It has long been an orthodoxy in economics that increases in the minimum wage will result in reductions in employment. Based on a strong theoretical construct with strong policy implications, this orthodoxy bas been buttressed by numerous studies showing that employment does decrease, particularly among the teen labor market. The policy implications are important because those opposed to minimum wage increases have a powerful ally in the orthodoxy. Instead of justifying their opposition on selfish and/or self-interest grounds, they need do no more than appeal to the orthodoxy on supposedly utilitarian grounds. And in the absence of strong evidence to the contrary, the orthodoxy not only achieves greater credibility, but it also serves as a powerful political tool. In recent years, however, new findings have emerged that do call into question the reigning orthodoxy. This, then, poses the following policy dilemma: What happens when new findings deeply rooted in social science methodology challenge a reigning orthodoxy just as deeply rooted in social science theory? And yet, this dilemma would appear to underscore a more critical question: To the extent that models and research do affect policy and the policy process, how is that process affected by new research? In very simplified form, under the orthodoxy, the policy maker when confronted with a question of what to do--particularly on the minimum wage--knew exactly what to do. The answer, after all, was dictated by the orthodoxy. But what does the policy maker do when confronted with new findings that effectively challenge this orthodoxy? This, of course, raises the larger question of just what the relationship is between social science and public policy. My concern in this paper is less with the merits of these recent findings than with the implications of such findings. Rather than presenting new findings, I intend to discuss the significance of new findings for the public policy process.
I argue that social science research that cal[s into question a reigning orthodoxy also opens up a whole array of public policy questions that ultimately have implications for public administration. By challenging an orthodoxy that has apparently been authoritative, that challenge effectively creates ambiguity. But contrary to the literature on decision making that holds ambiguity to be bad because it increases uncertainty (March 1988), I argue that greater ambiguity, particularly with regard to the minimum wage, ought to be regarded as a social good. It is a social good because it furthers the democratic process by forcing policy makers to respond and be accountable. Indeed, it forces them to in fact justify their policy actions in more coherent terms because the model that has for so long been dispositive now lacks the necessary authority to determine the policy outcome. No longer can a policy maker defer to the authority of the model; rather he or she must offer a reasoned rationale and justification for whatever policy outcomes are decided. Consequently, there are other directions for policy equally worthy of consideration. The policy maker, in other words, can no longer hide behind the authority of the model, which no doubt provides nice cover in a process laden with interest groups. For those who cling to the reigning orthodoxy, new data certainly are not going to alter their position because the orthodoxy is as inviolable as religious belief. But with new data, social science research has effectively provided conflicting results. Broadly speaking, the policy process is left open to experimentation and experimentation that may better capture the values of the political community.
The Minimum Wage Orthodoxy
The minimum wage orthodoxy is predicated on a neoclassical synthesis which maintains that in a competitive market, each worker receives the value of his or her marginal revenue product of labor (the amount of increase in the output that results from an increase in, say a unit of labor). If adding an additional worker results in a rise in total revenues, the firm's output will rise as a result. A wage floor, such as a mandated minimum wage, prevents the cost of labor from dropping below a set rate. With a minimum wage higher than the equilibrium wage, fewer workers will be hired than are willing to work, resulting in unemployment. As the cost of labor increases due to a mandated minimum wage that is higher than the market-clearing wage, firms will hire fewer workers and employment will drop. A minimum wage higher than the equilibrium wage results either in the layoff of those workers whose value is less than the minimum or the requirement that low-efficiency workers improve their productivity (Stigler 1946; Mansfield 1994; Schansberg 1996).
When applied to competitive labor markets in the real world, the disemployment effects predicted by the model are quite variable. Employment effects vary according to which segment of the labor market is being observed and the nature of the methodology being employed. And yet, despite the variation, employment effects are judged to be relatively small, particularly among adults (Brown, Gilroy, and Kohen 1982). It's among the teen labor market that disemployment effects are judged to be higher. In fact, most empirical literature in support of the theoretical construct is specifically centered on the teen labor market (Kosters and Welch 1972; Welch 1974, 1978; MWSC 1981; Meyer and Wise 1983; Neumark and Wascher 1992; Kosters 1996). Moreover, this focus on the teen labor market has only been buttressed by a federal Minimum Wage Study Commission in 1981 that found that a 10 percent increase in the minimum wage leads to a 1-3 percent reduction in employment, especially among teenagers. Since this commission report was issued, it has been the consensus among the economics profession that minimum wages lead to disemployment effects. Therefore, to the extent that the theoretical construct serves as the basis of the orthodoxy, this orthodoxy is only buttressed by this particular consensus.
The orthodoxy, however, is hot without its flaws. First and foremost, it assumes competitive markets and that in competitive markets, workers can determine the overall demand for labor simply by adjusting their wage demands. But in the real world, the demand for labor is based on output and aggregate demand for goods and services (Lester 1946, 1947; Palley 1998). Employers do not demand more labor because the price of labor has decreased but because consumers demand more of the goods and services that they produce. It is this need to produce more that leads employers to hire additional workers. Secondly, the orthodoxy is predicated on problematic data. Much of the data are based on time-series data which merely look at changing employment rates at various points in time, usually around when a minimum wage increase has occurred. Then through statistical testing, attempts are made to establish a correlation between an increase in the minimum wage and a change in employment. Statistical tests at best show correlations, and even strong ones, but they do not necessarily establish causal relationships. In the end, it cannot be established with authority that a particular minimum wage increase results in a disemployment effect--that it was specifically that increase that caused the disemployment effect, as opposed to other circumstances, i.e., an economic downturn, the loss of markets, and so on. It does not tell us anything about the behavior of firms per se; rather it provides a basis upon which their behavior can be inferred.
Even attempts to get around the inherent weaknesses of time-series data have their limitations. Janet Currie and Bruce Fallick (1996), for instance, used individual-level data from the National Longitudinal Survey of Youth (NLSY) to examine disemployment effects of minimum wage increases in 1979 and 1980. They also examined the effects of the minimum wage on individual year-to-year changes. Their approach was specifically to measure the effects by identifying a group of workers that would most likely be directly affected by the minimum wage and then compare their employment to the employment of a group of workers who were least likely to be affected. Affected workers were found to be 3 percent less likely to be employed a year later. At the same time, no evidence was found that minimum wage increases affected the wages of workers who remained employed a year later. Still, they were not able to establish that it was the increase in the minimum wage that was a direct cause of their not being employed a year later as opposed to other factors. In the end, all that can be made are inferences, and to the extent that it provides the basis for inference, it serves to support the theoretical construct contained in the traditional textbook. On the contrary, all that can be made is a plausible inference based on statistical tests bound to show a high likelihood. And yet, based on changes in employment patterns following, say, a minimum wage increase, a further inference is made about the behavior of firms, which only reinforces the reigning orthodoxy.
All of this raises the obvious question: why despite these flaws has the orthodoxy been so entrenched? There are several possible explanations. The first has to do with the composition of the minimum wage labor market itself. Because only a small fraction of the labor market actually earns the statutory minimum wage, the potential benefits are presumed to be so small that they could not possibly offset the more likely larger costs. Moreover, the benefits are presumed to be smaller still because most minimum wage earners are not the primary earners in their households but are secondary earners and by and large teenagers (Brown 1996; Burkhauser and Finegan 1989). So because these incomes are considered to be less consequential to the maintenance of their households, the potential...