Robert,
First...a couple of broad thoughts -- because I am not sure we are on the same page.
There are a few aspects to the "variable maturity" approach in DFAs 5YR Global fund <its not as simple as "shifting rates">:
1) evaluate 10 or more different
global yield curves (europe, aussie, jap, UK, canada, US, etc) for the steepest slopes (Fama/DFA assume term premiums are most reliable when yield curves are upwardly sloped). This could result in 100% US <2003>, or 0% US <1997>. This will be variable.
2) Within each yield curve, evaluate the buy
and sell decisions based on current prices (buy a 5 year bond with the intension of selling it in one year and buying another 5 year bond...or buy a 3 year and hold it two years...etc). This too will be variable.
3) Within each yield curve, evalate the spreads (and slopes) of corporate vs. gov't bonds, and opt for credit risk when yield spread is sufficient to justify the risk (coupon premium of 3% or more, or whatever the rule is). Finally, this also is variable.
...so, to that extent, I believe all 3 ("the variable maturity approach) contributed positively in some way to DFGBXs significant return in 1996 (and has detracted from returns recently as it was "too short" last year).
As I pointed out using the
hedged global bond indexes of 1-3 (ultra short), 1-5 (short), and 1-30 (medium/long with TERM of 0.5 -- or similar to when DFGBX has reached "maxium extension"), the shifting maturities of the fund from short to long midway through 1996 turned out to be very precient with hindsight...and using those indexes, its clear to see where most of the returns came from. No need to use the US indexes as the US curve was shaped differently and did not change as foreign curves did.
Also, remember that the bond indexes keep
constant exposures to the various bond markets, where DFGBX can isolate 2 or 3 contries where yield curves are particularly steep and avoid others that must stay in the index. In 1996, the fund only invested in about 5 bond markets, vs. the 10 or more in the index. Not every overseas yield curve was upwardly sloped --
thats likely where the extra return enhancement came from.
For 1996, from the above returns, the best return in the first 6 mths (2.52) added to the best return in second 6 mths (4.34%) comes to about 7%. My understanding was that the DFA 5 yr Global Fund has a maturity cap of 5 yrs (from Prospectus “It is the policy of the Portfolio that the weighted average length of maturity of investments not exceed five years”). So there still appears to be significant ‘non’ varying maturity related returns (from the above table comparison). There was a small credit premium in the second 6 mths of 1996 (LB 1-5 Gov returns were 3.79%, compared with the 3.87% returns for the LB 1-5 Gov./Credit returns). But this doesn’t seem to explain the close to 4% difference (10.8 for the DFA 5yr Global FI vs. the 7% for similar varying maturity fixed income). Interestingly the Payden Global Fixed Income return for the second 6 mths of 1996 was 6.41% (closer to the 8.12% for the DFA fund), which seems to suggest foreign (hedged) exposure may explain part (perhaps even more) of the difference. Again, the above analysis doesn't seem to suggest that most of the difference/value added (the largest of the 1991-2007 period) was due to varying maturity. Perhaps DFA have a more exacting analysis of this...but the above results are not too convincing.
Again, I have tried to be more specific about the sources of returns. When the fund was over 90% in foreign markets (but always fully hedges back to US dollars -- which eliminates currency risk and only exposes one to interest rate risk across bond markets), I am not sure the behavior of US markets mattered much at all over this period -- it was primarily the regional specific variable maturity decisions. Thats why I said above that I could see why Citi 1-5 YR World Bond Index maybe the best bogey. (even though ST Bond Index has closer credit charateristics to DFGBX)
PS: Using US bond factors in the FF two factor regression on a global bond fund complicates interpretation IMO, as would using US equity factors in a FF three factor regression on a global stock fund. Therefore difficult to draw strong conclusions (as the R^2 suggests)…
I don't think the stock factors : bond factors are completely fair. For one, most Int'l stocks are unhedged. Furthermore, globally hedged bond markets are more highly correlated that global equity risk factors.
The 60% isn't perfect, or the 95% we'd hope for or get with the 3F model, but we can certainly speak intelligently about more than half of the returns we are witnessing. With the shifting global maturities, shifting global markets, and shifting credit decisions, the strategy is obviously too robust and strategic for a simple 2 factor model. Local term and credit premiums across 10 different markets would likely be necessary -- a 20 factor fixed income model I guess.
Just as I believe global size, value, and equity premiums should be the same over time, I also think global term and credit risks should also be similar....just won't track closely every year. But from a risk/adjusted basis, I think ST Bond Index (not Global Gov't Index) is the appropriate bogey -- even with modest tracking error.
Yes, just trying to check the claims of significant superiority of the DFA variable maturity approach...
Well, 0.5% annual outperformance net of fees over global bond benchmarks,
when most bond indexes trail theirs by 0.25% or more is a pretty good start. In reality,
if the variable maturity strategy works as it should (and is intended)
over an entire interest rate cycle, we could see returns higher than ST, Int, or LT bonds.
Just image the period of 64-07. If a fund captured the "1-5YR return" from 64-81, and then closer to the "5YR return" from 82-07 <"variable*>, that would be a higher total return than short/medium/or long static portfolios without the significant volatility or inflation risk:
Code: Select all
1964-2007
1YR = +6.7%
*Variable = +7.7%
5YR = +7.4%
20YR = +7.5%
Depending on your expectations I guess, it seems to be working. Just my view.
Thats all for me, thanks for the Graham quote, I had been looking for something like that from him.
sh