In a number of recent posts, I’ve claimed that math is becoming increasingly important in the modern labor market. If this is true, it would have wide-ranging implications for how best to promote social mobility in low-income communities and whether or not we should teach advanced math to high schoolers.
My assumption was that the rise of the software industry and quantitative techniques in finance and academia made the increasing economic value of math more or less self-evident. I thought that with a little bit of digging, it would be easy to prove that advanced math is more important today than it was twenty years ago.
As it turns out, there is a lot less research on math use over time than I thought there would be. The Bureau of Labor Statistics has a page with data on “math careers”, but only actuaries, mathematicians, and “operations research analysts” fall into this tightly defined category.
Upon learning that my claim was based mostly on anecdotal evidence, I decided to take a look at the problem myself.
The BLS publishes Occupational Employment and Wage Statistics data each year, and records for this century are available on their website. For each of 900 or so occupations they identify, the BLS lists the number of people employed in that occupation and their median annual salary. In order to tackle the question of whether or not more people are working in math-intensive fields, I tagged each of the BLS occupation categories manually, indicating yes or no for the following labels:
Algebra (HS): The occupation regularly uses algebra at the level taught in a normal high school Algebra I course.
Algebra/Discrete: The occupation regularly uses techniques from a college-level algebra or discrete math course (ex. graph theory in software engineering).
Analysis: The occupation regularly uses techniques from a college-level analysis course (ex. some financial analysis).
Topology/Geometry: The occupation regularly uses techniques from a college-level topology or geometry course (ex. some aspects of engineering design).
Statistics/Data Science: The occupation regularly uses techniques from a college-level statistics course (ex. actuarial science, econometrics).
Number Theory: The occupation regularly uses techniques from a college-level number theory course (ex. cryptography).
There is of course a great deal of overlap between these labels. I am at best an amateur mathematician, so take my opinion on what goes where with a grain of salt.
My policy was to label very conservatively, applying a label only if I (and/or Google) thought the relevant subject was central to the occupation’s work. For example, I applied no label to “Family and General Practitioners”, even though most doctors need a working knowledge of statistics (and indeed must take much more math in their path through medical school). You can check my work on the labeling via the CSV file here. I am sure I mislabeled at least a few of the 925 occupations, but I think that the aggregate data can still tell us something.
Once the labeling was done, I ran a Python script on the BLS data for the years 2000 to 2020. It measured both what portion of the workforce was in an occupation requiring at least Algebra (HS) and also what portion of the workforce was in an occupation with at least one label beyond Algebra (HS).
Both graphs show a significant upward trend in use of math over the last twenty years. It is worth pointing out that the absolute percentage of workers using post-secondary math starts very low, at 5.78%, and only increases by about two points, to 7.81%. Even the high school algebra percentage is not so large, topping out at 17.1%.
My claim previously had been that both the absolute use of math and the rate at which it was increasing in importance were much more pronounced in high-paying occupations. In order to quantify this, I categorized an occupation as high-paying in a given year if its median annual salary was at least 1.5 times the median annual salary for all workers that year. I ran my analysis again on only these selected occupations.
Again we see a significant upward trend, but this time the absolute numbers are much higher as well. Most impressive is the data on the proportion of highly-paid workers using post-secondary math, which increased by over ten points from 21.7% in 2000 to 32.5% in 2020.
It should be noted that my approach to this data analysis has at least two weaknesses. First, it required me to subjectively label occupations. More importantly, it missed any intra-occupational changes in math usage (for example, the increasing reliance of medicine on statistics).
All that being said, it seems clear that advanced mathematical ability is an increasingly important skill in the modern labor market. All four of my tests showed significant upward trends in the proportion of people working in math-heavy fields. This was particularly pronounced in high-paying occupations, where close to a third of all workers are now employed in jobs that require some college-level math.
The fact that there has been significant change in professional use of math since 2000 indicates that education policy grounded in 20th century thinking may need to be reexamined.
Postscript
Out of curiosity, I also collected some data on the proportion of workers using each field of math in my tagset. Due to the small sample size, these graphs should be taken less seriously than the others, but I thought I would reproduce them here for general interest: