Evolutionary Economics and the Counterfactual Threat

 

Robin Cowan* and Dominique Foray**

April, 1999.

 

 

*MERIT, University of Maastricht, PB 616, 6200 MD Maastricht, The Netherlands.
r.cowan@merit.unimaas.nl

**IMRI, Université Dauphine, 75775 Paris Cedex, France.
dominique.foray@dauphine.fr

 

Abstract

We thank the participants of the International Seminar on Evolutionary Economics as a Research Programme in Stockholm, May 1997, for many helpful comments. We also thank Lorri Baier for many helpful substantive and textual comments.


L'Economie Evolutionniste et la Menace Contre-Factuelle

 

Les spéculations de nature contre-factuelle sont partie intégrante de la démarche scientifique. Ce texte examine la nature de l'exercice contre-factuel et les conditions dans lesquelles un tel exercice est considéré comme valide par la communauté scientifique. A la lumière de ces conclusions, nous analysons le cas particulier de l'économie évolutionniste. La nature même de l'approche évolutionniste et des fondations sur lesquelles elle repose, en particulier l'importance donnée à l'histoire et la façon particulière dont sont appréhendées les relations causales, nous amènent à mettre en évidence l'importance d'un exercice contre-factuel un peu particulier : l'histoire contre-factuelle. Nous discutons les forces et les faiblesses d'un tel exercice et reconnaissons qu'en dépit de son caractère hautement désirable en économie évolutionniste, une analyse historique contre-factuelle se révèlerait aussi excessivement délicate à conduire.


Historische "Was wäre wenn..." Analyse und evolutionäre Ökonomik

Implikatonen, denen falsche Prämissen zugrunde liegen, sind in jedem wissenschaftlichen Vorhaben allgegenwärtig. In diesem Aufsatz analysieren wir deren Eigenschaften und die Umstände unter welchen sie aufgestellt werden können. Im seinem weiteren Verlauf weiten wir unsere Analyse auf ihren Gebrauch in der evolutionären Ökonomik aus. Wir folgern, daß die Eigenheiten und Grundlagen der evolutionären Ökonomik, vor allem aber die Bedeutung, die historischen Prozessen beigemessen wird, sowie die kausale Konzeption, die den meisten theoretischen Arbeiten zugrunde liegt, ein wesentlich breiteres Konzept erfordern, nämlich eine historische Analyse, unter "Was wäre wenn..." Annahmen. Wir setzten uns mit den Stärken und inhärenten Problem dieser Methode auseinander und folgern, daß die eigentümlichen Aspekte, die aus dieser Methode ein wünschenswertes empirisches Instrument machen auch zu den Schwierigkeiten in ihrer Anwendung beitragen.



 

Evolutionary Economics and the Counterfactual Threat

This paper is concerned with a methodological question in evolutionary economics. In particular it examines the role and nature of counterfactual history as an empirical tool in evolutionary analysis of economic phenomena. As a method, counterfactual history has not met with universal approbation, and many neo-classical economists look upon it with considerable dis-favour. Part of the goal of this paper is to explain both why counterfactual history is appealing as a tool for evolutionary analysis, and why it may appear unappealing to someone deeply schooled in the neo-classical tradition.

Empirical analysis in any research field is entwined in theoretical analysis. That is, empirical work depends on theory for concepts, definitions and hypotheses all of which are used as foundations for empirical investigation. In significant parts of evolutionary economics empirical analysis raises no methodological issues. It effectively uses the same tools and criteria as are used in the neo-classical paradigm. To a great extent evolutionary and neo-classical analysis use the same concepts and definitions, terms and jargon, and are concerned with the same (classes of) phenomena. Both are concerned with relations between economic phenomena, and often these relations are expressed statistically. Thus empirical analysis involves examining data to see whether they are consistent with theoretical predictions. Statistical description and analysis of data, aggregated at various levels, form the essence of the work.

Some of the theoretical concerns which lie at the heart of evolutionary economics, however, diverge from the mainstream tradition and cannot help, therefore, to affect methodology. One feature which serves to contrast evolutionary from neo-classical economics is the presence or absence of multiple equilibria. Paradigmatic neo-classical models generate unique equilibria, whereas in evolutionary models multiple equilibria are generally accepted as a normal and untroubling outcome. Unique equilibria are consistent with other aspects of neo-classical economic analysis, in particular the comparative statics analysis. When an equilibrium is unique, the welfare question is restricted to whether this equilibrium can be made better. This is often answered through a comparative statics exercise, which in principle involves small adjustments to parameters, resulting in small changes to equilibrium values. For example, we can adjust a tax rate to move the current equilibrium to a point very near, and ask which of the two is preferred.

Recent theory, in particular in game theory and macroeconomics has indicated that we must take seriously the possibility that there may be multiple equilibria. A second welfare question arises because we introduce the possibility that one equilibrium may be better than the others. And if this is the case, it is further possible that the economy might rest in an equilibrium which is not globally optimal. One of the non-chosen equilibria may be better.

Mainstream welfare economics has tied its flag to the competitive market as the Panglosian structure that produces the best of all possible worlds. And of course if we are in the best of all possible worlds, there is no scope for regret. A perfect market allows no inefficiencies. This result has become a central dogma for neo-classical economists, or at least for the hard core. In the neo-classical rubric, when we observe a sub-optimal allocation in the world (which cannot be explained away as a mis-description), the vocabulary of explanation includes phrases like "market failure" or "market imperfection". Sub-optimal allocations indicate some sort of aberration--an outcome that can be considered "unnatural" in some way. Every neo-classical economist would agree that there are deviations from the perfectly competitive ideal and that some deviations cannot be fixed. But the point to note is that the starting point, and the thing taken as somehow the natural way of the world (if only governments would keep their hands to themselves) is competition, which, in the long run at least, produces ideal outcomes.

On the other hand, in evolutionary economics Potential Regret may be the outcome where multiple equilibria are common and need not be welfare equivalent. The presence of multiple equilibria raises the issue of selection. How do we get one equilibrium rather than another? For evolutionary economists the answer to this question is typically "History." Selection mechanisms are inherently historical. Historical processes can have the feature of path dependence, and this is commonly present in evolutionary models. Here is the moment at which the Potential Regret result arises. Had history taken a different course in the past, we would now be at a different, and better, position. The historicity present here implies that explanations which address the issue of Potential Regret will necessarily be historical. The questions "Is potential regret possible?", or "Is potential regret actual in this case?" effectively ask what would have been the case had history taken a different course. This is a counterfactual question, equivalent in this regard to the question "What would be the effect on employment if the tax rate were reduced by x%?" This paper is concerned with this type of counterfactual, and in particular how the general tenets of evolutionary economics change the nature of counterfactuals from those in neo-classical economics.

