Over the weekend, the New York Times published an online piece by economist Arindrajit Dube arguing for a higher federal minimum wage. Dube is a leading scholar associated with a revisionist view of how minimum-wage increases affect the employment of
low-skilled workers. The piece, accompanied by attractive Times infographics, is the latest in the paper’s “Great Divide” series on inequality. It is more even-handed than many treatments of the topic, but it nevertheless shades the discussion of several empirical questions in a way that makes the case for a minimum wage increase stronger than it is.
It is worth beginning with the central question of whether raising the minimum wage from $7.25 an hour to $9 or $10.10 (the leading proposals from Democrats) would reduce inequality. The existing research is not much of a guide. At best, the literature can say what impact increases in minimum wages within typical ranges in earlier time periods have had on inequality. To the extent that that evidence is consistent, generalizing from it is reasonable, although it is also reasonable to worry that the current recovery’s relatively high unemployment and weak demand might reduce the applicability of past experience.
But is the evidence consistent? A number of commentators have pointed to a chart showing that most studies of the minimum wage’s effects on employment find essentially none (see Figure 1). If raising the minimum wage increases the income of low-skilled workers without reducing their employment, then it almost surely reduces inequality. The chart is striking, but it ignores the issue that some research strategies are stronger than others. Dube gives some indication in his essay that the debate has not been settled, with David Neumark and William Wascher defending the neoclassical view against a new set of scholars. His description of that debate, though, suggests that the revisionist research clearly improves on the Neumark and Wascher studies. However, the two earlier this year argued that the “improved” research strategy Dube presents as the state-of-the-art is irrevocably flawed. A paper from July argues that this entire debate has been measuring the wrong outcome. It finds that minimum wage hikes reduce job growth. Dube has responses to both papers. Suffice it to say, the debate continues.
My point is not that a minimum-wage hike would have no impact on inequality—it’s that we have little basis for saying what that impact would be. Dube notes that earlier this year just one-third of economists in an expert panel agreed that raising the federal minimum wage to $9 would make it tougher for less-skilled workers to find jobs. He neglects to share that one-third disagreed with that claim and the remaining third were unsure, had no opinion, or chose not to respond. This is what research ambiguity looks like.
The basic facts about the minimum wage are also presented inaccurately in a way that strengthens the ostensible case for an increase. Dube claims that the federal minimum wage would be $10.60 today if it were at 1968’s all-time high. But that computation uses a cost-of-living adjustment that was rejected by the Bureau of Labor Statistics, the Census Bureau, and the Congressional Budget Office twenty years ago. Economists in these agencies have long known that the so-called “CPI-U” overstates inflation. When naïve or advocacy-minded researchers use it to raise, say, 1968 dollars to their 2012 purchasing power, the impact is to increase them too much, because prices have risen less than the CPI-U indicates.
Using the price index on which the Census Bureau relies for its historical analyses (the “CPI-U-RS”), the 2012 minimum wage would only have to have been $9.25 to have equaled its 1968 level. In other words, the bill sponsored by Senator Tom Harkin and Congressmen George Miller to raise the minimum wage to $10.10 would send it to a higher level than the U.S. has ever seen, and President Obama’s proposal would put it higher than any other year except 1968. Indexing the minimum wage to the cost of living would then keep it there (actually, indexing it to the CPI-U would increase it evermore).
But the CPI-U-RS is not the most accurate measure of inflation available either. It improves on the CPI-U in a number of ways, but it fails to fully account for the ability of consumers to substitute between goods and services as their relative prices change. The “chained CPI” improves on both of these indices, but it only goes back to 1999. Fortunately, it tracks nearly exactly a price index used in the national economic accounts that goes back to 1929, the “PCE deflator”. The PCE is favored by both the Congressional Budget Office and the Federal Reserve Board, and the Bureau of Labor Statistics clearly favors the chained CPI over the CPI-U-RS and CPI-U.
When this best cost-of-living adjustment is implemented, the result is that 2012’s federal minimum wage would only have to have been $8.32 to match its 1968 level. So even the President’s proposed minimum would exceed the previous high. In fact, the current minimum of $7.25 was nearly exactly the average—$7.30—from 1960 to 1980 before its steady fall during most of the 1980s.
Some might argue that we should raise the federal minimum wage to historic levels. As the Times infographic shows, it is just 37 percent of the average hourly wage of nonsupervisory production workers, down from 46 percent in 1979 (and from 53 percent in 1968). That reflects the fact that wages have risen over time while the minimum wage has fallen. Figures from the Economic Policy Institute (Tab 4B) show that the average wage rose by 17 percent from 1979 to 2011 using the PCE deflator. The value of the minimum wage fell 5 percent (Tab 4AD).
But that doesn’t mean that, broadly, pay for low-wage workers fell. We could leave the minimum wage at $7.25 in perpetuity and it would become a progressively smaller fraction of the average wage. Perhaps in one hundred years $7.25 would be just 10 percent of the average. But if so, it would almost surely be the case that no one would be working for $7.25 anymore. In other words, what matters is not whether an arbitrary threshold in the wage distribution is rising or falling but whether low-skill workers in general are seeing wage growth. Minimum-wage workers are poorer than they used to be, but there are fewer of them. In 1979, 8 percent of workers made the federal minimum wage (or less), but by 2011, just 3 percent did. Comparing minimum wage workers over time compares the worst-off 3 percent today to the worst-off 8 percent in 1979.
In contrast, comparing equally-sized groups of low-wage workers shows that they are better off than in the past. The Economic Policy Institute data indicates that the 10th percentile of hourly wages—the pay of the worker poorer than 90 percent of workers and better-paid than just 10 percent of them—rose 5 percent from 1979 to 2011. The increase was 12 percent for the 20th percentile and 16 percent for the median worker. This is not a dramatic rise in inequality between the poorest workers and the typical worker. The median worker made 1.8 times the worker at the 10th percentile in 1979, but the ratio rose only to 2.0 by 2011. Even the ratio of the 80th percentile of hourly wages to the 10th percentile rose just from 2.8 to 3.5.
Finally, note that a minimum wage worker in 1968 or 1979 paid higher taxes than she does today, despite payroll taxes going up. In no small measure, the decline in taxes for low-wage workers reflects the expansion of the Earned Income Tax Credit, now the largest antipoverty program for nonelderly Americans (benefitting one-quarter of American households), nonexistent in 1968 and practically so in 1979. Note, too, that today (unlike in 1968 or 1979) a large share of the workforce is subject to a higher state minimum wage than the federal minimum. In 2007, the federal minimum wage was only beginning to rise from its all-time low, but 70 percent of the workforce enjoyed a higher state minimum. Essentially no one did prior to the late 1980s.
What about the fact that “low-wage workers” (those earning under $10 per hour in Dube’s piece) are older and better educated today than in the past? Because the workers below any threshold are a smaller and smaller group as wages rise, we should not be surprised to see the demographics of low-wage workers change over time. But the finding that the share of low-wage workers that had attended college rose from 25 percent to 43 percent between 1979 and 2011 is less striking when you know that the share of all workers having attended college rose from 33 percent to 59 percent (Tables P-16 and P-17). Similarly, while the share of low-wage workers that are teens fell from 26 percent to 12 percent, the share of all workers that are under twenty fell from 9 percent to 4 percent (Table 5.2).
Different people will ultimately differ in their willingness to increase the minimum wage given the same information about its likely costs and benefits. Informed decisions, however, require accurate information.