Ezra Klein: Are Labor Markets Monopsonies?: Sadly, the best thing written on the blog today didn't come from me. Rather, it's a comment from Kathy G. arguing that labor markets don't look much like classical assumptions would suggest, and that the data -- and some emergent theory -- offers evidence that they're closer to a monopsony than the more traditional competitive-market-in-equilibrium model. Full comment below the fold:
Anyone who thinks mandated leave would inevitably lead to a decrease in employment or wages most likely has not taken any econ beyond Econ 101. Because if you believe that such results would inevitably follow, you clearly are not familiar with the empirical research on paid leave, nor do you understand economic theory beyond a very superficial and incomplete level.
I've studied the economics of paid leave fairly closely. I've read all the research and the first thing that needs to be said is that there is little evidence empirically that government-mandated paid or unpaid leave leads to a decrease in employment or wages.
Secondly -- and I can't emphasize this strongly enough -- economic theory does not offer any strong predictions about how government-mandated leave would effect employment or wages. If and only if we assume that the labor market is perfectly competitive and that there are no transaction costs, no externalities, and no market failures like imperfect information or adverse selection, then yes, mandated paid leave would have those negative effects, according to neoclassical economic theory.
But we don't live in a world of perfect competition, perfect information, zero transaction costs, etc. For example, employers may assume they know their employees' preferences and therefore they don't offer them more than a week or two of paid leave or vacation, even if that is not actually what employees want (and would be willing to take a pay cut to get). For family leave, in particular -- there is evidence that paid family leave increases women's employment, wages, and productivity (because it helps them maintain good job matches). It also saves the firm money in recruiting and training. And, if it really does do these things, the net result could be, not decreased employment and wages, but maybe even increased employment and wages.
Paid family leave can also create a positive externalities. For example, several studies show that paid family leads improved child health and decreased infant mortality. One particularly interesting (and very good) study showed that while government mandated, job-protected paid leave had a strong and significant positive impact on child health, leave that was unpaid and non-job-protected had no effect whatsoever on child health.
One final point -- a basic problem with the way the economics of labor policy is often studied is that it is assumed that labor markets are perfectly competitive. The perfect competition model may be useful in certain limited contexts, but it's a very flawed and misleading way of looking at the labor market as it actually exists. The basic problem with the competitive model is that it assumes that, if an employer cut wages by one cent, every employee would quit and go work for a different firm. But of course, that's not the way the world works. Imperfect information and significant transaction costs create strong barriers to employee mobility (and employee bargaining).
Influenced by Alan Manning's groundbreaking 2003 book Monopsony in Motion, some economists are beginning to view the labor market not as a perfectly competitive, but as a monopsony. Monopsony literally means "one buyer" (of, in this case, labor). When economists refer to a labor market as a monopsony, they generally don't mean it literally; instead, they use the term to refer the case where the labor supply to a firm is not infinitely elastic (in other words, not everyone will quit if wages are cut by one cent).
Manning and others argue that in situations where there are important frictions in labor markets, and where employers set wages, monopsony is a reasonable assumption. And you'll find, if you do the graphs and the mathematics assuming monopsony rather than perfect competition, the predicted results of various labor policies are different. For instance, the perfectly competitive model will predict that raising the minimum wage inevitably leads to a decrease in employment. But the monopsonistic model shows that employment may actually increase. Although, of course, even in the monopsonistic model, you can raise the minimum wage so high that employment decreases.
Monopsony is a better model for the labor market than is perfect competition, and not only because it's more theoretically compelling (since it incorporates the fact that there are transaction costs to leaving one's job and that employers, not employees, set wages), but also because it's a better fit for the empirical evidence. Research shows, for example, that the impact of the minimum wage on employment is mixed at best, and that it often increases employment. These results are consistent with the monopsony model but very much at odds with the assumption of perfect competition.
And one final note: since theoretically, you would model mandated paid leave similar to the way you'd model a minimum wage (because to the employer, mandated leave would be the same thing as a wage increase), the theoretical results of paid leave would be the same as the theoretical results of the minimum wage. Meaning: if you assume perfect competition (and no transaction costs, etc. etc.), mandated paid leave would definitely result in lower employment (or lower overall compensation). Whereas, if you assume monopsony, there is no strong prediction either way -- employment and wages could increase, decrease, or stay the same, depending on how much it cost the employer.
There is another possible explanation of low estimated negative effects of the minimum wage on employment -- efficiency wage effects. Most efficiency wage models are not like monopsony because they can't lead to a positive effect on employment.* In the most popular model workers shirk less because with a higher wage the job is more valuable and being fired for goofing off or pilfering is more costly. This can mean that, at the profit maximizing wage, a small change in wages (up or down) has a tiny negative effect on labor demand. Now the data really aren't good enough to tell us if an increase in minimum wage from a to b causes increased or decreased employment, so a tiny negative effect is consistent with the data. Note that the effect on profits would also be tiny (so long run effects via influencing investment should be tiny). The benefit to workers is not tiny, so if the minimum wage is near the profit maximizing wage an increase in the minimum wage would be good.
