Being an asset owner is hard! Plan management and risk oversight (by itself a nebulous and often thankless task) can dominate the agenda, leaving investment oversight as an operational must, rather than the rewarding intellectual exercise that we know and love.
For years, style boxes have been a cornerstone used to simplify understanding on the equity side of the ledger while fixed income has been left to its own devices (what exactly is “unconstrained” you ask?). This search for yield and (potential) risk mitigation through diversification is complicated even further given the explosion in multi-asset mandates (up 21% in 2016).
What we aim to do here is outline a process of applying factor analysis to fixed income and corporate bond assets, which was explored in a recent paper by Houwling and van Zundert. Our goal is simple: provide an illustration of the tools available and the regimented process needed to simplify fixed income investment and align it with the traditional equity approach long used by asset owners globally.
Factor Construction
While abundant in the equity space, factor-based benchmarks within fixed income are currently sporadic. Because of this scarcity, we had to begin our analysis with factor construction based on the method defined in the paper mentioned above, and screening for value, size, low risk, and momentum. These may look familiar to those focusing on equity portfolios, however in this case the factors were calculated using only characteristics from the bond market. Undoubtedly factor definitions could be improved through mixing characteristics from the equity market and linking data through the debt issuer, but that’s a topic for further conversation.
Houweling and van Zundert define factors as:
- Value: The deviation of current spread from the spread predicted by issue rating, time to maturity and the three month change in spread
- Size: The total public debt issued by the ultimate parent
- Low Risk: High-rated short dated paper
- Momentum: Six month excess return
Using a screening tool to narrow a universe by quantitative and qualitative factors much the same way a bottom-up active manager might, in the chart below, we've focused on investment grade debt of corporations on the S&P 500. Your analysis could be extended into the high yield or sovereign debt depending on your particular security mix.
Symbol |
Name |
Low Risk Factor |
Size Factor |
Value Factor |
Momentum Factor |
98385XAP |
XTO Energy, Inc. 5.5% 15-JUN-2018 |
1.0 |
5.0 |
10.0 |
10.0 |
94988J5A |
Wells Fargo Bank, N.A. 1.65% 22-Jan-2018 |
1.0 |
1.0 |
10.0 |
10.0 |
931142CJ |
Wal-Mart Stores, Inc. 5.8% 15-FEB-2018 |
1.0 |
3.0 |
9.0 |
10.0 |
931142CP |
Wal-Mart Stores, Inc. 4.125% 01-FEB-2019 |
1.0 |
3.0 |
10.0 |
10.0 |
90331HMU |
U.S. Bank N.A. 1.45% 29-JAN-2018 |
1.0 |
3.0 |
9.0 |
10.0 |
90331HMQ |
U.S. Bancorp 1.35 29-JAN-2018 |
1.0 |
3.0 |
9.0 |
10.0 |
882508AV |
Texas Instruments Inc. 1.0% 01-MAY-2018 |
1.0 |
10.0 |
10.0 |
10.0 |
842434CD |
Southern California Gas Co. 5.45 15-APR-2018 |
1.0 |
6.0 |
8.0 |
10.0 |
74153WCF |
Pricoa Global Funding | 1.9% 21-SEP-2018 |
1.0 |
4.0 |
7.0 |
10.0 |
74153WBZ |
Pricoa Global Funding | 1.6% 29-MAY-2018 |
1.0 |
4.0 |
9.0 |
10.0 |
71713UAQ |
Pharmarcia Corp. 6.5% 01-DEC-2018 |
1.0 |
3.0 |
10.0 |
10.0 |
717081AQ |
Pfizer Inc. 4.65% 01-MAR-2018 |
1.0 |
3.0 |
10.0 |
10.0 |
717081DW |
Pfizer Inc. 1.2% 01-JUN-2018 |
1.0 |
3.0 |
9.0 |
10.0 |
695114CH |
PacifiCorp 5.65% 15-JUL-2018 |
1.0 |
1.0 |
8.0 |
10.0 |
68389XAC |
Oracle Corporation 5.75% 15-APR-2018 |
1.0 |
2.0 |
9.0 |
10.0 |
68389XAN |
Oracle Corporation 1.2% 15-OCT-2017 |
1.0 |
2.0 |
10.0 |
10.0 |
With a universe defined and factor deciles calculated, the tricky business comes with constructing the factor-based benchmarks. To resolve this issue we can combine the screen above with the Axioma Portfolio Rebalancer. This tool combines the flexibility in defining factors with the ability to apply real-world constraints inherent in index construction (limiting turnover, min/max constituents and weights, transaction costs, etc.). The end-product being four single-factor portfolios.
