Bonds beat stocks over the last 15 years

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
alexfrey
Posts: 33
Joined: Tue Oct 29, 2013 6:02 pm
Location: San Francisco
Contact:

Re: Bonds beat stocks over the last 15 years

Post by alexfrey » Fri Jul 25, 2014 4:35 pm

Among other things, this example is also illustrative of the importance of initial conditions.

At the beginning of 1999, stocks were trading at a P/E of 33, a Shiller PE / CAPE / PE10 of 40 and a dividend yield of 1.3%, while ten year bonds yielded 4.7%. There are a number of ways to look at these numbers to gauge the relative attractiveness of bonds and stocks in 1999, here's two:
  • Taking the 'shiller earnings yield' as 1 / Shiller PE stocks had a 2.5% cyclically-adjusted earnings yield in 1999. Bonds yielded more than 2% above that. Compare the situation to today, where the Shiller earnings yield is ~3.9% and 10 year bonds ciome in at 2.5%, a +1.4% differential in favor of stocks. Stocks were very unattractive relative to bonds in 99, so it's no wonder they "lost."
  • Take the dividend yield and add the expected growth of dividends (let's say 4.5%), and you get about 5.8% potential return for stocks in 99, vs. 4.7% for bonds, assumming no changes in valuation. So that's a 1% risk premium, vs. historical average of about 6% (which is roughly what you get if you run the numbers at a 16x Shiller PE)
Point is, bonds can definitely outperform stocks over 15 years. But this is not a random occurence that is equally likely to occur over any 15 year period. Its heavily influenced by initial conditions, and the initial conditions in 99 were weighted against stocks.

berntson
Posts: 1366
Joined: Mon Oct 29, 2012 12:10 pm

Re: Bonds beat stocks over the last 15 years

Post by berntson » Fri Jul 25, 2014 4:48 pm

Kevin M wrote:
berntson wrote:I've always been puzzled by the idea that investors should invest in bonds because they "might" beat stocks over a long period of time.
I'm puzzled by framing it this way. I don't have a high allocation to bonds (well, really a lot more CDs) because they might beat stocks. I do so because I don't need to take the risk of a higher stock allocation.

Investing in stocks is a great way to have a shot at the brass ring. Investing in fixed income is a great way to keep the brass ring once you've won it. The balance between the two depends to some extent on where you are on your path.

As always, it's a matter of ability, willingness and need to take risk. This is one of the most important lessons I learned from Larry Swedroe about a decade ago.

Another thought: are you also puzzled that people invest only in large-cap blend because large might beat small and growth might beat value over a long period of time (not saying you do so, since I don't think you do).
Thanks Kevin for your thoughts. I understand using bonds to smooth out short-term volatility. Bonds are good for emergency funds and for investors who need to draw a stream of income from their portfolios. They're also good for delivering a smoother ride for investors. What I don't understand is holding bonds because they might beat stocks over the long-term. They might, but I don't see good historical precedent for their beating stocks by very much.

That's a nice analogy with small and value. I guess when it comes to small and value, I think there's always the risk that the premiums just disappear. So holding large stocks and growth stocks can make good sense, even for investors with a long time frame. I don't think that the equity premium will every be arbitraged away, even if it continues to shrink over time.

Johno
Posts: 1883
Joined: Sat May 24, 2014 4:14 pm

Re: Bonds beat stocks over the last 15 years

Post by Johno » Fri Jul 25, 2014 4:52 pm

denovo wrote:5.5 percent combined isn't great, but it's better than the mattress. Then again, real estate as measured by the Vanguard REIT Index was up about 11 percent a year annualized.
Emerging Markets also did a lot better over that period, eyeballing graph it's broadly similar for VEIEX to the REIT fund, maybe somebody has the actual number. However in the same general vein eyeballing a graph of developed market fund like VTMGX, that underperformed US stocks as well as US (high grade) bonds in the period (Vanguard's investor share junk fund VWEHX 15yr return was 6.26%).

