« Ray Fair: Interpreting the Predictive Uncertainty of Presidential Elections | Main | How Independent Was the Committee on Capital Markets Regulation? »

Thursday, November 30, 2006

Methodological Issues in Economics

Continuing a recent conversation on methodological issues within economics, this is part of a much longer article "The Philosophy of Economics" from The Stanford Encyclopedia of Philosophy. It was written by Daniel M. Hausman:

2. Six central methodological problems

Although the different branches and schools of economics raise a wide variety of methodological issues, six problems have been central to methodological reflection concerning economics:

2.1 Positive versus normative economics

Policy makers look to economics to guide policy, and it seems inevitable that even the most esoteric issues in theoretical economics may bear on some people's material interests. The extent to which economics bears on and may be influenced by normative concerns raises methodological questions about the relationships between a positive science concerning “facts” and a normative inquiry into what ought to be. Most economists and methodologists believe that there is a reasonably clear distinction between facts and values, between what is and what ought to be, and they believe that most of economics should be regarded as a positive science that helps policy makers choose means to accomplish their ends, though it does not bear on the choice of ends itself.

This view is questionable for several reasons. First economists have to interpret and articulate the incomplete specifications of goals and constraints provided by policy makers (Machlup 1969b). Second, economic “science” is a human activity, and like all human activities it is governed by values. Those values need not be the same as the values that influence economic policy, but it is questionable whether the values that govern the activity of economists can be sharply distinguished from the values that govern policy makers. Third, much of economics is built around a normative theory of rationality. One can question whether the values implicit in such theories are sharply distinguishable from the values that govern policies. For example, it may be difficult to hold a maximizing view of individual rationality, while at the same time insisting that social policy should resist maximizing growth, wealth, or welfare in the name of freedom, rights, or equality. Fourth, people's views of what is right and wrong are, as a matter of fact, influenced by their beliefs about how people in fact behave. There is evidence that studying theories that depict individuals as self-interested leads people to regard self-interested behavior more favorably and to become more self-interested (Marwell and Ames 1981, Frank et al. 1993). Finally, people's judgments are clouded by their interests. Since economic theories bear so centrally on people's interests, there are bound to be ideological biases at work in the discipline (Marx 1867, Preface).

2.2 Reasons versus causes

Orthodox theoretical microeconomics is as much a theory of rational choices as it a theory that explains and predicts economic outcomes. Since virtually all economic theories that discuss individual choices take individuals as acting for reasons, and thus in some way rational, questions about the role that views of rationality and reasons should play in economics are of general importance. Economists are typically concerned with the aggregate results of individual choices rather than with particular individuals, but their theories in fact offer both causal explanations for why individuals choose as they do and accounts of the reasons for their choices.

Explanations in terms of reasons have several features that distinguish them from explanations in terms of causes. Reasons justify the actions they explain. Reasons can be evaluated, and they are responsive to criticism. Reasons, unlike causes, must be intelligible to those for whom they are reasons. On grounds such as these, many philosophers have questioned whether explanations of human action can be causal explanations (von Wright 1971, Winch 1958). Yet merely giving a reason — even an extremely good reason — fails to explain an agent's action, if the reason was not in fact “effective.” Someone might, for example, start attending church regularly and give as his reason a concern with salvation. But others might suspect that this agent is deceiving himself and that the minister's attractive daughter is in fact responsible for his renewed interest in religion. Donald Davidson (1963) argued that what distinguishes the reasons that explain an action from the reasons that fail to explain it are that the former are also causes of the action. Although the account of rationality within economics differs in some ways from the “folk psychology" people tacitly invoke in everyday explanations of actions, many of the same questions carry over (Rosenberg 1976, ch. 5; 1980).

An additional difference between explanations in terms of reasons and explanations in terms of causes, which some economists have emphasized, is that the beliefs and preferences that explain actions may depend on mistakes and ignorance (Knight 1935). As a first approximation, economists can abstract from such difficulties. They thus often assume that people have perfect information about all the relevant facts. In that way theorists need not worry about what people's beliefs are. By assumption people believe and expect whatever the facts are. But once one goes beyond this first approximation, difficulties arise which have no parallel in the natural sciences. Choice depends on how things look “from the inside", which may be very different from the actual state of affairs. Consider for example the stock market. The “true” value of a stock depends on the future profits of the company, which are of course uncertain. In 1999 and 2000 stock prices were far above any plausible estimate of their true value. But what matters, at least in the short run, is what people believe. No matter how overpriced shares might be, they were excellent investments if tomorrow or next month somebody would be willing to pay even more for them. Economists disagree about how significant this subjectivity is. Members of the Austrian school argue that these differences are of great importance and sharply distinguish theorizing about economics from theorizing about any of the natural sciences (Buchanan and Vanberg 1989, von Mises 1981).

2.3 Social scientific naturalism

Of all the social sciences, economics most closely resembles the natural sciences. Economic theories have been axiomatized, and articles and books of economics are full of theorems. Of all the social sciences, only economics boasts a Nobel Prize. Economics is thus a test case for those concerned with the extent of the similarities between the natural and social sciences. Those who have wondered whether social sciences must differ fundamentally from the natural sciences seem to have been concerned mainly with three questions:

(i) Are there fundamental differences between the structure or concepts of theories and explanations in the natural and social sciences? Some of these issues were already mentioned in the discussion above of reasons versus causes.

(ii) Are there fundamental differences in goals? Philosophers and economists have argued that in addition to or instead of the predictive and explanatory goals of the natural sciences, the social sciences should aim at providing us with understanding. Weber and others have argued that the social sciences should provide us with an understanding “from the inside", that we should be able to empathize with the reactions of the agents and to find what happens “understandable” (Weber 1904, Knight 1935, Machlup 1969a). This (and the closely related recognition that explanations cite reasons rather than just causes) seems to introduce an element of subjectivity into the social sciences that is not found in the natural sciences.

