IS THERE CROSS-FERTILIZATION IN MACROECONOMICS? A QUANTITATIVE EXPLORATION OF THE INTERACTIONS BETWEEN DSGE AND MACRO AGENT-BASED MODELS Documents de travail GREDEG GREDEG Working Papers Series

Résumé En Fr

This paper compares Dynamic Stochastic General Equilibrium (DSGE) and Macro Agent-Based Models (MABMs) by adopting mainly a distant reading perspective. A set of 2,299 papers is retrieved from Scopus by using keywords related to MABM and DSGE domains. The interactions between the two streams of DSGE and MABM literature are explored by considering a social axis (co-authorship network), and an intellectual axis (cited references and bibliographic coupling). The analysis gave results that are neither consistent with a unitary structure of macroeconomics, nor with a simple dichotomic structure of alternative paradigms and separated academics communities. Indeed, the co-authorship network shows that DSGE and MABM form fragmented communities still belonging to two different larger MABM and DSGE communities rather neatly separated. Collaboration insists mainly inside the smaller groups and inside each of the two larger DSGE and MABM communities. Moreover, the co-authorship network analysis does not show evidence of systematic collaboration between MABM and DSGE authors. From an intellectual point of view, data show that DSGE and MABM articles refer to two different sets of bibliographic references. When a measure of paper-similarity is adopted, it appears that DSGE literature is fragmented in 4 groups while the MABM articles are clustered together in a unique group. Hence, DSGE approach is less monolithic than at the time of the New Synthesis: indeed, a large and a growing literature has developed at the margins of the core DSGE approach which includes elements of heterogeneous agent modelling, social interactions, experiments, expectations formation, learning etc. The analysis gave no evidence of cross-fertilization between DSGE and MABM literature whilst it rather suggests a totally dissymmetric influence of DSGE over MABM literature, i.e., only MABM modelers look at DSGE but not vice-versa. The paper questions the capacity of the current dominant approach to benefit from cross-fertilization.

This paper compares Dynamic Stochastic General Equilibrium (DSGE) and Macro Agent-Based Models (MABMs) by adopting mainly a distant reading perspective. A set of 2,299 papers is retrieved from Scopus by using keywords related to MABM and DSGE domains. The interactions between the two streams of DSGE and MABM literature are explored by considering a social axis (co-authorship network), and an intellectual axis (cited references and bibliographic coupling). The analysis gave results that are neither consistent with a unitary structure of macroeconomics, nor with a simple dichotomic structure of alternative paradigms and separated academics communities. Indeed, the co-authorship network shows that DSGE and MABM form fragmented communities still belonging to two different larger MABM and DSGE communities rather neatly separated. Collaboration insists mainly inside the smaller groups and inside each of the two larger DSGE and MABM communities. Moreover, the co-authorship network analysis does not show evidence of systematic collaboration between MABM and DSGE authors. From an intellectual point of view, data show that DSGE and MABM articles refer to two different sets of bibliographic references. When a measure of paper-similarity is adopted, it appears that DSGE literature is fragmented in 4 groups while the MABM articles are clustered together in a unique group. Hence, DSGE approach is less monolithic than at the time of the New Synthesis: indeed, a large and a growing literature has developed at the margins of the core DSGE approach which includes elements of heterogeneous agent modelling, social interactions, experiments, expectations formation, learning etc. The analysis gave no evidence of cross-fertilization between DSGE and MABM literature whilst it rather suggests a totally dissymmetric influence of DSGE over MABM literature, i.e., only MABM modelers look at DSGE but not vice-versa. The paper questions the capacity of the current dominant approach to benefit from cross-fertilization.