We begin by addressing the role of counterfactual conditionals in science in general, and then turn to their relationship with views of causation. This is followed by a discussion of two distinct views of counterfactual conditionals, the branching view, and the non-actual-possible-world view. The following section brings these abstract considerations back to evolutionary economics, addressing both the difficulties raised for counterfactual analysis, and suggestions regarding how to strengthen it.

Counterfactuals in Science

The prototypical theoretical result in mainstream economics is the comparative statics result: y/x = f(.), or that changes in the value of the dependent variable, y, are related to changes in the independent variable x by the function f(.). The number of patents filed will increase by 10% for every 25% increase in R&D spending. These results are usually presented as theorems or propositions derived from axioms. Similar results are presented in most sciences, though the degree of precision and extent of generality vary, both within and across sciences. What is key here is that these statements can all be re-cast explicitly as equivalent counter-factual conditionals, that is, conditionals in which the antecedents are false. Indeed, Nelson Goodman (1983) argues convincingly that six types of statements are equivalent, and that any statement of one type can be re-cast as a statement of any of the other five types. The types are counter-factual conditionals, scientific laws, dispositional statements (p.3), factual conditionals (p.4), statements regarding possible worlds (p.34), and causal statements (Goodman is not explicit about the last category but it is implicit in what he writes on pages 37 and 45). So the following statements are equivalent: If R&D spending were 25% higher than it is, patenting would be 10% higher. The ratio of the change in R&D spending to the change in patents is 2.5:1. Patenting has the tendency to increase (in the ratio 1/2.5) as R&D spending increases. Since patenting increased by 10%, R&D spending must have increased by 25%. In a world like ours but with 25% more R&D spending, we would observe 10% more patenting. Changes in patenting activity are caused by (inter alia) changes in R&D spending.

Counterfactual conditionals are ubiquitous in science. They can be, and are, used to express causal relations. Further, they are part of the link between theoretical and empirical work. Empirical work is successfully characterized as the investigation of counterfactual conditional claims. Do the data show that if R&D spending increases then patenting increases in turn?

Causation

Counterfactuals also appear in another place which is germane to this discussion: philosophical discussions of causation have often employed counterfactual conditionals as the key concept. Indeed for many years there were vigorous attempts to unpack causation entirely in those terms. The appeal is obvious: "C causes E." is very much like the statement "If C occurs then E occurs.", and modulo a couple of niggling problems, is identical in meaning. The niggling problems turned out to be insuperable when addressed purely in terms of counterfactual conditionals, so that endeavour has effectively been given up. Part of the difficulty was a tension between logical analysis and physical analysis. To most non-Humeans, causation does seem like a real feature of the world. Indeed it is the cement of the universe, to use Mackie's phrase. If one thing did not cause another then there would be no reason that the world is not totally chaotic, and therefore no reason, beyond observations of Humean constant conjunction, to believe that it might not suddenly become so. This is what logical analysis fails to capture. So we can note that understanding the statement "If C occurs then E occurs" as a causal statement involves understanding more than simply the formal logic involved. It involves understanding things about C and E, about what links them and how C "necessitates" the occurrence of E.

We can make this point a slightly different way. All counterfactual conditionals are true. Logically, they can be written as ¬C & (C Æ E). Such a statement has the truth value true. Logically "If R&D spending were higher than it is patenting would be higher than it is." and "If R&D spending were higher than it is patenting would be lower than it is." are both true, since R&D spending is not higher than it is. This makes it sound as if the logical conclusion of scientific endeavour is "Anything can happen.", (which of course is not inconsistent with many of the beliefs of the spiritual movement of the 1990s). But this is clearly not where science is headed or wants to end up. How do we distinguish between these two statements, accepting one and rejecting the other?

The first, brief part of the answer is that it cannot be done purely logically. Peter Gallison (1987) shows convincingly just how much this is true for experimental physics. Experimenters must decide when they have arrived at the correct result, and when errors introduced by faulty equipment and extraneous forces are not affecting the outcome of the experiment in a significant way. Gallison shows in a series of cases that while formal logical analysis is important, it is probably not the most important factor in making that decision.

The longer answer to distinguishing between acceptable and unacceptable counterfactuals begins by noting that we must give up on the truth or falsity of counterfactual conditionals. Instead, we follow Goodman and address their "assertability". We are willing to assert that increases in patenting follows increases in R&D spending, but we are unwilling to assert that a decrease in patenting would follow an increase in R&D spending. This removes the onerous burden of logical proof, but leaves open the question of what determines assertability. The problem, in Goodman's words, is "to define the circumstances under which a given counterfactual holds while the opposing conditional with the contradictory consequent fails to hold." (p.4). Assertability lies in our willingness to infer the consequent from the antecedent. There must, therefore be some strong link between the two, and this is clearly a link that goes beyond formalism. Evaluating a counterfactual conditional involves examining the link between the antecedent and the consequent. Is there a link beyond the formal links? Yes, and that is causation.

Goodman argues that in order to find a counterfactual conditional assertable, we need E to follow from C by a causal law. This conforms well with the intuitions which make counterfactual claims relatively common in everyday life. Asserting a counterfactual conditional is in essence making a claim about the regularity and coherence of the world. Causal forces connect events and provide the coherence that we seek. The reliability of a counterfactual arises from the same source, namely the coherent, consistent structure of events, even under Hume's minimalist, "constant conjunction" view of causation. To defend a counterfactual conditional is to explicate the causal connections among the events involved. If the causal account is judged acceptable, the conditional is judged assertable. This brings us to the nature of causation in economics.

