"If you are doing your own bond portfolio, make sure that you roll along the yield curve by selling bonds that are 1-2 years from maturity, and reinvest them. This is part of the magic that lets bond funds post excellent results ... the key is to maintain a constant avg maturity, and roll over bonds once their YTM drops to nil."
This is a rolling down the yield curve strategy, correct me if I'm wrong. However, it's not a shifting maturity strategy, but instead keeps constant average maturity. By selling bonds, you are now exposed to price risk, unlike with a buy and hold strategy. But if bonds are 1-2 years from maturity, that risk is small. Also, there will be transactional costs in selling bonds, which would be avoided with a buy and hold strategy. For this strategy to work, you'd need relatively liquid bonds. An attractive feature of this is that some of the yield will be cap gains, and in a taxable account, that's important.
How does this compare to GICs? One feature of GICs is that if you shop around, you can do well, relative to bonds. A disadvantage is that you're limited to 5 year terms, if you want the government guarantee. But since I don't want to hold fixed income more than 5 years, that's fine with me. A second important disadvantage is the illiquidity of GICs. If you need liquid fixed income, riding the yield curve makes sense. But if liquidity isn't important, would you be better off shopping for GICs?
I'm thinking of how one could implement this strategy. You could set up a 5 year ladder, but sell each ladder when it's 1 year (or possibly 2 years) from maturity. My guess is that it would work best with government of Canada bonds. I don't know about agency or provincial bonds.
One problem with this strategy is that it assumes a positive yield curve, which is usually but not always true. DFA has a somewhat similar strategy, except maturity is not kept constant. Maturity shifts, and will be extended, but only if if the longer maturity has at least 20 bp more per annum. In other words, the yield curve must be sufficiently positive. It is interesting to note that in the last edition of Hank Cunningham's book on bonds, he basically endorses DFA's strategy.
I like this strategy, as it helps deal with the negative expected return of fixed income in a taxable account. I would be interested in articles/research on this strategy.
Edited to include the following. In this strategy, you sell the bond, when the YTM is zero. There is interest remaining on the bond. But there's a capital gain on the bond, which if the bond is held to maturity, will be lost. So you sell the bond and take the capital gain instead of holding and getting the interest. So you'll get taxed on cap gains instead of interest, which cuts the taxation in half. OTOH, you'll have the transactional costs of selling the bond, which will decrease total return.
Last edited by 0okm9ijn; 2017-02-05 at 10:27 AM.
Yes, I am rolling down the curve and overall maintaining a fixed avg maturity of the portfolio. I consider the fees when I calculate the YTM on the candidate for sale. Because Govt bonds are so incredibly liquid, it makes very little difference. Yes there is price risk exposure but I think there's still a guaranteed return component. Every bond you purchase has a guaranteed YTM -- and rolling them doesn't change that (it can only boost the guaranteed part of the performance).
Here's how I think of it. There are two elements driving the price of the bond portfolio
(1) the guaranteed return, the aggregation of all individual bonds. This part is absolutely analogous to a GIC ladder, with staggered YTMs which collectively produce a guaranteed return
(2) the market price fluctuation of the bonds; the volatility. If GICs were liquid and had pricing, you would see this effect, but GICs hide this.
I think people are overly pessimistic about bond portfolios. The fact you have volatility (2) does not erase the guaranteed portion (1). And in this respect they are very different from stocks. Stocks have the volatility, and do not have any guaranteed portion. Bond funds do have a guaranteed return underpinning them.
Bad years of (2) can make people forget about (1), which is what's been happening since last year. You'll notice on the forums people gravitate towards short term bonds like XSB ... and yet those bond funds still use the same composition!
What are the differences between a GIC ladder, XSB, and VAB ? They are fundamentally the same thing and yet people have very different perceptions of them.
- people don't see the price fluctuations on GICs, element (2) above, and therefore think GICs are much safer
- XSB does have the volatility (2), but it's diminished, therefore people think XSB is much safer
The way I see it, for a long term holder the short term volatility (2) is irrelevant. Either way they will be getting the aggregate YTMs on the bonds, as the bond market offers it. And the whole point of the yield curve is that you are rewarded with higher yields farther down the yield curve. VAB is guaranteed to provide higher returns than XSB over the long term. With more volatility, yes, but its constituent bonds have higher YTMs.
