CBO and the minimum wage, Pt. 2

In a post yesterday, I reviewed a long list of ways in which Tuesday’s Congressional Budget Office (CBO) report embraced arguments made by supporters of the minimum wage. In this post, I want to make some observations on CBO’s analysis of the employment effects of the minimum wage, the aspect of the report that has received, by far, the most attention in the media.

In a major departure from earlier CBO analysis, the range of likely employment outcomes in the new CBO report includes zero.

Headlines have focused on CBO’s “central estimate” of the “change in employment” from an increase in the federal minimum wage to $10.10 –a loss of 500,000 jobs. But, the “likely range” in the CBO forecast runs from a “[v]ery slight decrease to -1.0 million workers.”

A mid-range estimate of 500,000 jobs lost, with a high-end estimate of one million jobs lost, is obviously bad optics for the proposed increase. Nevertheless, recognition in a CBO document that the “likely range” of employment effects effectively includes zero (a “very slight decrease”) is, as far as I can tell from reviewing several past CBO evaluations of the minimum wage, completely unprecedented.

Two CBO reports from the late 1990s, for example, assume that a 10 percent increase in the minimum wage would reduce employment of teenagers by between 0.5 percent and 2.0 percent, with a “smaller percentage reduction for young adults (ages 20 to 24).” (CBO, 1999, p. 4) A 2001 CBO report was not as explicit about its assumptions, but the estimated employment impact did not include zero (200,000 to 600,000 jobs lost).

Including zero in the range of plausible employment outcomes –for the first time ever– ought to feature more prominently in the discussion of the report and in the evaluation of the proposal on the table, especially considering that the proposal involves an increase in the minimum wage of almost 40 percent.

More than two decades of research that has questioned the negative employment impact of moderate increases in the minimum wage is slowly entering into standard analysis.

The CBO chose not to referee a deep divide in the economics profession and, instead, awkwardly split the difference on estimates of the employment effects.

The appendix to the CBO report provides details on specific assumptions about the employment effects of the minimum wage, but offers little on how CBO arrived at those specifics.

Two assumptions drive most of the employment results. The first is the assumption that a 10 percent increase in the minimum wage would reduce teen employment by 1 percent with a “likely range” from close to zero (“a very slight negative amount” p. 23) to as high as 2 percent. The second assumption is that the effect on low-wage adults would be “about one-third” (p. 25) of the estimated effect for teenagers.

The CBO cites numerous studies in connection with the choice of these “policy elasticities.” But, non-specialist readers won’t realize that none of CBO’s parameters actually appear in any of the studies cited. Probably the most prominent minimum-wage critics, David Neumark and William Wascher, for example, argue that a 10 percent increase in the minimum wage reduces teen (and less-skilled worker) employment by 1 to 2 percent –not the 0 to 2 percent used by the CBO. Other critics would put the range between 1 and 3 percent, for a mid-range of 2 percent. Meanwhile, the research by Arindrajit Dube, Michael Reich, Sylvia Allegretto, and William Lester –the group of economists that in recent decades has most informed minimum-wage supporters– puts the employment effect on teens as centered close to zero, with a 10 percent increase in the minimum wage associated with between a 0.6 percent decrease and a 1.3 percent increase in employment (this range taken from Allegretto, Dube, Reich, and Zipperer, Table 3, columns 5-8).

As Michael Reich has noted, CBO’s range lies somewhere between the two camps, with no explanation from CBO as to how it chose to weight the two very different sets of estimates. Siding with the critics of the minimum wage would have produced higher estimates of job loss than what CBO published. Siding with the large and growing body of research finding little or no employment effect would have produced much lower estimates of loss and not ruled out the possibility of job gains.

The CBO breaks with the existing research by assuming significant job loss for low-wage adults.

As I mentioned earlier, the CBO assumes that the employment effects on adults would be one-third of what they would be for teenagers. As the CBO notes, there is “much less research … on the responsiveness of adult employment to minimum-wage increases than on the responsiveness of teenage employment.” In fact, the idea that the minimum wage has essentially no effect on adult workers has long been close to the consensus view within the economics profession. In a large review of the literature at the beginning of the 1980s, for example, Charles Brown, Curtis Gilroy, and Andrew Kohen concluded that even the “direction of the effect on adult employment is uncertain in the empirical work, as it is in the theory” (p. 524) –and that was before the wave of research since the early 1990s that has questioned the negative employment impact of the minimum wage. Indeed, this view has been so standard, that the CBO studies from the late 1990s and early 2000s that I cited earlier appear to assume no employment effects on adults. Since the CBO concludes that 88 percent of workers affected by a minimum-wage increase are not teenagers, this unconventional assumption has a large impact on their final calculations.

Critics and opponents of the minimum wage agree that employment effects are not the only aspect of the minimum wage that should factor into decisions about the policy.