I believe the point is just a qualitative historical one: it's not guaranteed that stocks will outperform bonds over even extended periods, for anyone who didn't already realize that. And it could be true 15yrs from now. The fact that bond yields are so low doesn't mean it couldn't happen, it just means that if bond expected return=yield, then we'll be sadder earning 2.5% on bond and stocks for the next 15yrs than we were earning 5 something % for the last 15yrs, to the degree that comparison actually applies to one's portfolio (there are probably not that many people who invested almost all their money just in 1999, besides the fact that many people have assets other than US TSM and high grade bonds).
Last edited by Johno on Fri Jul 25, 2014 5:17 pm, edited 1 time in total.

User avatar
Kevin M
Posts: 10211
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: Bonds beat stocks over the last 15 years

Post by Kevin M » Fri Jul 25, 2014 5:14 pm

berntson wrote: Thanks Kevin for your thoughts. I understand using bonds to smooth out short-term volatility. Bonds are good for emergency funds and for investors who need to draw a stream of income from their portfolios. They're also good for delivering a smoother ride for investors. What I don't understand is holding bonds because they might beat stocks over the long-term. They might, but I don't see good historical precedent for their beating stocks by very much.
All of this still misses the point. Fixed income returns are less uncertain over any time period, not just short time periods. Thus if you apply the standard finance definition of risk as uncertainty of return, fixed income is less risky over any time period, long or short. Again, I don't hold fixed income because it might beat stocks over my remaining lifetime, but because it is more certain to meet my needs. I'm not just looking for a smoother ride, but one more certain of reaching my destination.

Sure, I probably would die with more money if I had a higher stock allocation, but I almost certainly will not run out of money before I die with a 30/70 stock/fixed portfolio. I probably would be fine with a 0/100 portfolio, but owning stocks adds some spice to life that probably will be a benefit, and also probably won't hurt me too much if the next 25 years happen to fall into the 5%-10% probability zone of under-performing the "risk free" return.
berntson wrote:That's a nice analogy with small and value. I guess when it comes to small and value, I think there's always the risk that the premiums just disappear. So holding large stocks and growth stocks can make good sense, even for investors with a long time frame. I don't think that the equity premium will every be arbitraged away, even if it continues to shrink over time.
Fair enough. If you just look at the historical data and the statistics, as in the Fama French paper linked by Robert T., the case for the small and value premiums seems about as convincing as for the equity risk premium. How good the stories are that support the premiums is another question. Clearly people don't agree on the stories.

Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)

berntson
Posts: 1366
Joined: Mon Oct 29, 2012 12:10 pm

Re: Bonds beat stocks over the last 15 years

Post by berntson » Fri Jul 25, 2014 5:44 pm

Kevin M wrote: All of this still misses the point. Fixed income returns are less uncertain over any time period, not just short time periods.
The return of a stock portfolio are more uncertain over long time periods mostly because there is more upside uncertainty, i.e. there is a huge range of possibilities for how much stocks will (say) beat the risk-free rate. I don't see that as a meaningful kind of risk. (A toy example. You offer me a choice between two bets. The first gives me a million if a coin lands heads and two million if lands tails. The second gives me a million dollars either way. The first bet has more dispersion (more uncertainty) but is no more risky than the second bet.)

What sort of risk does matter? Well, that will depend on an investor's goals. One kind of risk is failing to keep up with the risk-free rate. Looking at the linked Fama-French paper, the odds of that happening clearly go down over longer time periods.

Think about it this way. Suppose you know that you're going to sleep for 30 years (ala Rip Van Winkle). What sort of portfolio do you hold? Me, I would just put it in stocks and go take my long nap. No sense putting it in bonds because bonds "might" eke out a small win over those 30 years. Since I'll be sleeping, I also won't need the money and won't have to watch the volatility. On the other hand, real investors don't sleep for 30 years. They might need the money and they have to actually live through the volatility. So real investors may well want some bonds.

