Year to date stocks have been hot, bonds not. On a total return basis (as of 2/3/12 ) SPY is up 7.17 and TLT down -3.85. Should bondholders be nervous? For months now, pundits have been warning that interest rates can't go much lower, that inflation will eventually kick in, leading the bond bubble to burst, and that all those investors who piled into bond ETFs since 2008 will get burned.
What to do? Some investors are already trading in their bonds for dividend growth stocks. Yet it may be too early to bail out of bonds. The Fed has pledged to hold interest rates down for another year or two, and with some careful portfolio management, investors might be able to find more yield while lowering their interest-rate risk by paying attention to the duration of their bond holdings.
Duration is a measure of the sensitivity of bond prices to changes in interest rates. Bond prices move inversely to changes in interest rates, and the higher the duration, the more a bond's price will rise as interest rates fall and fall as interest rates rise. In anticipation of rising interest rates, bond investors can reduce their risk by shortening the durations of their portfolios.
The formula for calculating duration is complicated, but issuers of bond ETFs publish an "average effective duration" for all their offerings, and the data is widely available. TLT (iShares Barclays 20+ Year Treasury Bond ETF) has an average effective duration of 17.01 years (Morningstar data as of 2/3/12). This means that a 1% increase in interest rates will produce a 17% decrease in the ETF's market price. If interest rates are headed up, investors can reduce the negative effects on their portfolios by selling TLT and buying an ETF with a shorter duration, such as SHY (iShares Barclays 1-3 Year Treasury Bond), which has a duration of only 1.83 years.
But which ETFs have the lowest durations and how can they be combined to raise yield while reducing risk? To answer these questions we need some data. The table below lists all thirty bond ETFs with assets of more than $500,000 and at least three years of performance data, excluding inverse funds (ETFdb data as of 2/3/12).
SYMB | DESCRIPTION | ASSETS (MILLIONS) | 12-MO YIELD (M'STAR) | DURATION (M'STAR) | 3YR STDEV (M'STAR) |
iShares TIPS | 23,141 | 4.02 | 4.06 | 5.83 | |
iShares Inv Grade Corporates | 18,702 | 4.32 | 7.50 | 5.91 | |
Vanguard Total Bond Market | 14,812 | 3.08 | 5.09 | 2.82 | |
iShares US Aggregate Bd Fund | 14,571 | 2.84 | 4.39 | 2.67 | |
iShares Hi Yld Corporates | 13,253 | 7.48 | 4.32 | 10.78 | |
iShares 1-3 Year Treasuries | 10,852 | 0.81 | 1.83 | 1.00 | |
SPDR High Yield Bond | 10,682 | 6.86 | 4.46 | 13.22 | |
iShares 1-3 Year Credit Bonds | 9,324 | 1.90 | 1.83 | 2.18 | |
Vanguard Short-Term Bonds | 7,584 | 1.88 | 2.63 | 1.68 | |
iShares 7-10 Year Treasuries | 5,130 | 2.53 | 7.51 | 6.94 | |
iShares Intermed Credit Bond | 4,473 | 3.68 | 4.24 | 3.92 | |
iShares MBS Fixed-Rate Bond | 4,248 | 3.32 | 1.52 | 2.21 | |
iShares USD Emerging Mkts | 3,795 | 4.82 | 7.18 | 7.71 | |
iShares 20 Year Treasuries | 3,308 | 3.31 | 17.01 | 14.88 | |
Vanguard Intermed Bond Idx | 2,925 | 3.56 | 6.37 | 4.55 | |
iShares National Muni Bond | 2,704 | 3.25 | 6.99 | 4.80 | |
iShares Short Treasuries | 2,567 | 0.07 | 0.50 | 0.09 | |
iShares 3-7 Year Treasuries | 2,330 | 1.66 | 4.53 | 3.94 | |
SPDR Int'l Treasury Bond | 1,750 | 3.27 | 6.73 | 9.59 | |
SPDR Short Term Muni Bond | 1,490 | 1.43 | 2.79 | 1.73 | |
PwrShares Em Mkt Sov Debt | 1,406 | 5.43 | 8.64 | 8.14 | |
SPDR Int'l Gov TIPS | 1,255 | 4.45 | 9.91 | 11.39 | |
SPDR 1-3 Month T-Bills | 1,141 | 0.00 | 0.09 | 0.03 | |
iShares Credit Bond Fund | 1,105 | 4.07 | 6.44 | 4.83 | |
SPDR Municipal Bond Fund | 1,048 | 3.17 | 9.04 | 5.06 | |
PwrShares Hi Yld Corporates | 719 | 5.77 | 4.13 | 9.72 | |
