Tuesday, September 22, 2009

Asset Allocation

I appeared last Friday on a the PBS program WealthTrack, where the topic was asset allocation, in particular, as host Consuelo Mack put it, how to build an all weather portfolio. I was the skeptic of the group. I don’t think there is some magic asset allocation that protects you from the buffetings of financial storms without it also trimming your sails during fair weather. Here is an encapsulation of my views from the program.

Asset allocation and risk appetite
One of the participants, asset allocation guru David Darst of Morgan Stanley, proposed various portfolios to protect against a 100-year flood, 30 to 70-year flood, a 25-year flood, etc. Those portfolios boiled down to putting less in risky assets and more in bonds; the more severe the flood you anticipate, the less risk you take. Of course, that will do the trick. If by asset allocation you mean determining where to set your risk tolerance dial, we’re all on board.

Asset allocation is like clapping with one hand
But the discussion of risk tolerance highlights that we can only go so far with asset allocation if we only look at assets. What matters is assets versus liabilities, because the liabilities determine our risk tolerance and, related to that, our demand for liquidity. It is impossible to formulate an ideal asset allocation strategy without knowing the liability stream those assets are intended to meet. There is no one-size-fits-all for asset allocation. This reminds me of an FAJ article I did back in the 1980s with pension actuary Jeremy Gold entitled “In Search of the Liability Asset”.

Diversification works well, except when it really matters
We all know the argument from Finance 101: If you hold 16 uncorrelated assets, your risk will drop by a factor of four. Well good luck with that.

During a crisis, when diversification really matters, correlations aren't near zero (as if they ever are). All that people care about is risk and liquidity. All assets that are highly risky drop, all assets that are less liquid drop. No one cares about the subtlety of earnings streams. It is like high energy physics. When the heat gets turned up high enough, matter is just matter, the distinctions between the elements is blurred away.


This is not to say that one should not try to diversify, but rather that one should not think diversification will work magic. It is a given that a portfolio should not be limited to U.S. Treasuries and S&P 500 stocks, because while it should not be oversold, diversification does have some benefit. And, on the other side, unless someone is still living in the 1970’s, it borders on the intellectually dishonest to trumpet a diversified portfolio by using the S&P 500 as the bogey. A college kid can construct a portfolio that will beat the S&P 500 on a risk-adjusted basis, because there are so many more markets available now. A better approach is to look at a given asset allocation versus its nearby well-diversified neighbors, and try to understand why one is better than the other.

Commodities do not form an asset class
This sounds heretical given what we have seen oil and gold do recently, but a lot of the reason that has happened is precisely because people are treating them as an asset class when they are not.


Commodities are not assets. They are factors of production. They do not generate returns, they have no claim on production. They have supply that flows out at a nearly fixed rate short term, and they comprise very small markets compared to the financial markets. If pension funds all decided to put two percent of their capital into commodities, two things would happen. First, that two percent would be a rounding error in their returns, no matter how commodities behaved. Second, they would swamp the supply of the commodities for economic purposes – i.e. for their true role as factors of production. I agree with Michael Masters’ view that oil prices were pushed up by this sort of financial activity. I might quibble with one chart or another, I might not couch it in the loaded terms of speculation.
But the subsequent behavior of the market demonstrated that he was right and Goldman and others who took the opposing view were wrong.

Inflation-Linked Bonds
Which brings me to inflation-linked bonds. At the close of the program we all were asked for one investment recommendation. In one form or another we all focused on the same one: inflation-linked bonds. But I would not carve them out as a distinct asset class any more than I would commodities -- though unlike commodities, at least I think they are an asset. They are one of many assets that load on the inflation factor. If you have a long-term view, equities are also decent inflation hedge. After all, over time prices adjust, and so do earnings. And, as with commodities, the supply of inflation-linked bonds is low; there is a liquidity premium to pay.

