Low Unemployment: the Recipe for Higher Wages

Back in 2013, Jared Bernstein and I wrote a book called Getting Back to Full Employment: A Better Bargain for Working People (free download). The main point of the book was that low unemployment rates disproportionately benefited those who are most disadvantaged in the labor market. For this reason, we argued for using macroeconomic policy to get the unemployment rate as low as possible, until inflation became a clear problem.

At that time the unemployment rate was still close to 7.0 percent. It was still coming down from its Great Recession peak of 10.0 percent, but there were many economists, including many at the Federal Reserve Board, who argued that it should not be allowed to fall below a range between of 5.0–5.5 percent, because lower rates of unemployment could trigger spiraling inflation.

In fact, this was pretty much the consensus view in the economics profession. At the time, the Congressional Budget Office (CBO) put the non-accelerating inflation rate (NAIRU) of unemployment at 5.5 percent. The NAIRU is essentially the target rate of unemployment for policymakers since they want to prevent the accelerating inflation that would result if the unemployment went much lower. CBO’s numbers are also important in this respect, not only because it is seen as an authoritative source, but also by design it tries to produce estimates that are well within the consensus in the economics profession.

Our argument was directed at these people. We felt the evidence that unemployment rates this high should pose any sort of floor for macroeconomic policy were weak. We also pointed out that economists had been badly mistaken two decades earlier, in the 1990s, when they argued that the unemployment rate could not get below 6.0 percent without triggering spiraling inflation.

Fortunately, the Greenspan Fed ignored that view and allowed the unemployment to fall to 4.0 percent as a year-round average in 2000. This gave us the late 1990s boom, the only period of sustained real wage growth for those at the middle and bottom of the wage ladder since the early 1970s. Given the enormous gains from allowing the unemployment rate to fall further, we felt the Fed should take the small risk of accelerating inflation, and allow the unemployment to continue to decline below the conventional estimates of the NAIRU.

Thankfully, Janet Yellen, who was then Fed chair, agreed with this position. (It helped that our friends with the Fed Up Coalition were also pushing hard in this direction.) She held the Fed’s key federal funds rate at zero until December of 2015, at which point the unemployment rate had fallen to 5.0 percent. Since then, the Fed has had a path of moderate rate hikes (faster than I would have liked), that have not prevented the unemployment rate from falling further.

It hit 3.7 percent in September and November, the lowest reading since 1969. Its current rate of 3.9 percent is well below almost any estimate of NAIRU available at the time Jared and I wrote our book. And, we are still seeing very little evidence of any acceleration of inflation. On this score, that we could allow the unemployment rate to get well below 5.5 or even 5.0 percent, we seem to have been very much on the mark. We also have been right that the most disadvantaged in the labor market would be the main beneficiaries of low unemployment.

Starting with changes in unemployment rates, in 2013 the unemployment rate for blacks averaged more than 13.0 percent. The average for the second half of 2018 was just over 6.0 percent. The unemployment rate for Hispanics fell from more than 9.0 percent to 4.5 percent. For workers with less than a high school degree, the drop was from more than 11.0 percent to less than 6.0 percent. For workers with just a high school degree, the drop was from 7.5 percent to less than 4.0 percent.

While it would be foolish to celebrate unemployment rates that are still way too high, it would also be foolish not to acknowledge the enormous gains to the most disadvantaged segments of the labor market that have resulted from the low overall unemployment rate.

We see a similar story if we flip this over and look at the employment-to-population ratios (EPOP). At the time, there was a widely told story that workers had permanently left the labor force either because they no longer had the right skills for the jobs that were available, or alternatively that they just didn’t feel like working. The latter story was supposed to be especially the case for young men, with the argument being that they would rather look at Internet porn and play video games than work at low-paying jobs.

The overall EPOP for prime age workers (ages 25 to 54) stood at 76.0 percent in the fall of 2013, well below the pre-recession peaks of more than 80.0 percent. The drop from pre-recession levels was roughly 5.0 percentage points for men and 3.0 percentage points for women. In the most recent data, the EPOP for prime-age men is still more than a percentage point below its pre-recession peak, while for women it is slightly higher, although still well below peaks hit in 2000.

As for the porn-watching video game playing 25- to 34-year-old men, their EPOPs are up by roughly four percentage points from 2013 levels, although they are still two percentage points below pre-recession peaks. Anyhow, it seems that labor demand was a more important factor than the quality of video games in the drop off in employment.

