The risk of distributing risk
Lambda School and Bowie Bonds
Ranjan here, talking about risk management, income-sharing agreements, and David Bowie.
The ongoing Lambda School drama has one specific part I wanted to focus on: Income-Sharing Agreements (ISA). These are the contracts that effectively make going to the school "free" up front for the student. The terms are:
Lambda’s ISAs promise an alternative to traditional student loans by allowing students to defer tuition until they’ve landed a job that pays $50,000 a year or more. When that happens, they hand over 17 percent of their income until the $30,000 tuition is paid off. If students don’t find work within five years of completing the program, the ISA is automatically dissolved.
In theory, I love the idea of an ISA. If the student gets a good job and makes decent money, they'll pay a share of their income back to Lambda until they’ve paid back the listed tuition. It creates a massive incentive for the school to make sure the student finds a good job; otherwise the school doesn’t make a cent. Capitalism for the win!
What has come out as controversial is that Lambda was selling off the contracts to outside investors, primarily a company called Edly, a marketplace for ISAs.
Edly advertise that people can “earn monthly income while supporting students” while getting “10-15% target returns.” One of the Margins’ “theories of everything” is that zero interest rates have led to distortions in every imaginable sector. This is another tiny example. In a world starving for yield, 10-15% sounds amazing.
So should Lambda selling off the ISAs be a cause for concern?
Let me start by saying, this is the kind of structured finance that should be a thing of beauty. If there are investors willing to pay Lambda to assume the risk involved in the ISAs, why not sell them off? All debt is someone being compensated for assuming the risk of a future cash flow. Spreading the exact risk to the exact right people is the most efficient way to allocate capital.
People are clearly mad at Lambda for marketing the idea that, “we don’t get paid until you do”, when in fact they did. But is the reallocation of financial risk like this intrinsically a bad thing? Clearly, the 2008 Financial Crisis and the blowup of CDOs and MBSs left a bad taste in people's mouths over the chopping up and passing off of debt (note: I now get uncomfortable every time I write “MBS” and “chopped up” in a sentence). There was a long time that I thought this type of risk distribution was an unquestionably beautiful thing.
In college, while studying economics and learning more finance, I came across a relatively new innovation at the time: Bowie Bonds. I loved the idea as the combination of finance and rock music made me feel like I wasn't completely selling out by going off to Wall Street.
The way the bonds worked was:
Bowie bonds were first issued in 1997 when David Bowie partnered with Prudential Insurance Company and raised $55 million by promising investors income generated by his back catalog of 25 albums. The 25 albums, which were used as the underlying assets for Bowie bonds, were recorded prior to 1990 and included classics such as The Man Who Sold The World, Ziggy Stardust, and Heroes.
Bowie bonds, when issued, had a face value of $1,000 with an interest rate of 7.9% and a maturity of 10 years.
It's just beautiful. The archives of any established musician mean somewhat predictable future royalties - they're an income-generating asset. If you're David Bowie, your entire net worth is tied up in this one asset. If something bad happens, you're singularly screwed. Why not sell it off to a large number of investors and spread out the risk?
I vividly remember a good friend and I discussing the idea that we should "securitize ourselves." I had about 80k in student loans and was completely cash-strapped at the time. But I was also going into finance, so there was a decent probability I would be making good money.
Imagine getting someone to take over my loans, and maybe even giving me cash up front, for the right to a percentage of my future income. Maybe there wouldn’t be a cap on the payout, so the rich people investing in me would have significant incentive to leverage their own networks to get me a high-paying job, even mentoring me along the way.
It's not really different than a VC investing in a startup and working their network to see a return, right?
When I got into the world of finance, the notion of allocating risk in a highly targeted manner further seduced me that this financialization mindset was the greatest thing on earth.
Here's a very simple example in foreign exchange: A big US corporation has a plant in Thailand. They know they have to make a big cash disbursement in the Thai Baht in 9 months, but their corporate treasurer doesn't want to deal with currency risk. The bank trader offers to send them Thai Baht in nine months at a price agreed today. They’ve just sold off the currency risk to the specialist trader, and can go back to focusing on making widgets.
I bring up this currency example, because it's what I worked in and, I still believe, a genuinely good example of financial innovation. The globalized economy could not run without this constant chopping up and reallocation of risk.
When I first learned about mortgage-backed securities, they too seemed similarly brilliant. Pool up mortgages and chop them up. Sell the riskier ones to people who want a higher return. Imagine the extreme alternative, a single, local bank who assumes the entire default risk of every mortgage they issue. This is inefficient. The innovative new, financialized model would result in far more mortgages being issued, and consequently, more home owners.
