Richard Florida
by Richard Florida
Thu Jan 29th 2009 at 11:04am UTC

State of Human Capital

The U.S. Census has just released its latest assessment of human capital and educational attainment (h/t: Kevin Stolarick). It’s the first assessment to use information from both the Current Population Survey and the American Community Survey which enables analysis of historical trends and state-by-state comparisons. Lots and lots of interesting nuggets:

  • 27 percent of U.S. adults have a bachelor’s degree or above.
  • Those with a bachelor’s degree on average earned about $20,000 more a year ($46,805) than workers with a high school diploma ($26,894).
  • A slighly larger percentage of foreign-born residents (11 percent) had advanced degrees (that is a master’s degree or above) then those born in America (10 percent).

8 Responses to “State of Human Capital”

  1. steven Says:

    Not sure Lou Dobbs would enjoy this point

    “A slighly larger percentage of foreign-born residents (11 percent) had advanced degrees (that is a master’s degree or above) then those born in America (10 percent).”

  2. Michael Wells Says:

    While slightly more foreign born have advanced degrees, native born have more HS graduates, 88% to 68%. Immigrants come for opportunity at both ends of the education ladder.

  3. Mike L Says:

    “A slighly larger percentage of foreign-born residents (11 percent) had advanced degrees …”

    The report also says “Asian 19.6%” – what a huge boost for the USA!

  4. Buzzcut Says:

    I may have posted this before, but I’ve done regressions of the Census Bureau’s household income stats.

    The correlation coefficient for household income controlling for number of income earners, hours worked, and years of income is 0.996.

    Years of education is the strongest determinant of household income. Actually, I translated degree earned into years of education, so the degree you earned is the most important determinant.

    Interesting stuff. The regression equations I come up with are pretty startling. Households with 2 professionals with lots of hours make a lot of money indeed.

  5. Brian Says:


    Your regression looks weird. Why are there only 7 observations? What is the unit of analysis? Besides, the red flag is indeed your r-squared. One NEVER EVER sees an r-squared of .995. That immediately signals that something is amiss. If you could spell out the steps by which you estimated those regressions, we here could check your work.

  6. Buzzcut Says:

    I agree. The R squared is stupid high. Kind of like the Census Bureau made up the data. Maybe it is heavily processed.

    Keep in mind that the data is houshold income, and its broken down into quintiles. They report the median income per quintile. Actually, looking at the data again, they break houshold income into 7 income ranges (under $15k, over $100k, etc.). They give the median income for those in that income range, which is what I used in the regression.

    Then they break down the data. For example, for the people in each range, what is the median number of income earners per household. What is the median educational attainment (no HS degree, a HS degree, some college, a BA, some graduate school, etc. I translated that into a number of years of schooling, so I could regress it).

    For hours worked, the data is broken down by weeks per year worked(more than 50 weeks, 27 to 49 weeks, etc.). Again, I just translated that into a median so I could regress it. I actually translated it into a median hours worked per year, so that’s what you would put in the regression equation.

    I can’t claim to be a statistical genius or anything. I’m just an engineer, we’re not heavily into statistics like some of the social sciences. So I most certainly could be doing something wrong.

    On the other hand, as I wrote in the blog, I just started with what was on Wikipedia for household income, followed their reference back to the Census Bureau, found their excel file for the data from 2005, and went from there. I’m not doing anything that Richard couldn’t do in about half an hour.

    If I’m wrong, prove it.

  7. Mike L Says:

    Buzzcut, did you include the variance around the median in your regression? If not, the regression error (unexplained variance) is much too small, and the r-squared is much too high.

  8. Isaac B Says:

    I still find the data which correlates higher levels of education with higher income to be highly flawed and misleading.

    If the research Dr. Florida points to in his own research about correlating happiness and wealth is over-rated, or at least not anywhere as important as we have been led to believe then how does this support this idea?

    Perhaps more relevant to the battle between traditional assumptions about education held that are becoming less relevant to those raised in more developed nations, the (arguably) outrageous amounts of debt American young people are encouraged to take on to gain a university level degree they often don’t even use properly, is going to severly limit their choices and encourage them to be more materialistic…which in full circular logic will lead them to pursue less fulfilled lives, particularly as our spend at any cost system is collapsing.