While the number of people who were able to find work due to the drop in unemployment over the last five years is large, the number who have been able to secure pay increases because of the increase in bargaining power is even larger. And the gains have been especially impressive for many of those most disadvantaged segments of the workforce.

The Story on Wage Growth

The labor market first tightened enough for workers at the middle and bottom to start seeing real wage gains in 2014. Since then, the gains have been impressive.

Over the last four years, average weekly earnings, adjusted for inflation, have risen by 10.1 percent for workers at the cutoff for the bottom tenth of the wage distribution. They rose 8.5 percent for workers at the cutoff for the bottom quarter and 5.6 percent for workers in the middle. Increases in state and local minimum wages almost certainly contributed to the wage gains at the bottom, although low unemployment was undoubtedly the major factor higher up the wage ladder.

The path of real wage growth is shown in Table 1, with wages indexed to 100 for the fourth quarter of 2014. The data are for weekly earnings of full time workers (more than 34 hours a week) in the fourth quarter of each year.[1]

Table 1

Percentile of Wage Distribution

10           25           50           75           90               

2014                       100         100         100         100         100

2015                       103.3     103.4     102.8     105.1     104.1

2016                       104.6     105.2     104.0     104.9     106.0

2017                       105.7     107.1     102.8     104.3     105.3

2018                       110.1     108.5     105.6     106.9     108.8

Source: Bureau of Labor Statistics.

There are a couple of points worth making about the table. First, the changes, especially at the bottom end, are affected by the number of people working. As we pull more people into jobs, the ones being added are likely to have less education and experience than the workers who are already in the labor market. This means the 10thpercentile worker in 2018 likely has less education and experience than the 10thpercentile worker in 2014. That makes the 10.1 percent reported wage gain for the 10th percentile workers over this period even more impressive.

A second point is that the numbers are somewhat erratic. There is a substantial amount of measurement error in these data. For example, the 1.2 percentage point reported fall in the wage of the 50th percentile worker for 2017 may not be accurate.

The third point is that the price of oil is a big factor in the story. The sharp rise in the real wage in 2015 is partly due to a sharp drop in world oil prices that held inflation near zero. By contrast, the growth in 2017 was dampened by a partial reversal of the earlier drop. When discussing short-term changes in real wages it is important to distinguish what is attributable to nominal wage growth and what is attributable to changes in oil prices. We expect nominal wage growth to persist, whereas changes in oil prices tend to be erratic and are often reversed.

Differences by Gender

While in general the tighter labor market has disproportionately benefited the most disadvantaged, it has not helped to reduce the gender gap among those at the middle and bottom of the wage ladder. Table 2 shows the changes in wages for men and women separately at the 10th, 25th, and 50th percentile of the wage distribution.

Table 2

Percentile of Wage Distribution

Men                                                       Women

10           25           50                           10           25           50                

2014                       100         100         100                         100         100         100

2015                       103.6     102.7     102.4                     104.3     101.2     100.3

2016                       104.5     101.7     102.8                     105.8     103.0     102.4

2017                       109.1     102.1     102.8                     105.8     103.9     101.8

2018                       114.1     106.4     105.5                     108.7     106.4     102.8

Source: Bureau of Labor Statistics.

The gender gap by this measure has risen even as the labor market has tightened. It would require a more detailed analysis to determine the reasons for this increase, but in any case, this is an area where the tighter labor market has not had the effect of disproportionately benefitting the more disadvantaged group.

Differences by Race and Ethnicity

In contrast to the story with gender, the data do show that non-whites have seen more rapid wage growth than whites over the last five years. Table 3 shows wage growth at the median since 2014 for whites, Blacks, Hispanics, and Asians.

Table 3

Wage for the Median Worker

White                    Black                     Hispanic               Asian                            

2014                       100                         100                         100                         100

2015                       102.5                     103.1                     103.6                     113.3

2016                       104.7                     106.3                     105.3                     104.3

2017                       103.7                     102.7                     104.9                     105.9

2018                       106.0                     107.4                     106.8                     107.0

Source: Bureau of Labor Statistics.

