Lawrence A. Boland's Model building in economics: its purposes and limitations. Cambridge: Cambridge University Press, 2014, 298 pp.

Abstract

Review of Lawrence A. Boland's Model building in economics: its purposes and limitations. Cambridge: Cambridge University Press, 2014, 298 pp.The purposes and limitations of all types of models used in economics, past and beyond. This is the rather ambitious topic of Lawrence Boland's new book. The book reviews standard microeconomics, macroeconomics, game theory (including evolutionary), experimental economics, and econometrics (both micro and macro). In its mission statement, the book is declared methodology with a small 'm', to be distinguished with a capitalized Methodology discussed by Philosophers. Methodology discussed by Philosophers deals with lofty issues of 'Realism', 'Testing', and 'Explanation', whereas methodologists are interested in, well, such things as the realisticness of assumptions, the testing of models, and explanation. In the end, the biggest drawback of the book ends up being a seemingly aesthetic aversion to completely imagined capital letters. It is clear that the book discusses the very same issues that have vexed philosophers of science and philosophers of economics for at least the last few decades, and arguably for centuries, yet it refuses to engage these resources. This is a regrettable, but perhaps inevitable, consequence of the limits of any one person trying to cover the whole of economics and the relevant philosophy of economics.Boland begins with a thesis about a paradigm and generational shift in economic modeling around the beginning of the 1980s. Economists having their postgraduate studies in the 1980s or later no longer conceive models as being models of a theory, built in order to provide an instrument with which to empirically test that theory. Instead, model building began to be equated with theorizing itself. This shift is argued for in a concise, yet convincing manner. And this shift does not sit all that well with the Popperian spirit of Boland's Methodology (yes, with a capital 'M'). Although the book does not attempt to systematically build a single argument or an encompassing philosophical position, the overall message emerging from the discussion seems to be a plea for more interest in the realisticness and empirical testing of modeling assumptions.The book first goes through the stylized history of the division between micro and macroeconomics. It dates the beginning of modern microeconomics to Marshall and proceeds from there to discuss the need for, and the plausible form of, microfoundations. Standard criticisms against the use of representative agent constructs and the sterility of general equilibrium theory are also raised. The biggest faults located by Boland in theoretical macro models based on equilibrium assumptions are their lack of dynamics and any idea of adjustment process for prices. There is nothing new in these accusations, but Boland raises the issues with a peculiar twist. It would not, in fact, be enough to endow the model with dynamics by having the endogenous variables be determined by their lagged values and additional exogenous variables. This would be merely a combination of exogenous shocks and a predetermined time path, not truly endogenous dynamics resulting from the actions of free agents. Approvingly quoting Hicks, Boland holds that economics must not only include time, but be in time.The argument here is that the problem is not just in the lack of an empirically credible adjustment process, but in the fact that the conceptions of learning and expectation formation implicit in any proposed sketch of an adjustment process are flawed, because they have always been based on inductive reasoning. And inductive reasoning was demonstrated to be impossible by Hume. This foundational problem of the groundlessness of inductive learning is also raised with respect to agents within game-theoretical models: common knowledge of rationality either assumes a foundationally irrational inductive learning process or simply assumes away the whole 'problem of induction'. …