Piketty, required rate of return (r) and venture startups

The Traditional Chinese version of Prof. Piketty's homerun book « Capital in the Twenty-First Century » recently went on sales in Taiwan and Prof. Piketty made a quick trip to my homeland for a speech and a signing session. While his book created a heated debate across the whole world, I read those debates and felt that most were unwarranteed.

IMHO, the main difference between Pro- and Anti-Piketty thinkers often boils down to this: is required rate of return (r), one of the two pillars of the modern capitalism, the best metric to allocate capital in the new century?

 

Required Rate of Return

A recap about required rate of return (r). The basic assumption of Modern Portfolio Theory (MPT), and hence the modern capitalism, is that capital is emotionless and only seeks places where the risks involved are justified by the expected returns. In other world, less risky assets would have lower required rates of return (r) and vice versa.

MPT has come to dominate the modern financial thinking. It's logical, it's beautiful and if the historical volatility of an asset's prices (σ) is taken as a proxy for risks, it's also readily applicable. The problem, however, is that in the world of finance, persuasive and convenient tools tend to become self-fulfilling prophecies

For example, some researches have shown that since the birth of the famously beautiful Black-Scholes model, the prices of options and other derivatives converged toward what's predicted by the model and its follow-up versions over time. Now, is this a discovery of an underlying truth of the universe, or is it merely a result of every trader using the same tools for pricing, therefore the prices being forced to converge to the models? And as we know, when the Black Swan hit in 2008, of course everything fell apart.

The same thing goes with MPT. It's probably not a coincidence that since its birth in the 50's and its subsequent ascendance to prominence, the corporate world started to talk more and more about maximizing shareholder's values and ignoring everything else. In fact, shareholder is but one part of the capital. To the extent that the cost of debt reflects the risks of the free cash flow of a going concern and that it amplifies the volatility of the equity return, the board of director and the management team of a firm shall maximize the total return to the firm and everything else – environment, society and government issues – shall take care of themselves.

Modern capital then could be visualized as this emotionless machine that only sees risks and returns. It constantly surveys the investment landscape across the planet and will only allocate itself to an asset when it feels that the perceived risk is justified by the expected (possible) return. If it perceives Apple's shares as risky as Argentina's government bonds, it will give both the same yield and flow into these two distinctly different assets.  The machine couldn't care less about anything else.

There are obviously many problems in this view, even if you have managed to follow the beautiful logic behind it so far.

The first problem is a very practical one: how do you measure risks? The financial crisis of 2008 taught the whole financial world that it's very very difficult to measure risks, especially when an asset has a relatively short history and a huge tail risk. The quants will probably argue that now that we have the first set of data, it's a good start. The problem is: why shall the world suffer multiple crises so that the quants could have data to play with? And in any case the financiers will invent new derivatives that will have yet again different risk profiles that might need to be validated through yet another couple of new crises.

Thanks but no thanks.

Another problem is less obvious to most finance professionals but is probably what Prof. Piketty was trying to point out in contrasting returns on capital (r) with economic growths (g).

Note that for both r and g, there are both backward-looking (realized) and forward-looking (expected) definitions. This is where many less well-versed journalists got wrong in their comments on the book. The biggest mistake would be arguing that realized capital returns (r) cannot be larger than realized GDP growths (g) forever and that Prof. Piketty is merely stating the obvious.

In fact, in the book this kind of backward-looking comparisons are but a way to give clues about the wealth and income distribution (often with a delay in time). To see how it goes in the future, one has to wear the forward-looking glasses: what are the required rates of returns (r) and projected GDP growths (g)?

According to IMF World Economic Outlook, the global GDP growth is projected to be around 3.7% in 2014 (and 3.9% in 2015). We also know that the current global population growth is about 1.1%. The difference between GDP growth and the population growth stems, theoretically and hopefully, from the productivity growth, which is the only growth that can really elevate the living standard of human beings per capita.

Note that many researches have shown that living standards do not correspond 100% to happiness, and philsophers will even tell you that the pursuit of happiness is also in vain. But we'd be straying here.

Now in all likelihood, we shouldn't expect that productivity grow 2.6% (= 3.7% – 1.1%) every year. Most advanced countries see their productivity growths stagnate since the late 70's. And while new technologies such as the Internet and smartphones bring evolutions to how we do our jobs and manage our lives, many of those improvements are non-measurable or non-applicable to the so-called productivity.

Let's assume that we simply will be seeing an annual GDP growth of 3.7% going on forever – Prof. Piketty is more pessimistic about this in his book, especially over the long run. If we view the global economy as one big company, the annual global output (GDP) is then its free cash flow. And this free cash flow is growing by 3.7% per year. Now assuming we put a risk on this stream of free cash flow, what should it be? Well, if we assume that it will grow 3.7% forever, the risk will actually ZERO as we know very well all the cash flows going forward. This company is then risk-free and shall yield a return that's no more than the risk-free rate.

Is this assumption of 3.7% growth forever reasonable? Barring any major breakthrough such as cold fusion – FINALLY!! MARTY!! – or self-growing meat – Yuk!! – we shouldn't expect the global productivity to grow much faster than what they've gone through during the Industrial Revolution. And population growth is even more predictable as it's intrinsically tied to some self-reining human mechanisms such as society structures and health cares. Hence I would say overall assuming a 3.7% growth forever isn't too shaky, at least when compared to the growth projections in probably all the financial assets not named US Treasury Notes, Bills or Bonds.

 

Planet earth as a going concern and the distribution of its free cash flow

Now if we take this stream of predictable free cash flows (global outputs) that has a CAGR of 3.7% and we start to distribute it among its inputs, namely capitals and labors, we start to see clearly how required rates of return (r) move more free-cash flows (global outputs) to capitals instead of labors, which leads to wealth and income inequality.

