Some thoughts about correlation: Startups v.s. CDOs

Everyone knows that the last financial crisis was due to sub-prime loans. Most know that the main financial assets in question is CDO (Collateralized Debt Obligation). Some know that it's actually the short side CDS (Credit Default Swap) that triggered a lot of leveraged speculation. A few know that the market came crashing down because the financial model that everyone was using was broken.

However, few know that it was actually a key input into those models that really killed the market: the estimation of correlation among the default rates of the mortgages in the same pool.

The idea of CDO started with corporate bonds. Due to different industries and sectors, default rates of corporate bonds have limited correlation among them. There are also a lot fewer players.

When one applies CDO to home loans, one is facing tens of thousands of home owners, most of which make their living on salaries, which are correlated in general to the macroeconomy. As a result, the correlation should be very high in theory.

However, before the boom of CDO on home loans, there were limited data that track the individual cases of mortgage defaults. When mortgage CDO picked up, financial engineers did not have enough historical data to give a proper estimate of correlation among the mortgages. Most of them, therefore, seriously under-estimate the correlation, which significantly under-estimate the risks of even the most senior tranches, called Super Senior, of all those CDOs.

Hence when the housing market started to slow down, default rates shot up like crazy, surprising most of the investment banks in Wall Street that were in the game. The rest is history.

Correlation matters to startups and VCs as well.

In the dot-com bubble of the late 90's, many startups were doing B2B businesses that were basically selling services or products to other startups. When the public market was red hot, the VC funding was ample as well. When the VC funding was ample, the startups were buying services from other startups. However, at the very end of this value chain, the consumers hadn't picked up the slack yet. As a result, most of the B2B sales stayed just like as B2B sales among VC-backed startups, without the final undertakers that were the consumers.

If this is hard to understand, think about macroeconomics. GDP is either the total production or total consumption. The two have to equal (excluding import/export and other secondary components). B2B startups could produce as much revenue as they want from their startup clients. At some point consumers have to pick up the final bills. If the asymmetry is too large and the financial system suddenly decides that it's not gonna fund this anymore, we'll see domino effect.


In his highly applauded book « Hard Things about Hard Things », Ben Horowitz recounted his personal experiences at LoudCloud, the 2nd startup by him and Mark Andreessen, quite vividly:

The dot-com crash had spooked investors, so raising money wasn’t going to be easy, especially since most of our customers were dot-com startups.

The sales forecast for the quarter came in at $ 100 million, and things seemed like they might be okay. I felt confident that our sales forecasts would hold up given that previous forecasts had underestimated actual performance. And perhaps, I speculated, we could seamlessly migrate our customer base away from dot-com bombs to more stable, traditional customers such as Nike, our largest customer at the time. And then the wheels came off. We finished the third quarter of 2000 with $ 37 million in bookings— not the $ 100 million that we had forecast. The dot-com implosion turned out to be far more catastrophic than we had predicted.
— Ben Horowitz. The Hard Thing About Hard Things: Building a Business When There Are No Easy Answers (p. 19). HarperCollins

In fact, this was the main theme for the dot-com bubble in the late 90's. Most of the "dreams" being sold were far down the road. The market penetration of internet access was ridiculous. Most human beings did not own computers. Consumers were far behind the curve in picking up the final bills while startups were burning the VC money (and the public money after they went IPO) to build hardware and software infrastructures that would someday be worth something.

When the bubble bursted before that day came, everyone got burned, especially the B2B ones. Among the survivors, notably there were Amazon, eBay, PayPal and Google, all of which are either directly or closely consumer-related.

Today we live in a much different world. Smartphone has brought market penetration of personal computer to previously unseen level. 3G/4G, ADSL, cable modem and fibre reach out to a much larger percentage of the consumers than in the late 90's. We're therefore in a shorter cycle before the consumers pick up the final bills. And many consumer startups have been growing like crazy.

Still, there are many startups today that are what I call, value-chain startups, which provide a certain layer of the over all value chain. Many of them focus on productivity. Most of them have the potential to disrupt a dormant part of the old business value chain in their respective industries. And some are actually actively selling products/services to other startups.

This creates a huge correlation among many startups, notably in the SaaS sector. Almost all of the SaaS are tackling SME businesses initially. And due to the affinity in languages, their usual suspect of clients are usually vast amount of other startups.

This part of this bubble, if existent at all, will surely suffer the fate of many a B2B startups in the previous dot-com bubble.

As an entrepreneur you have to carefully analyze to see if your added value eventually get passed on to the consumers by your startup clients, or by the clients of your clients. If this doesn't seem to be happening, you have to be very careful about your cash flow management and really seriously review your product value proposition.

It could be the same for VCs. Many have adopted the verticals in doing investments. Those who pursue SaaS startups tend to have many B2B SaaS startups in their portfolio – some of them might even be using each other's service, essentially exchanging the VC money between them. Again, if there's no sign of consumers picking up the final bill down the road, the whole house of cards could come crumble down very fast when the funding momentum reverses.

Does elevator pitch work in hardware startups?

Fitbit numbers revisited and an old trick that works well