literature of labour economics, is taken into account. The difficulty and expense of generating relevant data through intensive methods also needs to be recognized.
But it has proved very hard to turn such insights into a clear set of propositions. What we would need is a statement about, say, different types of custom and how it is expected to influence pay, rather than explanations that are contingent on circumstance. To develop a research programme here one might need to develop a theoretical statement about different sorts of custom. Some customs, for example, have been reasonably strongly embedded in certain occupations while others are much weaker rules of thumb. One could then try to explain how different customs work when as many other factors as possible are held constant. Causal explanation might then be developed in a less inductive way than has hitherto been the case.
A note on method
In these two areas, in short, IR can demonstrate a research programme. But it has been inductive, and it has not lent itself to more formal and rigorous approaches that might speak to more conventional modes of analysis. The inductive approach reflects the established style of inquiry of old institutionalism: very detailed investigation of concrete cases. As Kaufman (2004b) shows, this approach lost influence in the US from the 1950s in favour of more formal model-building. It survived in the UK, for reasons including the relatively late institutionalization of the field but also the fact that concrete ‘labour problems’ at company and workplace level continued to suggest the relevance of detailed case study methods (see Edwards, 1995 and 2005a). The approach continues to thrive, but as argued in section 5 below it needs further development. To establish the content of this development, attention now turns to a key area in which IR research has developed important, but rather implicit, theoretical insights: the debate on UK productivity.
The debate on UK productivity
Tony Lawson (1997) takes as one of his central examples of realist analysis the case of Britain’s relative productivity problem. This is an excellent choice, for the issue has continued to shape academic and public debate (e.g. ESRC, 2004). Yet it will be argued that his analysis misses the opportunity of pressing forward a realist explanation, and in some respects it retreats from such explanation. If realist analysis is to gain any hold then it must be able to show, in terms comprehensible by conventional approaches, how it advances understanding. Going further than Lawson can help in this endeavour. It is notable that conventional approaches continue to pay scant attention to CR, the just mentioned ESRC publication, which is a compilation of leading empirical studies, being a case in point. If a conventional economist were to engage with Lawson’s analysis what would she find?
The conclusion in 1997 is much as it was when the analysis was first presented (Kilpatrick and Lawson, 1980): there is something in the character of trade unions in Britain that has retarded productivity. The explanation
focuses on a set of structures that empower workers . . . to exercise a significant effect on all manner of outcomes. The exercise of this power can be, and usually is, situationally quite rational and comprehensible, and is essentially defensive strength (Lawson, 1997: 257). Conventional economics would have little trouble with this view. As the ESRC (2004: 10) has it, ‘from about 1980, productivity began to rise in the UK, especially in manufacturing as large numbers of jobs were shed and the power of trade unions,