David Leonhardt, co-author of the New York Times story I critiqued here, disagrees with my critique. He has posted a defense of his story and a counter to my critique in the comments, but I thought it worthwhile to discuss the centerpiece of his defense. He says:
We stand by all the numbers in the story, and your posting contained a few errors. I get into detail below.
The larger question, however, is whether an objective analysis of
the economy can show the last few years to have been a period of
healthy growth in compensation for most workers. I think the answer is
no. Earnings for the median-earning worker — indeed, as we noted, even
for the worker at the 90th percentile — have trailed inflation over
the last three years, according to the B.L.S series on Usual Weekly
Earnings of Wage and Salary Workers. Benefits outpaced inflation
between 2003 and 2005, but they’ve trailed inflation over the last
year, according to the Employment Cost Index. Flat real compensation
for most workers is not what today’s incredible American living
standards were built on.
So you’re saying that earnings didn’t keep up with inflation for the last three years. Benefits exceeded inflation for three of the last four years. Last year, when incomes did outpace inflation, benefits didn’t.
That does not mean that real compensation is flat. It tells you NOTHING about real compensation.
You can’t look at earnings in one year and compensation in another. If tax policy or preferences or regulations or custom causes benefits to become a bigger part of compensation over the years, you’d expect wages and salaries to go down for that reason. And that is the history of the last 40 years–benefits becoming a larger part of total compensation, and wages and salaries becoming a smaller part. That’s not a coincidence. They’re related.
You have to look at total compensation. Compensation. Not one component one set of years and a different component in a different set of years. The whole package. Certainly you have to look at the whole package if you’re going to generalize about bargaining power or changes in standard of living over time.
There are other problems–compositional changes in the labor force distort conclusions. It’s really hard to measure inflation. We’ve talked about that already. But given the numbers you chose to highlight, they don’t make your case about either bargaining power or standard of living.
The panel data, the data that follows the same people over time shows people doing better, dramatically better over the last 20 or 30 years, contradicting the wage numbers that take a snapshot at a point in time, where different people are involved. Not just the top 1 or 5 or 10 or 20 per cent. But the median worker and poor workers are dramatically better off when you follow people over time.
David goes on to respond to my claims about bargaining power:
I also don’t think it’s an especially controversial notion that much
of the workforce has less bargaining power than it once did. To
generalize, perhaps overly, globalization and technological innovation
have raised the value of the work done by the roughly one-quarter of
adults who have bachelor’s degrees and reduced the value of the work
done by the three-quarters who don’t. As you note, this is about
alternatives. Employers have many more alternatives than they once did
to paying an American $20 an hour for blue-collar or basic
service-industry work. Alan Greenspan and Ben Bernanke have both
addressed this trend in some detail. Goldman Sachs and UBS have written
reports about the connection between it and the boom in profits.
Employers do have more alternatives. So do employess. It’s not the number of alternatives. It’s what they pay. And what they pay depends on the skills of the worker and various intangibles.
In my neighborhood in suburban Maryland, some folks hire people to come and clean their house for four or five hours, once a week or every other week. These women (and they are almost always women) often speak no English. They have no union. The minimum wage applies to them but is presumably hard to enforce. Some of these women may be educated, but without English, they are doing a job that requires little or no training or the use of any education they mave acquired.
Formally, these women have no bargaining power. None. Yet they earn $15-$20 an hour, often in cash. That is the going rate.
Why? Why do the people who hire these women pay so well. Because if they were offered less, the women would work elsewhere. These woman speak no English. They have no "bargaining power" in the traditional sense of the term. Yet the going wage is triple the minimum wage. What protects them is competition. Competition among home owners who value their work.
I do think globalization and other factors affect wages here in the United States. But the usual implication of the "labor share" discussion is that somehow, in recent years, corporations or rich people (the top 1% or 5% or 20%) are somehow keeping a bigger share of the pie than they have in the past. But if anything, competition among businesses is fiercer and faster and more furious than ever. That is good for consumers and good for employees even if it doesn’t always show up in the numbers in a transparent fashion.
Going back to those cleaning ladies. If the government required homeowners to provide health insurance and cleaning supplies and transportation to cleaning ladies (and if such a regulation could be enforced) their wages would certainly fall. Compensation and wages and benefits are tied together via competition in the marketplace as employers compete for workers.
On the details I got wrong:
The article did quote economists other than Jared Bernstein. (I claimed the Times story only quoted one economist.) In fact, the story quoted Ben Bernanke and Henry Paulson, as David points out. My mistake. But neither Bernanke or Paulson were quoted on the central issue of the wage/compensation data.
I also sloppily wrote that there is no BLS survey on median earnings. I meant wages. My mistake. But the important point is that there is no median data on compensation that is reliable. I will talk more about this and the labor share data (that David also argues I misinterpreted) in a later post.
Thanks to David Leonhardt for a thoughtful and careful response. Let’s keep the dialogue going.