Steve Reads

A discussion flared up on Facebook based around this article by Michael Ellsberg in the New York Times whose premise is that college isn’t worth the money, and that we’d be better off encouraging entrepreneurship.

I happen to have been looking recently at the data on this. Let me give you a spoiler: it’s not even close. It’s worth the debt load. Here are the numbers:

• Median income for males with bachelor’s degrees, 2010: $55,038 (does not count those with more than a bachelor’s degree, like doctors or lawyers)

• Median for males with associates degrees: $40,918

• Median for males with some college, no degree: $36,082

• Median for males who graduated from high school or got a GED: $30,232

(I could obviously include numbers for women in here, too. I’m leaving them out only for brevity’s sake.)

The mean incomes are similar, but even more striking:

• College degree, 2010, male: $70,567
• High-school degree or equivalent, 2010, male: $36,755

To dramatize this a bit, imagine I grabbed two men at random and asked them their incomes. The probability that the randomly selected college-educated male is earning more than the randomly selected high-school-educated male is about 74%. [1]. I imagine that if I ran a similar simulation — whereby I simulate our two graduates after each has worked for 40 years, and add up their accumulated earnings — that the results would be even more stark.

Fortunately, the Census Bureau has already done that work for me. On a quick scan, I can’t find “work-life earnings” for all males, so I’ll just compare white males. Over the course of his working life, a college-educated white male (specifically one with a bachelor’s degree, not a master’s degree, a professional degree, or a Ph.D.) will have earned about $2.3 million. His high-school-educated white-male partner will have earned $1.2 million.

I repeat: it’s not even close. It’s so not even close that I consider it irresponsible in the extreme for Ellsberg to cherry-pick some success stories (Steve Jobs, Bill Gates) and imply that students should strive to be like them, rather than encouraging them to take a rather more sure route to success. Had Ellsberg encouraged students, instead, to skip college and aim to be professional sports players, I hope we’d all be deeply offended. The piece he actually did write is no less offensive.

P.S. (25 October 2011): as various commenters have pointed out, this only shows correlation; it doesn’t show causation. It could well be that the people who have the drive to get a college degree are the same as those who have the drive to earn a lot of income — and that they’d earn a high income even without the college degree. It would be interesting to compare the lifetime earnings of those who got into college but chose not to go with those who got into college and went. I’ll see if I can find any interesting data in this direction later.

[1] — This is just an estimate. It’s based on a simplifying assumption, namely that the probability distribution of incomes is approximately lognormal (i.e., that the logarithm of income follows a Gaussian [bell-shaped] distribution). That assumption, combined with the estimated means and medians from the links above, gives the 74% number.

Thanks to Cosma Shalizi for pointing me to a simple approximation to the true income distribution, and noting how I could go from the estimated means and medians to an estimated probability.

Matt Yglesias makes the entirely correct point that We Underinvest In Infrastructure Because We Overinvest In War, Health Care, And Low Taxes, along the way noting the various estimates of how much we need to be spending over the next N years on bridges, roads, and whatever else. Various organizations give us various numbers:

“All of the numbers are so gargantuan large that they’re useless when you’re trying to communicate with the public,” said Roy Kienitz, undersecretary for policy at the Department of Transportation.

The American Society of Civil Engineers has estimated that an investment of $1.7 trillion is needed between now and 2020 to rebuild roads, bridges, water lines, sewage systems and dams that are reaching the ends of their planned life cycles. The Urban Institute puts the price tag at $2 trillion.

The fact that these numbers are so gargantuan is exactly why we shouldn’t be talking about them in raw terms. $2 trillion is an essentially unfathomable number. So let’s try it some other ways:

The IRS collected about $2.3 trillion in fiscal year 2009, across personal income taxes, corporate taxes, Social Security taxes, the estate (“death”) tax, Medicare part A, various excise taxes, and so forth. Individual U.S. states collected about $704 billion in 2010, of which 33.5% was income tax and a bit less was sales tax. Federal and state taxes altogether, then, come to about $3 trillion. So our $2 trillion infrastructure investment is about 2/3 of the country’s total tax bill. (The $3 trillion tax bill, to put that in some more perspective, comes out of a $14 trillion GDP. Taxes come to about 21% of GDP, in other words.)

U.S. 10-year Treasurys are available at historically low rates right now — right around 2.5%. So let’s say the government borrowed $2 trillion today, and paid it off at 2.5% over the next ten years. How much would the monthly payment come to? (There is probably a lot of complexity about government financing that I don’t understand, but I’m treating this exactly like a mortgage with a 2.5% interest rate. Feel free to correct me if there’s some important nuance I’m missing.)

The monthly payment, it turns out, would be about $18.8 billion. This is getting more fathomable. And while the various governments’ $3 trillion in receipts come from many classes of tax levied on many different types of people and businesses, let’s just simplify and say that there are 144.1 million taxpayers — which is the number of individual Federal income-tax returns. How much will each of those taxpayers have to pay every month to ensure that our bridges don’t collapse in a decade? The answer: $18.8 billion per month divided by 144.1 million taxpayers, or $130.43 per month.

Now that is a number I can understand. The cost of ensuring a non-crumbling infrastructure is $130.43 per person per month. How much of an increase is that over what we pay now? Well, we pay a total of $3 trillion per year, or a quarter-trillion per month, or about $1,731 per return per month. $130.43 is a 7.5% increase over what we pay now.

I can wrap my head around “my tax bill will increase by 7.5%.” The story gets even rosier if you consider that investors are currently willing to lend the Federal government money to be paid off in 30 years for about 3%. That lowers the monthly payment to about $60 per person, which is about a 3.5% increase in overall taxes.

All of this assumes, recall, that every dime of the $2 trillion would be new money — money that we wouldn’t spend anyway. That is obviously false: if a bridge collapses, we’ll presumably repair it. And presumably there’s a lot of roadwork that we’d already be doing. But assuming it’s all new money, the average tax bill would go up by somewhere between 3.5% and 7.5%. I can wrap my head around the concept “my tax bill will increase by 5%,” much more than I can wrap it around “$2 trillion over the next decade.”

Note also that, if we somehow managed to get a lot of high-income people into this country, we could lower the per-person expense even more. What counts as “affordable” depends to a great extent upon how many people live in this country, because “per-person expense” has both a numerator and a denominator. There are two obvious ways to get more high-income people into this country: either (1) allow in any immigrant who has an employer to sponsor him or is looking to get a Ph.D., or (2) encourage native-born Americans to have more babies. Ideas for how to make either (1) or (2) happen will have to occupy another post at another time.

Trillions are unfathomable, so it’s important to find some way to put them in context to make them fathomable. Switching from aggregates to per-capita numbers is one quick way. Switching from “costs over ten years” to “cost per person per month” is another way. Percentages are another good measure: percent of GDP, say, or percent of current tax revenues. Otherwise we get stuck in massive-number fatigue, which helps no one (except maybe the political party which takes terrorizing you over the budget as its reason for being).