Angels, entrepreneurs, and diversification: PART 3

Dimensions of risk

Hedging is all about understanding and responding to risk. Thus, it shouldn’t be surprising that a hedged economist isn’t content with the standard one-dimensional definition of risk. Financial economics usually defines risk as price volatility and builds from there. It is a powerful approach, and everyone can benefit from exploring where that assumption leads. PART 2 of the discussion of angels, entrepreneurs, and diversification was built around that approach. Therefore, some caveats are in order before proceeding to a more detailed look at the risk associated with non-publicly traded assets.
As powerful as the approach is, it has limitations. The approach assumes the price series contains all information that is relevant to risk. Let’s recognize two things the approach ignores and then move on. First, the approach ignores the fact that every price series also embodies market participants’ willingness to take risk (i.e., aggregate risk appetite or aggregate risk tolerances). Second, it ignores potential structural sources of risk (the interaction between series). Regarding the second point, it also ignores real or perceived CHANGES in potential structural risk. One can develop an entire theory of panics and bubbles around letting these two overlooked factors interact.
A bigger limitation of the approach is built in. The approach assumes that volatility is knowable and can be measured. When one delves into financial economics, one notices that assuming volatility is knowable creates a situation where further assumptions start flying fast and furious. A good deal of what is interesting in financial economics involves discussion, critique, and the mathematics surrounding volatility. The issue can also introduce all sorts of interesting issues like basic philosophy of science. Most importantly, it opens the discussion to a broader set of risk analysts using analytical techniques other than just financial economics.
Non-publicly traded assets, by definition, don’t have a robust price series. Further, the volatility in what price changes there are is plagued by discontinuities (i.e., big jumps). Consequently, conventional financial economics and investment theory are very limited with respect to what they can reveal. Yet, despite the limitations, conventional financial economics provides a common sense argument for including some non-publicly traded assets.
Less this posting be criticized for overlooking the obvious, let’s acknowledge a point. One can envision applying a variety of approaches to a portfolio of non-publicly traded assets. Any model designed for the analysis of non-continuous situations is a candidate. Survival, hazard, poison, event, and binary models are all types of models that could be used. These are the same types of models used to calculate expected returns on consumer debts, like mortgages. There is a very good case for the argument that they are better at selection than forecasting. However, if that’s your style, have at it. The contention here is that there is a more productive approach.
With that said, back to the issue. The phrase “some non-publicly traded assets” leaves a pretty big range. It almost raises more questions than it answers. So, first let’s look at some risks that are unique, or at least more acute, with non-publicly traded assets.
Let’s start by reviewing some risks that have already been discussed.
First, lack of liquidity is inherent in the fact that they aren’t traded. But, overlooked is the fact that their lack of liquidity means tax planning options are thereby constrained.
Second, accessibility has been mentioned previously. Most people will have access to far too few opportunities that they can consider. That creates tremendous potential for concentration risk.
Third, the lack of pricing data and nature of the volatility of non-publicly traded assets are only the tip of the iceberg when it comes to the scarcity of information. Startups have no history. Further, there is a tremendous potential for information disparities where one party in the transaction has an information advantage. However, the information disparity can exist with any asset.
Fourth, while implied by the lack of liquidity, the risks associated with exit strategy are sufficient to deserve a separate mention. But, people often overlook the associated dilution risk.
While the listing above is far from exhaustive, it is sufficient to introduce the rationale behind another type of risk. Investing in non-publicly traded assets, especially startups, requires a different set of skills from investing in public assets (also different from the skills required to earn a high income). Thus, the argument in PART 2 that financial criteria are irrelevant to the issue of accreditation.
Even without going into the skills required, the point indirectly implies what is probably the greatest risk. Behavioral economists are fond of pointing out that over confidence among investors leads to all sorts of investment mistakes. It is interesting but almost useless information since lack of confidence is another reason for investment mistakes. However, in the case of investing in startups there is reason to believe investors who actual invest in startups may be more prone to overconfidence than investors in general.
The evidence is indirect. But, right off the top, the option of not investing in startups is probably the default of many who lack confidence; it is just so much easier than investing at all. Therefore, there is probably a self selection process that favors the confident.
Second, the characteristics of those who start businesses as the way to access this asset class definitely supports the contention that they are either very confident or have very high risk tolerances. Specifically, as a group, people who start business display an above average willingness to take on concentration risk and leverage risk.
