The War on Cost of Capital
Section 1: What does “Cost of Capital” Mean and Why is it Important?
Section 2: Why is Cost of Capital is Coming Down in Credit Markets
Section 3: How Data Gathering Companies and Supply/Demand Will Drive Down Cost of Capital in Other Asset Classes
My question to you: in a world with no risk, and efficient supply and demand, how will we price our cost of capital?
Section 1: What does “Cost of Capital” Mean and Why is it Important?
Cost of Capital is defined by how much money I need to make in order for me to justify an investment.
For example, if I put my money in government bonds, that’s about as safe as it gets. So if you want me to invest in something other than a government bond, you have to offer me a higher return to justify that risk.
When you hear people say “interest rates are at zero” or crazy shit like that, they’re basically talking about the rate of return they could get by making the safest type of investment possible.
And Cost of Capital is insanely important. It’s what drives every investment decision. When a venture capitalist invests in your startup, they are thinking about their investor’s cost of capital. Your VC needs to earn a return on her investment of some amount that, after many of her other investments fail, is higher than her LP’s cost of capital.
In our credit fund, we have hurdles (meaning I don’t make any money unless we hit this return) ranging from 12% to 15%. So our cost of capital ranges from 12% + the fees I charge to 15% + the fees I charge.
That’s the minimum return I need to get. And the way we got to that number is by comparing the perceived risk of our investments and compared that risk to other potential investments one of our investors could make.
Just like with any product, we had to figure out our pricing strategy and that is where we ended up.
Section 2: Why is Cost of Capital Going Away?
I keep using the word “risk” because that’s the derivative of “cost of capital.” My cost makes up for my risk. But what if risk starts going away?
We’ve started building machines that are really good at improving the certainty of the decisions we make (in case you forgot, risk and certainty are inversely correlated…) . IBM’s Deep Blue helped increase our certainty around what decision we should make during a chess match. And IBM’s Watson computer is helping doctors increase their accuracy in medical diagnoses.
Waze helps us understand which traffic route we should take based on traffic. This mitigates our risk of being late.
Drones are gathering data about landscapes and situations at war, mitigating our risk that soldiers will die on the battle field.
And perhaps less noble, but certainly relevant, alternative lending platforms and emerging tech-enabled insurance companies are using data to better underwrite loans and predict whether or not they will get paid back.
The better these companies get at underwriting risk, the safer they will be perceived to be. And the safer they are perceived to be, the lower return investors on those platforms will demand. And each subsequent year thereafter, as technology gets better and better, and the test of time increases our confidence intervals, cost of capital on these platforms will go down.
Imagine I, as the investor have money on Prosper and my returns have been pretty good. I’ve probably been getting back a low teens return, which compared to the risk (people have been doing consumer lending for a long time and Prosper seems like a stable enough originator/servicer) is a good deal. But what happens if too many other people think: “wow, that’s a good deal!” too?! Other people will invest and then there will be too much capital, and then Prosper won’t have to give me as high a return anymore. The supply of capital will become so great that on a risk adjusted basis the investment no longer becomes worth it.
Now, as the inspired, motivated, and entrepreneurial investor I’ll start looking for new places to put my money. And I’ll likely find a new, weird place to put it. Like… farm lending, or receivable factoring, or title lending.
And here’s where things get interesting. While most of the innovation around OnDeck, LendingClub and Prosper were around customer acquisition (acquiring a customer online for less money and giving them a better, faster experience), these newer companies are using previously unobservable data to materially improve their underwriting processes. They are mitigating risk in places where that risk was often too difficult to assess, so previously required a very high and in many cases, abusive, return on investment. Payday lending and title lending are two good examples of abusive products.
So these companies are making types of investments that used to be risky, less risky, and are offering a high rate of return not because the expected default rates will be high, but because of two driving forces:
(1) They are new (and new introduces the risk of what we don’t know we don’t know). We price “new” into the risk formula.
(2) And Because of Supply and Demand: The investors who are able to invest in new stuff only can if they take the time to do diligence on these new investment opportunities. But taking all that time costs money (you gotta pay your team to do the diligence), and it’s only worth taking the time if the investor can make a big investment (like, $50M or more). There are very few people willing to make “new” and “small” investments. Very few.
So the return on investment (or cost of capital) comes at the hands of the “risk of the unknown” and a limited supply of capital willing to chase those opportunities.
But that’s temporary, as the platforms become older and bigger, more capital will come in, and the returns will once again be priced more on perceived risk than on supply and demand of capital.
So we’ve established that cost of capital is driven by perceived risk, and by the supply and demand of capital. And we’ve also established that computers are good at gathering and processing data that increases certainty/mitigates risk. So it would seem that over time cost of capital will be driven more by supply and demand than by risk. But I think computers will be just as good at assessing supply and demand in real time and adjusting in lieu of it. Orchard and Even are positioned well to help.
So risk will go down, and supply and demand will become more efficient, thus driving down the expectations people will have for needing to get paid back.
In theory, if we knew everything and were 100% certain what would happen with our capital, and supply and demand became efficient, how would we price capital? I don’t know.
Section 3: Cost of Capital Will Go Down in Other Asset Classes Too and VC Firms Will Be Able to Get Away With “Mediocre Returns”
My cost of capital out of our venture capital funds are higher than out of our credit funds. Why? Yeah, you get it now, because of risk.
The other reason is a limited supply of capital. The only people who can invest in a venture capital fund are those who can put up $250k at a time and not touch the money for another 10 years. And even more than that, they have to be okay with the idea that all that money may disappear. It’s capital at its greatest risk. So the risk reduces the supply, which perpetuates itself.
And why are the investments risky? It’s because we have so little data about how the investments will do. Why do we have so little data?
(1) Because the companies are very new, and have done less, so there is less action of the company to assess.
(2) We have to hold on to our equity for ~7 years — so while we can potentially predict what will happen in year 1, it’ll get harder in year 2, and almost impossible in year 7.
(3) The investments are so small that we can’t hire a team of analysts to assess every single deal — it’s often just one person doing diligence and often only for a couple of weeks.
These three factors show that there are limits to gathering data and only a limited number of servers (human brains) to process that data.
But if computers could help us gather data better (which they have — we have Linkedin that lets us diligence a founder, we have CB Insights and Mattermark which give us industry trends, social media comments and reviews of the product etc.) we could be better at making investments.
And a higher % of our portfolio companies would succeed, and our perceived risk would go down, and people would expect a lower return.
To take a less extreme example, let’s think about traditional private equity.
Traditional private equity used to require a low 20’s IRR. And now, if you can earn low double digits life is pretty good.
(1) Returns in public markets have gone down as public markets have become more efficient, so “comparing one return to another” has helped private equity.
(2) Private equity is being seen as less risky than it used to be viewed because it’s existed longer, and established firms can show a track record of consistent returns.
(3) That change in perceived risk has allowed more potential investors to come into the market with bigger dollars, increasing the supply of capital.
I think over time as it gets easier and easier to find liquidity in traditionally illiquid market via secondaries, or via a blur between the lines of public and private companies, more and more capital will come into the market.
But machines will also continue helping us increase the certainty of our investments (just like firms like Two Sigma and Renaissance Capital have done so in public markets).
And so that combination of greater supply and less risk via an improved underwriting process via data from companies like Mattermark and greater supply from innovate products such as “liquid alternatives” will lower people’s expectations of what they need to earn.
So my question remains: in a world with no risk, and efficient supply and demand, how will we price our cost of capital?