MPT now and in the future

[Architect]

Unfortunately, it seems Modern Portfolio Theory didn't work too well this year. Comment from CALPers

"In theory, asset allocation models are meant to mitigate risk through diversification. If one bucket, say equities, falls, then the fixed-income or private equity investments should offset the decline. But when markets fall more or less in tandem, the model not only fails but also puts other assets at risk of the capital needs of others. So much for theory."

Comment from Gary Shilling (hope he doesn't mind)

But as we’ve noted continually in Insights for more than 10 years, there are tremendous amounts of hot money flowing around the world. And whether it’s managed on the basis of fundamental factors, momentum, technical analysis, etc., it all tends to end up on the same side of the same trade at the same time.

further

So when stocks get clobbered, as they have since October 2007 (Chart 14), and force out hot money, it will also retreat from otherwise unrelated long positions in, say, grains, to conserve capital. Many institutional investors believe in the Modern Portfolio Theory of diversification, but erroneously thought that alternative investments would have zero or better still, negative correlation with their basic equity holdings (Chart 15). They also became convinced that commodities and foreign currencies were asset classes like equities and bonds, and merited 5%, 10% or 15% of their portfolios. They’re learning the hard way that all those correlations have proved to be close to 100% and that commodities and currencies aren’t asset classes but speculations. benefit pension plans at the end of 2007 was $60 billion. It’s now estimated at over a $300 billion deficit, which will require substantial company contributions and hits to earnings and book values next year. The 100 largest pension funds lost $120 billion in October, and probably at least as much more company contributions. Colleges and universities that have “pre- spent” their investment returns are especially devastated by 30% or 40% portfolio declines even though most average their asset values over several years in determining their withdrawals.

Locally, MyCo's pension (glad to still have it) tanked this year. The costs of refunding the loss is being pushed down to the local groups, taken directly out of the division profits.

Comments? Do you think the hedgies will wash out enough to decrease asset correlations, or is this a more permanent condition? Given the ease of pushing money around the world now (even by ordinary investors with web browsers), I'm not sure the easy days of MPT are coming back.

 

[Meadbh]

Excellent question. I suspect the answer is that new theory is required to explain asset correlation in a global financial universe. Definitely worth a PhD thesis or two.

 

[FIREd]

"In theory, asset allocation models are meant to mitigate risk through diversification.
"

And the theory has been proven right so far! Treasuries are up (some are up a lot!), GNMA (and agencies) are up, bonds as an aggregate (total bond market index) are up, cash is up... Even most of the bonds which are down YTD, are down only marginally compared to stocks.

If one bucket, say equities, falls, then the fixed-income or private equity investments should offset the decline.

I don't think that bonds are supposed to offset the decline completely, they are supposed to offset some of the decline, and they have done their job as far as I can tell! It would take a very conservative AA for bonds to completely offset stock declines even in a typical downturn... I think this is an unreasonable expectation.

VG's 500 index fund is down about 39% YTD
An index portfolio composed of 50% stocks / 50% bonds (VTSMX + VBNFX) would be down only about 19% YTD... Diversification is not working?

 

[Architect]

This is my suspicion. MPT worked great in the latter half of the 20th, but since 1988 it seems to be broken. This is why I'm considering a barbell.

Investing is an important, if there was just one formula that worked now and forever, well that would be easy wouldn't it? Too easy, considering how everybody is trying to game the same system it seems. Another way to look at it is the Fallacy of Composition. Maybe MPT (or any system) can only work if not everybody does it.

The trick might be to use a system out of favor, then if it becomes what the MBA's are all getting taught, time to find another.

Excellent question. I suspect the answer is that new theory is required to explain asset correlation in a global financial universe. Definitely worth a PhD thesis or two.

 

[Architect]

The only bonds that are up to any degree are Treasuries, or very near Treasuries. Even Fannie/Freddy has been hit, though less than other bonds. Corporates, Muni's have all been slammed, as nobody knows who is going to go bankrupt. You may have lost less on bonds, but they're hardly uncorrelated.

But are Treasuries and cash an 'asset class'? I don't think so - bonds in total are an asset class, and in aggregate they're not doing too well. Certainly not as well as they should. My 401k PIMCO fund (the best I could pick given the choices for this turn) is zero to -1% for the year, despite kicking out nice coupon every month. And it's almost all in Fan/Fred!

And the theory has been proven right so far! Treasuries are up (some are up a lot!), GNMA (and agencies) are up, bonds as an aggregate (total bond market index) are up, cash is up... Even most of the bonds which are down YTD, are down only marginally compared to stocks.

 

[FIREd]

But are Treasuries and cash an 'asset class'? I don't think so - bonds in total are an asset class, and in aggregate they're not doing too well. Certainly not as well as they should.

How much were they supposed to go up?

Cash is definitely an asset class. Some people consider Treasuries an asset class (see Bob Clyatt's rational portfolio for example) because they can behave differently from other bonds. As an aggregate bonds are up about 2% this year (VBMFX). It may not be a lot, but it sure beats a 40% drop in equities!

But when markets fall more or less in tandem, the model not only fails but also puts other assets at risk of the capital needs of others. So much for theory.

According to this quote, the model fails because markets have fallen in tandem. That's simply inaccurate.

 

[Architect]

How much were they supposed to go up?
Cash is definitely an asset class. Some people consider Treasuries an asset class (see Bob Clyatt's rational portfolio for example). As an aggregate bonds are up about 2% this year (VBMFX). It may not be a lot, but it sure beats a 40% drop in equities!

If I could have done a buy and hold of individual bonds/CD's in my 401k I would have had a 5% return.

You make a fair point though, but I'll still maintain there's a strong correlation between all asset classes excepting Treasuries. And for cash I'll just note the buck rally this year! (it depends on what cash you're talking about. Those who took the commonly voiced advice to invest overseas didn't appreciate that rally)

According to this quote, the model fails because markets have fallen in tandem. That's simply inaccurate.

I don't follow. Rather I would say the model fails because of (many other reasons). Then markets fall in tandem, which makes the problem worse.

 

[FIREd]

If I could have done a buy and hold of individual bonds/CD's in my 401k I would have had a 5% return.

You make a fair point though, but I'll still maintain there's a strong correlation between all asset classes excepting Treasuries. And for cash I'll just note the buck rally this year!

I think that when people read "low correlation" they think "negative correlation", meaning when one asset class goes down, the other one has to go up. Low correlation may mean that when one asset class goes down, the other one goes down a lot less.

I don't want to be too insistent but there was no strong correlation between stocks and bonds this year. You would have a strong correlation if EVERYTHING went down 40-50% this year. It's simply not the case. Even the worst of bonds (high yield corporates) had a correlation of "only" +0.7 to stocks this year (some will argue that high yield corporates behave more like stocks than bonds even under normal circumstances anyways). Investment grade bonds (intermediate term), which have been "slammed" had a correlation of only +0.25 to stocks YTD, hardly a strong correlation. Mid term munis had a correlation of only +0.06 to stocks YTD. On and on. I think it is important to keep things in perspective.