On the trail of a dubious stat

Banging the drum about income inequality in her Washington Post column last week, Katrina vanden Heuvel advanced this fairly amazing factoid:

In 1978, according to [Robert B.] Reich, a “typical male worker” made $48,302, while the typical top 1 percenter earned $393,682, more than eight times as much. In 2010, even as overall gross domestic product and productivity increased, the average male worker’s wage fell to $33,751. Meanwhile, the average top 1 percent earner was making more than $1.1 million — 32 times the average earner.

This should strike anyone with even a passing familiarity with the US economy as extremely questionable. Are we actually supposed to believe that the typical male worker has seen their earnings plunge by almost $15,000 -- a 30 percent drop -- since the halcyon days of the late 70s? Does that really make sense?

If one performs a google search for "male 1978 $48,302" virtually every result that comes up has Reich's name associated with it, which would suggest that it's a figure that, if he did not simply pluck it out of thin air, arrived at by some method of statistical alchemy. While I initially decided to leave these questionable numbers alone after first reading vanden Heuvel's column last week -- as correcting the economic misinformation spewed by Reich and Nation columnist vanden Huevel could be something close to a full-time job -- last night I noticed the stat repeated by the @urbandata twitter feed:

Seeking once again to establish where this number came from, I sent the following tweet to @urbandata:

@urbandata how is "typical male worker" defined? Where does Reich gets his data from? Looked for this info last week to no avail.
— Colin (@richisglorious) September 16, 2013

Getting radio silence, and disturbed that these stats were being further propagated, I then sent the following tweet to vanden Heuvel:

@KatrinaNation @RBReich how is “typical male worker” defined and what is the source for the stats re: typical male worker income? Gracias.
— Colin (@richisglorious) September 16, 2013

Shen then gave a one-word response:

@richisglorious @RBReich Median
— Katrina vandenHeuvel (@KatrinaNation) September 16, 2013

According to the US Census Bureau (Table P-5), median male income in 2010 was $33,221 (2011 dollars). This isn't the same as the Reich/vanden Heuvel figure of $33,751, but it's fairly close. What is not close, however, is the median male income figure for 1978. Rather than the $48,302 claimed by Reich/vanden Heuvel, the Census Bureau provides a figure of $34,596 -- a discrepancy of over $13,000! I sought to alert vanden Heuvel to this fact:

@KatrinaNation @RBReich in that case the data is completely wrong, as http://t.co/xG3t88UQE9 lists 1978 male median inc as $34.6K, not $48K
— Colin (@richisglorious) September 16, 2013

An answer has thus far not been forthcoming. Additional searches for where Reich could have produced his figure for money made by male workers in 1978, produced this post from the New York Times' Economix blog regarding male and female wages. This chart in particular is fairly interesting:

As can be seen, earnings for male workers since 1970 -- rather than falling to less than $34,000 -- have consistently remained above $40,000. The $33,000 figure cited by Reich and vanden Heuvel only makes sense if one looks at earnings for all men (not just workers), which appears to be approximately $33,000 for 2010 (of course, since Reich/vanden Heuvel specifically said male "workers" rather than all men, this provides zero vindication/validation).

Furthermore, even if we take the all men figure, the income for this group in 1978 appears to be around $43,000 for 1978 rather than the $48,000 claimed, so they are still wrong. The only possible explanation, which still doesn't make much sense, is comparing stats for male workers in 1978, which appears to be about $45,000, against all males in 2010, which is a case of apples and oranges. 

Either way, it seems that Reich has a duty to explain where his figures -- which appear to be both at odds with those of the Census Bureau and the Brookings Institution -- come from. More likely, given the lack of supporting evidence, it seems he owes a correction.