Diversification has been called the only free lunch in investing. But how many stocks do you need to be adequately “diversified”? Some say forty; some say as few as ten, provided they’re in different industries.
But wait … it’s a trick question. No matter how many stocks you own, you’re not diversified if all you own is stocks.
Diversification at the next level up … across asset classes – cash, stocks, bonds, commodities, realty – is where it counts most. How investors allocate among asset classes accounts for the majority of their investment performance. But is there an “optimum” of diversification? Assuming the objective is to reduce risk, is there a point where it reaches zero?
Having heard the mantra that there is no such thing as investing without risk, your first instinct might be to answer no. And to the extent that no one asset is risk-fee, you would be right. But at least theoretically, there is such a thing as a risk-free combination of assets; a risk-free portfolio. And it can be at least approached in practice.
It’s usually referred to as the “global market portfolio”. In its theoretical ideal, it consists of a proportional share of all the world’s assets. It is risk free in the sense that it reduces the risk of loss due to your selection of investments to zero.
Now there could be a world war or natural disaster which destroys a sizable share of these assets. But in such a case, you’re no worse off than your fellow man. Your proportional share of purchasing power would continue to be the same as before. And at the opposite extreme, you’re assured to equally benefit from the discovery of a cure for cancer or abundant source of cheap energy. In any case, your risk had nothing to do with your selection of assets, the thing you’re concerned with when constructing an investment portfolio.
The more practical problem is that you cannot assemble a true global market portfolio. Much of the world’s assets are privately owned … they don’t trade in public markets. You’ve heard of private equity and debt, but real estate is a more common example. You can’t own a proportional share of your neighbor’s privately owned home.
But you can own your own, similar, home. And some real estate does trade in public markets. So even if you don’t, a facsimile is available. And much of the world’s assets, especially stock and bonds, are publicly traded.
Another issue to contend with is that most authorities’ interpretations of the global market portfolio exclude currencies. Being as currencies are the ultimate public assets, this isn’t a practical constraint; it’s a private bias. Currencies have an notorious tendency to depreciate, so they are excluded as undesirable investments. But they do occasionally appreciate, and excluding them inherently increases risk. So even a common take on the GMP takes some risk in pursuit of performance.
This is an example of the common sense that investing isn’t driven by risk alone. And it certainly isn’t my intention here to suggest that attempting to improve on the global market portfolio is without merit. Investment goals are foremost an individual matter. But the GMP is the gold standard against which to measure any portfolio. Comparing it with the global market portfolio clarifies exactly what risks it contains. If you’re overweight an asset or asset class and underweight another, it’s an objectively definable risk.
Because a true global market portfolio is a theoretical ideal, you might be tempted to throw up your hands and write it off as irrelevant. But an ideal gas is also a theoretical ideal, yet closely approached by real gasses and a highly valuable tool in physics and engineering. It might also seem dauntingly complicated. But remarkably simple approximations exist. One such simple approximation is Harry Browne’s Permanent Portfolio. It’s not a stretch to say that any properly diversified portfolio is in some sense an approximation of the GMP.
This approach to risk is fundamentally different than in most finance, where “risk” is defined in terms of volatility, and volatility in turn defined relative to currency (cash). It’s also fundamentally superior. The trouble with the volatility-relative-to-currency model is that it assumes currency itself is risk free. As anyone who attempts to invest solely in cash can tell you, it’s nothing of the sort. The only sense in which it might be said to be low risk is that you have a low risk of retirement. The risk of losing purchasing power is near certain.
This led to the popular but deeply flawed concept of tradeoff between “risk” and “return”. You could increase your return (relative to cash) but only by assuming more risk (relative to cash). At “risk” of revealing my disdain for such popular garbage (including the tendency to include only stocks and bonds), the global market portfolio approach is in comparison a free lunch, because even a zero risk portfolio carries not-too-shabby returns … you can at least maintain purchasing power. Whereas in following the conventional conception, zero risk leads to negative returns.
Yet most of the time, a global market based portfolio will also have relatively low (cash based) volatility. The exceptions would be in extremes of inflation and deflation, where cash itself is gyrating wildly. And in such instances the lack of volatility of cash is merely an illusion based on using the currency itself as a unit of account, an illusion that soon passes.
So looked at in terms of correlations, similar conclusions follow. The less similar the assets, the greater the diversifying power. Financial assets dependent on discounted future cash flows, like stocks and bonds, have more in common than often portrayed and correlations are inconsistent, depending on the drivers dominating at any given time. Changes in the time value of money affect them similarly but changes economic vigor tend to affect them oppositely. Physical assets like copper and gold are fundamentally similar but can also diverge markedly; copper trades strikingly inversely to Treasuries but gold more positively.
Either way, the fundamentals of diversification are similar. Include at least some of each major asset class – cash, bonds, stocks, commodities and realty. The composition of the global market portfolio itself changes, but requires little rebalancing; as relative values evolve, your allocation tends to follow. They also vary according to source; as noted, most exclude currencies. And as also suggested, it’s more important to know what the global market portfolio is than to try to emulate it … it’s main value is in knowing what risks you’re taking in pursuit of your investment goals.
Financology’s home grown approximation of the Global Market Portfolio, tailored to American investors by virtue of the use of US Treasuries for the bond allocation and of ETFs available in the US, is the Capital Portfolio in the Model Portfolios section. The Income Portfolios can also be thought of as variations tuned to their particular objectives.
Some other sources:
What would be the closest approximation to the realty component – Owning a home debt free, Owning a home with a 30-yr mortgage (or any mortgage), or owning rental property ?
owning a home outright is tied to local market conditions, so is at best a distorted model of global real estate. owning a home with a mortgage = owning a home, being short a bond, and owning a call on that bond [via the possibility of refi]. rental property will again be subject to distortion by local factors.
Thanks rtchoke. Real estate as a class is especially difficult to approximate. As jk points out, any would be a crude representative at best. The class includes everything from the Imperial Palace in Japan, the Taj Mahal, and Chinese high rises all the way down to Detroit slums. Forest and farm land. You can’t own a share of all these things.
This just highlights the inattainablity of the GMP ideal in practice, however useful for portfolio analysis and design.
On the other hand, the securities markets, particularly stocks and bonds, being mostly publicly traded, can be tracked closely through ETFs like VT and BND. VT is global, and while it’s possible to do global bonds (eg BNDX), the practical value for national investors is debatable due to their currency-linked nature. Most investors want their bond allocations to reduce volatility in their national currency terms. This is why the Financology Model Portfolios use only US bonds. International investors generally have different menus of ETFs to choose from anyway.
There are ETF portfolios that aim to approximate the broader GMP. I believe Portfolio Charts and The Capital Speculator linked in the post have them. (For the latter see eg “Major Asset Classes…” where it’s referred to as “GMI”). Financology’s do too, except target realty just to the extent normally represented in stock indexes (~3%-4%), any more being left to the reader, who might already have substantial realty.
It depends on what your objective is, but in general if you own a home, counting it towards your real estate allocation is sensible. However imperfectly representative it may be, it’s an actual part of the global real estate market, and of your financial picture. A renter might add real estate through ETFs like VNQ, VNQI, REET or even individual realty stocks or rental properties. May as well make it something you like. Then just accept that it’s close enough.
If the home is mortgaged, count the debt as a negative position in bonds (you’d need more bonds to offset it). Or better yet, pay off the debt before building a big securities portfolio. The main exception would be if it happens to be one of those now-extinct ~3% mortgages…