I think what has elevated inflation-linked bonds from the category of “asset” to that of “asset class” is memories of the 1970s, a heyday for inflation-linked bonds. If you could have held them during the stagflation period, you would have looked golden; they would have given you a Sharpe Ratio of over 1.0 while many other assets was flat-lining. If I were building a simulation to beat the market on an historical basis, I would add in inflation linked bonds just for the pop they would give in that decade.

7 comments:

  1. It's difficult to get the professional advisors to recognize that the world is different and that the magic of diversification and asset allocation are significantly diminished.

    Today new tools are need to control risk. My favorite is the collar, but that's not the only choice.

    ReplyDelete
  2. Rick writes: "During a crisis, when diversification really matters, correlations aren't near zero (as if they ever are). All that people care about is risk and liquidity. All assets that are highly risky drop, all assets that are less liquid drop. No one cares about the subtlety of earnings streams. It is like high energy physics. When the heat gets turned up high enough, matter is just matter, the distinctions between the elements is blurred away."

    Since this is so well known, why do we keep having financial meltdowns in which diversification/correlation was used as the major avoidance of risk?

    In the subprime market, people rediscovered this by finding out that all housing markets can move together - downward.

    I don't get how this institutional knowledge keeps getting lost.

    ReplyDelete
  3. This comment has been removed by a blog administrator.

    ReplyDelete
  4. http://venturepopulist.com/2009/06/hybrid-portfolio-theory/

    There is an answer.

    ReplyDelete
  5. "Since this is so well known, why do we keep having financial meltdowns in which diversification/correlation was used as the major avoidance of risk?"

    I'll take a stab, Michael. Because the mathematical models are easy, and it allows advisers to sell investments that promise more return than they can likely deliver.

    True risk reduction almost always involves lowering the return objective, and since Americans have not been great savers, they try to achieve their goals through aiming for higher returns, and thus adding unintended risk.

    ReplyDelete
  6. Perhaps it is a mistake to say asset allocation does not work in a crisis. I think it is safer to say it does not work as it is currently taught in a crisis. As it is currently taught, you place many asset classes in the portfolio at various weightings to get a certain target std. deviation which is used as the measurement of risk/volatility....

    To get to a more truthful use of asset allocation one must look as a more basic definition; mixing non-correlated asset or low correlated assets in order to lower the risk. Now, isn't that the same thing you ask? Well, no. Asset allocators commonly use the long term numbers for asset classes...long term correlations, long term standard deviation...big, BIG mistake. First, we are in an ever changing world--and frankly, not every financial number reverts to a long term mean. Emerging countries become developed, developed empires crumble, trade across borders becomes freer and then cycles to constricted, stable currencies de-stablize... There are many a research paper at university level, professional level, whatever that observe the increasing correlation of a wide variety of asset classes. Logic would therefore cause one to conclude that the effectiveness of allocation models relying on old, aged numbers is going to yield unsatisfactory results. I showed to co-workers that the tenant of asset allocation held true in 2008 if properly applied. There were 2 asset classes non-correlated to equities (and by equities, I mean ALL equities domestic and intl), and those were US Treasuries and Cash. Everything else was near extreme correlation well before the fireworks of Sept-November of 2008. So if you took a S&P500 etf, a treasury etf, and cash and mixed them to an appropriate std. deviation (let's say half of the S&P500 for a risk adverse individual), you would have gotten more than acceptable results even in the peak of the crisis.

    The fallacy of allocation modeling is the thought that the inputs are static and timeless. There is no such thing.

    ReplyDelete
  7. On commodities: it is an asset class because a). it has real return tracking the inflation. b). it is not entirely correlated with other existing asst classes.

    On the other hand, the 'new normal', as what Bill Gross and others termed it, absolutely requires a diversification not just over static assets.

    The best piece of paper I ever read is the diversification over asset classes and strategies. I suggest everyone to take a look at here: http://seekingalpha.com/article/151754-diversify-asset-classes-and-investment-strategies

    ReplyDelete