Contemporary Directions in Economic Methodology

The past half-century has witnessed the emergence of a large literature devoted to economic methodology. That literature explores many methodo­logical approaches and applies its conclusions to many schools and branches of economics.

4.1 Popperian approaches

Karl Popper's philosophy of science has been influential among economists, as among other scientists. Popper defends what he calls a falsificationist methodology (1968, 1969). Scientists should formulate theories that are “logically falsifiable” - that is, inconsistent with some possible observation reports. “All crows are black” is logically falsifiable, since it is inconsistent with (and would be falsified by) an observation report of a red crow. Second, Popper maintains that scientists should subject theories to harsh tests and should reject them when they fail the tests. Third, scientists should regard theories as conjectures. Passing a test does not confirm a theory or pro­vide one with reason to believe it. It only justifies continuing to employ it (since it has not yet been falsified) and devoting increased efforts to attempting to falsify it (since it has thus far survived testing). Popper has also written in defense of what he calls “situational logic” (which is basically rational choice theory) as the correct method for the social sciences (1967, 1976). There appear to be serious tensions between Popper's falsificationism and his defense of situational logic, and his discussion of situational logic has not been as influential as his falsificationism.

Given Popper's falsificationism, there seems little hope of understand­ing how extreme simplifications can be legitimate or how current economic practice could be scientifically reputable. Even if economic theories were logically falsifiable, the widespread acceptance of Friedman's methodo­logical views insures that they are not subjected to serious test. When they apparently fail tests, they are rarely repudiated. Economic theories, which have not been well tested, are taken to be well established guides to policy, rather than merely conjectures. Some critics of neoclassical economics have made these criticisms (Eichner 1983). But most of those who have espoused Popper's philosophy of science have not repudiated mainstream economics.

Mark Blaug (1992) and Terence Hutchison (1938, 1977, 1978, 2000), who are the most prominent Popperian methodologists, criticize particular features of economics, and they both call for more testing and a more critical attitude. For example, Blaug (1980, ch. 14) praises Gary Becker (1976) for his refusal to explain differences in choices by differences in preferences, but criticizes him for failing to go on and test his theories severely. However, both Blaug and Hutchison understate the radicalism of Popper's views and take his message to be merely that scientists should be critical and concerned to test their theories.

Blaug's and Hutchison's criticisms have sometimes been challenged on the grounds that economic theories cannot be tested, because of their ceteris paribus clauses and the many subsidiary assumptions required to derive testable implications (Caldwell 1984). But this response ignores Popper's insistence that testing requires methodological decisions not to attribute failures of predictions to “interferences.” For views of Popper's philosophy and its applicability to economics, see de Marchi (1988), Caldwell (1991), and Boland (1982, 1989, 1992).

Applying Popper's views on falsification literally would be destructive. Not only neoclassical economics, but all known economic theories would be condemned as unscientific, and there would be no way to discriminate among economic theories. One major problem is that one cannot derive testable implications from theories by themselves. To derive testable impli­cations, one also needs subsidiary assumptions or hypotheses concerning distributions, measurement devices, proxies for unmeasured variables, the absence of various interferences, and so forth. This is the so-called “Duhem-Quine thesis” (Duhem 1906, Quine 1953, Cross 1982). These problems arise generally, and Popper proposes that they be solved by a methodological decision to regard a failure of the deduced testable implica­tion to be a failure of the theory. But in economics the subsidiary assump­tions are dubious and in many cases known to be false. Making the methodological decision that Popper requires is unreasonable and would lead one to reject all economic theories.

Imre Lakatos (1970), who was for most of his philosophical career a follower of Popper, offers a broadly Popperian solution to this problem. Lakatos insists that testing is always comparative. When theories face empirical difficulties, as they always do, one attempts to modify them. Scientifically acceptable (in Lakatos' terminology, “theoretically progressive”) modifications must always have some additional testable implications and are thus not purely ad hoc. If some of the new predictions are confirmed, then the modification is “empirically progressive,” and one has reason to reject the unmodified theory and to employ the new theory, regardless of how unsuccessful in general either theory may be. Though progress may be hard to come by, Lakatos' views do not have the same destructive implications as Popper's. Lakatos appears to solve the problem of how to appraise mainstream economic theory by arguing that what matters is empirical progress rather than empirical success.

Developing Thomas Kuhn's notion of a “paradigm” (Kuhn 1970) and some hints from Popper, Lakatos also developed a view of the global structure of what he called “scientific research programs.” Lakatos emphasized that there is a “hard core” of basic theoretical propositions that define a research program and are not to be questioned within the research program. In addi­tion members of a research program accept a common body of heuristics that guide them in the articulation and modification of specific theories. These views were also attractive to economic methodologists, since theory development in economics is so sharply constrained by prior commitments and specific heuristics. The fact that economists do not give up basic theo­retical postulates that appear to be false might be justified by regarding them as part of the “hard core” of the neoclassical research program.