So that's another way to put my position. I don't know why Rip Van Winkle would want any bonds. :beer

User avatar
Robert T
Posts: 2534
Joined: Tue Feb 27, 2007 9:40 pm
Location: 1, 0.2, 0.4, 0.5
Contact:

Re: Bonds beat stocks over the last 15 years

Post by Robert T » Fri Jul 25, 2014 6:20 pm

.
The earlier linked Fama-French paper indicates that a 15 year period of underperformance of bonds over stocks should not be too surprising - as there is a non-zero probability that this can happen,even over longer time horizons. And as we have seen from history it has happened in the US three times in separate (non-overlapping) time periods over the last 100 years. The same is true of the size and value premiums - that there is a non-zero probability that they could be negative for long periods of time.

At least for me, data of underperformance over long periods (eg 15 yrs) of small cap vs. large cap, and value vs. growth are not evidence that the longer term premiums have disappeared (as seems to be sometimes suggested) - just as the 15 years of underperformance of stocks vs. bonds it not evidence to me that that stocks will not outperform over the longer term. There is a non-zero probability of underperformance even over a 50 year time horizon (as per the Fama-French linked paper), but also a fairly high probability that the premiums will be positive.

In finance there are no guarantees, only probabilities, and you go where those are highest (Bernstein quote).

FWIW - I have a 75% stock allocation with a small cap and value tilt. Periods of underperformance should be expected, and not be surprising (and should not lead to conclusions that all these premiums have now disappeared so I should change my allocation) - but should reinforce the need to stick to an allocation over the longer term - this is well illustrated in the linked Fama-French paper. If there is any action to be taken, it should rather be to save more.

Robert
.

berntson
Posts: 1366
Joined: Mon Oct 29, 2012 12:10 pm

Re: Bonds beat stocks over the last 15 years

Post by berntson » Fri Jul 25, 2014 6:24 pm

Robert T wrote: At least for me, data of underperformance over long periods (eg 15 yrs) of small cap vs. large cap, and value vs. growth are not evidence that the longer term premiums have disappeared (as seems to be sometimes suggested) - just as the 15 years of underperformance of stocks vs. bonds it not evidence to me that that stocks will not outperform over the longer term. There is a non-zero probability of underperformance even over a 50 year time horizon (as per the Fama-French linked paper), but also a fairly high probability that the premiums will be positive.
I like this way of putting the point. Even when it comes to stocks and bonds, long periods of underperformance are not only possible but not all that improbable. So investors should not see such underperformance as reason to abandon well thought out strategies.

User avatar
Kevin M
Posts: 10211
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: Bonds beat stocks over the last 15 years

Post by Kevin M » Fri Jul 25, 2014 7:03 pm

berntson wrote: The return of a stock portfolio are more uncertain over long time periods mostly because there is more upside uncertainty, i.e. there is a huge range of possibilities for how much stocks will (say) beat the risk-free rate.
Although it's true that historically the right tail has been fatter than the left tail, at least for US stocks, that doesn't mean it had to be that way, or that it will be that way in the future. Incidentally, the right tail has also been fatter for small-value stocks, but my skepticism about the future resembling that past is what keeps me from holding "the Larry portfolio", although I do tilt to small and value.
berntson wrote:One kind of risk is failing to keep up with the risk-free rate. Looking at the linked Fama-French paper, the odds of that happening clearly go down over longer time periods.
True, but according to FF the odds of it happening are still way above zero, which is a direct contradiction of your point about the only risk being by how much stocks are likely to beat the risk-free rate.

I think for most people the marginal utility of wealth decreases as wealth increases. Once you have enough wealth to last the rest of your lifetime, be it 20 years or 30 years or more, without taking significant risk, why take the risk, even if it is small? On the other hand, one could be so wealthy that having 100% stocks also is fine (think Bill Gates, Warren Buffet), but that ain't me.

Here's another thought: if stocks were not risky over a 30-year period, why would anyone buy a 30-year treasury (be it nominal or TIPS)? I think that by definition stocks must be risky over any time period for there to be an equity risk premium.

Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)

berntson
Posts: 1366
Joined: Mon Oct 29, 2012 12:10 pm

Re: Bonds beat stocks over the last 15 years

Post by berntson » Fri Jul 25, 2014 7:29 pm

Kevin M wrote: Although it's true that historically the right tail has been fatter than the left tail, at least for US stocks, that doesn't mean it had to be that way, or that it will be that way in the future.
That's fair. I'm beginning to think that different attitudes towards equity risk have a lot to do with whether an investor takes the last one hundred some odd years to give us a reasonable sense of the sorts of risks faced by investors. Given that there were two world wars, a cold war, and a global depression, the last century would seem to have had its fair share of tail risk. And long-term equity investors (with reasonable diversification) did fine.
Kevin M wrote: True, but according to FF the odds of it happening are still way above zero, which is a direct contradiction of your point about the only risk being by how much stocks are likely to beat the risk-free rate.
That wasn't quite what I said. My claim was that most of the extra volatility over long time periods comes from upward volatility. I'm sure there's still downward volatility, but I'm not convinced that it's less than the volatility for nominal bonds. Maybe inflation adjusted bonds.

I think for most people the marginal utility of wealth decreases as wealth increases. Once you have enough wealth to last the rest of your lifetime, be it 20 years or 30 years or more, without taking significant risk, why take the risk, even if it is small? On the other hand, one could be so wealthy that having 100% stocks also is fine (think Bill Gates, Warren Buffet), but that ain't me.
Kevin M wrote: Here's another thought: if stocks were not risky over a 30-year period, why would anyone buy a 30-year treasury (be it nominal or TIPS)? I think that by definition stocks must be risky over any time period for there to be an equity risk premium.
This is a great question. I'll have to think about this more, since I'm not sure that I have a good answer. One thought is that 30-year bonds are useful for hedging short-term volatility because they get a bigger "bounce" when interest rates drop. I've considered owning 30-year treasuries myself for precisely this purpose. But if I was in a Rip Van Winkle sort of situation (going to sleep for 30 years with no need for short-term stability), I wouldn't consider buying them. The odds of equities significantly trailing 30-year treasuries over the next 30 years just isn't that high in my expectation.

I think we may actually mostly agree. I agree that investors who have "made it" and are drawing income off their portfolio should have bonds (and possibly a lot of them) because they have lots of short term (less than 20 year) risk. If an investor has "made it", but is certain she won't use the money for another 30 years, bonds would play a much less important role.

User avatar
Kevin M
Posts: 10211
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: Bonds beat stocks over the last 15 years

Post by Kevin M » Fri Jul 25, 2014 8:03 pm

berntson wrote: Given that there were two world wars, a cold war, and a global depression, the last century would seem to have had its fair share of tail risk. And long-term equity investors (with reasonable diversification) did fine.
It depends on what you mean by reasonable diversification. True for a globally diversified portfolio, but not true for a number of individual countries. Most investors have very strong home bias, which tends to decrease global diversification. US stocks were one of the big winners over the last century; may not be the same over the next century.

Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)

berntson
Posts: 1366
Joined: Mon Oct 29, 2012 12:10 pm

Re: Bonds beat stocks over the last 15 years

Post by berntson » Fri Jul 25, 2014 8:39 pm

Kevin M wrote:
berntson wrote: Given that there were two world wars, a cold war, and a global depression, the last century would seem to have had its fair share of tail risk. And long-term equity investors (with reasonable diversification) did fine.
It depends on what you mean by reasonable diversification. True for a globally diversified portfolio, but not true for a number of individual countries. Most investors have very strong home bias, which tends to decrease global diversification. US stocks were one of the big winners over the last century; may not be the same over the next century.
Agreed. A nice argument for holding international equity at (more or less) market weight!