iShares Int Gov/Credit Bonds | 715 | 2.61 | 3.90 | 2.52 | |
PwrShares Insured Nat Munis | 648 | 4.34 | 12.80 | 6.52 | |
SPDR TIPS | 646 | 3.87 | 8.53 | 5.95 | |
Vanguard Long-Term Bond | 632 | 4.18 | 14.21 | 8.44 | |
Total | 166,956 |
Together these thirty EFTs have almost 167 billion dollars in assets and represent the bulk of the bond EFT market. Note that the yield quoted is 12-month yield, not SEC yield. Besides yield and duration, the table includes each ETF's three-year standard deviation, a measure of its volatility (Morningstar data as of 2/3).
To visualize the relationship between these variables, the figure below shows a plot of 12-month yield by duration for the above data.
(Click to enlarge)
The first thing to notice is the three outlying data points at the top (HYG, JNK, and PHB), which have much higher yields than other ETFs with similar durations. They represent junk bond ETFs which must compensate investors for higher default risk with higher yields. Although such ETFs have a place in some portfolio strategies and have attracted a lot of money, they are omitted from the following analysis because of their high volatility as measured by three-year standard deviation.
Looking at the remaining twenty-seven ETFs, we see that four of them (MBB, TIP, EMB, and PCY) offer higher 12-month yields relative to their durations than all the others. Drawing a line through these four data points produces an "efficient frontier" or upper boundary which arches over the other data points, all of which lie below and to the right of the line.
Any ETF whose data point lies below and to the right of the efficient frontier is providing less yield for the same duration or more duration for the same yield (or both) than any ETF on the efficient frontier. No combination of the twenty-seven non-junk-bond ETFs can produce higher yields for a given duration than those defined by the efficient frontier. Any combination of duration and yield on the efficient frontier can be produced by a combination of two of the four ETFs that lie directly on it. The efficient frontier is thus an envelope that can't be pushed, unless you add in junk bond ETFs.
Application 1: iShares Barclays Intermediate Credit Bond ETF (CIU)
To see what this means for reducing risk, let's look at the case of CIU, iShares Barclays Intermediate Credit Bond ETF. It holds mostly corporate bonds (81%) and yields 3.68% with a duration of 4.24 years. If interest rates rise 1%, its market price will decline 4.24%, all other things being equal. How can we use the efficient frontier to build an alternate portfolio of ETFs with the same or better yield but lower duration?
Limiting our choices to the three ETFs on the efficient frontier with the lowest duration (MBB, TIP, and EMB) and combining them in increments of 5% with a minimum 5% share produces 171 possible combinations, 154 of which have a yield greater than 3.68%. The table below summarizes the best of these results.
% CIU | % MBB | % TIP | % EMB | YIELD | DUR | STDEV | NOTES |
100 | 3.68 | 4.24 | 3.92 | ||||
60 | 30 | 10 | 3.68 | 2.85 | 3.85 | LOWEST DUR | |
65 | 5 | 30 | 3.81 | 3.35 | 4.04 | ||
65 | 10 | 25 | 3.76 | 3.19 | 3.95 | ||
65 | 15 | 20 | 3.73 | 3.03 | 3.85 | ||
65 | 20 | 15 | 3.68 | 2.88 | 3.76 | ||
70 | 5 | 25 | 3.73 | 3.06 | 3.77 | ||
70 | 10 | 20 | 3.69 | 2.91 | 3.67 | LOWEST STDEV |
As the table shows, a 60% MBB, 30% TIP, 10% EMB combination provides the same yield as CIU (3.68%) while reducing the duration from 4.24 to 2.85 years. The 3-year standard deviation is also lower for most of these combinations, the best one being 70% MBB, 10% TIP, 20% EMB with a yield of 3.69%, duration of 2.91 years, and standard deviation of 3.67. Both of these combinations provide the same yield as CIU with much less interest-rate risk and less overall volatility.
