Elderly Employment during the Recession: Supply or Demand?

Copyright, The New York Times Company

The supply of elderly workers increased during the recession, and employment rates of the elderly increased along with it. This result is difficult to reconcile with Keynesian characterizations of the labor market.

Last week I showed how employment per capita had increased slightly among elderly people since 2007, while employment per capita in the general population plummeted. I, and other economists previously, have concluded that employment of the elderly deviated so much from the general population because of changes in elderly labor supply.

Recession-era supply episodes like these are important to identify, because they can prove or undercut Keynesians’ fundamental argument (so far unproved) that supply does not matter during a recession or during a “liquidity trap” such as we’ve experienced since the recession began.

(One Keynesian rhetorical trick is to “prove” the supply claim by pointing to the existence of unemployment. Of course, unemployment exists in large numbers, but that does not tell us whether, and how much, labor supply affects employment rates. Only the latter indicates the value, if any, of Keynesian policy prescriptions.)

In reaction to my post last week, Dean Baker, co-director of the Center for Economic and Policy Research, offered some analysis of weekly earnings and concluded that the demand for elderly workers increased during the recession even while it plummeted for everybody else. Supply played little or no role, he said.

Generally, I agree that wage rates are an important variable for gauging the relative importance of supply and demand (see, for example, my analysis of labor supply during the summer season, which featured hourly wage rates as one of three key indicators). But in this case, Dr. Baker’s weekly earnings results ran into a few contradictions.

First, what was the demand shift experienced by elderly workers that was so large as to completely offset the demand shift purportedly experienced by the rest of the population? Perhaps elderly workers picked up some of the hundreds of thousands of jobs that teenagers and other unskilled workers lost thanks to the minimum wage increases? Dr. Baker did not say.

Second, a huge increase in demand for the elderly might be expected to reduce their unemployment rates, yet the chart below shows how unemployment rates for the elderly followed very much the same time pattern as unemployment rates for the general population. (According to the Census Bureau, every person is either employed, unemployed or not looking for work, which is why unemployment and employment can move in the same direction). Dr. Baker’s purported demand shift is not visible in the unemployment data.

Third, Dr. Baker’s own data suggested that something other than demand was driving most, if not all, of the earnings changes he measured. Elderly employment per capita increased only 4.5 percent. Demand increases by themselves usually increase wage rates about the same amount that they increase employment, which in this case would be well less than 10 percent (for example, during Christmas season, when demand increases employment rates more than it increases wages, not less).

But Dr. Baker measured a weekly earnings increase of at least 17 percent, which tells me that weekly earnings were increasing for reasons other than demand. (This may have been because elderly people were working more hours per week, took on jobs with greater responsibility or because the composition of the elderly work force changed in the direction of greater skill. See this paper for a technical analysis of these forces.)

Labor supply, rather than labor demand, readily explains why elderly unemployment rates tracked the general population’s during the recession, while elderly employment did not: the willingness (or necessity, if you want) of working had changed more for the elderly than for the rest of the population. The fact that labor supply really does matter during recessions means that Keynesian policy prescriptions will not deliver what they promise.

Why Hasn't Employment of the Elderly Fallen?

Copyright, The New York Times Company

While employment rates have fallen sharply among the general population, they have not done so among the elderly. This result is difficult to reconcile with Keynesian characterizations of the labor market.

The red line in the chart below displays an index of the per capita employment for the general population. For example, a value of 93 for 2010 means that the fraction of people employed in 2010 was 7 percent less than it was in 2007, before the recession began. The red line shows what we all know by now: many fewer people have jobs now than did a few years ago.

The other two series in the chart are for specific age groups: ages 65-69 and ages 70-74. Both groups have a somewhat greater fraction working now than in 2007 (the increase is even more for ages 75+, but that group is small, so it is omitted from the chart).

Recent studies have looked at the labor-market experiences of the elderly during the first half of the recession. The authors emphasize that, while the recession by itself might reduce elderly employment, the elderly have become increasingly willing to work. I agree.

Many elderly people, for example, saw the market values of their homes and retirement assets plummet in 2008 and feel they can no longer afford to be retired. Naturally, many of them react by looking for work.

The blue and green lines in the chart show that the elderly have been much more successful than the general population at obtaining and retaining jobs.