The table does show some modest narrowing in the wage gap at the medians, with the median wage for blacks and Hispanics rising somewhat more rapidly over this period. This is true with the median for Asian workers also, although the median wage for Asians is somewhat higher than the median for whites. While the greater relative gains for Blacks and Hispanics are not huge, at least the gap is getting smaller, as opposed to the gender gap over these years. It is also worth noting that since blacks and Hispanics have accounted for a disproportionate share of the employment gains over this period, the changing structure of the employed is likely to have more of an impact on the wage distribution for workers in these demographic groups.

In other words, if we assume that the newly employed workers are mostly towards the bottom end of the wage distribution (since they have less education and experience) the 50 percentile white worker in 2018 may have been the 48th percentile white worker in 2014. By contrast, the 50 percentile black or Hispanic worker in 2018 may have been the 45th percentile black or Hispanic worker in 2014. It would require an analysis of the microeconomic data to determine the extent to which this could be the case, but on its face, this composition story could be a factor in why median wages for Blacks and Hispanics did not rise more rapidly.

It is worth noting the big jump in the median wage for Asians from 2014 to 2015 and then the fall in the subsequent year. This almost certainly did not happen in the world and just reflects the erratic nature of the data. As we get to smaller groups, the quality of the data deteriorates and we are more likely to see sharp jumps or falls that do not correspond to trends in the labor market.

Differences by Educational Attainment

Workers with lower levels of educational attainment have done considerably better than workers with just a college degree over the last four years, and even slightly better than workers with advanced degrees. Table 4 shows the changes in median wages by levels of educational attainment over the last four years.

Table 4

Wage for the Median Worker by Educational Attainment

LTHS                      HS                          SC                           Col                         Ad. Deg.[2]                                                    

2014                       100                         100                         100                         100                         100

2015                       101.8                     103.5                     97.8                        100.6                     102.7

2016                       103.4                     102.8                     100.2                     100.9                     103.1

2017                       103.6                     103.0                     99.4                        99.1                        101.7

2018                       103.6                     105.3                     103.7                     99.8                        103.3

Source: Bureau of Labor Statistics.

The picture in the table goes directly at odds with the conventional wisdom that changes in technology are leading to a relative increase in the demand for workers with more education and reducing the demand for less-educated workers. Over this four-year period, workers with just a college degree have fared the worst, actually seeing a small drop in their median wage. This is consistent with the story of many new college grads struggling with debt, as the jobs they are able to find do not pay them a wage high enough to pay off their loans.

By contrast, workers with just a high school degree have seen the largest increase in their pay, with the real median wage rising by 5.3 percent over this period. To be clear, this is not an argument that workers should not try to get more education. The median pay for a worker with a college degree is far higher than for a worker with just a high school degree, but over the last four years, this gap has been getting somewhat smaller.

Low Unemployment is Good News for Workers

The last four years are the only period, apart from the low unemployment years of the late 1990s, when workers at the middle and bottom of the wage distribution have seen sustained real wage growth. The story is turning out pretty much as Jared and I had expected when we wrote our book in 2013. Low unemployment rates give relatively disadvantaged workers increased bargaining power, allowing them to share in the benefits of economic growth. There are even reports of employers looking to hire people with disabilities and criminal records, people they would never look to hire in a labor market with high unemployment.

The fact that workers at the middle and bottom have been seeing healthy wage gains for the last four years does not mean the economy is great. There are still many people who cannot afford health care, rent, child care for their kids, and other necessities of life. Four years of decent wage growth does not undo the effects of forty years of rising inequality.

But it is important to recognize that at least things are moving in the right direction for now. If the current sub-4.0 percent rate of unemployment can be sustained, or ideally even pushed lower, it will mean that workers in the middle and bottom will continue to share in the gains from economic growth and make up some of the ground lost in prior decades.

As a practical matter, there are many policies that we might like to see the government pursue to help those at the middle and bottom. But having a full employment economy is by comparison relatively simple. We just have to keep the Federal Reserve Board from slamming the breaks on the economy with high interest rates. There are many good things we want the government to do, but in this case, we just need to keep it from doing something bad.


[1] Data are taken from the Bureau of Labor Statistics Usual Weekly Earnings Seriesand are adjusted by the Consumer Price Index. It is also worth mentioning that these are weekly earnings. This should correspond closely to hourly earnings since there has been virtually no change in average weekly hours over the last five years.

[2] LTHS – less than high school, HS – high school, SC – some college, COL- college only, Ad. Deg. – advanced degree.

This article originally appeared on Dean Baker’s blog.

Dean Baker is the senior economist at the Center for Economic and Policy Research in Washington, DC.