So, while open to debate, pointing out the importance of not being overconfident probably does no harm. Especially in the context of a posting that is a part of an argument that more investors should give startups more consideration and a role in their asset allocation.
Enumerating the risks can only go so far in helping one decide on an asset allocation. Ultimately, it’s a very personal decision based on risk tolerance and the total portfolio. However, generally one can find some empirical evidence to inform the decision. That isn’t easy with respect to startup investing.
The first approach we’ll call the Bogle approach after John Bogle the founder of the Vanguard Group. Bogle advocates cap weight index funds. Vanguard started the first S&P 500 index fund which was quite ground breaking at the time. Currently, Bogle has suggested total market index funds are a better alternative. These are cap weighted funds. So, weights are fluctuating with capitalization. The logic of the funds would be to just let the weighting for non-public assets fluctuate with the portion of the economies total capitalization represented by non-publicly traded businesses.
However, the entire logic of index funds falls apart once one considers non-public assets. First, a major argument for indexing, the low cost, doesn’t translate. Second, indexing involves a default objective. It implies the objective of keeping up with the market. Third, it contradicts another element of the Bogle approach, namely rebalancing. At a minimum it assumes rebalancing within equities is irrelevant.
The Bogle approach isn’t a lot of help. Yet, it does turn up some important leads. The contradiction between rebalancing and indexing across equities highlights the importance of checking for systematic differences between the behaviors of asset classes. PART 2 went though the logical reason for difference in the behavior of traded and non-public assets. Quantitative data is by definition scarce; we’re talking non-public assets. However, there are some potentially similar public asset classes.
The Motley Fools (http://www.fool.com) recently discussed some research related to the role of small cap stock in a portfolio (see: http://www.fool.com/retirement/general/2010/04/13/scary-stocks-worth-buying-anyway.aspx ). It was such serendipity to come across this while writing this posting that a lengthy quote seems appropriate.
“If there's one guy who should know, it's Ibbotson. He's made a career out of compiling, analyzing, and publishing historical data on the returns of different asset classes. In an interview he did with Barron's back in 2006, Ibbotson maintained that adding small-caps to a portfolio would actually reduce risk -- volatility, in other words -- as long as the total proportion of small-caps to other assets was kept under a reasonable limit, 20% to 25%.
Still skeptical? So was Fool Robert Brokamp, the advisor for the Fool's Rule Your Retirement service. He broke out some sophisticated analytical tools and tested the theory, using the Vanguard 500 Index Fund (VFINX) as a proxy for the blue-chips and a small-cap fund as a proxy for, well, the small caps. And sure enough, a portfolio that was 80% blue chips and 20% small-caps would have had higher returns and less volatility over the past 10 years than just the blue chips alone.”
Some research exists that break out micro caps from small cap. (Please accept my apology I can’t find the reference). It also concludes around 20-25% small AND 5% micro cap. So, publicly traded assets support the belief that small company open new risk frontiers. Now, if non-public assets are even less correlated with large caps than small caps publicly traded companies (as argued in PART 2), the bang for the buck should be even greater. That would imply that the same risk frontier (i.e., risk/return tradeoff) can be achieved with a smaller allocation to non-public assets. Alternatively, the same allocation can achieve a much better risk/return profile. Unfortunately, that’s about as far as that approach takes us.
A different approach is to look at survey results. There is some relevant data. However, the issue is who should be included in the survey. There are quite a few instance where the research design used was to define a target for the survey, and then, to let the reader decide whether the target population is relevant. Most target an audience that is composed of the “rich.” Rich is various defined; then the researcher surveys them about investment behavior. The Federal Reserve’s Survey of Consumer Finance is also relevant. But, all of these sources have what is known as survival bias. If a bunch of non-rich followed the same investment approach, we would never know it just from the research.
Despite the limitations, the survey research is informative. Most of the surveys, in fact every one I’ve seen, show a much larger portion of assets in non-public companies than is characteristic of investors in general. That fact, seems to support the conclusion that more than survival bias is at work.
Figures from 15% to 50% show up depending on what range of asset classes are considered (e.g., owner occupied first residences and non-rental second home, life insurance, pension values, and other assets classes aren’t treated uniformly). Results also vary depending on the level of detail of breakdowns within asset classes (e.g., are startups broken out from other private companies, are other “alternative assets” included, etc). Finally, how the target population is defined is obviously important. But, even with these limitation, the surveys strongly suggest most investors’ portfolios are sub-optimal.