Of course, investors in CIU might have good reasons for wanting to buy corporate bonds rather than MBB, TIP, and EMB, especially if they believe corporates will outperform mortgage bonds, TIPS, and emerging market bonds in the coming months and years. In that case they are betting that the potential outperformance of corporates will compensate for the higher susceptibility of their investments to interest rate increases.
Application 2: iShares Barclays Aggregate Bond (AGG) and Vanguard Total Bond Market EFT (BND)
These two ETFs are popular choices for passive indexers who use them as proxies for all or most of the domestic investment-grade bond market. Both are attractive for their relatively low volatility as measured by 3-year standard deviation: 2.67 for AGG and 2.82 for BND. Yet both are situated well below the efficient frontier, each having a duration higher than its yield. The question is, can we find a combination of MBB, TIP and EMB that provides a higher yield and lower duration, without increasing the standard deviation too much?
The table below shows the best results from 171 possible combinations.
%AGG | %BND | %MBB | %TIP | %EMB | YIELD | DUR | STDEV | |
100 | 2.84 | 4.39 | 2.67 | |||||
100 | 3.08 | 5.09 | 2.82 | |||||
55 | 35 | 10 | 3.71 | 2.97 | 4.03 | |||
60 | 30 | 10 | 3.68 | 2.85 | 3.85 | |||
65 | 20 | 15 | 3.68 | 2.88 | 3.76 | |||
65 | 25 | 10 | 3.65 | 2.72 | 3.66 | BEST? | ||
70 | 10 | 20 | 3.69 | 2.91 | 3.67 | |||
70 | 15 | 15 | 3.65 | 2.75 | 3.58 | |||
70 | 20 | 10 | 3.61 | 2.59 | 3.48 | |||
75 | 10 | 15 | 3.62 | 2.62 | 3.40 | |||
75 | 15 | 10 | 3.58 | 2.47 | 3.30 | |||
80 | 10 | 10 | 3.54 | 2.34 | 2.12 | BEST? |
From a yield point of view, all of these combinations are good, providing 3.54% to 3.71%, at least 70 basis points more than AGG and 63 basis point more than BND. But the reduction of duration is even better, lower by at least 1.42 years for AGG and 2.12 years for BND.
From a purely statistical perspective, the last combination (80% MBB, 10% TIP, 10% EMB) is optimal, since if offers both the lowest duration (2.34 years) and standard deviation (2.12). However, its 80% allocation to MBB doesn't offer sufficient diversification. As a rule of thumb, it makes sense to allocate no more than two-thirds of capital to one component of a three-ETF portfolio. From this point of view, the 65% MBB, 25% TIP, 10% EMB combination may be the best, since it provides a higher yield and much lower duration than both AGG and BND, while increasing standard deviation by only 0.85 or 0.99 to 3.66.
Passive indexers holding AGG and BND may object that a portfolio consisting of only MBB, TIP, and EMB doesn't represent the total bond market, and they are right. But if and when interest rates rise, their holdings will be more negatively affected than any the combinations described above.
Conclusion
No one knows what interest rates will be two years from now. If the Fed has its way, things may not have changed very much. But if the market wins out, inflation heats up, defaults increase, and sovereign bonds come under further attack, rates could be much higher. Paying attention to the duration of your bond ETFs before this happens could protect your portfolio from significant losses, while earning you higher yields in the meantime.
Disclaimer: This article is for informational purposes and is not a recommendation to buy or sell any investments. Do your own research and due diligence before making any investments. Good luck!
