Why bonds when one's investment horizon is long?

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Dude2
Posts: 1777
Joined: Fri Jun 08, 2007 3:40 pm
Location: FL

Re: Why bonds when one's investment horizon is long?

Post by Dude2 »

Money Market wrote:I'm curious for the veteran members of this board/the previous morningstar group. This board existed when the 2008/9 crisis hit. How many people were making threads or asking a lot of questions regarding their 100% stock portfolio plummeting? I'm wondering because after an 8 year bull-run, it seems like everyone has the mentality of 100% stocks after seeing all the good times we've had. Psychologically, people are more inclined to invest more and take more risks at the near-highs of a bull market. I could not stomach a 100% portfolio myself since the risk-adjusted returns and volatility and lack of ability to rebalance makes it overly risky for me.
Yes, I think this is what bubbles are made of. All of a sudden people start to drink the Coolaide that stocks are where it's at, and they don't want to miss out. They buy and buy and buy, so that valuations do not matter. Eventually there is a correction. Same thing with real estate. Rates are going to rise, oh no, better immediately buy a house or you'll miss out. The cattle go to the trough and valuations get all messed up. Time and again the cycles repeat. People mostly just need to be real with themselves. You will never hear a gambler tell you stories about the times they lost money. Do 100% stock people really stay the course, or do they simply just not remember the times they capitulated?
TBillT
Posts: 1001
Joined: Sat Sep 17, 2011 1:43 pm

Re: Why bonds when one's investment horizon is long?

Post by TBillT »

I see that Gary Shilling is still recently predicting the 30yr T-Bond will go down to 2.0% interest vs. 2.8% current. So that represents quite a large potential gain for Bonds. The basic argument for Bonds is that the 35-yr bond bull market is apparently still alive. Not sure what happens when it's over, or if we ever see the end in the near term future. The reason people listen to Shilling is to get the overall trend, and he is still in Bonds, mainly long 30-yr bonds.
User avatar
TD2626
Posts: 637
Joined: Thu Mar 16, 2017 3:40 pm

Re: Why bonds when one's investment horizon is long?

Post by TD2626 »

gordoni2 wrote:Time is a diversifier, but a weak one. Or to put it another way, you probably need a much longer investment horizon than you might imagine.

A conservative assumption (used in the figure below) might be that going forward stocks will return 4.5% arithmetic real return, an annual standard deviation of 16.8%, and no long term auto-correlation effects (for a 3.2% expected geometric real return). Then over 50 years, the standard deviation in annualized return will be roughly 1/sqrt(50) times 16.8%, or 2.4%. Now you need +/- two standard deviations to generate a 95% confidence interval. So a 95% confidence interval on annualized real return after 50 years would be roughly -1.6% to 8.0%.

Thus even after 50 years stocks might plausibly return a negative real return! You might not want to take that risk. Fortunately, long-term inflation indexed bonds currently pay a guaranteed 0.9% real with very little risk.

Image
This is a great example of what people mean when they say stocks are less risky over the long run.
Yes, they are, sortof. I mean, the 50 year distribution clearly is much less wide than the 10 year distribution. However, the effect is far weaker than most people imagine. It's good that this post so clearly points out both the risk of negative real returns after even 50 years.

The "standard deviation of the mean" is the best way to look at things in my opinion - dividing risk by 1/sqrt(# of years). Realizing fully that this represents the distribution of possible CAGRs over such a long timespan is concerning.

This is a case where risk is lower, but only to statistics people who can interpret these sort of metrics and charts correctly. Blindly throwing out "risk is lower" to beginners is dangerous as the real picture is so complicated.

For example, this picture assumes risk = standard deviation. It also assumes no reversion to the mean. It further assumes that the distribution of stock market returns has no skew or excess kurtosis. All of these simplifying assumptions are important to understand when looking at this sort of analysis. However, taking them into account is difficult - and it's possible that auto-correlation and kurtosis risk effects end up partially cancelling, allowing the Gaussian approximation to hold.

The chart is really great and I found it quite helpful.
User avatar
grabiner
Advisory Board
Posts: 35265
Joined: Tue Feb 20, 2007 10:58 pm
Location: Columbia, MD

Re: Why bonds when one's investment horizon is long?

Post by grabiner »

Money Market wrote:I'm curious for the veteran members of this board/the previous morningstar group. This board existed when the 2008/9 crisis hit. How many people were making threads or asking a lot of questions regarding their 100% stock portfolio plummeting?
You can go back and see for yourself. Here's a link to the posts on the Personal Investing forum from 71,000 posts ago, which as of now (June 25, 2017) was October 2008.

viewforum.php?f=1&start=71000

You can scroll backwards or forwards to see that we had several threads about panicked investors, and just as many who were afraid of starting investing in a bear market as there are now who are afraid of starting in a bull market.

