experiment offers a chance to retain jobs, they may co-operate with it. Such conditions are unlikely in themselves to be sufficient for this result. Thus if an experiment is one of a series of failed managerial promises, or if it changes little in reality, acceptance is likely to evaporate rapidly. But where there seems to be concrete change, acceptance can occur. For example, Wright and Edwards (1998) report a case in which managerial intent was demonstrated by the imposition of teams on unwilling line managers and supervisors; in addition, specific and meaningful rights were given to teams, which were thus far more than a rhetorical device.
An important part of this line of research was the demonstration that teams did not have pre-defined causal influences. They had sets of potentials that could be released, sometimes in unintended ways. One study for example found that the language of empowerment used by managers was taken up by workers, who began to hold managers to account against the new image of a democratic organization and who took the rhetoric of empowerment as a means to make their own demands (Rosenthal et al., 1997). Such findings are strongly consistent with the arguments of CR about the indeterminacy of causal powers.
Pay determination and the causal powers of legislation
If there is one issue that illustrates the strengths and weaknesses of institutional analyses in IR, it is the setting of pay. For over 100 years, institutionalists have noted the inability of conventional economic theory to explain the diversity of pay, notably why the ‘same’ occupation in the same labour market can be rewarded differently and why externally mandated pay rises, such as those associated with minimum wages, have not had the expected effects in terms of reduced numbers of jobs.
It is also true that conventional explanations are not always convincing. Consider the two results just noted. The first is commonly rationalized in terms of efficiency wage theory: paying more than a market rate helps to secure worker commitment and induces discretionary effort (see Groshen, 1991). The second is usually explained in terms of monopsony power: employers had this power, and the effect of a rise in wages was to cut the super-normal profits they were making (Card and Krueger, 1995). Yet such explanations are profoundly functionalist in character, assuming that an answer must lie in rational models: if employers pay above ‘market rates’ there must be a functional explanation. Rather than asking how employers in fact behave, they are constrained by extant economic models. Monopsony seems to fit the logic of these models, and because it is thus consistent the assumption seems to be that an explanation has been provided. But it has not. Independent evidence for efficiency wage behaviour or monopsony is not provided, and in the latter case at least it seems unlikely to be available: minimum wages cover low-wage and competitive industries, and it is not clear how such competitive firms have monopsony power. The explanation was simply the most plausible within economic theory, but that does not necessarily make it a good one.
As Rubery (1997) has argued, existing models of wage setting tend to assume that the process is orderly and rational, whereas in fact it is often chaotic and unplanned. An illustration is the study of small firms in competitive markets that were affected by the arrival of the UK National Minimum Wage (Gilman et al., 2002). Given that they were in competitive sectors, and also that institutionalized pay setting through collective bargaining was absent, these firms might be expected to be paying ‘market rates’. Yet they do not have formal pay-setting arrangements such as structures that specify rates for a job. This absence of formal institutions means that pay is set by