Yet Lakatos' views do not provide a satisfactory account of how economics can be a reputable science, despite its reliance on extreme simplifications. For it is questionable whether the development of neoclassical economic theory has demonstrated empirical progress. For example, the replacement of “cardinal” utility theory by “ordinal” utility theory in the 1930s, which is generally regarded as a major step forward, involved the replacement of one theory by another that was strictly weaker and which had no addi­tional empirical content. Furthermore, despite his emphasis on heuristics as guiding theory modification, Lakatos still emphasizes testing. Science is for Lakatos more empirically driven than is contemporary economics (Hands 1992). It is also doubtful whether research enterprises in economics have “hard cores” (Hausman 1992b, ch. 6). For attempts to apply Lakatos' views to economics, see Latsis (1976), and Weintraub (1985). As is already apparent in Blaug and de Marchi (1991), writers on economic method­ology have in recent years become increasingly disenchanted with Lakatos' philosophy.

There is a second major problem with Popper's philosophy of science, which plagues Lakatos' views as well. Both maintain that there is no such thing as empirical confirmation (for some late qualms of Lakatos, see Lakatos 1974). Popper and Lakatos maintain that evidence never provides reason to believe that scientific claims are true, and both deny that results of tests can justify relying on statements in practical endeavors or in theoretical inquiry. There is no better evidence for one unfalsified proposition than for another. Someone who questions whether there is enough evidence for some proposition to justify relying on it would be making the methodo­logical “error” of supposing that there can be evidence in support of hypotheses. With the notable exception of Watkins (1984), few philo­sophers within the Popperian tradition have faced up to this implausible consequence.

4.2 The rhetoric of economics

One radical reaction to the difficulties of justifying the reliance on severe simplifications is to deny that economics passes methodological muster. Alexander Rosenberg (1992) maintains that economics can only make impre­cise generic predictions, and it cannot make progress, because it is built around folk psychology, which is an inadequate theory of human behavior and which (owing to the irreducibility of intentional notions) cannot be improved. Complex economic theories are valuable only as applied math­ematics, not as empirical theory. Since economics does not show the same consistent progress as the natural sciences, one cannot dismiss Rosenberg’s suggestion that economics is an empirical dead end. But his view that it has made no progress and that it does not permit quantitative predictions is hard to accept. For example, contemporary economists are much better at pricing stock options than economists were even a generation ago.

An equally radical but opposite reaction is that of Deirdre McCloskey (1985, 1994), who denies that there are any non-trivial methodological standards that economics must meet.

McCloskey’s studies of the rhetoric of economics have been valuable and influential (1985, esp. chs. 5-7), but much of her work consists not of such studies but of philosophical critiques of economic methodology. These are more problematic, because the position sketched in the pre­vious paragraph is hard to defend and potentially self-defeating. It is hard to defend, because epistemological standards for good science have already infected the conversation of economists. The standards of pre­dictive success which lead one to have qualms about economics are already standards that many economists accept. The only way to escape these doubts is to surrender the standards that gave rise to them. But McCloskey’s position undermines any principled argument for a change in standards. Furthermore, as Rosenberg (1988) has argued, it seems that economists would doom themselves to irrelevance if they were to sur­render standards of predictive success, for it is upon such standards that policy decisions are made.

McCloskey (1985, ch. 9) does not, in fact, want to insulate economists from all criticisms. For she herself criticizes the bad habit some economists have of conflating statistical significance with economic importance. Sometimes McCloskey (1985, ch. 2) characterizes rhetoric descriptively as the study of what in fact persuades, but sometimes she characterizes it normatively as the study of what ought to persuade. And if rhetoric is the study of what ought to persuade, then rhetoric is not an alternative to methodology: it is methodology. Questions about whether economics is a successful empirical science cannot be conjured away.

4.3 “Realism” in economic methodology

Economic methodologists have paid little attention to debates within philo­sophy of science between realists and anti-realists (van Fraassen 1980, Boyd 1984), because economic theories rarely postulate the existence of unobservable entities or properties, apart from variants of “everyday un­observables,” such as beliefs and desires. Methodologists have, on the other hand, vigorously debated the goals of economics, but those who argue that the ultimate goals are predictive (such as Friedman) do so because of their interest in policy, not out of interest in epistemological and semantic puzzles concerning references to unobservables.

Nevertheless, there are two important recent realist programs in economic methodology. The first, developed mainly by Uskali Maki, is devoted to exploring the varieties of realism implicit in the methodological statements and theoretical enterprises of economists (see Maki 1990a, 1990b, 1992). The second, which is espoused by Tony Lawson and his co-workers, mainly at Cambridge University, derives from the work of Roy Bhaskar (1978) (see Lawson 1997 and Fleetwood 1999). In Lawson's view, one can trace many of the inadequacies of mainstream economics (of which he is a critic) to an insufficient concern with ontology. In attempting to identify regularities on the surface of the phenomena, mainstream economists are doomed to failure. Economic phenomena are, in fact, influenced by a large number of different causal factors, and one can achieve scientific know­ledge only of the underlying mechanisms and tendencies, whose operation can be glimpsed obscurely in observable relations. Maki's and Lawson's programs have little to do with one another, though Maki (like Mill, Cartwright, and Hausman) shares Lawson's and Bhaskar's concern with underlying causal mechanisms.