First we note that there are actually very few asset classes currently available that have a required rate of return that's lower than the 3.7% growth rate.

The US Treasury yields currently start at 0.02% for 3 months and 3.05% for 30 years. The next risky asset in line would be corporate bonds. If we take 10-year US Treasury Bonds as basis, the 10-year A+ Corporate Bond would probably be priced to yield 2.33% + 0.72% = 3.05%. And if we take the long-term view, then a 30-year A+ Corporate Bond would yield 3.05% + 0.95% = 4%, just above the 3.7% GDP growth projections.

Now these bond assets are as safe – less risky – as you can find on this planet, but they're already asking for a return that is at the same level of the GDP growth. Once we move to required rates of return in other assets such as equities, the numbers jump up easily to 10%, 20% or even 30%.

You might argue that expected returns do not equal to realized returns. This is very true, but required rates of return are what drive the allocations of capital, not realized returns.

If an asset is perceived to have a risk profile that shall compensate the capital with a 30% expected return, the capital won't go into this asset if it expects the asset to yield lower returns in the future. Even if the asset, to the capital's surprise, yields 50% in the next year, the capital will remain doubtful and wait and see. Only when the asset builds a track record of 10 or 20 years, or more precisely, a large enough sample of regular transaction prices and returns, will the capital start to see it serioursly through the risk-return profiling glasses.

And if the capital will only go into assets that yield what they require, the people operate the assets, whether it's managers running companies or governments printing bonds, will be forced to allocate, on a global and collective level, proper annual returns to the capital to keep them in the assets.

And like we demonstrated above, most of the assets required higher returns than the 3.7% global GDP growth. This is why intuitively, whatever the realized rates of return for capital are, the wealth and income inequality are definitely gonna continue to rise given how modern capitalism functions.

 

What does this have to do with venture startups?

Or maybe we shall rephrase the question: what does it take for new and better technologies to be invented, which would eventually elevate our productivities and benefit all human beings on a per capita basis?

A short and optimistic answer as we know today is the Startups-VC ecosystem pioneered by Silicon Valley, which as an assset class today carries in general a 30% required rate of return (while yielding in reality much lower on a global basis). 

The question is: what's the risk involved? Is 30% required rate of return just right, too high or too low?

Note that if measuring risks in public stocks is already difficult enough as industry structures and economics policies shift, it's much much more difficult to measure risks in venture investments since all breakthrough technologies, by definition, are unprecedented and therefore do not have any past history that could allow the financiers to crunch numbers.

Just a few examples: before Paypal there was no real on-line payment companies and therefore no way to gauge the risk-return profile of investing in such startups; before Google there was no search-based advertisments and therefore no way to gauge the risk-return profile of investing in such startups. The same argument applies to of course Youtube, LinkedIn, Salesforce.com and Facebook, so on and so on.

How about taken as a whole? If we pool all VC returns into an asset class, can we find a risk-return profile that can allow the bean-counting financiers to allocate their capital properly?

Well, yes and no. The yes part is that's exactly what the institutional investors have bee doing (or hoping to do) for the past two decades, by looking at VCs as an asset class that can extend their Efficient Frontier. However, it has been found that the distribution of VC returns is so skewed and so irregular that it's very difficult to assign a required rate of return that justifies the risks taken. In fact there are many arguments that except a couple of top funds, which in general get better entrepreneurs coming to them due to the winner effect, most other tens of thousands of VC firms on this planet simply don't yield returns that are worthy of their risks.

More and more academics are therefore calling into questions VCs a valid asset class. The impact of this questioning, if validated, is at best that VCs shall not stand alone as an asset class and shall probably be lumped into the entirely different PE bunch – or worse, no capital shall ever go into VCs except those top ones, which by the way are usually locked to new LPs and therefore inaccessible.

Now what does this imply if it comes true?

The real question is: if we leave the world to purely risk-return optimizing, emotionless capital that operates on MPT, will we still have innovations?

To answer this, one can examine all the successful startups that have forever changed our lives. One will find that they were ALL sponsored by VCs during their pre-public phases. There's no exception, zero. At the same time, if we examine, or try to examine, the 99 times or 999 times more VC-sponsored startups that eventually died, we would instinctly conclude that investing in VCs (and therefore startups) doesn't make sense.

However, without enough capital venturing into the high-mortality-rate world of startups, will there still be real ground-breaking innovations? 

Or put it this way, without crazy people like Elon Musk, will GM, Ford and Toyota have come up with a real electric car like the ones sold by Tesla today? Maybe, but more likely not. As Peter Thiel highlighted in his Zero to One, most large corporates at the maturing stage fall into internal rent-seeking traps. Since it's hard to fathom the potential long-term upside of a risky small project, most managers would only fight to do projects that have big budgets and yield moderately better end products. Worse still, many more would not even bother working on moderately better products, but would spend their time engaging in politics to maximize their own rental income, whether it's windowed offices, corporate jets or golden parachutes.

Humanity is such that it's very clear we need daring people to venture outside the corporate world to bring true innovations, and we, probably and hopefully, need venture capital money to support such endeavours.

If this is acceptable to you, my dear readers, then the required rate of return on venture startups shouldn't really be a concern for you. And maybe it's time to re-examine the properness of leaving the capital allocations to the risk-return maximizing machine called MPT. For this, reading Prof. Piketty's book, in whatever language, would be a good start.

[Bookmark] Uber revenue numbers

Sam Altman and the CS183B course at Stanford – free resources