The last few years has seen a major shift in minimum wage policies as states like California and New York and cities like Seattle and Washington, DC, have passed laws phasing in a $15 minimum wage, targeting bolder increases than had been the pattern over the last two decades. Federal proposals such as the Raise the Wage Act of 2017, which would bring the minimum wage to $15 by 2024, are consistent with these bolder local approaches. This is bold, as it mandates a 71.9 percent real increase in the minimum wage. But even this bold increase only brings the minimum wage to 29 percent above its 1968 purchasing power, despite productivity nearly doubling over the same timeframe. Many policymakers have concluded, in effect, that years of modest or no change in the minimum wage have created a situation where bold increases are required just to move the minimum wage closer to being a living wage.

This shift in policy now necessitates a rethinking of the flawed way we have evaluated minimum wage increases over the last few decades. There have been, in particular, two flaws. One was first highlighted by economists Fiedler, Howell, and Luce, who noted that public discussion of various minimum wage proposals, including among economists, tends to center solely on whether there will be any job loss. In effect, minimum wage proposals are being evaluated according to a “no job loss” criterion, which specifies that the “right” wage floor is the one that previous research has demonstrated will pose little or no risk of future job loss, anywhere.” A no job loss criterion implies that if a minimum wage increase has any costs (whether loss of employment or work hours), regardless of benefits, then it is inappropriate.

The second flaw in our evaluations is that “job loss” mischaracterizes the actual dynamics of the low-wage labor market, and embeds the discussion in an Econ 101 world where there are only two employment situations after a minimum wage increase: people who continue to work and those who never see work again. In fact, as discussed below, if there is a reduction in overall hours worked it will likely take the form of many people affected working fewer hours over a year’s time while earning much higher annual wages.

Let’s review these flaws in greater detail. First is our argument that critics who cite claims of job loss to reject bolder minimum wage increases are using a distorted evaluation framework, focusing only on the potential costs of raising the minimum wage and ignoring the benefits of raising low-wage workers’ total earnings. This framing is distinctly different from the way we evaluate every other type of economic policy, and biases the public against the bold approach that would help low-wage workers out of the deep hole they are in after decades of inadequate increases in the minimum wage.

We can illustrate the problem using the Congressional Budget Office analysis of the effect of a 2013 proposal to raise the federal minimum wage to $10.10 by 2016. Media coverage of the study highlighted CBO’s estimate of 500,000 jobs lost. Even if that estimate were correct—and there are many reasons to doubt that much job loss from such a modest increase—this implicit “no job loss” criteria distorts the outcome of the policy. CBO’s results, in fact, showed that 17.0 million workers would be directly affected and 97.1 percent of these workers would be clear winners (remaining employed and earning 14.5 percent higher wages). In contrast, the 500,000 jobs lost implies that 2.9 percent of the affected workforce would lose out. The average affected worker would enjoy an 11.6 percent increase in annual earnings. This more complete description of the findings shows that the benefits of the policy far outweigh the estimated costs.

The second flaw is that characterizing the costs as 500,000 jobs lost mischaracterizes the outcomes because it assumes a world with only two outcomes: either a worker maintains employment or loses the ability to find a job. This ignores the high degree of churn in the low-wage labor market over the course of a year, giving the misleading impression that a given pool of workers would lose their jobs and have no earnings over an entire year. We provide evidence that the low-wage labor market already exhibits a high degree of churn—workers constantly move in and out of employment. This implies that low-wage workers who experience employment reductions are actually experiencing reduced total annual hours of work, through fewer weeks worked and longer spells in between jobs, rather than a job loss that lasts the whole year. The more churn there is in the low-wage labor market, the more widely the job losses will be spread among workers in the form of lost hours, and thus the more likely that workers will come out ahead—working fewer annual hours at a higher hourly wage, and thus earning more over the course of a year. Thus, the share of workers with lower annual earnings will be far less than even the 2.9 percent found in CBO’s analysis of “jobs lost.”

As we move forward to assess the many local and state initiatives with bold minimum wage increases we must shift from an evaluation framework that only considers costs and ignores benefits. We must also move away from the Econ 101, point-in-time labor market description that has only those employed and those excluded from work. In fact, over the course of a year the low-wage labor market has a high churn among both employees and employers. Consequently, the best approach to examining the outcomes of minimum wage increases is to examine the number of workers with higher or lower annual earnings, allowing for possible outcomes such as working fewer annual hours but earning greater annual earnings. New bold policies requires us to update our thinking.