[Update, November 17, 2013. If you read this post, *please* take the time to read all the comments, 60+ at present, which may stretch onto several pages. Readers and the GET itself have provided very important clarifications and perspectives (including some wrong or misleading ones) and it’s important to read *all* the comments. Thank you.]
Just over three years ago, I wrote a post entitled “The Washington State 529 Program (GET) Offers an Overlay.” The post has since been the most-commented on the blog (many other comments having been lost when I switched to WordPress, unf.). However, the post’s original thesis (as stated in its title) is now quite wrong, and I no longer recommend the GET and have not for several years.
This post will be mostly qualitative, as I do not have the time to produce a more rigorous analysis; I apologize, and feel free to add your take, qualitative or quantitative, in the comments.
I. The GET’s predominant characteristic is that it is an unfunded defined benefit investment scheme, which creates several inherent tensions.
There are two parts here: “unfunded” and “defined benefit.” “Unfunded” means that, like Social Security, the system uses “pay as you go,” at least in part. This means that, moreso than other investment types, the performance of the investment is contingent upon future buy-ins. (Cynics would say “greater fools,” but that’s not really fair; lots of things are unfunded but entirely legitimate and not foolish.)
“Defined benefit” means that the scheme is promising some particular return. Usually, when one says “defined benefit” one speaks of pension plans that have numeric formulas for determining or projecting specific dollar payouts; here, the payout is linked to state university tuition rates. This means that, versus other investment types, the GET plan assumes the investment performance risk (mostly, but see below).
II. Comparison to insurance.
When you buy an insurance policy, the insurer makes money in two ways. There’s underwriting profit, which comes from charging you slightly more than the actual expected (probability- and time-adjusted) value of paying off a potential future claim. (E.g., there’s a 1% chance that your house burns down and they have to pay $250k, so they charge you $3000 for this expected $2500 liability, and bank $500.) Then, there’s float profit, which comes from taking the money they sit on, and earning some investment return in the interim.
Now, insurance as an industry has been around a long while. (If you want to know more, the colorful Andrew Tobias has written a book that is actually a quite engaging history and critique of the insurance business, believe it or not.) And for “real” insurance, that is, casualty and life insurance (not health “insurance” which is, in my opinion, a vast and crass misnomer), the system works really quite well: the value of the asset that is insured is pretty well scoped out by the contract and by the market system.
Another fun benefit arises when you have big money at risk in casualty insurance: you’ve now created big entities that have a vested monetary benefit in making the world safer and less prone to fires, theft, flood damage, untimely deaths, etc. This is because, except in fraud situations, the common adversary in insurance transactions is misfortune; that is, both you and the insurer would rather your house not burn down (though technically you sort of “win” back your premium if it does). So insurance companies send out workplace safety inspectors, and mail you free cell-phone headsets, and offer discounts for driver’s ed and sprinkler systems and whatnot, and so they lower their expected payouts, increase their underwriting profits, and you stay a bit safer. Win-win-ish.
With the GET, there are some key differences. One is that the “casualty” being insured is your kid going to college (or you otherwise spending the dough). This is virtually certain to happen, because even if your kid decides at 19 to go on tour with an all-handbell Steely Dan cover choir and eschew higher ed, you’ll find a nephew or neighbor kid or someone else to use the funds. So there’s very little uncertainty about the fact and timing of payout.
The risk here is how much UW tuition is going to cost when your kid turns 19. That’s the risk that GET notionally takes on your behalf. It’s worth paying some reasonably large underwriting premium not to have to think about that risk (but see below).
III. GET is a governmental scheme and that can get wacky.
So, if GET were an independent entity, say a bank or insurance company, that said “no matter what UW tuition is, in X years, we here at Bear Stearns Lehmann Bros AIG Acme Bank will pay you that amount, in return for $Y today,” you’d think of the risk like so: Will this entity be able and willing to make good on its promise in X years?