Causation in Economics

There is a distinct difference in the view of causation implicit in evolutionary economics from that implicit in neo-classical economics. In neo-classical economics, the fundamental goal is to explain phenomena in terms of equilibria. Something is explained when it is shown to be held in place by countervailing forces. The economy (firm, market) exhibits behaviour X because there is no incentive to deviate from that behaviour, and in fact, deviations are punished. The explanation of X consists in finding the forces that impinge on X and showing that they are in balance. How these forces came to impinge on X, or how they came into balance is not generally considered part of the explanation. It is assumed that some process would produce this result, but that process is rarely spelled out (Hausman, 1990). This is not quite as bleak, or perhaps oblique, as it sounds, for it is consistent with other important aspects of the neo-classical approach. The final result of an analysis is to find the effects of some independent variable on a dependent one, R&D spending on patenting for example. Demands of rigour imply that this is to be produced analytically and formally, which means that it is expressed as a comparative statics result. But for a comparative statics result to be reliable, the equilibrium must be unique. Uniqueness implies that the path to the equilibrium is in a very real sense unimportant. If this equilibrium is the only place we can end up, and if we are in or close to equilibrium most of the time (otherwise its usefulness as an explanatory tool would necessarily be called into serious question) what difference does it make how we get there? It is not even that interesting as a curiosity, since the path to equilibrium must be a relatively short one.

Causation, when seen this way, is called "sustaining causation" and the neo-classical economists' concerns are with sustaining causes. There is another type of causation, though, referred to as "originating causation". Evolutionary economists tend to be much more interested in originating causes.

The process by which an event, state or phenomenon comes to exist is central to explanation in evolutionary economics. The general goal is to explain the movement of something over time, or to explain the existence of a phenomenon in terms of how it came to be. This is common to all evolutionary sciences (Nelson, 1996). Explanations of, for example, the location of an industry would involve a dynamic model of location choice, in which choices made by firms this period depend on the location choices of previous actors. The process might evolve to a stationary distribution, but also might never do so, if the choices of later actors sufficiently upset the choices of earlier actors that they provide incentives to change location. The cause of the current distribution of activity is seen as historical. We must look back at decisions made in the past, and trace their effects on following decisions, as the choice of one actor affects the incentives of other actors. To understand a phenomenon, then, we must understand the process by which it came to be. There is a sequence of events leading to this outcome, and we must be able to trace this sequence, and the causal links between the events in it, before we "really understand" why this phenomenon exists.

Notice that this is all part and parcel of the possible existence of multiple equilibria. When multiple equilibria exist, the selection mechanism gains prominence. Selection mechanisms (in the world) are inherently historical. If there are two possible outcomes, one cannot simply look at sustaining causes to explain which is selected--by definition both are sustainable. One needs to address origins.

This discussion of causation impinges on counterfactual conditionals in a crucial way. We argued above that unpacking a counterfactual conditional involves examining causal connections. This is the process by which a counterfactual is judged--if the causal story is judged acceptable, or credible, the counterfactual is deemed assertable. Clearly, then, the nature of causation will determine what is unpacked during this enterprise. In a science in which causation is ahistorical, examination of the counterfactual claim will not involve history. Contrarily, if causation is viewed as inherently historical, then examining a counterfactual claim will necessarily involve history in one way or another.

We can make the same argument in a slightly different way. We pointed out above that a counterfactual claim is implicitly a causal claim, but this is deeper than simply a semantic identification. A claim about a counterfactual condition is about the same thing as a causal claim. Thus the counterfactual will contain within it the nature of causation. What this means is that while in neo-classical economics simple counterfactual conditionals will be the norm, in evolutionary economics, counterfactuals involving history will be common. Empirical work then, will involve historical excursus: hence the counterfactual history.

This discussion of causation makes contact with the solution of Conrad and Meyer (1964) to the problem of counterfactual conditionals. As noted above, the difficulty arises from logical problems with the counterfactual statements themselves. The solution, Conrad and Meyer suggest, is to look for statements that are co-tenable with the counterfactual statement. Consider, for example, the statement, "If there had been no Civil War, slavery in the southern US would have been dismantled peacefully." There is no way to prove this analytically, nor is there a direct empirical "proof". But this statement would only be true if slavery was unprofitable at the time of the war. Thus we can implicitly test this statement by testing the profitability of slavery. Notice, however, that this approach makes an assumption about the causal forces at work. It assumes that the non-occurrence of the Civil War would not have affected the profitability of slavery, nor would it have set in play any other events that would have over-whelmed the effects of the (non)profitability of slavery. Both of these assumptions may be valid in this case, but by making them explicit we see that this solution runs the risk of being too static. It fits easily with the view of counterfactuals as describing a world as similar as possible to the actual world at the time under examination, but it may not fit so well if one's view is more historical, namely that the counterfactual world must be the outcome of some feasible path from some point in the history of the actual world.

We turn now to a discussion of these two views of counterfactual statements.

Two Views of Counterfactuals

The philosophical literature contains two view of counterfactuals. David Lewis (1974), for example, sees a counterfactual conditional as describing a possible but not an actual world. There are many worlds which in principle could exist, the actually existing world being only one of them. A counterfactual conditional is merely describing another. One might be a world just like ours, but in which R&D spending is 25% higher than it is in ours. Lewis's view is that to do science effectively, which of course involves examining counterfactual conditionals, this alternative world should be as much like ours as is possible, and in fact he describes a formal metric. His idea is largely ahistorical since within it is no constraint that it be possible to get to the alternative world from some point in our own actual history. It is simply a world running parallel to our own in some alternative universe, and which we examine when doing so might tell us something interesting about our own.

There is a second view of counterfactuals, defended by Jon Elster (1978, ch. 6), which involves a branching view of history. History is seen as a tree, in which each decision taken represents a branching point. We proceed up the tree, moving higher and higher, never descending, as decisions are made and history unfolds. A counterfactual analysis involves moving back down the tree to some branching point, and examining a branch not taken. Sometime last year a decision was taken about this year's R&D budget. The analyst returns to the point at which that decision was taken and asks "If at that point a different decision had been taken, how would history have unfolded?". Bring this alternative history up to today's date and ask how the alternative present differs from our own.

Having observed these two views of counterfactuals, we must ask to what extent two possible worlds can differ. Leibniz presents an extreme view since for him, every entity (monad) contains within it the entire world. Thus to change one aspect of the world involves changes to every single entity in it. Less anachronistically, if the world is a very densely connected system, then to change one thing in it will involve changing the things directly connected to it, following which we must change the things connected to those, and so on until we have changed virtually everything in the world. This makes the "possible worlds" approach to counterfactuals problematic, since it may not be possible to imagine a world close to ours in the desired sense.

