When a smart professional like Ronit thinks that my analysis of the College’s use of leverage is wrong then either, a) I am wrong or, b) I haven’t explained things very well. Since a) is inconceivable, I will give another shot at explaining things below. Summary: Leverage is bad. The College should not have borrowed $267 million dollars. It should have spent down the endowment instead. At the time these decisions were made, reasonable people (like the ones who run Williams) thought it was a good idea. And things might have worked out. But, knowing what we know now, there is no doubt that the College would have been tens of millions of dollars better off if it had not borrowed so much. Should we have realized that 2 years ago? I did.

Ronit writes:

What Morty probably said was that “our debt isn’t leverage”; what this means is that Williams is not borrowing money in an attempt to juice returns from the equity markets. He is, in a sense, refuting David Kane’s argument to that effect in this thread, where he compared Williams College’s taking on debt to the actions of highly leveraged property speculators. I guess maybe Morty reads Ephblog after all!

Not closely enough. Let’s use a simple example to keep the math easy. Imagine that it is June 30, 2007 and the College has a $2 billion endowment and no debt. We are about to start a $200 million building project. We have two options. First, spend that money from the endowment, leaving us with $1.8 billion. Second, borrow that money by issuing $200 million in bonds at, say, 3% interest. Make the example easier by assuming that we spend the whole $200 million on July 1, 2007. Assume that the debt comes due in one year. What should the College do?

[Digression: Perhaps some knowledgeable readers could give us more background on the details of college borrowing. My understanding is: First, colleges (and other non-profits) can borrow money by issuing bonds that are tax free to the lenders, just like municipal bonds. Second, the college can’t just borrow whatever it wants. It can only do so in conjunction with capital projects like new buildings. Third, there are a lot of messing details in terms of when the money arrives from lenders, what happens to it before Williams spends it and so on.]

The conventional wisdom was summarized by the (appropriately named) “Simple” two years ago.

Because Williams’ cost of equity is higher (i.e. return on the endowment) than its cost of debt. Especially if the interest paid is tax exempt.

Come on Kane, don’t you work in finance?

The point was reinforced by “johnatrisk” earlier this month.

Seriously. I give up. It’s really only the last shred of intellectual integrity that gives me the energy to type this. Since the comment that disturbed you so much is credited to Simple (I really wish I know who that was, and shake his or her hand) I’m going to stick mostly to declarative sentences.

Debt is not bad. Issuing debt is not “the logic of the condo flipper.” I mean who writes this stuff? Debt issuance has many associated costs and benefits, including but not limited to taxes, financial distress, and is the solution for many investment distortions related to agency conflicts. There is a lot of research on this.

I really can’t believe that I have to defend borrowing, but I guess it’s come to this. Lastly, I can’t even address the “loss” that you calculated for the university on its endowment.

Because finance is about EXPECTED returns, and not REALIZED returns.

Now I think my head is about to explode. I get so tired of explaining this to people who wait for an event to rehash some comment that stuck in the craw 2 years ago. Really, if we have to start here, financial literacy in this country is at an alltime low.

Indeed, but not for the reasons that johnatrisk thinks. Let’s go back to our stylized example, working in nominal terms. After one year, the College pays off the debt at a total cost of $206 million. But what happened to the endowment during that year. Working in EXPECTED returns, i.e., making our best forecast as of June 2007, we might expect the endowment to return 8%. By June 2008, the endowment would be at $2,160,000. Once we pay off the debt, we are left with an endowment of $1,954,000.

Compare that outcome with what would have happened if we had just spend down the endowment in June 30, 2007. In that case, the endowment would have still earned 8%, but on a starting value of only $1.8 billion, resulting in a value of $1,944,000. The difference between this value and the endowment value in leveraged scenario is $10 million. In other words, the College makes $10 million dollars by borrowing $200 million for a year rather than spending the money immediately if the endowment returns 8%. Was it reasonable for the College to expect an 8% return in June 2007? I don’t think so, but hold off for now on what expectations were reasonable then.