In their book Minimum Wages, critics David Neumark and William Wascher write:

“But the existence of disemployment effects does not necessarily imply that minimum wages constitute bad social policy. As with many government rules and regulations, a higher minimum wage entails both benefits and costs. Thus, the question is not whether there are any costs to a higher minimum wage, but instead whether the trade-offs between the costs and the benefits are acceptable…” (pp. 141-42)

And minimum-wage supporter Jared Bernstein makes a similar point: “even if [critics] are right…the beneficiaries far outweigh those displaced.” (Or see liberal columnist Harold Meyerson’s tweet: @HaroldMeyerson: CBO: Minimum wage hike will help 33 workers for every 1 it hurts. Pretty damn good ratio.)

Several commentators have made a more forceful version of this argument, suggesting that if the minimum wage isn’t causing some amount of job loss, it probably isn’t being set high enough. The unconventionally liberal Matt Yglesias, for example, writes:

“If the White House genuinely believes that a hike to $10.10 would have zero negative impact on job creation, then the White House is probably proposing too low a number. The outcome that the CBO is forecasting—an outcome where you get a small amount of disemployment that’s vastly outweighed by the increase in income among low-wage families writ large—is the outcome that you want. If $10.10 an hour would raise incomes and cost zero jobs, then why not go up to $11 and raise incomes even more at the cost of a little bit of disemployment?”

This view is shared, in almost identical terms, by the not-so-liberal Josh Barro in a post titled
If Your Minimum Wage Increase Doesn’t Raise Unemployment, You Didn’t Raise The Minimum Wage Enough:

“…a minimum wage increase can cause a modest rise in unemployment and still be a good policy idea, so long as it has more than offsetting positive effects. And the minimum wage trade-off presented by CBO looks awfully favorable. For every person put out of work by the minimum wage increase, more than 30 will see rises in income, often on the order of several dollars an hour. Low- and moderate-income families will get an extra $17 billion a year in income, even after accounting for people who get put out of work; for reference, that’s roughly equivalent to a 25% increase in the Earned Income Tax Credit.”

We can only ask CBO to lay out the likely consequences of particular policies. Once trade-offs are involved, we need to make the value judgments that CBO can’t make for us. Much of the media coverage has hyped the mid-range job-loss number and what that number means for the political prospects of proposed increase, but the same coverage has done little or nothing to explore any trade-off between higher incomes and fewer jobs.

We need to have a realistic understanding of the low-wage labor market.

A range of people –young, old, men, women, white, black, Latino, Asian, full-time and part-time, less-educated and college-educated– work in low-wage jobs, many for large parts of their working life. But, an important feature of low-wage jobs is that they tend to have high turnover. Even if half the workers in a low-wage workplace are in stable long-term jobs, the other half of positions might turnover completely once or even twice in a year.

High turnover is an important context to keep in mind when evaluating the costs and benefits of the minimum wage. Even if the CBO’s central estimate of job loss is correct, very few low-wage workers will receive pink slips. Given high turnover, employers who want to reduce employment are much more likely to make any adjustments implied by the CBO estimates through attrition –failing to replace a few percent of the workers who leave on a regular basis.

Workers looking for jobs at the new, higher minimum wage may be looking in a slightly smaller job pool, for a slightly longer period of time. But, when they find a job, it will pay substantially more than the job they would have found somewhat more quickly at the old, but lower minimum wage. Given this reality and the CBO numbers, which suggest that the minimum wage yields a large net transfer of income from employers to low-wage workers as a group, it is hard to imagine that any low-wage workers would be worse off on an annual basis after the minimum-wage increase. (As my colleague Dean Baker puts it, unlike many other policies, including trade agreements, patent protection, or fiscal austerity, there are no “designated losers” with the minimum wage.)

Whenever we’re talking about employment effects, we need to be sure that the conversation includes macroeconomic policy.

In the current context of high unemployment, the easiest way to make up for negative employment effects of any policy is to be sure that we are pursuing appropriately expansionary macroeconomic policy. To a first approximation, labor-market institutions such as the minimum wage, unemployment insurance benefits, and unions determine the distribution of wages, benefits, and incomes, while macroeconomic policy determines the level of employment. There may be circumstances where labor-market institutions begin to act as important constraints on employment, but it is hard to argue that we are anywhere near there now, or even that we have been anywhere close in the last three decades. (For example, we saw no signs of rising inflation at the end of the 1990s and into 2000, even when the unemployment rate hovered for an extended period near 4 percent.) If opponents of the minimum wage are genuinely concerned about the fate of low-wage workers, they should be pushing for appropriately expansionary macroeconomic policy, not fighting policies that make low-wage workers as a whole substantially better off.

One Comment

  1. […] I stumbled on this excellent discussion of all the above points by John Schmitt, one of the premiere synthesizers of minimum wage […]

Leave a Reply