Philosophy of Economics.

This book, which reprints a double issue of the philosophy journal Erkenntnis, contains 13 papers from a 1987 conference. The essays raise a variety of conceptual issues. How sound are economic arguments? What is the nature of economic explanation and prediction? How empirical is general equlibrium theory? Are contending macroeconomic theories empirically distinguishable? How are economic theory and data related? What roles do abstraction, specialization, deduction, and approximation play in economic modeling? To what extent is economics science?

Hausman analyzes a class of economic arguments including famous claims by Muth that firm expectations and economic theory predictions coincide, and by Becker and Friedman that firms don't racially discriminate. He concludes that such arguments are logically valid, but that economists rarely check their often implausible premises. Nelson formulates necessary conditions for an average to explain a single event well, and argues that the conditions don't hold in economics. So, for example, average consumer behavior doesn't explain the behavior of an individual consumer well. Nelson also claims that though we have empirically useful econometric models of market demand, we don't yet know if individual consumer behavior explains those models.

Rosenberg argues that economic theory makes only qualitative predictions and asks if those are enough for science. He notes that classical macroeconomics implies that labor market equilibrium is possible while Keynesian macroeconomics implies that labor market disequilibrium is possible. Balzer and Haendler relate ordinary least squares parameter estimation in an econometric market demand model to a philosophical theory of measurement. They argue that a natural scientist repeatedly remeasures the same situation, justifying application of probabilities and statistics. In contrast, a social scientist remeasures a quickly changing situation - e.g., econometric time series are not stationary.

Common knowledge that p is true means that everyone knows that p is true, everyone knows that everyone knows, ad infinitum. Bicchieri demonstrates that the existence of a unique equilibrium solution in some simple games implicitly assumes the absence of common knowledge of things like players' strategies. She argues that in real situations such common knowledge; rather than player bounded rationality, altruism, or imperfect information; may explain why players don't choose the apparent unique equilibrium solution. De la Sienra notes conceptual problems that arise in trying to employ mathematical learning theory to model how competitive firms arrive at equilibrium prices and outputs.

M. and D. Pearce claim that science is ideas while technology is partly things, e.g., machines. So, they argue, Kuhn's ideas of scientific paradigm and revolution may not help to understand technological change, which is evolutionary in their view. Diederich shows that Marx made mathematical mistakes in his attempt to explain market prices with the concept of surplus value. He then indicates how to relate prices to surplus value in an input-output model of a production economy which includes the commodity labor power. Sneed sketches how to allocate resources over N time periods to solve scientific problems so as to maximize an objective function supposed to capture the extrinsic value of the solutions.

Janssen forms a macroeconomic model with three equations and adds an equation to make it classical, or another to make it Keynesian. From his analysis of the models he concludes that, barring questionnaire evidence, whether unemployment is voluntary or involuntary is not an empirical question at the present stage of applied economic research. Janssen and Kuippers argue that reconstruction of general equilibrium research requires two imperfectly linked theories - of individual and market demand. They note that questionnaires are a possible application of the first theory from which, however, economists shy away.

Nowak tries to account for the roles of abstraction, deduction, and approximation in economic modeling. First abstract an ideal economy and deduce some conclusions about it from accumulated theories of background knowledge. Gradually make the economy more concrete by allowing variables previously abstracted to zero to take their actual values, deducing conclusions about the economy at each stage. In the final stage, deduce the relations of an econometric model which approximate those of the actual economy. Hamminga offers a contrasting account. Deduce general conclusions about the economy from conjectured general features. Then specialize generalizations and deduce from them an econometric model whose relations approximate those of the economy.

Do economists fail to check premises of arguments out of laziness or a propensity for prejudgment (as Hausman suggests), or because the premises are often not specific enough to check? For example, do the general equilibrium or production theories of argument premises imply any empirically testable predictions about firms? The possibilities of labor market equilibrium and disequilibrium are compatible predictions, which therefore fail to distinguish classical and Keynesian macroeconomics. Both predictions are not specific enough to check against a real economy. These difficulties, rather than their qualitative nature, leave the predictions of contending macroeconomic theories just shy of science.

Do Balzer and Haendler want to claim that applying probabilities and statistics in econometrics is not justified? Can explanation succeed without statistical theory? For example, does just repeatedly refitting possibly different functions to an emerging nonstationary time series help to explain it? By what criterion of empirically useful do we have empirically useful econometric models of market demand?

General equilibrium research aims to prove existence, uniqueness, and stability of equilibrium from assumptions on individual consumers and firms. The necessity of a two theory reconstruction suggests that so far the research has failed to reach its goal for uniqueness and stability. Are the demand functional forms of an econometric model deduced from (or even compatible with) economic theory? Doesn't theory only imply some of the variables to include in the model while intuition provides the rest? Is such a process, together with repeated function refitting, a successful form of explanation? Deductive explanation in economics seems a little realized ideal, borrowed from natural sciences which employ it more successfully.

Philosophy of Economics ends with a useful bibliography of previous work by its authors. Not each paper will interest, nor every argument convince, all economists. The book suffers from common conference volume maladies - e.g., insufficient editing and revision, and lesser papers by important figures. Also, many papers enclose the economist's mathematics within a model theory structure from advanced mathematical logic that, with little training, is hard to penetrate. Taken as a whole, however, the essays provoke considerable thought about the most important conceptual issues in economics.