Manager Selection
Having created the four single-factor portfolios we looked across the ETF market for U.S.-based funds investing in Investment Grade Corporate debt. Regressing the returns of each of the funds against the four single-factor return streams it is possible to expose the underlying factor tilts inherent in the market.
Interestingly over the period selected very few funds seemed to behave in line with the Momentum factor and none with the Value factor. This could be down to an issue of sample selection bias, since the ETF in this universe are more likely to invest primarily in higher rated, longer term investment grade paper. Additionally, the clustering in the Size quadrant makes market sense, as the ETFs would be more inclined to purchase the paper of larger borrowers.
Attribution Analysis
A typical asset owner or multi-manager selects funds and managers with the view of gaining exposure to investment grade corporate U.S. debt. Since any combination of the above funds would represent such exposure, we put together an example fund using the six highlighted managers. Running this market weighted composite through a balanced attribution model where the first allocation decision is calculated top down, we can see a negative allocation effect. The main issue here, when looking at multi-asset or fixed income mandates, is it’s not immediately clear how to remedy the allocation to each manager to optimize return.
|
Port. Avg. Weight |
Factor. Avg. Weight |
Port. Total Return |
Factor. Total Return |
Factor Allocation Effect |
Shift Effect |
Twist Effect |
Spread Effect |
Selection Effect |
Total Effect |
Total |
100.00 |
100.00 |
10.03 |
7.94 |
-1.02 |
1.56 |
1.30 |
-1.29 |
1.55 |
2.09 |
|
|
|
|
|
|
|
|
|
|
|
Manager 1 |
15.20 |
-- |
3.28 |
-- |
-- |
0.04 |
0.13 |
0.04 |
-0.04 |
0.16 |
Manager 2 |
27.69 |
-- |
5.08 |
-- |
-- |
0.11 |
0.14 |
-0.11 |
0.44 |
0.57 |
Manager 3 |
8.48 |
-- |
16.90 |
-- |
-- |
0.09 |
0.26 |
-0.25 |
0.20 |
0.30 |
Manager 4 |
23.81 |
-- |
18.04 |
-- |
-- |
1.24 |
0.87 |
-0.33 |
0.90 |
2.68 |
Manager 5 |
10.84 |
-- |
12.02 |
-- |
-- |
-0.05 |
0.05 |
-0.18 |
0.16 |
-0.01 |
Manager 6 |
13.97 |
-- |
8.90 |
-- |
-- |
0.14 |
-0.14 |
-0.47 |
-0.11 |
-0.58 |
|
|
|
|
|
|
|
|
|
|
|
If we were to maintain the original allocation to Investment Grade, but group out the managers by their factor tilts (running each pair against the single-factor portfolios created earlier) we can then dissect the allocation effect further. What became evident is that the implicit heavy allocation to Low Risk managers appears to be driving the negative allocation effect.