Given the historical nature of the enterprise of evolutionary economics, it seems that the branching view of counterfactuals is more apt.

But in its most general form, a counterfactual analysis amounts to postulating that something is different than in fact it is, and examining what follows from this difference. Two difficulties arise. What is it feasible to postulate as a difference? How do we tell what follows? The key to addressing both of these difficulties lies in there being theory that can be employed in the analysis. Theory confines a counterfactual analysis in two ways. First, it restricts what can be postulated as the counterfactual antecedent--we cannot ask what would be the case if anti-gravity machines existed for example. Second, it determines what follows from those postulates.

Branching

The branching approach to counterfactual analysis consists in choosing a branching point in our actual history, and tracing an alternative history that would follow from a different branch having been taken. The branching view is common among economic historians. When discussing the "new economic history", Fogel (1973, p. 139) states, "In order to determine what would have happened in the absence of a given institution, the economic historian needs a set of general statements that will allow him to deduce a counterfactual situation from institutions and relationships that actually existed." Elster (1978, p.181) is more explicit, claiming that there are three conditions that must be met to make such any counterfactual analysis compelling. The antecedent of the counterfactual conditional must be feasible in a static sense; that is, it must describe a state that is internally consistent. The claim that had railroads not existed, GDP would only be slightly smaller, needs, as one necessary condition, that a world without railroads does not contain logical contradictions. The antecedent must also be "insertable"; there must be some feasible historical path from a real historical state that would produce the state described in the counterfactual antecedent. Given the history of the world to some point, 1840 say, it must be possible to trace a feasible history which includes no logical or physical impossibilities, from that point to the one in which the antecedent--railroads do not exist in 1890--obtains. Finally, the consequent must be linked to the antecedent by a dynamic theory. (This linking typically involves going back to the original branching point and tracing the history. Tracing history is obviously an exercise that involves dynamic analysis, and here is where the introduction of a dynamic theory takes place.) It is clear that all three of these conditions are governed by theoretical considerations: internal consistency is a theoretical notion; the feasibility of a hypothetical history implies that it does not contradict held theories of how the world works; and the connection between antecedent and consequent is, if it is to be compelling, based on accepted causal theories.

The importance of the role of theory here introduces what Elster (p. 184) considers to be the basic paradox of counterfactual analysis. The stronger is the theory (that is to say the more it restricts what can happen) the better grounded is the "deduction" of the consequent from the antecedent. But on the other hand, the stronger is the theory, the more restrictive are the conditions of insertability, that is, the smaller is the set of antecedents that are consistent with "what we know about the world". Any counterfactual analysis must avoid both risks of vagueness and risks of absurdity. An analysis is vague when the model connecting consequent to antecedent is too loose, or not specified completely--it allows too many things to happen. Absurdity arises when the antecedent runs awry of a tight theory, that is, a theory that is highly specific about what can happen, and so when used to examine the counterfactual antecedent, reveals a state that is internally incoherent or absurd.

The branching view of counterfactuals implies a stronger notion of consistency than the possible worlds view. To logical and physical consistency the branching view adds historical (or originary causal) consistency as a requirement. Thus to ask what would be the effect on GDP if railroads did not exist in 1890 is not to ask about the world in the instance that railroads were vaporised on the first of January 1890, since some of the remaining physical and institutional structures would be historically incongruous. Historical consistency rules that out as an avenue of inquiry, and demands instead that we consider the possibility that they were not introduced in 1840, nor between then and 1890. Notice that this permits or accommodates some looseness in the statement of the counterfactual antecedent. Whether the antecedent is stated as "railroads do not exist in 1890", "railroads were not introduced in 1840, or any time up to 1890", or "railroads were never invented", all demand the same analysis. The same theories apply in each case, assuming the counterfactual consequent in each case refers to the state of GDP in 1890.

Counterfactuals and Neo-Classical Economics

One of the strengths of neo-classical economics, and something which makes the counterfactual analysis used there seem innocuous, is that the theory is very strong. Its strength places severe restrictions on what can be assumed as antecedents. The primitives of the theory, utility functions and production functions, are tightly constrained by the analytical tradition: a small number of functional forms is acceptable, and all have the same regularity properties. Thus to a great extent, the only possible counterfactual antecedents have to do with changing parameter values in the models. The fact that models traditionally contain unique equilibria for any acceptable parameter values implies that consequents follow very tightly from counterfactual antecedents. In fact the theory has been specifically constructed so that consequents follow antecedents by the laws of formal logic or mathematics. Tighter connections are not available even in principle. Further, the kind of causation accepted as the standard in neo-classical theory is sustaining causation, from which it follows that comparative statics analysis tells a complete story. We can place this in the context of Elster's three conditions for counterfactual analysis. Because changes in parameter values are by definition exogenous to the model, they lie outside the scope of economics, and so many changes are acceptable. The only constraints are internal, logical consistency, and that the change not create a violation of the functional form regularity conditions that most models contain. The "course of history" that produces the parameter changes is exogenous to the analysis, and so an economist has little, if anything, to say about it. Further, since the equilibria are typically unique, all possible histories will lead to the "new equilibrium". The link between consequent and antecedent is deductive, which is the strongest kind of link we know. So if the tenets of the theory are accepted, counterfactual analysis will be a very strong tool.

It is clear that any deductive science will be able to make very strong counter-factual claims, since theory provides a strong link between consequent and antecedent. If one agrees with the premises of the basic theory, one cannot argue with the counterfactual claims that are drawn from it. Some of evolutionary economics fits this mould. Specifically, it is often possible to derive, using an evolutionary model, results about patterns of behaviour. Technology choice models, for example, derive results about proportions of agents using different technologies. And evolutionary game theory models typically have this feature.

