Last month’s jobs report from the Department of Labor reported that average earnings grew by 2.8 percent over the year ending in August. This is an improvement on wage growth, but subtracting inflation of maybe 2.2 percent from this figure yields wage growth that will double inflation-adjusted take home pay every 115 years or so. Over the past few years, three explanations have emerged for persistently slower wage growth. They are not mutually exclusive. Let me explain them in turn from benign to very problematic.

In a market economy, workers are paid something akin to the incremental value they bring to the workplace. Because different skills have different value, wages differ by worker skills, education and occupation. Businesses don’t want to pay that much, but to hire workers in competitive labor markets they must do so. In this way, the power of markets for products or services and the workers who produce them determine wages. This market discipline also informs non-market labor contracts, forcing government to pay close to market wages for workers who don’t produce anything tangible, like soldiers, police officers, professors and teachers. Our wages are calibrated on what our earning power might be in private sector jobs. 

Understanding this dynamic through data is an imperfect effort. Wage data is far less simple to collect and omits many things that matter. Fringe benefits, for example, cost the employer but are invisible in most federal statistics. So, too are job quality issues, like safety or worksite unpleasantness. Thus, much of the problem associated employment with might simply be due to missing data points. 

Health care is the easiest to example to explain. A minimal health care plan costs in excess of $6.00 an hour, and grows at a much faster rate than inflation. So, a typical 3.0 percent increase in health care costs could account for almost a quarter of average wage increases last year. Importantly, this could also contribute to wage growth differences between high- and low-income workers, since health care benefits are largely the same for both groups. 

The second hypothesis about stagnant wages is that the unemployment rate is really much higher than reported because there are so many workers out of the labor market. If true, this means that employers are feeling less wage pressure because there is an invisible labor supply steadily moving back into jobs. Recent job growth numbers support this hypothesis, since job growth has averaged more than twice the natural growth of labor supply.

The third explanation for stagnant wages is the emergence of something called monopsony power in recent years. Monopsony power is the ability of firms to control worker wages. This could be occurring in many places as the market concentration among businesses has grown. With a greater share of workers working at a smaller share of businesses, it is easier for individual firm wage setting decisions to carry over across the whole labor market. This has long been the case in some small occupations; that it might now be the case across whole regions is troubling. There is increasing evidence that this is a real problem.

The cause of rising monopsony power is likely varied. The self-destruction of America’s labor unions removes some bargaining power from workers. The rapidly declining number of small- and medium-sized businesses also shifts power from workers to large firms. Technology may also be to blame. There has been an explosion of human resource consulting firms that aggregate wage data across many companies. They can then advise firms on wages, acting as an outside agent in wage collusion.

Likewise, local economic development officials have long participated in wage surveys, whose prime purpose is really to help keep down wages among local firms. My center stopped performing this work due to ethical concerns about the practice. Finally, there is some evidence of outright collusion among firms. While I suspect this is rare, I am troubled that such seemingly benign activities as regional meetings of human resource officers might be fertile opportunities to set wages. I even heard a first-hand account recently of one meeting of local businesses where there was an active discussion about keeping wages down. I should have to remind no one that such a conversation is likely a perseviolation of federal anti-trust laws. 

It is impossible to know just why wage growth has been so sluggish. If I were forced to weight these three causes, I’d say the argument citing fringe benefit growth and excess supply of workers accounts for 90 percent of slow wage growth. However, too many serious labor market researchers have evidence of monopsony power to discount this cause. 

The next year or two will be important in understanding what is happening with wages. If wages rise as the invisible labor force returns to job sites, we can worry less about monopsony. If not, the US is due for a very difficult conversation about anti-trust enforcement and wage growth. That will animate our elections and political discourse for a long time.