The] whole is more than the mere sum of its parts … these elements are related to each other in a particular manner … [and] the existence of those relations which are essential for the existence of the whole cannot be accounted for wholly by the interaction of the parts but only by their interaction with an outside world … Like others (Kontopoulos, 1993; Weissman, 2000), Hayek was clear that relations between individuals must be included in our social ontology. Unfortunately the aforementioned advocates of ‘microfoundations’ are imprecise about the nature of the foundations upon which we have to build. I ask them to clarify their position on (v)–(viii), (x) and (y) above. Without such clarification we are invited into a swamp of ambiguity and rhetorical bias. In particular, to be even-handed, it is important for them to acknowledge the validity of statements (v)–(viii), as well as their valid emphasis on the individual. Statement (iv) emphasizes causal mechanisms and says we ‘should fundamentally be concerned’ with ‘intentional human action and interaction’. But why is this individual level ‘fundamental'? As scientists, aren't we also obliged to examine the ‘nuts and bolts’ of individuals as well? Aren't we also required to explain the causes behind individual capacities and intentions, or do we regard these as somehow uncaused, or beyond the reach of science? So far the answer of our authors to the final question is unclear. In another paper, Felin and Foss (2011) argue for ‘free will’, some causal ‘wiggle room’, or ‘indeterminacy’. Hence do they regard intentions as (partially) uncaused, thus beyond the reach of (any eventual) causal explanation? Are intentions entirely caused or not? Personally I follow others in claiming that they are (Bunge, 1959; Hodgson, 2004; Veblen, 1919). Do Felin and Foss by contrast believe in the uncaused cause? Does their rhetorical elevation of the individual among explanantia result from a belief that human intentions are somehow privileged in nature as causally undetermined? Again they are unclear. Science is about causal explanation. And all science tries to explain wholes partially in terms of their components. But complex wholes cannot be entirely explained in such terms. If they could, then there would be no such thing as social science. We would all have to be subatomic physicists, attempting to explain social and other phenomena entirely in terms of the most elementary subatomic particles. Different sciences exist precisely because this goal is beyond our reach: relations and interactions involve a multiple-layered ontology with emergent properties (Humphreys, 1997) and processes that take place through time (Winter, 2012). To settle on the individual as the ‘fundamental’ unit of analysis is problematic. Why stop at the level of the individual? Why not get down to the neural structures in the brain? Or the biochemistry of the human organism? Or atomic physics? The same arguments concerning explanatory reduction from the macro to the micro, and from groups to individuals, apply equally to explanatory reduction from individual to gene, gene to molecule, molecule to atom, and so on. If we can reduce explanations to individuals, then why not further reduce them? To avoid this double standard one must either accept multiple levels of analysis, or reduce everything to the lowest ontological level – the realm of physics. If we admit multiple levels of analysis then there is no reason to assume that the social realm has one level only. Organizations involve emergent properties that render them additional units of analysis, while fully acknowledging that their very existence depends on individuals – just as the existence of individuals depends upon atoms. Although Abell, Felin, and Foss point to some genuine limitations of the routines literature, their proposed ‘microfoundations’ strategy is unclear and misdescribed. They rightly stress the importance of understanding individual psychology and motivations, but this point has already been stressed by many authors (including myself) and we do not need biased ‘microfoundations’ rhetoric to establish this point. Accordingly, when Linda Argote and Yuqing Ren (2012) identify ‘transactive memory systems’ as part of the ‘microfoundations’, they do not claim that they exhaust the explanantia. David Teece (2012) does not have to reduce explanations entirely to individuals to make the point that individual entrepreneurship can be vital. Michael Cohen (2012) points to the crucial role of habit – much in line with my own work in this area (Hodgson, 2004, 2008) – but does not claim that habit is the sole unit in terms of which everything can be explained. ‘Microfoundations’, if it is a meaningful and operative term, cannot mean complete micro-reduction. Abell, Felin, and Foss underplay the role of relations or structures in explanation and are unclear whether explanations can or should be in terms of individuals alone. Alternative accounts of routines do exist that are more in line with reputable philosophy of science.