But, there are two complications here. One to the upside, one to the downside.
On the upside: GET notionally is backed with the full faith and credit of the State of Washington. There are critiques here to be made, such as the fact that the backing is in statute, and not in the constitution, and that given sufficient political will, the legislature or the people by initiative could decide that a bunch of upper-middle-class tax-dodgers need to pay for their class’s sins and confiscate, dishonor, or otherwise do bad stuff to the GET. But, probably, if the state is doing its usual stuff and the roads are paved and the ferries are sailing, the GET will get paid.
On the downside: GET is run by the same people who decide what to charge for tuition. Yep, that’s right: the political interconnections between the GET leadership and the state financial and higher-ed communities are significant. If the GET program should face a shortfall, any of the following options might start to look appealing:
- Recharacterize a lot of the UW “tuition and state-mandated fees” to be not-quite “state-mandated” fees.
- Keep an artificially low in-state tuition (perhaps making up the below-market rate by capping in-state attendance and jacking up out-of-state tuition).
- Do some clever calendar-changing with trimesters / semesters / years / half-courses / whatever that effectively keeps nominal tuition low.
Look, this isn’t saying that anyone is corrupt, and I’m certainly not a Norquistian starve-the-beast type. But consider what you’re playing for in this game. You’re hoping that GET gives you more (risk-adjusted, at least) than a self-managed 529 plan would return in the public markets. The only way that will happen is if tuition rise at a rate so much faster than the market return that it catches up to and beats the “underwriting” premium.
If that happens, then GET will be way behind, because all they’re doing is investing in 60% stocks, 40% TIPS. Their options then will be to: 1. increase inflows (get more signups or charge a bigger premium), 2. get help from the state’s general fund (if it is politically available, which we should think likely), or 3. take some measure to limit outflows (pressure the university system to limit “tuition and state-mandated fees.”
To their credit, the GET leadership has started to jack up inflows, and is riding a wave of public disaffection with the stock markets and mutual funds to charge an enormously higher premium (underwriting profit), which is good for the plan’s solvency (but bad for those buying in today).
IV. Well, smartass, why did you recommend it in the past?
In 2007, when my niece was born and I looked into GET, the S&P was flirting with 1600 and attractive valuations were hard to find. Risk premia were at all-time lows and P/E multiples at all-time highs. Bubble-callers smarter than myself were ranting about real estate. Investing on my own for an 18-year maturity seemed like a tough nut to crack, timing-wise.
At that time as well, as my prior post’s table points out, GET offered a more reasonable spread between purchase price and payout value. In 2000, the premium was a reasonable 8%; in 2007, a rich but defensible 19%. Today, the premium is a whopping 36%! (Payout value, $85.92, buyin cost, $117)
I might point out that assets under management at GET have ballooned to $1.3 B over the past year, at the same time as fear has driven individual investors out of equity markets. Therefore, things are going to look like smooth sailing for the next several years at GET. The real problems are going to be years down the road, when investment performance has lagged and all of these new buy-ins become new payouts.
V. What’s the big point here?
Well, perhaps I missed the big point back in 2007. Yes, it looked like a good idea then; you’re probably still getting the best of it if you bought in 2007-2008.
But the bigger point is about defined benefit plans. The management of such plans seems to be an activity fraught with roadblocks to true honesty. By “honesty,” I mean with a truly conservative and best-estimate view of what returns will look like, and what the ability to meet future needs requires of the plan.
For us as citizens, the message is that we need to apply oversight and demand hard-headed thinking, unless we want the near-certainty of having to fund notionally private pockets out of the public purse (see PBGC).
For us as investors, it means eschewing magic bullets, and being duly skeptical when we are promised a return without its associated risk. (It also means jumping at opportunities when they are truly underpriced, as GET was for its first 8 years.)
I’d love to hear your stories about GET or defined benefit plans, and how you’ve thought about the associated risks.