On paid maternity leave, there is an important reason why it might not be introduced by the private sector even if it is Pareto improving -- adverse selection. If one firm has good maternity benefits, it will attract women who are planning to have kids soon. This is not good for the firm. This is a purely private cost to the firm (they will work somewhere) so forcing all firms to off paid leave can improve efficiency (it is easy to write a model in which it makes everyone better off). Adverse selection models often have gross inefficiency (moral hazard models as the efficiency wage model above less so).
*one exception to the pattern that minimum wage increases cause reduced employment is more a renaming of a monpsony model than anything else. One efficiency wage effect is reduced quits (the turnover efficiency wage model). It is assumed that firms can fill the vacancies, so quits do not imply permanently reduced employment. However they do imply training costs. If, in addition, one notes that vacancies are not filled instantly, the reduced employment at low wages due to temporary vacancies due to quits imply lower employment. Thus adding an other than magically instant matching of unemployed workers and vacant jobs makes the turnover efficiency wage model a model in which employment increases in wages. This is my interpretation of the monopsony model.
Posted by: Robert Waldmann | July 25, 2007 at 06:30 PM
There is another possible explanation of low estimated negative effects of the minimum wage on employment -- efficiency wage effects. Most efficiency wage models are not like monopsony because they can't lead to a positive effect on employment.* In the most popular model workers shirk less because with a higher wage the job is more valuable and being fired for goofing off or pilfering is more costly. This can mean that, at the profit maximizing wage, a small change in wages (up or down) has a tiny negative effect on labor demand. Now the data really aren't good enough to tell us if an increase in minimum wage from a to b causes increased or decreased employment, so a tiny negative effect is consistent with the data. Note that the effect on profits would also be tiny (so long run effects via influencing investment should be tiny). The benefit to workers is not tiny, so if the minimum wage is near the profit maximizing wage an increase in the minimum wage would be good.
On paid maternity leave, there is an important reason why it might not be introduced by the private sector even if it is Pareto improving -- adverse selection. If one firm has good maternity benefits, it will attract women who are planning to have kids soon. This is not good for the firm. This is a purely private cost to the firm (they will work somewhere) so forcing all firms to off paid leave can improve efficiency (it is easy to write a model in which it makes everyone better off). Adverse selection models often have gross inefficiency (moral hazard models as the efficiency wage model above less so).
*one exception to the pattern that minimum wage increases cause reduced employment is more a renaming of a monpsony model than anything else. One efficiency wage effect is reduced quits (the turnover efficiency wage model). It is assumed that firms can fill the vacancies, so quits do not imply permanently reduced employment. However they do imply training costs. If, in addition, one notes that vacancies are not filled instantly, the reduced employment at low wages due to temporary vacancies due to quits imply lower employment. Thus adding an other than magically instant matching of unemployed workers and vacant jobs makes the turnover efficiency wage model a model in which employment increases in wages. This is my interpretation of the monopsony model.
Posted by: Robert Waldmann | July 25, 2007 at 06:30 PM
There is another possible explanation of low estimated negative effects of the minimum wage on employment -- efficiency wage effects. Most efficiency wage models are not like monopsony because they can't lead to a positive effect on employment.* In the most popular model workers shirk less because with a higher wage the job is more valuable and being fired for goofing off or pilfering is more costly. This can mean that, at the profit maximizing wage, a small change in wages (up or down) has a tiny negative effect on labor demand. Now the data really aren't good enough to tell us if an increase in minimum wage from a to b causes increased or decreased employment, so a tiny negative effect is consistent with the data. Note that the effect on profits would also be tiny (so long run effects via influencing investment should be tiny). The benefit to workers is not tiny, so if the minimum wage is near the profit maximizing wage an increase in the minimum wage would be good.
On paid maternity leave, there is an important reason why it might not be introduced by the private sector even if it is Pareto improving -- adverse selection. If one firm has good maternity benefits, it will attract women who are planning to have kids soon. This is not good for the firm. This is a purely private cost to the firm (they will work somewhere) so forcing all firms to off paid leave can improve efficiency (it is easy to write a model in which it makes everyone better off). Adverse selection models often have gross inefficiency (moral hazard models as the efficiency wage model above less so).
*one exception to the pattern that minimum wage increases cause reduced employment is more a renaming of a monpsony model than anything else. One efficiency wage effect is reduced quits (the turnover efficiency wage model). It is assumed that firms can fill the vacancies, so quits do not imply permanently reduced employment. However they do imply training costs. If, in addition, one notes that vacancies are not filled instantly, the reduced employment at low wages due to temporary vacancies due to quits imply lower employment. Thus adding an other than magically instant matching of unemployed workers and vacant jobs makes the turnover efficiency wage model a model in which employment increases in wages. This is my interpretation of the monopsony model.
Posted by: Robert Waldmann | July 25, 2007 at 06:32 PM