These findings contradict the Keynesian narrative of the labor market, in which the marketplace fails to recognize the degree to which people would like to have a job. (The Keynesian narrative helps rationalize, among other things, the assertion that unemployment insurance did not reduce employment during the recession, because “what’s limiting employment now is lack of demand for the things workers produce. Their incentives to seek work are, for now, irrelevant.”)

Employment, even during a recession, is not solely the result of lucky few finding available positions. All else being the same, the market tends to create and allocate jobs for those people who are most interested in working.

That’s why, if government is to avoid making employment any less than it has to be, it’s so important to pay attention to the incentives created by taxes, subsidies and government regulations.

Where a Minimum-Wage Increase Would Bite

Copyright, The New York Times Company

Last week I explained how I estimated that the July 2009 federal minimum-wage increase reduced national employment by about 800,000. That 800,000 is the sum of employment impacts for each of several demographic groups and can therefore be used to estimate which groups were most affected by the wage increase.

For each group distinguished by age and part-time status, I calculated “impact”: the percentage gaps between actual per capita employment after July 2009 and the employment I estimated to occur absent the minimum-wage increase.

For example, I found the part-time employment impact among teenagers to be negative 14 percent: that is, 14 percent fewer teenagers were working part time since July 2009 as a result of the federal minimum-wage increase.

The basic economics of the minimum wage suggests that employment impacts will be greatest among groups with the lowest hourly wages. Teenagers employed part time have particularly low hourly wages, and the hourly wages for teenagers employed full time are not much greater. The percentage impact on people 20 and over working full time should be essentially zero, because 98 or 99 percent of them would make more than the minimum wage anyway (see Table 1 of this paper).

The chart below compares, by group, the minimum-wage increase’s employment impact (vertical axis) to a measure of the hourly wages of the workers in the lowest-paid quarter, shown on the horizontal axis). The diagram shows that impacts are negative, or essentially zero, and that the size of the impact is progressively smaller for groups with higher wages – exactly as economic theory predicts.

Compare, for example, the minus 13 percent impact for teenagers in part-time jobs (25 percent of whom were earning less than $5.17 an hour in 2008 — note that a -0.13 log change is essentially a 13 percent reduction) with the zero percent impact for full-time workers 20 to 54 (75 percent of whom were earning more than $11.76 an hour).

Teenage employment rates have fallen to 25 percent from 35 percent over the last four years. Much of that drop cannot be attributed to the federal minimum wage, but my estimates show that the federal minimum wage was a significant factor.

Profs. William E. Even and David A. Macpherson also studied the recent minimum-wage increases for the Employment Policies Institute, using a different methodology. They focused on teenagers (not distinguishing part time and full time), and their state-level model included three federal minimum-wage increases (2007, 2008 and 2009).

They found that those three increases cost 114,000 teenage jobs (I found 800,000 for all age groups combined, only some of which were teenage employment effects; Prof. David Neumark had estimated that teenage and young-adult employment would be reduced 300,000 as a result of the July 2009 minimum-wage increase).

A minority of states were not affected by federal minimum-wage increases because their own state-legislated minimum wage would exceed even the new higher federal minimum (the federal minimum prevails over the state minimum whenever the former is greater).

Thus, whatever jobs were lost because of the federal increase would be lost in the states — 31, after the 2009 increase — and the District of Columbia (which sets its minimum to be $1 above the federal minimum, so when the federal minimum jumps, do does the district’s). Professors Even and Macpherson’s state-level estimates confirmed that this was, in fact, the case.

For the same reason that my model separately considered teenagers and part-time workers, Professors Even and Macpherson followed their teenagers study with a study of 16-to-24-year-old men with less than a high school diploma. They found that each 10 percent increase in a federal or state minimum wage decreased white employment by 2.5 percent, Hispanic employment by 1.2 percent and black employment by 6.5 percent.

For the time being, at least, it seems that increasing the federal minimum wage again would give a few workers a small raise — but at the cost of eliminating entire paychecks for the young, less educated and least skilled.

Kudlow Clip

Teen Employment Increases Almost One Million

Based on my view that labor supply always matters, last month I predicted that teen employment would increase the summer, much like summers past, largely because summer is when teens become available to work. When more people are available to work, jobs are created.

