In case you were born yesterday, just fell off the back of a turnip truck, and have lived in blissful ignorance of the possible abuses of economic data, let me direct your attention to yesterday’s New York Times front‐​pager by Steven Greenhouse and David Leonhardt, titled “Real Wages Fail to Match a Rise in Productivity,” and the blog aftermath. George Mason economics professor Russ Roberts is none too pleased.

I shouldn’t be upset when the New York Times news division writes a intellectually dishonest story that plays to the biases of its readership base. But today’s front‐​page above the fold story on wages depresses and surprises me anyway. Maybe it’s because one of the authors, David Leonhardt, is a good reporter with good economic intuition. (I can’t speak for the other author, Steven Greenhouse.) But I suspect the source of my dismay is simply the knowledge that this article, despite its inadequacies will be met with nods of agreement around the breakfast tables of America.

Even around the breakfast tables of famous economists! Berkeley economics professor Brad DeLong–a vociferous, self‐​appointed arbiter of the quality of economics journalism–gives Greenhouse and Leonhardt a total pass, excerpting their article, and simply calling them “thoughtful and reliable.”


Here’s Roberts again:

Let me repeat the key sentence [from the article]:

The median hourly wage for American workers has declined 2 percent since 2003, after factoring in inflation.

That’s a very strange sentence for many reasons:


1. Why would you use a measure of compensation that ignores benefits, an increasingly important form of compensation?


2. Why would you use 2003 as your starting point when the recession ended in November of 2001?


3. There are no government series that I know of on median earnings. Where did those data come from?


There’s a chart accompanying the article. It tells the reader that the median hourly pay data are from the Economic Policy Institute. The Economic Policy Institute has a policy agenda. Their main issue is the alleged stagnant or falling standard of living of American workers. They support a higher minimum wage and the strengthening of labor unions.


… for every year since the recession of 2001, real hourly compensation has actually increased. It’s up since 2003 as well. And this year it’s up quite dramatically…


As I have mentioned here before–the standard claims you hear about labor’s share declining come from using wages without other forms of compensation. When you include benefits, labor’s share is virtually a constant at 70% of national income and has been steady since the end of World War II …

Greenhouse and Leonhardt thoughfully and reliably rely on the Economic Policy Institute! But they could have thoughtfully asked other people, too. For instance, they could have asked David Altig, vice president and associate director of research at the Cleveland Fed Bank. Altig signs on to Roberts’ criticism, and adds his own, noting that a decrease in labor’s share of growth does not necessarily imply an increase in capital’s share, and that a lot turns on which Bureau of Labor Statistics data series you look at.


Harvard’s Greg Mankiw reinforces Roberts’ and Altig’s diagnosis of the main error: failing to recognize that total compensation–cash wages plus benefits–is the relevant measure of real wage growth. Mankiw also relates the importance of using the correct price index, and of not comparing average productivity to median wages.


I wonder if DeLong truly thinks Greenhouse and Leonhardt were adequately thoughful and reliable in this piece. If so, I wonder what, if anything, he thinks is wrong with Roberts’, Altig’s, and Mankiw’s analyses.