I previously explained that intelligence is not very highly correlated with income. Based on analysis of the General Social Survey, using educational degree, age, and score on the Wordsum vocabulary test as a proxy for intelligence, I concluded that one’s educational degree is far more important than intelligence when it comes to predicting income, and there even seemed to be some mysterious evidence that high IQ might cause lower income.
This week I decided to analyze the data in a different way, restricting the analysis to respondents with the same highest level of educational attainment.
Here is the shocking conclusion: in the recent years of the GSS (1991 to 2004), for people whose highest level of educational attainment is a bachelor’s degree, there is a negative correlation between intelligence and income. In the 1998 to 2004 data, each point higher on the Wordsum test causes a $1,200 decrease in income.
This trend seems to be increasing with time. Before 1991, there is a slight positive correlation and each point higher on the Wordsum test causes a $500 increase in income.
From 1991 to 1996 we observe the negative correlation, and each point higher on the Wordsum test causes a $700 decrease in income.
This negative correlation only exists for people who have a bachelor’s degree as their highest level of educational attainment. For respondents whose highest degree is high school, there is a very solid positive correlation. In the 1998 to 2004 data, each point higher on the Wordsum test causes a $1700 increase in income. This is not something to get too excited about. There’s a vast difference in ability between someone in the top 5% (Wordsum 10) and someone below the median (Wordsum 5), yet using the figure above the smart person only earns an extra $8,500 per year. This amount corresponds 31% of the median income of $27,500.
There is also a positive correlation for respondents with a graduate degree. In the 1998 to 2004 data, each point higher on the Wordsum test causes a $1,400 increase in income. Because there are only 167 cases in this regression, the t-statistic is only 0.726, so this is a weak correlation. Because the median income for respondents with a graduate degree is $55,000, a five point difference on the Wordsum test corresponds to only 13% of the median income.
So what’s my explanation for all this?
It’s easy enough to explain why people with only high school diplomas earn more if they have a higher score on the Wordsum test. They are being paid slightly more for their higher level of ability. For example, they are more likely to end up in skilled blue collar professions which earn decent salaries.
But why is there a negative correlation for people with only bachelor’s degrees? The obvious conclusion is that what a person majors in at college is far more important in determining future income than how smart the person is. We can also conclude that as intelligence increases, a student is more likely to choose a less financially rewarding major.
This makes sense if you think about it. People of average intelligence who attend college are only attending for the practical benefits of a better career. People of average intelligence don’t appreciate college as a learning experience.
People with higher intelligence enjoy learning, and because of this they may choose a major that seems intellectually interesting (like Roman History or 19th Century English Literature) yet has no practical use.
This negative correlation surely indicates that young people are getting bad career advice from their parents and schools. If there was an efficient system for sorting young people into appropriate majors, surely the more intelligent students would be directed towards fields where their intelligence would most benefit society.
Finally, let’s explain the positive correlation for people who have a graduate degree. In this case, I suspect that higher intelligence predicts a more economically useful major. Specifically, a large percentage of graduate degrees are given to school teachers, because many states either require or strongly encourage their school teachers to get graduate degrees in education. It is well known that school teachers have lower SAT scores (and thus lower intelligence) than other college majors. Respondents with a higher Wordsum score are less likely to be a school teacher, and this explains the very small positive correlation between Wordsum and income.
The final conclusion is that choice of college major and graduate program are the major determinants of future income. Intelligence is only important to the extent that, as intelligence decreases, the universe of educational programs that the student is capable of completing becomes smaller.