(iii) Owing to the importance of human choices (or perhaps free will), are social phenomena too “irregular” to be captured within a framework of laws and theories? Given human free will, perhaps human behavior is intrinsically unpredictable and not subject to any laws. But there are, in fact, many regularities in human action, and given the enormous causal complexity characterizing some natural systems, the natural sciences must cope with many irregularities, too.

2.4 Abstraction, idealization, and ceteris paribus clauses in economics

Economics raises questions concerning the legitimacy of severe abstraction and idealization. For example, mainstream economic models often stipulate that everyone is perfectly rational and has perfect information or that commodities are infinitely divisible. Such claims are exaggerations, and they are clearly false. Other schools of economics may not employ idealizations that are this extreme, but there is no way to do economics if one is not willing to simplify drastically and abstract from many complications. How much simplification, idealization, and abstraction is legitimate?

In addition, because economists attempt to study economic phenomena as constituting a separate domain, influenced only by a small number of causal factors, the claims of economics are true only ceteris paribus — that is, they are true only if there are no interferences or disturbing causes. What are ceteris paribus clauses, and when if ever are they legitimate in science? Questions concerning ceteris paribus clauses are closely related to questions concerning simplifications and idealizations, since one way to simplify is to suppose that the various disturbing causes or interferences are inactive and to explore the consequences of some small number of causal factors. These issues and the related question of how well supported economics is by the evidence have been the central questions in economic methodology. They will be discussed further below in Section 3 and elsewhere.

2.5 Causation in economics and econometrics

Many important generalizations in economics are causal claims. For example, the law of demand asserts that a price increase will (ceteris paribus) diminish the quantity demanded. Econometricians have also been deeply concerned with the possibilities of determining causal relations from statistical evidence and with the relevance of causal relations to the possibility of consistent estimation of parameter values. Since concerns about the consequences of alternative policies are so central to economics, causal inquiry is unavoidable.

Before the 1930s, economists were generally willing to use causal language explicitly and literally, despite some concerns that there might be a conflict between causal analysis of economic changes and “comparative statics” treatments of equilibrium states. Some economists were also worried that thinking in terms of causes was not compatible with recognizing the multiplicity and mutuality of determination in economic equilibrium. In the anti-metaphysical intellectual environment of the 1930s and 1940s (of which logical positivism was at least symptomatic), any mention of causation became highly suspicious, and economists commonly pretended to avoid causal concepts. The consequence was that they ceased to reflect carefully on the causal concepts that they continued implicitly to invoke (Hausman 1983, 1990, Helm 1984, Runde 1998). For example, rather than formulating the law of demand in terms of the causal consequences of price changes for quantity demanded, economists tried to confine themselves to discussing the mathematical function relating price and quantity demanded. There were important exceptions (Haavelmo 1944, Simon 1953, Wold 1954), and during the past generation, this state of affairs has changed dramatically.

For example, in his Causality in Macroeconomics (2001) Kevin Hoover develops feasible methods for investigating large scale causal questions, such as whether changes in the money supply (M) cause changes in the relate of inflation P or accommodate changes in P that are otherwise caused. If changes in M cause changes in P, then the conditional distribution of P on M should remain stable with exogenous changes in M, but should change with exogenous changes in P. Hoover argues that historical investigation, backed by statistical inquiry, can justify the conclusion that some particular changes in M or P have been exogenous. One can then determine the causal direction by examining the stability of the conditional distributions. Econometricians have made vital contributions to the contemporary revival of philosophical interest in the notion of causation. In addition to Hoover's work, see for example Geweke (1982), Granger (1969, 1980), Cartwright (1989), Sims (1977), Zellner and Aigner (1988).

2.6 Structure and strategy of economics

In the wake of the work of Kuhn (1970) and Lakatos (1970), philosophers are much more aware of and interested in the larger theoretical structures that unify and guide research within particular research traditions. Since many theoretical projects or approaches in economics are systematically unified, they pose questions about what guides research, and many economists have applied the work of Kuhn or Lakatos to shed light on the overall structure of economics (Baumberg 1977, Blaug 1976, Blaug and de Marchi 1991, Bronfenbrenner 1971, Coats 1969, Dillard 1978, Hands 1985b, Hausman 1992, ch. 6, Hutchison 1978, Latis 1976, Jalladeau 1978, Kunin and Weaver 1971, Stanfield 1974, Weintraub 1985, Worland 1972). Whether these applications have been successful is controversial, but the comparison of the structure of economics to Kuhn's and Lakatos' schema has at least served to highlight distinctive features of economics. For example, asking what the “positive heuristic” of mainstream economics consists in permits one to see that mainstream models typically attempt to demonstrate that an economic equilibrium will obtain, and thus that mainstream models are unified in more than just their common assumptions. Since the success of research projects in economics is controversial, understanding their global structure and strategy may clarify their drawbacks as well as their advantages.

    Posted by on Thursday, November 30, 2006 at 07:07 PM in Economics, Methodology | Permalink  TrackBack (1)  Comments (22)


    TrackBack URL for this entry:

    Listed below are links to weblogs that reference Methodological Issues in Economics:

    » Brad DeLong has the best Economics discussion of the week from Marginal Utility

    Brad DeLong (last seen getting battered and bruised here) shows why one must fight on as he gets into the details of Paritha Dasgupta's discussion of the Stern report, which Arnold Kling (as usual) hacks into misinterpretation. [Read More]

    Tracked on Friday, December 01, 2006 at 12:21 PM


    Feed You can follow this conversation by subscribing to the comment feed for this post.