Reviewed by
Charlotte Bruun
Aalborg University, Denmark
Whereas agent-based computational economics started out in the mid-nineties as a new methodological approach to mainly microeconomics, Macroeconomics from the Bottom-up proves that agent-based macroeconomics is finally maturing as a field of research. Why it has taken so long for agent-based macroeconomics to mature becomes evident in the introduction, where a very ambitious research agenda is presented. It is the goal of Delli Gatti and colleagues not only to compete with the dominant Dynamic Stochastic General Equilibrium (DSGE) modeling tradition and its standards for empirical validation. They also aim to study financial dynamics following Keynes and Minsky.The authors present a critique of the "mainstream weltanschauung" which is mainly focused on methodological issues, and my main concern is whether a new methodological approach can replace a weltanschauung? This concern is related to the fact that Axtell (2005), among others, already introduced the agent-based modeling technique into the weltanschauung of general equilibrium theory. One can discuss the success of this approach, but Axtell demonstrated that general equilibrium theory and agent-based modelling are not mutually excludable. There need not be more to agent-based macroeconomics than the introduction of a new tool which initiates a reconsideration of assumptions and theorems within the old weltanschauung. But, when Delli Gatti and colleagues refer to the old weltanschauung, we are left with the impression that the aim is to build up a new weltanschauung.
Whereas Macroeconomics from the Bottom-up is very rich of methodological discussions, the authors are not so clear when they present the economic theory on which they base their model. The reader has to search to find the small pieces of information that can be collected into a macroeconomic theory. For example, they are keen on rule-based behavior as opposed to optimization. However, no matter which methodology is eventually chosen, there must be something else beneath the exact method applied, driving behavior. If not, what is to be optimized or what generates rules? In the Bottom-up Agent-based Model (BAM), it is difficult to identify the goals individuals are claimed to seek (p.7).
With respect to the macroeconomic set up, I encounter similar problems. Firms need to pay out monetary wages before they sell their product and, thus, they need to finance production. A credit institution is introduced to serve this purpose. However, how can firms generate aggregate monetary profits, if profits are determined as receipts minus expenditures? Firms directly get back the part of the wage bill that is spend on consumption and, by an (apparently) implicit stock market, they get back the money saved out of the wage bill by selling equity. But, the stock market of the BAM does not have the function of creating a "conventional valuation" emphasized in a quote by Keynes (p.3), i.e. there is no stock market with forward looking agents placing a monetary valuation on future profit generating abilities of firms. No wonder that the system needs to get rid of some of this accumulating debt through bankruptcies. But, are bankruptcies by big firms really responsible for all financial turmoil?
With the focus on presenting a new methodological approach to macroeconomics, and insisting on empirical validation, such foundational problems of economics are relegated to the background. But, is being able to replicate stylized facts, or even empirical data, sufficient? It could be argued that a more direct confrontation with foundational theoretical problems of economics, would provide the resulting model with a theoretical consistency that may linger even after stylized facts or conventions have broken down. That in a truly constructive approach understanding can be just as important as data replication.
To sum up, I believe that this book should be read neither for the agent-based macroeconomic model it presents, nor for its postulated capability of replicating stylized facts. Rather it should be read for its methodological discussions and applications. Macroeconomics from the Bottom-up can persuade many students that agent-based modeling is one of the best options that economists have for understanding what is currently happening in the real world, i.e. for understanding financial crisis.
References
Return to Contents
of this issue