But you will also find threads of people who had set their risk tolerance correctly. I was 90% stock at the time (with the risk of 100% stock because I overweight small-cap, value, and emerging markets), and when the market crash made that 86% in October 2008, I sold some bonds in my retirement account to get back to 90% stock.
Wiki David Grabiner
dbr
Posts: 46137
Joined: Sun Mar 04, 2007 8:50 am

Re: Why bonds when one's investment horizon is long?

Post by dbr »

One might like to peruse the "Plan B" discussions: https://www.google.com/search?sitesearc ... g&q=plan+b
User avatar
arcticpineapplecorp.
Posts: 15014
Joined: Tue Mar 06, 2012 8:22 pm

Re: Why bonds when one's investment horizon is long?

Post by arcticpineapplecorp. »

niners9088 wrote:
munemaker wrote:
niners9088 wrote:All past performance says that a 100% stock portfolio has better returns over long periods then any mix below 100% stocks.
Academic studies show a 90/10 portfolio (with rebalancing) outperforms 100% equities.
Can you share any links? I'd be interested to read them as I haven't seen any to this point.
I was going to ask the same thing. I thought, well, that "could" be true if for example you invested during a bond bull market. In fact, I started looking at past bogleheads posts to see if there was already such an article/link posted but didn't find any.

But that wasn't even the case from 1970-2009...which included the great bond bull market (numbers below according to trevh, source: viewtopic.php?p=601694#p601694):

100% stocks turned $10,000 into $492,393.54 while
90/10 turned $10,000 into $479,571.22

There might be a certain period where it's true that 90/10 beat 100% stocks (like during a recession you're clearly going to do better holding bonds) but that doesn't make it true all the time. The nature of that thread above is that valuations matter. So it "could" be true but it depends on what stocks and bonds earn going forward. Holding bonds should mean that your portfolio won't drop as much as the total stock market during bear markets (so 90/10 will beat 100% then) but the opposite is true during bull markets...any portfolio that holds bonds will keep the portfolio from doing as well as the total stock market. It's a tradeoff:

If stocks lose 50% and bonds don't lose (or gain) anything:

100% stock loses 50%
90/10 loses 45%

If stock market goes up 50% and bonds don't lose (or gain anything):

100% stock gains 50%
90/10 gains 45%

Bonds keep your whole portfolio from doing as badly as the stock market in down years, but then has to keep your whole portfolio from doing as great as the stock market in up years. Another way of saying it is "bonds limit your downside...but they also limit your upside". How much upside are you willing to give up in order to limit the downside risk?

Here's an interactive vanguard tool that shows the different gains/losses over time using 100% vs 90/10, etc.:

https://personal.vanguard.com/us/insigh ... about-risk
It's hard to accept the truth when the lies were exactly what you wanted to hear. Investing is simple, but not easy. Buy, hold & rebalance low cost index funds & manage taxable events. Asking Portfolio Questions | Wiki
User avatar
Kevin M
Posts: 15750
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: Why bonds when one's investment horizon is long?

Post by Kevin M »

TBillT wrote:
runner540 wrote:
TBillT wrote:Cap gains is one reason.
I believe 30-yr Treasuries have out-performed S&P 500 by 5.5x since 1980.
<snip>
TBillT, cam you please explain/show your calculations for Tbills outperforming S&P 500?
If you know me, I am quoting preeminent American economist A. Gary Shilling.
https://seekingalpha.com/article/403683 ... lling-says <no it is still alive!>
We can approximate the return of 30-year Treasuries using 30-year yields from FRED to calculate annual returns. https://fred.stlouisfed.org/graph/?g=ecZr.

First, a few clarifying comments.

The 15.2% yield mentioned in the article was the yield on 9/30/1981 and 10/26/1981--these were the highest daily yields ever reached. On 1/2/1981 it was 12.01%, and on 12/31/1981 it was 13.65%. I'll use annual average yields in my calculations. The average yield for 1981 was 13.45%.

If you hold a 30-year zero-coupon bond until maturity, you earn exactly the initial yield. All of the return comes from capital gain. For example, the price of a 30-year bond with a yield of 15% is 1.51 (par = price at maturity = 100), and the annualized return is 15% =(100/1.51)^(1/30)-1.

A 30-year bond bought in 1981 would have matured in 2011, so you obviously cannot compute 35-year return for a 30-year bond held to maturity. I think what Shilling is talking about is rolling 30-year bonds monthly or annually. I'll do the calculation for rolling a 30-year zero-coupon bond annually.

Annual average yield data is not available for 2003-2005, because the 30-year constant maturity series was discontinued on February 18, 2002, and reintroduced on February 9, 2006. For these years I just use the yield from 2002 (so return = yield), which is pretty close to the return from holding the bond for the additional three years then resuming the annual rolling.