This morning the Labor Department reported that teen employment had increased about 900,000 from April to June. I expect that July will show still higher teen employment.

How to Create Jobs and Cut the Deficit

Copyright, The New York Times Company

Another federal minimum-wage increase would not, as some proponents promise, create jobs, but would reduce employment.

A few months ago, The New York Times editorialized that America’s minimum-wage workers “need a raise.” The Center for American Progress said a higher federal minimum wage “encourages spending, investment and economic growth.”

Many economists expect the minimum wage, if it has any effect, to raise employer costs and thereby reduce employment, especially among people who are likely to work in minimum-wage jobs, like part-time workers.

Federal and state minimum wages have changed a number of times over the years, and each of those instances provides an opportunity to test the employment-reducing hypothesis. For the episodes before 2007 (which I’ll refer to as the historical minimum-wage increases), many statistical studies of minimum-wage effects are summarized in books by David Neumark and William L. Wascher and David Card and Alan B. Krueger.

Not everyone interprets the historical evidence the same way. The New York Times and the Center for American Progress cited some of that evidence to alleviate concerns that a minimum-wage increase would reduce employment.

Even those who believe that historical minimum-wage increases did little, if anything, to reduce employment are still likely to appreciate that a minimum wage that was high enough — a hypothetical $100 an hour, for example — would significantly depress employment. The real disagreement is whether historical minimum wages were high enough, and the economic situations right, for those increases to destroy a large number of jobs.

The most recent federal minimum-wage increase, on July 24, 2009, had the potential to have different effects than its predecessors. Adjusted for inflation (and deflation), by 2009 the real federal minimum wage had been raised to a level 32 percent higher than it had been in 2006 — nowhere near our hypothetical height, which we all agree would be destructive — but perhaps high enough to have some of those effects.

In addition, during a recession hiring decisions may be especially sensitive to employment costs, though some economists say recessions make employment less sensitive to wages.

It’s also easy to exaggerate the effects, good or bad, of the federal minimum wage, because seasonal workers, employees who rely on tips and others are exempt from it, and a few states have minimum wages above the federal minimum.

For these reasons, I have been studying the 2009 federal minimum-wage increase by itself and trying to separate the effects of the recession from the effects of the wage increase (see this recent publication for more details; my next posts will point to some other findings from the 2009 increase).

Part-time and teenage employees are especially likely to have hourly wages near the federal minimum. The red line in the chart below displays seasonally adjusted national part-time employment by month, from the Census Bureau’s monthly household survey. Before July 2009, part-time employment increased by about three million during the recession and that month reached the peak level of part-time employment.

To investigate the possibility that the July 2009 wage increase stopped further increases in part-time employment and perhaps affected other employment categories, I estimated a monthly model of national part-time and full-time employment per capita for each of 12 demographic groups distinguished by race, gender and age (white and nonwhite, male and female, and 16 to 19 compared with 20 to 54 and 55 and over), using data from before the increase.

I used the model to forecast part-time and full-time employment for each demographic group for August 2009 through December 2010. The aggregate deviation of the part-time predictions from the actual was added to the red line to arrive at the aggregate part-time prediction shown as the chart’s blue line.

After falling 9.3 million during the recession through July 2009, aggregate full-time employment fell another 1.8 million by the end of the year and remained below July 2009 levels at the end of 2010. Some people dismissed from their full-time jobs probably had trouble finding another suitable one, and some of them worked part time while they searched.

Consistent with this story, my estimates predicted that part-time employment would have continued to increase during the second half of 2009 because, before the increase, part-time employment tended to increase with full-time job losses.

The actual and predicted data depart dramatically beginning in September 2009, with actual part-time employment 1.2 million below predicted part-time employment by December and averaging 975,000 part-time positions below what was predicted over the months August 2009 to December 2010.

A small number of these job losses were offset by possible increases in full-time employment (relative to what it would have been – more on this next week). I find the total job loss from the July 2009 minimum-wage increase to be about 800,000.

If raising the minimum wage reduced employment by 800,000, cutting it back to its early 2009 level is likely to increase employment by 800,000. That would add a bit to government revenue as some of those people moved from unemployment benefits to tax-paying workers.