Romney was referring to
this study (the link appears to be a summary of the study; I could not find the actual 50+ percent number in there, but it is what websites cite as the origin of Romney's claim) by Paul Harrington and Neeta Fogg of Drexel University, which appeared in the
Higher Education Review in 2011.
The Monkey Cage, a blog written by actual political scientists who know this stuff (social science studies) better than journalists, has given
an analysis of the claim.In particular, they note:
"t is only fair to note that another post in The Atlantic by Jordan Weissmann notes that in 2000 this figure was “at a low of 41 percent, before the dot-com bust erased job gains for college graduates in the telecommunications and IT fields.” Obviously an increase from 41% to 54% of recent college graduates being unemployed or underemployed is hardly trivial, but only Weissmann seems to have provided the comparison which shows that even in boom times, 40% of recent college graduates have troubles in the job market. Note that the 2000 figure is from before the dot.com recession, when the overall unemployment rate was about 4%. "
So basically, under Harrington's standards, even in 2000, 40 percent of college graduates would be unable to find work.
The blog goes on to note:
"Fogg and Harrington define underemployment (which they call mal-employment) as a college graduate not being employed in an occupation “which utilize(s) the skills and knowledge that are commonly thought to be acquired through a college education.” (p. 55-56). While that article does not specify what such occupations are, it does note that the occupations “generally include profession, technical, managerial and high-level sales occupations….” (p. 55). Of course many of these require an advanced degree, so recent college graduates under age 25 are unlikely to be eligible for such occupations."
In another paragraph, the authors suggest that 'mal-employment' (what they measure) is any work that is 'substandard in some way.'
In any case, 'can't find work' is obviously incorrect when it comes to these people.