Counterfactuals and Evolutionary Economics

Not all of evolutionary economics can be characterized this way though. Many of the models yield probabilistic results. This is not surprising given the underlying view that many processes are contingent. Probabilistic results are not a difficulty when two conditions are satisfied. The first is that there is a deducible relationship between the probability distribution and some exogenous variable (a form of a comparative statics result); and second, that the phenomenon under investigation is manifest in a way that permits estimation of the entire distribution. When both of these are fulfilled, the counterfactual conditional relating changes in the parameter to changes in the probability distribution can be examined since both are observable. It is common, though, that the second condition fails, and that we cannot observe distributions, but rather a very small number of realizations of the distributed variables. This is typically the case with potential regret results. In the technology choice literature for example, a typical result which is provable analytically, is that with non-zero probability, the system will standardize on an inferior technology. When turned into a counterfactual conditional referring to a particular instance we get: "Had we standardized on gas-graphite nuclear power reactors, we would now be better off." This claim seems much stronger than anything present in the theory which only claims that if there are enough cases of technology competitions, some of them will have the result that an inferior technology wins. It makes no claims about nuclear power, nor about this particular realization of its development. If we were to confront the theory with the nuclear power case, what would be involved?

On the "possible worlds" view of counterfactuals, we need to imagine a world in which we simply replace all the light water reactors with gas-graphite reactors, and ask whether it would be a better world. This is the thought experiment of a comparative statics analysis, and makes perfect sense if history does not matter. A (formal, analytic) map from technology type to net benefits from use will be enough to tell us whether gas-graphite would have been better. An ahistorical map renders the possible worlds approach sensible.

On the branching view of counterfactuals, we must go back in history, probably to about 1960, and change several of the decisions made at that point, and then re-write the history to the present. This is the approach taken if historical events do make a difference. And this is the approach appropriate to evolutionary economics in a great many instances. This follows from the presence of positive feedbacks in evolutionary models. Changes in decisions can be magnified, and the system can re-enforce those decisions. Again, though, if an analytic map from technology type to benefits from use existed, the counterfactual conditional would be strong again, even if history were involved in that mapping. What renders this attempt infeasible, from the point of view of many evolutionary economists, is the importance of randomness, coupled with positive feedbacks. When these are both present, at the level of analysis typically conducted by economists any outcome is under-determined by the data, even in an equilibrium model. A central tenet of evolutionary economics is that there are many sources of indeterminacy in any economy: the identities of interacting agents, which will necessarily be beneath the vision of the analyst; the process of learning, which is often unpredictable; the presence or details of innovation which are, again, unpredictable, are all examples. This indeterminacy implies that model outcomes (at least the details of them) are under-determined, and further, that the actual determination of them is historical. This indeterminacy weakens the link between antecedent and consequent in the counterfactual. Evaluating it no longer is just a matter of formal logic, it now involves judgment.

It is important to notice that under-determination notwithstanding, history is not unconstrained--that would make investigation impossible. But history is only loosely constrained.

We argued above that law-like statements can be re-expressed as counterfactuals, and that thereby, the investigation of law-like statements will be implicitly at least, an investigation of counterfactuals. Analysts investigate, therefore, the causal link between antecedent and consequent of the counterfactual conditional. If causation is viewed historically, so will be the counterfactual and so the investigation of it. But if history is only loosely constrained, many things can happen, which means that in all probability there is some link between consequent and antecedent. But "many things can happen" does not mean that everything happens with equal probability. Nonetheless, the fact that many things have non-trivial probabilities implies that tracing a tight link between an event (or counter-event) and its consequences may be problematic, the more so as the two are removed from each other in time. History unfolds by choosing among possible possibilities. That is, history makes a choice about what today is like. With every possible today, there is associated a set of possible tomorrows. If agents are learning (often in an unpredictable way) and agents are innovating (in an unpredictable way) and interacting with other agents (in an unpredictable way), the choice of today's state does not place strong restrictions on the set of tomorrow's possible states. This means that tracing the effects of a (counterfactual) event becomes less secure as the tracing occurs over longer times.

The most compelling counterfactual analyses exist when the theory on which they are based is tightest, and most restrictive about what can take place. One of the intuitive strengths of evolutionary economics is the premise that in a system as complicated and complex as any modern economy must be, any theory that too tightly constrains what might happen will leave out very important aspects of economic activity. But this makes counterfactual analysis very difficult. If history matters, then counterfactual analysis will often be historical. Without severe theoretical constraints on what history can produce, though, the analysis runs the risk of being far from compelling.

We can summarize this section briefly. To accept a counterfactual conditional we demand an argument that it is possible to infer the counterfactual consequent from the counterfactual antecedent plus auxiliary conditions. This inference must be causal, so causes and the causal structure must be understood (Goodman, 1978). Hausman points out that for a comparative statics analysis to be coherent, the economist must believe that there is a process that would connect the states before and after the change in the exogenous variable. He argues that for a variety of reasons this process is not of particular interest in general. But, as Tunzelman (1990, p. 296) argues, counterfactual histories "represent an empirical working out of comparative statics", that is, they are an explicit rendering of Hausman's process. Evolutionary economics contains a general belief that the process between initial and final states is important, due to the under-determined nature of outcomes, and so counterfactual history will be an important part of explanation. To come to the same conclusion from a different direction, we note that virtually all sciences embrace some form of causality, and use that notion as part of their explanations and to support their predictions. For evolutionary economics, since it is generally accepted that "cause" refers to originating or genetic causes and that the causal structure is temporal or historical, the argument supporting the counterfactual conditional will necessarily be temporal or historical. But one of the underlying beliefs of many evolutionary economists is that outcomes are under-determined; in general any basic data present more than one possible outcome. Thus theory does not tell us what will happen, it only restricts us to a set of possibilities.

Evolutionary Economics and Counterfactual History

The discussion in the previous section seems to be driving toward the conclusion that since empirical counterfactual analysis is so difficult to perform convincingly in evolutionary economics, this branch of economics is fated to be non-empirical and therefore, highly speculative, and presumably, of little practical value. In this section we argue that we need not, in fact be driven to this conclusion. We present four different routes away from it--three "in practice", and one "in principle".

In practice 1: Weak Counterfactuals and Appreciative Theory

Counterfactual analysis can be used to buttress appreciative theory. The aim of appreciative theory is less to formalize general laws, which could be used in predicting the future, than to use general, well-understood theoretical concepts to understand either the past or the present. Evolutionary economics takes almost as a precept the idea that understanding is necessarily historical. Thus to understand the present we must know and understand the history that brought us here. History can be seen as a tree--we follow the structure up the trunk, moving onto different branches, but never moving down. As time passes, the economy is continually eliminating future branches, and thereby producing the present. Understanding this process involves showing either that there were no branching points, and thus the present state is inevitable given the starting point, or identifying the branching points (those at which there was a non-trivial probability of taking a different route), and showing why the economy followed the route it did.

