The most painful part of writing The Case Against Education was calculating the return to education. I spent fifteen months working on the spreadsheets. I came up with the baseline case, did scores of “variations on a theme,” noticed a small mistake or blind alley, then started over. Several programmer friends advised me to learn a new programming language like Python to do everything automatically, but I’m 98% sure that would have taken even longer – and introduced numerous additional errors into the results. I did plenty of programming in my youth, and I know my limitations.
I took quality control very seriously. About half a dozen friends gave up whole days of their lives to sit next to me while I gave them a guided tour of the reasoning behind my number-crunching. Four years before the book’s publication, I publicly released the spreadsheets, and asked the world to “embarrass me now” by finding errors in my work. If memory serves, one EconLog reader did find a minor mistake. When the book finally came out, I published final versions of all the spreadsheets underlying the book’s return to education calculations. A one-to-one correspondence between what’s in the book and what I shared with the world. Full transparency.
Now guess what? Since the 2018 publication of The Case Against Education, precisely zero people have emailed me about those spreadsheets. The book enjoyed massive media attention. My results were ultra-contrarian: my preferred estimate of the Social Return to Education is negative for almost every demographic. I loudly used these results to call for massive cuts in education spending. Yet since the book’s publication, no one has bothered to challenge my math. Not publicly. Not privately. No one cared about my spreadsheets.
The upshot is that I probably could have saved a year of my life. I could have glossed over dozens of thorny issues. Taxes. Transfers. The effect of education on longevity. The effect of education on quality of life. The effect of education on crime. How unpleasant school is compared to work. Instead of reading multiple literatures to extract plausible parameters, I could have just eyeballed and stipulated for every tangential issue. Who would have called me on it?
Notes and commentary.