Note that the math is actually much simpler than described above because we can look just at the borrowed money and the difference between its interest rate and the endowment rate of return. We can ignore the $1.8 billion that is in the endowment in either case and just look at the marginal contribution of the $200 million in borrowing. The $10 million in “profit” is just 8% – 3% = 5% (the spread in rates) multiplied by the $200 million.

Alas, the endowment may not have REALIZED returns of 8%. Instead, assume that the endowment drops 20%. (The endowment from June 30, 2007 to June 30, 2008 was actually down 1%. But returns since then have been much worse than 20%. I am just using a simple example here.) In that case, the, we have the same borrow of $200 million but now the interest rate spread is negative: -20% – 3% = -23%. And -23% of $200 million is $46 million. In other words, using borrowed money to pay for new buildings instead of spending down the endowment costs the College $46 million in this scenario.

The exact details of what the College has done over the last few years are unclear. For example, I can’t easily calculate the average interest rate that the College is paying on its debt, and the College has not released its recent endowment returns. But there is no doubt that the College’s borrowing over the last year has been in the hundreds of millions and that the endowment’s return over that time period has been significantly negative. The College has lost tens of millions of dollars because it borrowed so much money. If the College had known, in 2006 or before, the future rate of endowment return, it would not have borrowed so much money.

So, it is obvious that, ex post, the College’s actions were stupid. But were those choices stupid ex ante? That is a much harder question. Since the College is run by smart Ephs (and since other colleges made similar choices), I wouldn’t call those decisions “stupid.” It is always hard to know what the future will bring. After two decades of more or less continuous bull markets, it was easy to expect the good times to continue. If the College’s endowment had, in fact, returned 8% over the previous 25 years, wasn’t it reasonable to assume, in 2006, that 8% was a fair forecast for the future? If, given that forecast, wasn’t it smart to borrow at 3%? If you can make 8% and borrow at 3%, you have a money machine!

And that was the problem. Despite 25 years of excellent returns, it was deeply suspect of Williams (or any college) to assume that it could make 8% returns (or 5% real returns after inflation) forever. After all, world GDP growth is definitely lower than 5%, probably closer to 3%. If world GDP grows at 3% and the Williams endowment grows at 5%, then, with mathematical certitude, Williams will eventually own the entire world. Since that outcome seems unlikely, one of our assumptions is false. It may not have been obviously stupid for the folks that run Williams to borrow hundreds of millions of dollars in 2006. Who knows? In some alternate universe, it might have worked out! But it was stupid (it is still stupid!) to assume, for planning purposes, that the Williams endowment could, over the long term, grow at 5% in real terms. That is impossible.

The lesson? Make sure that your assumptions about the world have some connection to reality. Don’t assume 5% real returns just because everyone else assumes that. Note the psychological bias in favor of such pleasant assumptions. The higher the EXPECTED return, the more money that we can spend now!

Should Williams learn from this mistake and pay off the debt now? I don’t know. The logic of the above argument applies just as well today as it did in 2006. But, at the same time, you hate to sell low after buying high. In order to pay off that debt now, the College would have to liquidate much of its endowment holdings at a substantial loss. If I were on the Investment Committee, I would probably vote to do so — ideally by buying the debt back at a discount; I bet that there are some motivated sellers out there. But I could also see the merits in holding tight, in giving the markets a chance to bounce back.

But there is simply no excuse for failing to learn from this mistake, for serious Ephs like Simple and johnatrisk to not realize that, knowing what we know now, the College would have been better off in 2006 with zero debt. It is absurd for Morty and the Trustees to pretend that 5% real returns is a plausible assumption for the long term returns of the endowment. Where are those “careful stewards” when we need them? Going forward, Williams should assume something much closer to 3%.

[There is a tiny chance that I am wrong in this overview of the College’s finances. Perhaps there is some reason why borrowing hundreds of millions of dollars was a good idea, even knowing what we know now about the fall in global financial markets. Corrections/justifications welcome!]

Print  •  Email