In brief, efficiency wage theory argues the following. Employer and employee are playing a game. Employer would like the employee to work harder than she usually does. Thus, the employer offers employee a higher wage in order to entice more effort. The problem is, the employer can always go back on his promise by not paying the higher wage at the end of the day. However, then the next day employee comes into work, she will simply resort to a lower level of effort. Hence, a "commitment problem" exists. Commitment to a higher wage and higher effort can only happen if both sides keep their promises to work hard over a long period of time.
There are many ways of criticizing the efficiency wages literature. Neoclassical and other conservative economists say commitment, in the sense of the above story, simply cannot last, since employee may stop exerting high effort at any time. Also, employer may, at any time, go back on his promise to pay a higher wage for more effort -- and he might just hire a new employee if the previous one quits. Marxists say the employment relationship is not a "game" where commitment must be maintained in the labor process -- the labor process is ultimately about class struggle between the owner of the means of production (employer) and those from whom the employer extracts profits above and beyond their compensation (employee). Historians may point to the fact that other conditions -- sociological, historical, political -- are more important for determining the formation of employment contracts (such as the "efficiency wage" contract) at any given time.
I should add that both the Marxists and Neoclassicals have a separate political issue with the conclusions of the theory, aside from any details of the theory itself. For example, some Marxists don't like the fact that the policy implication of the efficiency wage theory is, higher wages for workers. Marxists are primarily interested in the disestablishment of labor markets and of capitalism itself. Thus, any policy that stresses higher wages still affirms the underlying "good" of capitalism. (Neoclassicals don't like it because they want wages to be determined by market forces -- not "soft" forces such as reputation or loyalty.)
The historical critique is especially powerful because it has a lot of sharp things to say about the famous examples to which efficiency wage supporters love to draw attention. For example, the popular model of welfare capitalism prior to the Great Depression highlights the "high road" labor relations pursued by employers and employees in terms of generous benefits, high wages, and good working conditions in the absence of strong unions. This "commitment equilibrium" broke down during the Great Depression simply because the economic pressures prevented employers from keeping their promises, leading to a forced new equilibrium of a highly unionized workforce and bargaining between employers and employees enforced by the state.
This historical critique of this view (particularly, these are my own thoughts) draws on the fact that there is a long history of state involvement in labor issues prior to the New Deal, and since unions were not legal during this time period, employees were not completely free to bargain. Contracts were therefore not primarily about commitment issues between the employer and employee -- they were about the uncertain legal possibilities faced by employees. Higher wages may have existed for some workers, but they were not the norm in any industry. Nor were they offered in order to induce "commitment", since commitment by labor was not exactly voluntary, given the uncertainty revolving around whether they would be successful in strikes (due to both state enforcement and various legal instruments used against employees during this time). When employment law is understood in a broader legal context, seeing contract formation as an issue of enforcing commitment does not hold up to the legal facts.
To be honest, I sympathize with all three of the critiques. The neoclassicals have it right by saying that commitment is difficult to achieve. Marxists have it right by saying that the employment relationship shouldn't be framed as a problem of forming a risk-sharing contract. Historians have it right by saying that many other factors influence contract formation. I also tend to agree with the main political concern of the Marxists in my view that labor should not be a commodity. The problem is that the efficiency wage view is the dominant view of mainstream economics at this time. The same theoretical framework used to derive efficiency wage equilibria has been used in everything from the evolution of property rights, to calls for the regulation of the financial sector. How to chip away at such a hegemonic theory?
Let's go back to this idea of commitment, the centerpiece of it all. Perhaps the efficiency wage theorists have a point: not everything can be accounted for in a contract. Since the employer can't completely see everything the employee does, in order to entice the employee to do those nonobservable things in a loyal, respectable manner, the employer will have to offer some wage premium. The game theory literature on this says that commitment will occur in the following situation.
Employer offers a higher wage to employee at the conclusion of employee's week on her job as long as employee works harder than she would have otherwise. Suppose employer contemplates cheating employee. Clearly, the employer would gain in this scenario [A]: employer pays for cheap but may end up getting a lot of effort out of employee if she believes in the employer. However, there is a caveat [B]: if employee is cheated, she is less likely to work harder in the following periods, leading to lower profits for the employer. :(
CRUCIAL: Commitment by employer will be ensured as long as benefits to employer of [A] are less than the costs to employer of [B]. That is, commitment is ensures as long as employer doesn't have a real economic interest in cheating the employee, especially if the worker will screw the employer in later periods if employer cheats her.
And now we reach the clincher, the real nail in the coffin that drives home a central point of some of the most central works of legal history of the last 35 years: in what scenarios might we imagine the costs of [B] declining? That is, how can we disrupt this equililbrium? Let's see...
[B] If employee is cheated, she is less likely to work harder in the future. But what if costs of enforcing her to work harder in following periods was low enough?
If costs of enforcing her to work harder in following periods was low enough...
Sounds awfully like eminent domain law, doesn't it? Here's a little quiz: how much were mill owners in western New York paid for disruption of water flow caused by the building of the Erie Canal?
TO BE CONTINUED