User avatar
Robert T
Posts: 2534
Joined: Tue Feb 27, 2007 9:40 pm
Location: 1, 0.2, 0.4, 0.5
Contact:

Re: Bonds beat stocks over the last 15 years

Post by Robert T » Fri Jul 25, 2014 9:03 pm

Kevin M wrote: Here's another thought: if stocks were not risky over a 30-year period, why would anyone buy a 30-year treasury (be it nominal or TIPS)? I think that by definition stocks must be risky over any time period for there to be an equity risk premium.
Reminds me of Fama-French response to question on long-term government bonds outperforming the S&P500 for the 40 year period ending March 2009.
  • " ... Over any finite horizon, no matter how long, there is a positive probability that the [equities over bonds] premium will be negative. Without this long-term uncertainty, we would see simple arbitrage opportunities between stocks and bonds. Suppose, for example, we all know stock will beat bonds over every 40 year period. If stock lose to bonds over the next 39 years, who would buy bonds in the 40th year? In fact, in that case we could make as much money as we want in the 40th year with no risk by simply shorting bonds and using proceeds to invest in stocks."
Robert
.

User avatar
Kevin M
Posts: 10211
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: Bonds beat stocks over the last 15 years

Post by Kevin M » Fri Jul 25, 2014 10:28 pm

Robert T wrote: The earlier linked Fama-French paper indicates that a 15 year period of underperformance of bonds over stocks should not be too surprising - as there is a non-zero probability that this can happen,even over longer time horizons. And as we have seen from history it has happened in the US three times in separate (non-overlapping) time periods over the last 100 years. The same is true of the size and value premiums - that there is a non-zero probability that they could be negative for long periods of time.
Thought it would be good to get the 15-year probabilities of risk premium under-performance from the FF paper into this thread:

Equity: 15%
Size: 19%
Value: 10%

Note that if one makes decisions based on historical results, at least as represented by the 49-year period covered in the FF paper (1963-2011), there is less chance of underperformce in the value premium than in the equity premium, but the size premium is somewhat riskier.

Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)

HenryPorter
Posts: 161
Joined: Sat Dec 05, 2009 2:09 am

Re: Bonds beat stocks over the last 15 years

Post by HenryPorter » Fri Jul 25, 2014 10:50 pm

Ran these calcs:

http://www.buyupside.com/calculators/do ... te+Results

http://www.buyupside.com/calculators/do ... te+Results


I ran the same funds with just a lump sum of $5000 and the bond fund aced the stock fund by about $200 over that 15 years. I think a safer investment methodology when we have market highs is to DCA instead of lumpsum and maybe add extra when major dips happen. We had some major dips since 1999. If that is market timing, OK.

Tamales
Posts: 1297
Joined: Sat Jul 05, 2014 10:47 am

Re: Bonds beat stocks over the last 15 years

Post by Tamales » Sat Jul 26, 2014 12:09 am

berntson wrote:My claim was that most of the extra volatility over long time periods comes from upward volatility. I'm sure there's still downward volatility, but I'm not convinced that it's less than the volatility for nominal bonds. Maybe inflation adjusted bonds.
I guess you could use something like the Ulcer Index (or Martin Ratio) to get more focus on the downside.
http://www.tangotools.com/ui/ui.htm

Stockcharts.com seems to have the ulcer index available as an indicator under SharpCharts, but I haven't played with it yet.

User avatar
grayfox
Posts: 4936
Joined: Sat Sep 15, 2007 4:30 am

Re: Bonds beat stocks over the last 15 years

Post by grayfox » Sat Jul 26, 2014 2:33 am

Robert T wrote:.
Annualized return 15 years to 7/23/2014

Vanguard total stock market (VTSMX) = 5.25%
Vanguard total bond market (VBMFX) = 5.37%

The equity premium can be negative for long periods of time (as can the value, size, and I would argue, momentum premium). The benefit from exposure to these factors comes from long-term discipline to stay the course.

Obviously no guarantees going forward.

Robert
.
It happens.

I can look up the price of a 15-year Treasury.
http://online.wsj.com/mdc/public/page/2 ... nav_2_3020

2029 May 15 STRIPS has price 63.434/63.526 bid/ask. So if I put $63,434 into STRIPS I will have $100,000 on May 15, 2029. More or less guaranteed.

If I put the same amount, $63,434 into some stock index fund, let's say VTWSX Total World Stock Index and re-invest distributions, what is the chance that I will have less than $100,000 on May 15, 2029? (i.e. default-free bond beats stocks)

Certainly not zero. Probably somewhere around 10% to 20% chance.
80% or 90% chance I will have more than $100,000, so it's a pretty good bet, but no way is it a sure thing.

