My friend Sarah Constantin just wrote a post about wage statistics. These are very important, and she deserves applause for bringing them up. However, they must also be used carefully when deciding what job you should pursue. Let’s explore why, taking doctors as an example.
“What Jobs Pay Best? Doctors. Definitely doctors. The top ten highest mean annual wage occupations are all medical specialties. Anesthesiologists top the list, with an average salary of $235,070…. Bottom line: if you want a high-EV profession, be a doctor.”
Statistically, this is totally true. But if you want to become a doctor, and make that high salary, there’s a catch. In fact, there are a lot of catches. (Wages for any job should be ‘adjusted’ up or down for these factors, but doctors make a good case study.)
The first catch is, of course, medical school. Since you must go there to be a doctor, and it’s very expensive, a doctor doesn’t ‘really’ make the salary listed – part of the money must be used for repaying med school tuition. The same is true for college tuition.
The second catch is usually called ‘opportunity cost’. Consider a cashier vs. a doctor. The cashier makes less, but can start making money at age 16. The doctor doesn’t start until 30, on average – he does a lot of work (in school) for which he isn’t compensated. If people start ‘working’ at age 16 and retire at 65, the cashier is paid for all of those hours, while the doctor is only paid for two thirds. To make it a fair comparison, the doctors’ wages must be ‘spread out’, to cover the years he isn’t paid anything.
The third catch makes the first and second catches a much bigger deal than they seem. That’s discount rates. If I offer you $1 in 2040 in exchange for $1 now, you probably say no. Money in the present is worth more than money in the future, which is why loans charge interest.
For example, $250,000 in med school tuition isn’t that much, spread over a doctor’s career. However, the tuition is due now, while the career earnings may not arrive for thirty or forty years. How much this matters depends on what your ‘effective interest rate’ is. If it’s 5%, then each dollar of med school tuition must be ‘repaid’ by five dollars of additional salary thirty years out, in order to make it a ‘fair’ deal.
There’s also simple number of work hours. If doctors make a lot, but must work eighty hours a week, then a ‘fair’ comparison is to someone who is (eg.) working two forty-hour-a-week cashier jobs.
And finally, we get to more ‘subjective’ (but still very important) factors. Consider IQ. Getting into med school is hard; suppose, picking a number out of a hat, that you must have at least a 115 IQ to enter. If the average manager has a 100 IQ, then if you go into management, you’ll probably make more than the statistical median – the ‘real’ comparison, between doctor and the management job you’d actually have, is less favorable to doctor than it seems. Likewise, if the average physicist has 130 IQ, a doctor going into physics will most likely do worse than median, so the ‘real’ comparison is more favorable to doctor. The same is true for work ethic, charisma, initiative and all the other intangibles.
Summing it all up, one gets a net present value calculation: the EV of a job option is equal to the sum of discounted future cash flows, both positive (like salary) and negative (like college tuition). NPV is how good companies make business decisions, and I think we can do well if we also use it for career decisions.
You’re being very thorough, but even your proposed analysis still has biases:
The average salary of a doctor is not the same as the expected average salary for a person trying to become a doctor. Some of the latter will be unable to get into med school, unable to make it through med school, unable to find jobs afterwards, etc, and there are obstacles in other career paths which are qualitatively analogous but have greatly differing quantitative failure rates. To use a better example from the linked article, “airline pilot” might be an excellent job, but anecdotally it’s not one that you can jump straight into without an indefinite period of inferior piloting jobs first.
Expected value should be measured in utility, which is typically varies very sublinearly with dollars. Even if we were to take average salaries from sets without survivorship bias, those professions with similar averages but higher variance could be less preferable to anyone who can’t confidently predict that they’ll be above average. E.g. the legal profession is now having problems here: the mean salary for a new lawyer is still great, but the distribution is now bimodal, with the majority of lawyers earning well under the mean balanced by a minority earning far more.
And of course the final unavoidable flaw is just “past results are not an indicator of future performance”. I sometimes think I’d like to see “what are your expected earnings in these careers” be a mandatory class for freshmen in high school, but not only would that information fail to account for ways in which the past fails to predict the future (Which jobs are going to be automated away next? You’ll never find out from historical data), but it could even be a self-defeating prophecy: students move into the career paths which are most lucrative at year X, which creates oversupply and crashes those careers’ salaries at year X+15.
Useful data on expected earnings given a career path seems to be nonexistent or at best, not accessible to the average high school student. Seems the colleges have an interest in obfuscating this, they sort of sell a dream of a better future, which you buy into at 17 or 18, before your brain has even fully developed the ability to make and be motivated by long term decisions / outcomes. Curious how to better align this, such as a pricing model for college that is tied to future earnings of the graduates – like taking an equity position in their students.