Yet, we all know how that turned out. The financial crisis had a huge impact on how I see the world. Possibly the biggest lesson I took away was, the more distant a risk becomes, the more distorted all the related incentives become.
Picture a small-town mortgage from a local bank. I'm not even sure this exists anymore, and I'm kind of imagining a Norman Rockwell painting, but the type of thing where they personally know every single borrower. The bank's well-being is directly tied to whether the borrower repays the debt. They're incentivized to find good borrowers. They're even somewhat incentivized to make sure property values in the town rise.
But then a big New York bank comes in and buys up their entire mortgage portfolio, along with the portfolios of hundreds of other small banks. Now these banks have already been compensated and care a bit less whether the borrower repays. In fact, they're incentivized to generate more and more mortgages, knowing they can sell them off. The big NY bank chops up their giant portfolio into various risk tranches, and sell these pieces off to big pools of money with names like the Ontario Teachers Pension Fund (seriously, they have over $200bn in assets).
At every step of the way, the people owning the risk are further and further away from that original homeowner. They have no idea what's happening in their lives or what it's like to live in their neighborhoods or even whether the borrower has a job or steady income. Even if they wanted to, they most likely couldn't figure it out.
This reminded me of Can’s piece on empathy (and his work at Uber):
in our well-meaning attempts to roll up thousands of lives in a single data point on a Graphite dashboard or a Kafka log, it’s easy to lose sight of what is at stake. I had never been responsible for the lives of hundreds like an air traffic controller is, but it wasn’t too far off.
Back to ISAs
Which brings us back to the Lambda ISAs. At first glance, it doesn't seem that slimy for Lambda to be selling off the ISAs to a specialized marketplace. If there are people with more specialized risk appetites, why not let them buy these up? It almost necessarily means you can bring in more students because Lambda is now taking in cash up front, and not bearing the associated risks. More people learning to code!
But once again, it's the incentives. If Lambda owns the ISA, they own the risk. It's how they marketed themselves forever (this is from 2018):
“To some degree — if a student doesn’t get a job, that’s on us and we shouldn’t get paid,” Lambda School cofounder Austen Allred told VentureBeat.
Their homepage even advertised (as per NY Mag):
“We don’t get paid until you do, so we’re in this together, from your first day of classes to your first day on the job.”
The moment Lambda started selling off the contracts, their incentives completely flipped. Their income was then derived from generating ISAs, meaning, the more, the better. Enroll as many students as possible. Of course, if no one ends up getting jobs, investors will eventually stop buying the Lambda ISAs, but that would all happen a long ways down the road.
People are right to be angry that Lambda was selling off the ISAs. It broke their core promise.
P.S. Everything is Risk Management
Note #1: Can’s last piece about how shopping on Amazon sucks had me thinking about the distribution of risk, independent of the Lambda story. It reminded me of a horrifying CNBC investigation into how Amazon regularly sells expired baby formula.
When I walk into a local store and buy baby formula, the shopkeeper clearly has assumed the risk that the formula is not expired (hahaha, just kidding - there are no local stores anymore). But, in the Amazon scandal, it's not even Amazon selling the formula, but a third-party seller. The risk for that baby formula has been chopped and distributed into so many steps that I have no idea who the real seller is and Amazon certainly won't be held liable.
A well-run marketplace is, at its core, an efficient redistribution of risk. The marketplace handles certain elements (user reviews, tech infrastructure) while passing off certain risk to both the seller and buyer. When it works, it’s the good kind of risk distribution (Amazon is not this kind).
Note #2: This tweet also had me thinking about far-flung risk management:
The same way financial innovation created mortgage-backed securities which enabled a massive scaling in home ownership, globalized supply chains have led to a previously unimaginable explosion in commerce. Everything from production to inventory management to delivery has been outsourced to specialized, low-cost providers in various geographies. Everything is so cheap and comes to you so fast.
It’s why I’m terrified about the impact of the Coronavirus on the global economy. Like mortgages in the 2000s, the risk has been spread so far and wide and thin. We have no visibility into how an interruption like this plays out. The market has been shockingly sanguine about the potential risks, even with Apple et al starting to ring the alarm bells, but the most dangerous part to me is just how unknown the unknowns are.
Note #3: On the Lambda School story, if you weren’t aware of it, make sure to start with The Verge’s The High Cost of a Free Coding Bootcamp.
And then NY Mag’s Lambda School’s Misleading Promises, which gave the amazing anecdote:
Before I left, I asked Allred about the thing that had first attracted my attention about Lambda: Why did he post on Twitter so much? Taking off his hat, he told me that he “doesn’t really have friends,” and used Twitter as a way of having some social interaction. A moment later, he clarified that he felt “an obligation to do everything I can for the students, in a way that I just don’t feel towards friends.”