There is even some chance that you end up with less than the initial $63,434. It can happen.

All you can say about $63,000 world stock index investment for 15 years is there is some distribution of possible outcomes. Maybe the outcome can range from $50,000 to $250.000. Contrast with the Treasury bond where the outcome is exactly $100,000.

User avatar
packer16
Posts: 1045
Joined: Sat Jan 04, 2014 2:28 pm

Re: Bonds beat stocks over the last 15 years

Post by packer16 » Sat Jul 26, 2014 6:35 am

Kevin M wrote:
Robert T wrote: The earlier linked Fama-French paper indicates that a 15 year period of underperformance of bonds over stocks should not be too surprising - as there is a non-zero probability that this can happen,even over longer time horizons. And as we have seen from history it has happened in the US three times in separate (non-overlapping) time periods over the last 100 years. The same is true of the size and value premiums - that there is a non-zero probability that they could be negative for long periods of time.
Thought it would be good to get the 15-year probabilities of risk premium under-performance from the FF paper into this thread:

Equity: 15%
Size: 19%
Value: 10%

Note that if one makes decisions based on historical results, at least as represented by the 49-year period covered in the FF paper (1963-2011), there is less chance of underperformce in the value premium than in the equity premium, but the size premium is somewhat riskier.

Kevin
If you extend the period out to 20 or 30 years over the period 1871 to 2012 the probability of stocks beating bonds goes up to 96% and 99% respectively. For a 10 year period it is closer to 78%. These are from Siegel's latest stocks for the long term. What is also interesting is the st dev of real returns for periods over 15 year is less for stocks than it is for bonds. Which implies that on a real basis stocks may be less risky than bonds (ie. that stocks are better at protecting holders against the largest long term risk (inflation)).

Packer
Buy cheap and something good might happen

berntson
Posts: 1366
Joined: Mon Oct 29, 2012 12:10 pm

Re: Bonds beat stocks over the last 15 years

Post by berntson » Sat Jul 26, 2014 12:00 pm

packer16 wrote: If you extend the period out to 20 or 30 years over the period 1871 to 2012 the probability of stocks beating bonds goes up to 96% and 99% respectively. For a 10 year period it is closer to 78%. These are from Siegel's latest stocks for the long term. What is also interesting is the st dev of real returns for periods over 15 year is less for stocks than it is for bonds. Which implies that on a real basis stocks may be less risky than bonds (ie. that stocks are better at protecting holders against the largest long term risk (inflation)).
I'll post the chart packer16 is referring to incase anyone is interested. This is data up to 2006, so it doesn't include the recent outperformance of bonds. With eight more years of data, we now know that the 100% at the bottom of the chart isn't really 100%. Extraordinary things happen ever so often. :beer

[Edit: Looks like I have an older version of the book....but this is the gist of it.]

Image

User avatar
Kevin M
Posts: 10211
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: Bonds beat stocks over the last 15 years

Post by Kevin M » Sat Jul 26, 2014 12:12 pm

berntson wrote: My claim was that most of the extra volatility over long time periods comes from upward volatility. I'm sure there's still downward volatility, but I'm not convinced that it's less than the volatility for nominal bonds. Maybe inflation adjusted bonds.
I wanted to get back to this point. My concern is not so much volatility, which I think brings to mind uncertainty of returns over shorter periods, like a month or a year, but uncertainty of return over a long holding period. This is illustrated in this chart that I shared in another recent thread on the long-term risk of stocks. It is from this excellent paper by forum member John Norstad: Risk and Time, which I think would be well worth reading.

Image

Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)

User avatar
Kevin M
Posts: 10211
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: Bonds beat stocks over the last 15 years

Post by Kevin M » Sat Jul 26, 2014 12:47 pm

Here's the table from the Fama-French paper that shows their calculated values for the probability of the three risk premiums underperforming bills. The fourth row for each premium (e.g., Prob < 0, E(RM-RF) uncertain) includes the estimate of the uncertainty of the true expected value, as well as the uncertainty due to variance of investment outcomes; it's the more realistic estimate.

