Welcome! Please Log In
Go
Essentials Popular Topics Favorite Forums & Blogs Join Discuss to setup a list of your favorite forums/blogs.
Discuss > Single Post
FF Bond Model & HY backtests
Robert T 11-28-2006, 4:47 AM | Post #2285250 | 
0  
.
0Brian0,

FF two bond factors

I agree that more attention has been given to the FF three (equity) factors than their two bond factors (which I think was the focus of post 10). But their bond factors haven't done too badly at explaining variations in bond performance (except for below Baa). The results using data from 1963-1991 presented in their earlier paper shows the term and default factors explained 79 to 98 percent of the variation in monthly bond returns over the period (except for below Baa).


Two Factor Model R^2

1-5 year Government bonds.....79%
6-10 year Government bonds....87%
Aaa rated bonds...............97%
Aa rated bonds................98%
A rated bonds.................98%
Baa rated bonds...............90%
Below Baa bonds...............59% (incl. all 5 factors)

Source: Fama-French (1993): Common risk factors in the returns on stocks and bonds.


The FF model is far from perfect - but provides a reasonable framework for long-term investment decisions (at least for me).

HY in backtested portfolios

Here are a few back-tested portfolios. The results suggest that adding HY in place of 5 yr. T-Notes raised the portfolio standard deviation but not the annualized return. This is for both a global total market, and value and size tilted portfolio. Shifting 28% and 36% of 5-yr T-Notes to equities in each portfolio respectively yielded the same standard deviation as a portfolio with HY, but with higher returns. The results also indicate that tilting a portfolio to small and value stocks had a signficantly large impact on returns than shifts among fixed income.


1979-2005

37.5:37.5:25
..........................Annualized return...Std Dev.
CRSP1-10:EAFE:5 yr T-Notes.......12.0..........12.7
CRSP1-10:EAFE:HY.................12.0..........13.8

A stock:bond mix of 82:18 of the first portfolio would have yielded the same standard deviation as the second (13.8) but the annualized return would have increased to 12.2.

37.5:37.5:25 - with a small and value tilt (loadings of 0.2 and 0.4 respectively)

.........................Annualized return....Std Dev.
CRSP1-10:EAFE:5 yr T-Notes.......14.1..........12.4
CRSP1-10:EAFE:HY.................14.1..........13.7

A stock:bond mix of 84:16 of the first portfolio would have yielded the same standard deviation as the second (13.7) but the annualized return would have increased to 14.6.

[CRSP1-10:EAFE are taken from Ken French's website. HY is the Vanguard High Yield Corp. Fund - annual data for the HY fund only go back to 1979].

Robert

Originally posted in thread: 54885
View Complete Thread
Reply Quote
  • Favorites
  • Flag
  • Contact
  •  
    © Copyright 2019 Morningstar, Inc. All rights reserved. Please read our Terms of Use and Privacy Policy.