We refer to this exercise as one of weak counterfactual history, since it involves minimal knowledge or assertions about the non-actual branches. In this regard it is only necessary to argue that the branches not taken would lead to a different present. Even more generally, weak counterfactual analysis is restricted to understanding the major events and chains of decisions which, coupled with processes that magnified rather than damped their effects, can be considered as having played a role in disconnecting some sub-regions of the tree from the branches followed by actual history.

Advantages of a weak counterfactual argument are clear. Accepting its goals renders Elster's conditions less important, and thus permits economists to treat a large number of problems in those terms. Drawbacks also seem clear however. Weak counterfactual analysis does not give welfare results, and in a sense can only argue that the actual outcome is not the only possible one. This does permit the claim of potential regret, but cannot assert more; it will not make a case for actual regret. (See Foray, 1997.) For that we need something stronger.

In practice 2: Strong Counterfactual Analysis

Appreciative theory, or appreciative analysis more generally, is sometimes considered not enough. In particular, economists are often concerned about making welfare comparisons between two different states. This exercise involves knowing something about the nature of both states. It is not generally difficult to know about the actual present. But to compare it to a counter-factual present can be tricky. If, following the arguments given above, we adopt the branching view, rather than the "non-actual possible worlds" view of counterfactuals this comparison is made more difficult, as we must compare our state not simply with another state having minimal variations from it, but rather with the state that would exist now following some counterfactual change at some point in the past.

The obvious difficulty of the exercise is that strong counterfactual analysis consists of (re)-vitalizing something that did not happen. In contrast to weak counterfactual argument, the restrictions discussed above must be given much more consideration in the analysis. The theory or principles employed in strong counterfactual argument must, therefore, i) constrain the world of potential developments; and ii) use strong laws on the basis of which both to construct and to develop the alternative paths "selected" from the set of possible paths.

The first principle is relatively easy to illustrate. Consider the hypothesis that environmental problems would be less severe if the internal combustion engine did not dominate personal mobility technology. There are, currently, four possible automobile propulsion sources: internal combustion; lead-acid batteries; fuel cells; and steam. An argument that internal combustion is an inferior technology (on the environmental criterion) could only be built using as competing technologies steam and lead-acid batteries. This follows from the fact that the important events determining the outcome of the competition took place at the turn of the century, and the fuel cell technology did not then exist. Thus the important branching point in the history was a three-way branching in which the internal combustion branch was taken. This is related to Elster's insertability criterion. The fuel cell option is not insertable at the crucial point in time. But this is, rather than a weakness, a strength of the analysis. A small set of options implies that a small number of branches need be traced to address the issue of whether the present state is an optimal one. Bifurcation points with few branches make the case for regret stronger, by deriving it from a point with relatively few options.

The second condition, that counterfactual history be based on strong laws, is more difficult to illustrate. A causal story is simplest and most compelling when there is a single causal factor that clearly dominates all other considerations. In this case a relatively tight link between antecedent and consequent can be drawn, because the construction of the chain of events is tightly controlled and constrained.

The law must be strong in two senses: it must be general (that is to say, robust); and important. Generality is straightforward. The law must cope with a variety of spatial and temporal contexts. Many macro-and micro-economic laws are general in this sense.

The other dimension, importance, is less simple. We need a law "important enough" to cause the counterfactual construction we base on it to be, in effect, isolated from the influence of other laws. It must be over-ridingly strong. It clearly follows that the strength of a law cannot be dissociated from the context of the enquiry. Network effects, for example, can create forces which underlie a strong law in the field of telecommunications technologies. But these same forces will not be "important enough" to build a counterfactual history about, for example, regional polarization, since it is not plausible that the dynamics generated by network externalities could be isolated from the effects of transportation costs, infrastructure externalities, local labour markets and so on. In this dimension, "importance" implies that the ceteris paribus clause can be plausibly invoked.

There is another problem which must be faced, namely the risk of circularity. Any economic system, agent or phenomenon will have many forces acting in or on it. The discussion immediately above had to do with identifying and examining the effects of the strongest forces. This implies that forces can be identified as primary or secondary. Concentrating attention on primary forces (of which one hopes there are few), connected to strong laws will not generate misleading results. The identification of primary and secondary forces is one typically done in the abstract, or theoretically, and here lies the possibility of circularity, or perhaps assumption of the result. Evolutionary theory, in common with other theories of complex systems, typically has the feature that positive feedbacks, and the interconnections between different parts of a system, can magnify what appear to be secondary considerations into primary ones. Unfortunately, determining whether or not this is a relevant consideration can only be done using counterfactual history. So again the very feature that makes counterfactual history a vital tool makes it a tricky one to use. We can see this in the discussion of Robert Fogel's (1964) work on US railroads. One central issue of Paul David's (1975) criticism is that Fogel ignored this possibility. Fogel made a priori judgments about which forces were primary and which secondary, and then showed that when the primary ones were considered, US railroads had a certain effect on GNP. One interpretation of David's arguments is that if the forces considered secondary were put into a complete counterfactual history, they would have been acknowledged to be as important as the primary ones.

In practice 3: Stronger theory

We have just argued that whether or not a law is strong depends on context. This might suggest that a general change in context could have the effect of increasing the strength of laws generally. In other words, it may be possible to make strong counterfactual history more compelling by looking at theory slightly differently, reconsidering what questions are feasible, what count as respectable answers, or what phenomena it is appropriate to analyse, if these have the effect of making the theory in general stronger. If, for example, the theory is seen to be more deductive, and in particular more deterministic, this reduces the "anything can happen" problem. But because of the underlying tenets of evolutionary economics, to adopt such a strategy implies thinking about economics problems in a slightly different way.

Neo-classical economics attempts to make predictions of the behaviour of economic agents, either as individuals or as aggregations. Predictions tend to be very detailed, in which, even in work on aggregates, actions of individual agents are specified. In models of perfect competition all agents are infinitesimally small, and thus as individuals have no influence on outcomes. It follows that generic market predictions are equally predictions about the actions of each agent. At another extreme, we observe models with few enough agents that all can be traced and the non-anonymity is tractable. Evolutionary models, though, often have a large but finite number of heterogeneous agents in them, and these agents are affected by each others' actions. When this is true, a detailed prediction of agent behaviour is simply not possible. It is this large population of heterogeneous agents, interacting with and affecting each other that causes the serious under-determination of evolutionary models. Even if there is no ontological certainty, epistemological uncertainty can be extreme.