Keep in mind that these numbers are calculated statistics based on a single 49 year period (1963-2011)--not actual historical results. Obviously we don't have enough independent 50-year periods for valid statistical analysis of actual 50-year periods.

Table 2: Probability that an average premium is negative in periods of different length
Image

Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)

User avatar
ginmqi
Posts: 243
Joined: Sat Feb 22, 2014 4:35 pm

Re: Bonds beat stocks over the last 15 years

Post by ginmqi » Sat Jul 26, 2014 12:51 pm

Kevin M wrote:I wanted to get back to this point. My concern is not so much volatility, which I think brings to mind uncertainty of returns over shorter periods, like a month or a year, but uncertainty of return over a long holding period. This is illustrated in this chart that I shared in another recent thread on the long-term risk of stocks. It is from this excellent paper by forum member John Norstad: Risk and Time, which I think would be well worth reading. [/url]

Kevin
I've come across that piece before and glad I am revisiting it. So if it's true that there is in fact no time diversification, and in fact this is one of the most pervasive things being told across all forms of media: that young investors should be more into equity than elderly investors.

If that is actually a fallacy, as Norstad argues. Then what is the practical, real-world decision that would impact actual real world portfolios? Should we go 50/50 then?

Or is this simply another academic exercise to mathematically show, basically, that there are in fact more risks than meets the eye and so investors should simply tread more carefully and be less in equities? But then with bonds we are trading equity risk with all the risks that come with bonds (credit/call/interest rate/inflation/re-investment). And then how would this impact real world investment decisions?

Ugh, I feel like there are always "truths" in investment..and then there's always a paper/academic to show why those "truths" are actually false. :|

As a young investor who currently subscribes to the mantra of being aggressive in equity when I am young...this is quite concerning. Should I change my IPS and dramatically reduce my equity???

ks289
Posts: 626
Joined: Sun Mar 11, 2012 12:42 pm

Re: Bonds beat stocks over the last 15 years

Post by ks289 » Sat Jul 26, 2014 1:08 pm

Kevin M wrote:
berntson wrote: My claim was that most of the extra volatility over long time periods comes from upward volatility. I'm sure there's still downward volatility, but I'm not convinced that it's less than the volatility for nominal bonds. Maybe inflation adjusted bonds.
I wanted to get back to this point. My concern is not so much volatility, which I think brings to mind uncertainty of returns over shorter periods, like a month or a year, but uncertainty of return over a long holding period. This is illustrated in this chart that I shared in another recent thread on the long-term risk of stocks. It is from this excellent paper by forum member John Norstad: Risk and Time, which I think would be well worth reading.

Image

Kevin
Great analysis in the article. I still believe that the take home message was that despite countering the dogma that risks decrease with time, human capital allows those with a long time horizon to accept the risk of investing aggressively in stocks. Essentially the end result recommendation is the same for most stock investors whether we say risks decrease or increase with time -the human capital element is key.
I also felt that the labeling on the chart overstated the risk argument since the blue area (dog food section) includes the risk free 6% bank account which ended up 1.26 standard deviations below the mean.

Tamales
Posts: 1297
Joined: Sat Jul 05, 2014 10:47 am

Re: Bonds beat stocks over the last 15 years

Post by Tamales » Sat Jul 26, 2014 1:10 pm

Kevin, re: "...uncertainty of return over a long holding period. This is illustrated in this chart that I shared in another recent thread on the long-term risk of stocks..."

I might be off-point, but can't you see essentially the effect of increasing risk with time via the monte carlo simulator tool on portfolio visualizer, where the max and min lines are expanding in a funnel shape, outward? But also, you can affect the slopes and crossover points pretty significantly with different portfolio compositions, and arguably a portfolio which stretches the blue "min balance" line out as far as possible (which represents the "bad side" of risk) is one of the more desired attributes?

