Structure, Agency and Historical Institutionalism

Abstract

As be®ts two if its principal exponents, Hall and Taylor's recent article `Political science and the three new institutionalisms' provides a meticulous and provocative review of the many faces of the `new institutionalism' and a distinctive contribution to the growing literature in this area in its own right.* It provides an important opportunity to consider again the strengths and weaknesses of contemporary institutionalism and to raise the question of how its many insights might be more fully incorporated within the British political science mainstream. While careful to distance themselves from the idea that a `crude synthesis' of rational choice, sociological and historical institutionalism is `immediately practical or even necessarily desirable' (p. 957), they suggest that a dialogue between them is both necessary and crucial. We argue that the prospects for such a dialogue are more limited than Hall and Taylor suggest. For, rational choice and sociological institutionalisms are based on mutually incompatible premises or `social ontologies'. Moreover, in identifying two social ontologies ± the calculus and cultural approaches ± within the historical institutionalist canon (and hence in reconstructing historical institutionalism in rational choice and sociological terms), we argue that Hall and Taylor do a considerable disservice to this distinctive approach to institutional analysis. While this view of historical institutionalism makes it appear `pivotal' to future dialogue between institutionalisms, such a reading neglects the potentially distinctive social ontology of this approach. This may leave historical institutionalism prone to precisely the tendential structuralism characteristic of much institutionalist analysis, while giving a super®cial impression that the approach has already overcome this problem. We argue that if institutionalism is to develop to its full potential, it must consider the relationship between structure and agency, on which Hall and Taylor merely touch, as a central analytic concern.