The one-year return for holding a zero-coupon 30-year Treasury is = (1+Y30)^30/(1+Y29)^29-1. I assume a flat yield curve, which is reasonable since the yield curve typically is pretty flat at the long end. So for the return for 1981-1982, I take the 1981 yield = Y30 = 13.45%, and the 1982 yield = Y29 = 12.76%, and the return = (1+13.45%)^30/(1+12.76%)^29-1 = 35.41%.

For the 35-year period 1981-2016, the annualized return (geometric mean) is 16%, and the growth factor is 180. I calculated the geometric mean by applying the GEOMEAN function to the annual returns. The growth factor can be computed directly using the PRODUCT function, or as (1 + 16%) ^ 35.

We can get the returns for the S&P 500 from Portfolio Visualizer asset class returns (during the period of interest Vanguard 500 Index is used). Backtest Portfolio Asset Class Allocation. We see that the annualized return is 11.3%, and the growth factor is 42.3.

I think it's the ratio of growth factors (cumulative growth) that the article is referencing. The ratio of growth factors is 180.2 / 42.3 = 4.3. So the 30-year Treasury did indeed have a significantly higher return than the S&P 500, but perhaps not quite as large as indicated by the article. Perhaps if you cherry pick the start date when the yield was at its peak of 15.2%, you'd get the higher number.

Note that in the PV chart linked above I included the long-term Treasury asset class, for which the annualized return was 9.85% with growth factor of 26.75, so much lower than what I've calculated for 30-year Treasuries. PV uses 30-year Treasury yields for 1978-1986, but uses the Vanguard long-term Treasury fund for 1987+. The long-term Treasury fund currently has an average duration of about 17 years, while a 30-year zero-coupon Treasury has a duration of 30 years, and I think this explains the lower return of the fund in the falling-yield environment.

30-year zero-coupon bonds are potentially high-return securities, but like stocks, they also are very high-risk. The standard deviation of the 30-year zero annual returns is 30%, compared to 15% for the S&P 500 over the time period being discussed.

The returns of the last 35 years, starting from yields in the 12-15% range, cannot be achieved over the next 35 years, with the yield now at about 2.7%. A 30-year zero-coupon bond bought today and held to maturity will earn 2.7%.

Kevin
If I make a calculation error, #Cruncher probably will let me know.
dbr
Posts: 46137
Joined: Sun Mar 04, 2007 8:50 am

Re: Why bonds when one's investment horizon is long?

Post by dbr »

I guess if you buy a 30 year bond exactly at the top of an extreme outlier in over 100 years of US interest rates and compute the growth factor to maturity of the bond, then you have a pretty good chance of outperforming just about anything. I wonder if people appreciate how extreme and uncharacterizable the history of US interest rates over the last hundred years really is: https://www.google.com/imgres?imgurl=ht ... A..i&w=300 and http://www.cnbc.com/2016/11/17/200-year ... chart.html
User avatar
Kevin M
Posts: 15750
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: Why bonds when one's investment horizon is long?

Post by Kevin M »

Kevin M wrote: Note that in the PV chart linked above I included the long-term Treasury asset class, for which the annualized return was 9.85% with growth factor of 26.75, so much lower than what I've calculated for 30-year Treasuries. PV uses 30-year Treasury yields for 1978-1986, but uses the Vanguard long-term Treasury fund for 1987+. The long-term Treasury fund currently has an average duration of about 17 years, while a 30-year zero-coupon Treasury has a duration of 30 years, and I think this explains the lower return of the fund in the falling-yield environment.
I did some research on this. I think PV is using year-end returns, not annual average returns, and they are assuming a coupon bond, not a zero-coupon bond. I think they may also be doing a little yield curve tweaking rather than assuming a flat yield curve, but this doesn't have a huge impact.

I get annual returns for 1982-1986 (when PV uses FRED data) very close to PV by using year-end values and assuming par bonds (coupon = yield upon purchase) with a flat yield curve.

Par bonds had much lower returns than zero-coupon bonds for 1981-2016, whether I use annual average or year-end returns. Using annual average returns, the average annual return for par bonds is 10.4% and the growth factor is 31.9. Using end of period data, annual average return is 9.65% and the growth factor is 25.2, so very close to the PV value for this period. Either way, lower returns than the S&P 500, and much lower than a 30-year zero-coupon bond.

As expected, standard deviation also is much less with par (coupon) bonds.

Kevin
If I make a calculation error, #Cruncher probably will let me know.
TBillT
Posts: 1001
Joined: Sat Sep 17, 2011 1:43 pm

Re: Why bonds when one's investment horizon is long?

Post by TBillT »

Kevin - Thank you for the analysis. The point Shilling would make if he buys a zero coupon bond today at 2.8% and sells at 2% yield, he gets a large cap gain. As an investment manager, he has no "interest" in the yield, just the potential cap gain. Of course, Shilling is making money by predicting the interest rate direction (generally correctly). Sort of the antithesis of Bogle passive approach.
Post Reply