Whether or not under-determination is important, though, is crucially dependent on the level of analysis. Where in the world it will rain today is certainly under-determined by my knowledge of the weather system. That it will rain somewhere is not. And with a small amount of research one could almost certainly predict fairly accurately how much rain will fall today. Thus we can reduce considerably the degree of under-determination simply by changing the level or nature of analysis. While the details of the actions of individual agents may be under-determined, general patterns of behaviour may not be. This was an idea present in the work of Hayek (1967, for example), and based on roughly the same arguments--there is too much information present in the economy to be processed by any analyst. Any agent will have more or different information than any analyst, thus the goal of predicting detailed agent-level behaviour is un-reachable. There will be general tendencies, however, which are predictable, due, in effect, to the central limit theorem.

Consider models of firm location. A firm chooses a location based in part on the locations of other firms. Locating near another firm on the one hand permits encroaching on its market, but on the other reduces one's monopoly power. In early models of such a system, (Eaton and Lipsey 1975, is a good example) the question was where firms would locate in a certain space. It quickly became apparent, though, that in many models attempts to expand the model beyond very small numbers of firms, or beyond one dimension, created intractable problems, partly because of the economic process involved. (Note that the goal was not to ask where a specific firm, Firm A, would locate but rather the weaker one, whether there would be any firm at all at some location X.) The question was un-answerable, possibly in principle. A different question, though, and one which involves a different level of analysis, is whether firms will cluster or spread out. This is of course related to the central issue in Hotelling's (1929) original paper. It is also possible to ask how large clusters will be, and how close together they will be. Here the questions do not refer specifically to firm behaviour, but rather to the properties of the system as a whole, and in particular the pattern or distribution of location. It is possible to address these questions analytically. By changing the unit of analysis we have recovered strong theory and the analytical results that are produced using deductive tools, yet the models underlying these results retain the evolutionary processes inherent in the economic decisions made by firms.

A second approach to making evolutionary theory tighter is to make the analysis statistical. The "interacting agents literature" models large numbers of (heterogeneous) agents who interact both within and outside the market. Population properties evolve as agents respond both to their own past actions and to the actions of other agents. Here, theory seeks to make predictions about statistical properties of populations, rather than about the properties of individual agents. (This literature is reviewed in Kirman 1996, and Durlauf, 1996.) Schelling (1978) refers to analysis which explores the relationships between behavioural characteristics of individuals and those of the aggregate. Developing counterfactual arguments is dependent on our ability to calculate aggregate data as determined by individual actions. This is indeed possible for some classes of social and economic interactions. Schelling uses a simple example to illustrate: "if we know that every driver, on his own, turns his lights on at sundown, we can guess from our helicopter we shall see all the car lights in a local area going on at about the same time." (Schelling, p. 13). In more complex cases, however, agents' behaviour or agents' choices depend on the behaviour or the choices of other agents; and those situations often do not permit any simple summation or extrapolation to the aggregates. Then counterfactual constructions must look at the system of interactions between individuals and their neighbourhoods.

Both approaches, by changing the focus of analysis, produce strong theory and analytical results. They both reduce the historicity present in the abstract research, and permit counterfactual analysis which is more like comparative statics in appearance. Tight theory and tight counterfactual conditionals are possible.

In Principle: A Different Criterion?

There are occasions in which tightening the theory in one of these ways is the right approach. But there may be phenomena for which it is not appropriate, and some looseness in the theory, and consequent looseness in the counterfactual empirical claims are either desirable or necessary. When this is the case, the arguments given above indicate a problem for empirical evolutionary research. It may fall prey to the observation of Kenneth Arrow (1995): "While the theory is pretty good, empirical evidence is frequently weak." But this may be too hasty. One cannot judge the empirical endeavours of one approach by standards derived from a different theoretical approach. Standards to judge success must be consistent with the task undertaken.

Intuitive arguments against counterfactual history are that it tends to be based on theory which is so loose that it allows anything to happen. Of course if one changes the past one can change the present. Any counterfactual claim can be asserted given the weakness of the theory involved. Therefore, because the results are "loose" and are based heavily on the judgment of the analyst, they are of little (or no) value. But this argument is to say that "this apple makes bad orange juice." Neo-classical economics has opted for an axiomatic approach involving deductive theory. Counterfactual conditionals are deduced under standard assumptions, and empirical investigation involves checking the values of parameters (traditionally, checking only that they are not zero). Predictions about behaviour are arrived at deductively, and are supported by finding an elasticity in the right range. Is the elasticity of patenting with respect to R&D spending equal to 0.4 (or more generally, positive)? Answering this question in the affirmative corroborates the theory with which the question was generated. The argument is of the form, "If assumptions 1 through n hold, then the elasticity of patenting with respect to R&D is equal to 0.4." Measuring the value of the elasticity can provide very strong indications regarding the theory, and whether the theory is valuable in understanding innovation.

When counterfactual conditionals are not (formally) deduced from axioms or assumptions, this simple argument form is not available. Knowing an elasticity neither proves nor disproves the theory, nor does it permit one to predict the future. The latter follows from the under-determination typical of the approach we are considering.

If one opts for a non-deductive theory, empirical analysis will be of a different nature, and it follows that the criteria by which it is judged will also be different. In particular, the assertability of a counterfactual necessarily rests on a different type of argument. To judge the success of empirical investigations of evolutionary theory by the criteria used to judge empirical investigations of neo-classical theory is to jump too fast. To reject evolutionary theory on the grounds that it provides no elasticities to be measured (or perhaps more precisely that it suggests that knowing elasticities is far from the whole story), for example, or to reject an instance of it on these grounds, involves first justifying the criteria by which it is being judged. On the one hand, if criteria appropriate to judge neo-classical theory are being used to reject evolutionary theory altogether, (on the grounds that the empirical justification involves counterfactual history) this is circular. On the other hand, if the criteria are being used to reject a particular instance of evolutionary analysis, then there is a major uncompleted (and largely un-noticed) task, namely showing that the criteria for empirical success are those of deductive (neo-classical) theory. This has yet to be done in economics.