4.4 Economic methodology and social studies of science

Throughout its history, economics has been the subject of sociological as well as methodological scrutiny. Many sociological discussions of economics, like Marx's critique of classical political economy, have been concerned to identify ideological distortions and thereby to criticize particular aspects of economic theory and economic policy. Since every political program finds economists who testify to its economic virtues, there is a never-ending source of material for such critiques.

The influence of contemporary sociology of science and social studies of science, coupled with the difficulties methodologists have had making sense of and rationalizing the conduct of economics, has led to a socio­logical turn within methodological reflection itself. Rather than showing that there is good evidence supporting developments in economic theory or that those developments have other broadly epistemic virtues, method­ologists and historians such as D. Wade Hands (Hands and Mirowski 1998, Hands 2001), Philip Mirowski (2002), and E. Roy Weintraub (1991) have argued that these changes reflect a wide variety of non-rational factors, from changes in funding for theoretical economics, to political commitments, personal rivalries, attachments to metaphors, or mathematical interests.

Furthermore, many of the same methodologists and historians have argued that economics is not only an object of social inquiry, but also a tool of social inquiry. By studying the incentive structure of scientific disciplines and the market forces impinging on research (including of course research in economics), it should be possible to write the economics of science and the economics of economics itself (Hull 1988, Hands 1995, Leonard 2002).

Exactly how, if at all, this work is supposed to bear on questions concerning how well supported are the claims economists make is not clear. Though eschewing traditional methodology, Mirowski's monograph (1990) on the role of physical analogy in economics is often very critical of mainstream economics. In Reflection without Rules (2001), Hands maintains that general methodological rules are of little use. He defends a naturalistic view of methodology and is skeptical of prescriptions that are not based on detailed knowledge. But he does not argue that no rules apply.

4.5 Detailed contemporary studies

The above survey of approaches to the fundamental problems of apprais­ing economic theory is far from complete. For example, there have been substantial efforts to apply structuralist views of scientific theories (Sneed 1971, Stegmuller 1976, 1979) to economics (Stegmuller, Balzer, and Spohn 1982, Hamminga 1983, Hands 1985c, Balzer and Hamminga 1989). The above discussion does at least document some of the disagreements con­cerning how to interpret and appraise economic theories. It is not surprising that there is no consensus among those writing on economic methodology concerning the overall empirical appraisal of specific approaches in eco­nomics, including mainstream microeconomics, macroeconomics, and econometrics. When practitioners cannot agree, it is questionable whether those who know more philosophy but less economics will be able to settle the matter. Since the debates continue, those who reflect on economic methodology should have a continuing part to play.

Meanwhile, there are many other more specific methodological questions to address, and it is a sign of the maturity of the sub-discipline that an increasing percentage of work on economic methodology addresses more specific questions. There is a plethora of work, as a perusal of any recent issue of the Journal of Economic Methodology or Economics and Philosophy will confirm. Some of the range of issues currently under discussion were mentioned above in §2. Other areas of current interest include:

1. Feminist economics. Although more concerned with the content of economics than with its methodology, the recent explosion of work on feminist economics is shot through with methodological (and sociological) self-reflection. The fact that a larger percentage of eco­nomists are men than is true of any of the other social sciences and indeed than several of the natural sciences raises methodological questions about whether there is something particularly masculine about the discipline (Nelson 1996). Since 1995, there has been a journal, Feminist Economics, which pulls together much of this work.

2. Economic models. A century ago economists talked of their work in terms of “principles,” “laws,” and “theories.” Nowadays the standard intellec­tual tool is a “model.” Is this just a change in terminological fashion, or does the concern with models signal a methodological shift? What are models? These questions have been discussed by Cartwright (1989, 1999), Hausman (1992b), Maki (1991), Morgan (2001), Morgan and Morrison (1999), Rappaport (1998), and Sugden (2000).

3. Experimental economics. During the past generation, experimental work in economics has expanded rapidly. This work has many different objectives (see Roth 1988) and apparently holds out the prospect of bridging the gulf between economic theory and empirical evidence. Some of it casts light on the way in which methodological commit­ments influence the extent to which economists heed empirical evi­dence. For example, in the case of so-called preference reversals (where people claim to prefer A to B but are willing to pay more for B), econo­mists devoted considerable attention to the experimental findings and conceded that they disconfirmed central principles of economics. But economists were generally unwilling to pay serious attention to the theories proposed by psychologists that predicted the phenomena before they were observed. The reason seems to be that these psycho­logical theories do not have the same wide scope as the basic principles of mainstream economics (Hausman 1992b, ch. 13). The methodological commitments governing theoretical economics are much more com­plex and much more specific to economics than the general rules proposed by philosophers such as Popper and Lakatos.

The relevance of experimentation, however, remains controversial. There are many questions about whether experimental findings can be generalized to non-experimental contexts and, more generally, concerning the pos­sibilities of learning from experiments. See Camerer (2003), Guala (2005), Hey (1991), Kagel and Roth (1995), Plott (1991), V.L. Smith (1991), and Starmer (1999).

5.