|
Port. Avg. Weight |
Factor. Avg. Weight |
Port. Total Return |
Factor. Total Return |
Factor Allocation Effect |
Shift Effect |
Twist Effect |
Spread Effect |
Selection Effect |
Total Effect |
Total |
100.00 |
100.00 |
10.03 |
7.94 |
-1.02 |
1.56 |
1.30 |
-1.29 |
1.55 |
2.09 |
|
|
|
|
|
|
|
|
|
|
|
Factor - Low Risk |
42.89 |
33.35 |
4.50 |
2.64 |
-0.60 |
0.14 |
0.26 |
-0.08 |
0.40 |
0.13 |
Manager 1 |
15.20 |
-- |
3.28 |
-- |
-- |
0.04 |
0.13 |
0.04 |
-0.04 |
0.16 |
Manager 2 |
27.69 |
-- |
5.08 |
-- |
-- |
0.11 |
0.14 |
-0.11 |
0.44 |
0.57 |
Factor - Momentum |
32.29 |
33.29 |
17.65 |
8.50 |
-0.27 |
1.33 |
1.13 |
-0.58 |
1.10 |
2.71 |
Manager 3 |
8.48 |
-- |
16.90 |
-- |
-- |
0.09 |
0.26 |
-0.25 |
0.20 |
0.30 |
Manager 4 |
23.81 |
-- |
18.04 |
-- |
-- |
1.24 |
0.87 |
-0.33 |
0.90 |
2.68 |
Factor - Size |
24.82 |
33.36 |
10.86 |
12.47 |
-0.15 |
0.09 |
-0.10 |
-0.64 |
0.05 |
-0.75 |
Manager 5 |
10.84 |
-- |
12.02 |
-- |
-- |
-0.05 |
0.05 |
-0.18 |
0.16 |
-0.01 |
Manager 6 |
13.97 |
-- |
8.90 |
-- |
-- |
0.14 |
-0.14 |
-0.47 |
-0.11 |
-0.58 |
By actively managing the exposure and reweighting the portfolio toward the Momentum and Size factors, there is an immediate improvement in the allocation effect and subsequently the total performance of the fund.
|
Port. Avg. Weight |
Factor. Avg. Weight |
Port. Total Return |
Factor. Total Return |
Factor Allocation Effect |
Shift Effect |
Twist Effect |
Spread Effect |
Selection Effect |
Total Effect |
Total |
100.00 |
100.00 |
12.56 |
7.94 |
2.47 |
1.46 |
1.09 |
-2.13 |
1.72 |
4.62 |
|
|
|
|
|
|
|
|
|
|
|
Factor - Low Risk |
12.67 |
33.35 |
6.21 |
2.64 |
1.18 |
0.09 |
0.06 |
-0.02 |
0.25 |
1.57 |
Factor - Momentum |
36.78 |
33.29 |
17.57 |
8.50 |
0.21 |
1.16 |
1.18 |
-0.84 |
1.17 |
2.88 |
Factor - Size |
50.55 |
33.36 |
11.41 |
12.47 |
1.07 |
0.20 |
-0.14 |
-1.27 |
0.30 |
0.16 |
Running the overall reweighted fund against the market benchmark it’s clear that significant alpha can be generated across the investment horizon by incorporating the factor-selection decision into the allocation process.
![significant alpha can be generated across the investment horizon by incorporating the factor-selection decision into the allocation process significant alpha can be generated across the investment horizon by incorporating the factor-selection decision into the allocation process](https://insight.factset.com/hs-fs/hubfs/1)Insight/2017/Insight_11.2017/11.07.2017_CorporateBonds/significant%20alpha%20can%20be%20generated%20across%20the%20investment%20horizon%20by%20incorporating%20the%20factor-selection%20decision%20into%20the%20allocation%20process.png?width=1024&name=significant%20alpha%20can%20be%20generated%20across%20the%20investment%20horizon%20by%20incorporating%20the%20factor-selection%20decision%20into%20the%20allocation%20process.png)
As the search for additional return drives asset owners further and further afield into investments in non-traditional asset classes, focusing on strategic factor allocation in the corporate bond portfolios might be simpler solution to their woes. When allocating balanced mandates the ability to incorporate equity-equivalent style analysis for fixed income funds can reap significant alpha. While traditional benchmark providers may not yet cater for this type of allocation, the tools are available to asset owners and multi-managers to both identify and measure managers who display consistent factor exposures in their investment strategies. While the combination of such managers may yield significant fund tracking error due to inadequate benchmarks in the short run, the long-term risk adjusted benefit from adopting such a strategy is certainly worth it.