One of the best-explored economic phenomena is the tendency of retail prices to end in $0.99. This oddity, the subject of ample academic and lay interest, stems largely from consumers’ left-digit bias: we notice the dollar amount and ignore the cents. It’s the sort of fun behavioral quirk that keeps Freakonomics and Radiolab in business. As Roger Sterling told us in Mad Men: “I’ll tell you what brilliance in advertising is: 99 cents.”
It turns out a similar bias occurs in how employers pay workers. Instead of wages falling evenly along a distribution, hourly pay tends to clump in round numbers. In about 40 states, $10.00 is the modal, or most commonly occurring, wage. Meanwhile, tasks on Amazon’s micro-freelance platform MTurk tend to bunch around 10-cent increments.
Since retail prices confront us every day, it makes sense that economists and popular media have explored the subject deeply. But it might be a reflection of our focus on people as consumers rather than workers that price biases have received ample attention, while the corresponding phenomenon in the labor market has not.
As Arindrajit Dube, Alan Manning and Suresh Naidu argue in a new paper, round-number wage bunching is evidence not of worker bias toward round numbers, but of the market power of employers. In other words, it’s not irrational individual preferences driving the pattern, as in the case of 99-cent prices, but the ability of employers to transcend the competitive pressures of the market.
First, a quick note why wage bunching matters. In pure neoclassical theory, workers receive the marginal product of their effort—that is, employers pay workers exactly the extra value their labor adds to raw materials. Here, there’s no reason for numbers to bunch. In actually existing mainstream economics, however, researchers do recognize that labor markets have frictions, and workers don’t really get their marginal products. Yet competitive pressures still lead to a rough equilibrium wage that clears the market. Again, no reason for all the round numbers.
So how do we explain the presence of bunching? As Dube et al write, it could be one of two things. Option A is the familiar left digit bias: employers attract more workers by raising a wage of $9.95 to $10.00 than by raising a wage of $10.00 to $10.05. It’s the inverse of the retail phenomenon; instead of a $10 item being offered for just below $10, we see a sub-$10 wage bumped up to $10. In this case, we’d expect to see relatively fewer offered wages in the vicinity just below round numbers.
Option B is employer frictions: for one reason or another, employers just like offering even-numbered hourly wages. While the authors don’t explore why these preferences might exist, they note the possibility of “administrative costs, inattention, limits on manager cognition, or norms that constrain wage setting behavior.”
For the second option to be the case, employers would need to have some market power. Instead of simply being captive to the competitive forces and irrational worker preferences that determine the going rate, option B presumes that firms have significant price-setting power. In the jargon, this is known as monopsony. Literally speaking, that’s just one buyer of labor in a market who sets the going rate; in actual practice there can be many buyers.
So how do we find out whose bias drives the round-number phenomenon? Dube et al scraped data from a few sources. Because commonly used surveys like the CPS might be inaccurate due to the rounding of respondents, the authors got a hold of direct wage date from unemployment insurance payroll tax records in two states (WA and MN), which provided virtually exact hourly wage numbers. The distribution unmistakeably shows bunching at $10 and other round numbers:
To determine whether it’s option A or B—worker bias or employer power—requires some more involved math. Yet the intuition is simple. If workers had a bias toward taking round-numbered offers, we’d see an asymmetric dearth in the distribution of wages just below $10 an hour, but not just above $10. Employers would find that the uptick in recruitment that occurs when you bump a $9.95 wage up to $10 is worth the 5 cents extra in pay. On the other hand, if bunching came from the employer side, the dearth in wage offers around the $10 mark would be symmetric; we’d see relatively few $9.95 offers and relatively few $10.05 offers.
After some statistical techniques, the authors show it’s the latter. The “missing mass” in the distribution—the relative lack of wage offers around the bunched number—comes from both above and below $10. Since we’d expect asymmetry in the worker-bias scenario, the results point to employer wage-setting power.
There’s one further question: accepting that the bunching comes from the firms, is it more about employers failing to maximize profits (by offering even numbers rather than whatever number is most profitable), or straight-up monopsony?
As the authors conclude, it takes only a small sacrifice on the firm’s end to suggest a lot of market power: “When there is sizable market power, it requires only a modest extent of optimization error to rationalize substantial bunching in wages.” In other words, small deviations from the most profitable wage rate—offering $10 rather than, say, the profit-maximizing $9.92—can be evidence of a large amount of market power.
The broader conclusion, that employers have more power than workers, is one of those findings in economics that strikes anyone outside of the field as mere commonsense. Employers set wages where they will, subject only partly to the forces of competition that are supposed to give workers power, like quitting for better-paying jobs or flat-out refusing to take a measly $10. Yet while it goes against neoclassical theory, monopsony has become an increasingly recognized fact of our economy.
One final note: it’s worth emphasizing that monopsony theory does not require every employer to be a behemoth and every labor market to be a virtual company town, as is sometimes assumed. While there’s plenty of evidence this is often the case, the basic theory of monopsony refers only to the relative power employers have over workers in wage-setting. Alan Manning, coauthor of the round-number paper and the guy who wrote the book on monopsony, is clear on this: even in a model of perfectly competitive and identical firms, employers might exert market power if there exist realistic frictions in the market, e.g., difficulty finding a new job. The operative questions are how much monopsony exists, and to what degree firm size and concentration contribute to it.
Owen Davis 