Thursday, 31 January 2008

Why Not All Equities? Why Hold Bonds at All in a Portfolio?

Fact: Over a suitably long period, like twenty years, equities have always outperformed bonds/fixed income investments. This is well documented in such books as the Equity Risk Premium by Bradford Cornell, which I will be reviewing soon (e.g. see this Google book extract from page 71). If you are a long term investor such as a person in your twenties starting to save for retirement, why not just forget about bonds, grit your teeth through market downturns and go for a 100% equity portfolio?

Here are some reasons:
  1. Danger of selling out at the wrong time - if you went through the 40% decline in equities from 2001 to 2003 and did not sell, then you will be ok but believe me it ain't easy. Could you stick out a ten year period like the 1970s when equities went nowhere? As the famous economist John Maynard Keynes once said, "The market can stay irrational longer than you can stay solvent." (reference in the Wikipedia entry here; go look, there are so many other juicy quotes). This quote raises the other way of bailing out too early - you cannot absolutely know what your holding period will be, despite your initial intentions. Your personal circumstances may cause this to happen; suppose you decide you want to use the money to buy a house or to give it away to an important cause. Suppose you die and your family needs to use the money for living expenses. It would be better to avoid the most severe dips if you can without sacrificing returns, which is exactly what bonds can do in a portfolio with equities, as I explain below.
  2. A portfolio with bonds and equities will have higher performance and lower volatility than an equity-only portfolio! Yes, Michael, you can have your cake and eat it to. This surprising counter-intuitive result I have previously written about last May in Portfolio Magic ... 3+1=5. The beneficial effect is not confined to bonds - international equities, real estate and commodities have also been shown to do the magic. That's why my portfolio, the structure of which is shown at the bottom of this blog, is built as it is. The effect comes about through the combination of assets with positive returns whose returns are not correlated (i.e. don't move in sync, the best situation being when they are negatively correlated so that one goes up when the other goes down). Two great books which explain and demonstrate this effect with real data are Roger Gibson's Asset Allocation (reviewed here; see Chapter 8 The Rewards of Multiple-Asset Class Investing) and Richard Ferri's All About Asset Allocation (reviewed here; see Ch4. Multi Asset Class Investing). Have a look at my portfolio: I find it interesting and reassuring that the only holding that has actually gone up is DJP the iPath DJ-AIP Commodity Index while all my equities are down. The bond holding AGG is down slightly on the chart because it doesn't include the cash interest payments I have received.
So, I'm suffering through this downturn with everyone else but I have confidence that I will be further ahead in the long run. As Keynes also said, "The long run is a misleading guide to current affairs." Fool that I am, I expect to be here for a long run yet.

2 comments:

Michael James said...

Your first point about possibly having to sell at a bad time is quite sensible. I deal with this by having adequate cash reserves and investing any money I might need in less than 3 years in bonds. Everything else goes into stocks. The higher returns I expect to get in the long run more than compensate for the possibility that I'm way off in guessing when I'll need the money.

Your second point falls into my area: math. In the example of your May 7th post, the main investment has expected return r1=0.137 with standard deviation s1=0.147. The investment that we are considering mixing in has expected return r2=0.103 with standard deviation s2=0.243. Let p be the coefficient of correlation between the two investments. Then the math says that mixing in some of investment 2 will improve expected compounded returns if

r2 > r1-s1*(s1-p*s2)

If the investments are perfectly negatively correlated (p=-1), and we plug in the numbers, we get

.103 > .081

which is true. If the two investments are uncorrelated (p=0), then the inequality is false, and we shouldn't mix in the second investment.

I have looked around for plausible numbers for returns and volatility of stocks and bonds, and in all cases, bonds don't pass the test. The all stock portfolio gives a better compounded return.

John said...

Michael

As a former math (Waterloo) student and current investor a different way to see this is the hidden power of rebalancing.

If you blindly keep some keep some percentage of you portfolio in XBB or TDB909 then during the crashes (ie 911, the last 2 months) you will automatically buy equities when the prices are low. As they recover you move you money back into XBB (just following the % allocation)

After a few cycles of this (and there will be many in your investing career) you get the hang of keeping some power dry for the next inevitable market crash.

As your portfolio grows you keep less in XBB and simply buy on the margin during the darkest days and sell to cover your loans in the recovery.

Bottom line - You need to model a random wave function)))

Wikinvest Wire

Economic Calendar


 Powered by Forex Pros - The Forex Trading Portal.