Where I think you do have to be careful is picking the appropriate avg maturity such that (2) does not swamp (1). Personally I am running my portfolio a bit shorter than 10 year avg maturity.
Last edited by james4beach; 2017-02-06 at 07:10 PM.
I should add, this is consistent with the rule of thumb that the best estimate you have for the forward return of a bond fund is the YTM of the entire portfolio, what I describe as (1) above.
Originally Posted by james4beach
More accurately stated, historically there has been a very high correlation between the US aggregate bond performance over 10 years and the initial YTM at purchase time (it's 10 years because that's the avg maturity of the portfolio). This is consistent with what I'm calling (1) in my model. The second factor (2) is transient and averages out to zero.
Last edited by james4beach; 2017-02-07 at 12:04 AM.
Originally Posted by james4beach
Table 2: Interest-Rate Cycles and Stock/Bond Annualized Returns
Bond Maturity/ Stock Index
1964-1981 (interest rates UP*)
1982-2013 (interest rates DOWN*)
Cash (1Mo T-bills)
OK, it looks like I was wrong about this point. I'm surprised at those results over such a long period, 1964-1981. I agree that it does show that there can be periods where short end of the curve (like XSB) consistently outperforms the longer end of the curve (like VAB).
I can corroborate what you demonstrated from this data source. With that data I can also find a 17 year period where the short end of the curve outperforms... wow. I am surprised it can be such a long period.
Note that this isn't inconsistent with what I wrote about the guaranteed return + volatility. The return is still positive as the bonds are still providing the guaranteed returns as predicted. But I was clearly wrong about the superior returns in the longer end of the curve.
Some observations / Devil's Advocate
* Even in this worst case scenario, the bond returns are still positive across different maturities
* To think that you're better off at shorter maturities, you must believe that interest rates are going to consistently rise, a lot. We're talking about a historical period where rates went up from about 4% to 16%.
* If you think we're entering a period like that, you don't want to be invested in stocks either. The economy has been dependent on low interest rates/consumer borrowing since the 90s, and US govt will default if borrowing costs soar like this. In Canada, it would be nothing short of catastrophic to the housing market, and probably collapse the banks as a result after mass consumer and business defaults. It would end the ~ 25 year credit bubble.
* Since nobody can predict where interest rates are headed, on average you get better performance at longer maturities. Don't rule out Japan-style ultra low rates for 20 years.
Some actions I will take
You've changed some of my thinking on this. I now see that shorter maturities can outperform longer maturities, even over decades. For now I will continue tracking VAB (10y) but will reduce my portfolio maturity if there are signs interest rates are steadily heading higher. Currently, I am not convinced they are heading higher.
But I will not rule out shifting my portfolio towards XSB / VSB if it looks like we're in a period of steadily rising rates.
Here was the error in my reasoning: yes the yield curve rewards you with higher yields at longer maturities, but you're also trapped at those yields for a longer time. That's fine when yields aren't changing much. BUT when yields are rapidly rising, you are better off at the short end of the curve because the bonds get flushed out of the portfolio more rapidly and replaced with even higher YTMs.
Originally Posted by james4beach
Nice article showing aftertax real return in USA from cash, bonds and stocks from 1928-2012. Obviously some assumptions about taxes. In a taxable account, I conclude that bonds and cash are for risk management. In a tax advantaged account, you could have a positive expected return. But since it's a while until I retire, I own stocks in my tax advantaged accounts, to get growth.
It sounds like you're 100% in stocks. Are you comfortable with the ~ 50% drawdown situation and the possibility for being negative for a decade or more? Everyone says they're ready for that, until it happens.
You can't take 1928-2012 in the US as an indication of what will happen 2017-2080. In the 20th century the American economy was a huge winner and inflation was high. Special circumstances played a major part, like giving up the gold standard and the length of time it took to figure out how to control inflation. The future will not be the same as the past.
Originally Posted by 0okm9ijn
What we can glean from the 20th century is that stocks handle inflation well compared to bonds and cash. Oh well, we kinda knew it anyway.
mordko, do you foresee (or think there's a reasonably good chance) of a prolonged period of low inflation ahead?
I agree that stocks definitely handle inflation well. Personally I think low inflation or even deflation are possibilities on the horizon -- which is a reason I keep an allocation to cash & bonds.