The response to Arrow (1995), quoted above, must be the following. Perhaps it is true that by the standards of neo-classical analysis the empirical evidence is weak. But if you admit that the "theory is pretty good", you must accept that the standards applied to neo-classical empirical analysis do not apply. Different standards do. Whether or not an counterfactual conditional is accepted as empirical evidence depends not on the statistical significance of parameter estimates but rather on whether an historical, causal explanation is compelling and thus whether the counterfactual is assertable. The evidence looks much stronger when judged on its own grounds.

 

References

Arrow, K. (1995). "Economics As It Is and As It Is Developing: A Very Rapid Survey" in Albach and Rosenkrans (eds.) Intellectual Property Rights and Global Competition Sigma: Berlin.

Arrow, K. and T. Skitovsky, eds. Readings in Welfare Economics, London: Allen and Unwin, 1969.

Bassanini, A. and G. Dosi (1998). "Competing Technologies, International Diffusion and the Rate of Convergence to a Stable Market Structure", IIASA Interim Report IR 98-012.

Bunge, Mario (1979). Causality and Modern Science, 3rd ed. New York: Dover Publications.

Conrad, A.H. and J.R. Meyer (1964). The Economics of Slavery and Other Studies in Economic History. Chicago: Aldine Publishing Co.

Cowan, Robin and William Cowan, (1998). "Technological Standardization with and without Borders in an Interacting Agents Model" MERIT Research Memorandum, # 2/98-018

Cowan, Robin and Dominique Foray, (1998). "Counterfactual History and Evolutionary Economics" MERIT Research Memorandum, # 97-012

Cowan, Robin and Philip Gunby (1996) "Sprayed to Death: Path Dependence, Lock-In and Agricultural Pest Control" Economic Journal, vol 105.

Cowan, Robin, and Mario Rizzo (1996), "Genetic Causation and Modern Economic History" Kyklos,.

David, P.A. (1975) "Transport Innovations and Economic Growth: Professor Fogel on and off the Rails" in Technical Choice, Innovation and Economic Growth: Essays on American and British Experience in the Nineteenth Century Cambridge: Cambridge University Press.

David, P.A. (1988). "Path Dependence: Putting the Past into the Future of Economics" Technical Report #533, Institute for Mathematical Studies in the Social Sciences, The Economics Series, Stanford University.

David, P.A. and Shane Greenstein, (1990). "The Economics of Compatibility Standards: An Introduction to Recent Research", Economics of Innovation and New Technology, vol. 1, pp.3-41.

Durlauf, S. (1996). "Statistical Mechanics Approaches to Socioeconomic Behavior", Santa Fe Institute Working Paper 96-08-069.

Eaton, C. and R. Lipsey (1975). "The Principle of Minimum Differentiation Reconsidered: Some New Developments in the Theory of Spatial Competition", Review of Economic Studies, Jan., vol 42.

Elster, J. (1978). Logic and Society: Contradictions and Possible Worlds, Toronto: John Wiley and Sons.

Fogel, R. (1964). Railroads and American Economic Growth: Essays in Econometric History, Baltimore: Johns Hopkins Press.

Fogel, R. (1973). "The Specification Problem in Economic History", in P. Temin (ed.) The New Economic History, Penguin. reprinted from Journal of Economic History, vol 27: 283-308, 1967.

Foray, D. (1989). "Les Modèles de compétition technologique: une revue de la littérature" Revue d' Economie Industrielle, vol. 48, pp.16-43.

Foray, D. (1997). "The Dynamic Implications of Increasing Returns: Technological Change and Path Dependent Inefficiency", International Journal of Industrial Organization. vol. 15.

Foray, D. and A. Grübler. (1989). "Morphological analysis, Diffusion and Lock-Out of Technologies", Research Policy vol. 19(6).

Gallison, Peter , (1987) How Experiments End, Chicago: University of Chicago Press.

Genovese, E. (1962). "The Significance of the Slave Plantaion for Southern Economic Development", Journal of Southern Economic History vol. 28: 422-37.

Goodman, Nelson (1983). Fact, Fiction and Forecast, 4th edition, Cambridge, Harvard University Press.

Goodman, Nelson (1978). Ways of Worldmaking. Indianapolis : Hackett.

Hausman, Daniel (1990). "Supply and Demand Explanations and their Ceteris Paribus Clauses". Review of Political Economy, vol. 2, p. 168-87.

Hayek, F. von (1967). Studies in Philosophy, Politics and Economics Chicago: University of Chicago Press.

Hirsch, W. (1952). "Manufacturing Progress Functions" The Review of Economics and Statistics vol 34.

Hotelling, H. (1929). "Stability in Competition", Economic Journal, March, vol 39.

Kindleberger, C. (1989). Economic Laws and Economic History, Cambridge: Cambridge University Press.

Kirman, A. (1996). "Economies with Interacting Agents" European University Institute.

Lewis, David. (1974) Counterfactuals Harvard University Press.

Liebowitz, S. and Steve Margolis, (1995). "Path Dependence, Lock-In and History", Journal of Law, Economics and Organization, April.

McAfee, R.P., (1983). "American Economic Growth and the Voyage of Columbus", American Economic Review, vol 73(4), p.735-740.

McCloskey, D. (1989) "Counterfactuals" in The New Palgrave.

Mill, J.S. (1967) A System of Logic, London: Longman.

Nelson, Richard (1996) "Recent Evolutionary Theorizing about Economic Change" Journal of Economic Literature vol 33, p. 48-70.

Putnam, Hilary (1987). The Many Faces of Realism LaSalle, Ill: Open Court.

Reder, Melvin W. (1982). "Chicago Economics: Permanence and Change." Journal of Economic Literature, vol. 20.

Samuelson, P. (1950), "The Evaluation of Real National Income" Oxford Economic Papers, 2:1-29.

Schelling, T. (1978). Micromotives and Macrobehaviors, New York: Norton.

Tunzelman, G.N. von (1990), "Cliometrics and Technology" Structural Change and Economics Dynamics, vol 1(2): 291-310.

Verdoorn, P. (1956). "Complementarity and Long-Range Projections" Econometrica vol. 24.

Wright, T.P. (1932) "Factors Affecting the Cost of Airplanes", The Journal of the Aeronautical Sciences vol. 3, Jan.