User avatar
Kevin M
Posts: 10211
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: Bonds beat stocks over the last 15 years

Post by Kevin M » Sat Jul 26, 2014 1:44 pm

ginmqi wrote: If that is actually a fallacy, as Norstad argues. Then what is the practical, real-world decision that would impact actual real world portfolios? Should we go 50/50 then?
<snip>
As a young investor who currently subscribes to the mantra of being aggressive in equity when I am young...this is quite concerning. Should I change my IPS and dramatically reduce my equity???
The answer is what ks289 mentioned: the ratio of your human capital to financial capital is huge when you are young, which justifies taking more equity risk. John's paper discusses this, but here's a blog post I wrote this topic: Human Capital, Financial Capital.

Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)

User avatar
Kevin M
Posts: 10211
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: Bonds beat stocks over the last 15 years

Post by Kevin M » Sat Jul 26, 2014 1:53 pm

ginmqi wrote: As a young investor who currently subscribes to the mantra of being aggressive in equity when I am young...this is quite concerning. Should I change my IPS and dramatically reduce my equity???
I want to add that I advise my adult children in their late 20s to be 80% in stocks, and they are. My equity allocation is 30%. My 85 year old Mom's is 20%.

It comes down to ability, willingness, and need to take risk. Having lots of human capital increases ability to take risk, and most young folks with little financial capital need to take risk to have a shot at a decent retirement. That leaves willingness to take risk as the moderating factor (i.e., will you stick with your plan when stocks drop 50% or more?).

Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)

User avatar
ginmqi
Posts: 243
Joined: Sat Feb 22, 2014 4:35 pm

Re: Bonds beat stocks over the last 15 years

Post by ginmqi » Sat Jul 26, 2014 2:06 pm

Kevin M wrote:
ginmqi wrote: As a young investor who currently subscribes to the mantra of being aggressive in equity when I am young...this is quite concerning. Should I change my IPS and dramatically reduce my equity???
I want to add that I advise my adult children in their late 20s to be 80% in stocks, and they are. My equity allocation is 30%. My 85 year old Mom's is 20%.

It comes down to ability, willingness, and need to take risk. Having lots of human capital increases ability to take risk, and most young folks with little financial capital need to take risk to have a shot at a decent retirement. That leaves willingness to take risk as the moderating factor (i.e., will you stick with your plan when stocks drop 50% or more?).

Kevin
Great, I think I can look at the issue better now. Thanks for the insight. The human capital is definitely a large part. Being that I am a 28yo going into a fairly job-secure employment field with above average income and I am a very prodigious saver, I am not as panicky as in the prior post. I am definitely a believer in the steadfast, hold out for the long run investor. A stock market "correction" of 50% or more in my early years would simply let me know that equity is on sale and it's time to buy buy buy :mrgreen:

I am still only 100% equity simply because of the fact that I just started investing last year (started my career and hence my income stream last year, was in school for all the years prior) and the fact that I have not read/understand enough about bonds to put my money in them yet (as Buffett says...only invest in things you understand).

And IMO, bonds are quite a bit more complex to understand and invest in than equity but it's always not nearly discussed as much by a cursory view of the forum thread topics. I've gone through several recommended general investment books (Bernstein, Swensen, Bogleheads guide) and now have recently finished The Bond Book, and am currently going through a couple other bond books. And hopefully as my mind feel comfortable grasping and explaining bonds then I will start to move into bonds...hopefully very soon.

Another problem with the title of this thread is that Vanguard total bond market index is not and should not be used synonymously with "bonds." Since a bond mutual fund, IMO, is quite a different animal than holding individual bonds and even within bonds there are so many types that it would be more useful to compare different bond assets (mutual fund, ladders, bar bells, corporates, munis, US treasuries, different durations, credit ratings, etc. etc.) to equity returns.

User avatar
jimb_fromATL
Posts: 2278
Joined: Sun Nov 10, 2013 12:00 pm
Location: Atlanta area & Piedmont Triad NC and Interstate 85 in between.

Re: Bonds beat stocks over the last 15 years

Post by jimb_fromATL » Sat Jul 26, 2014 2:11 pm

MathWizard wrote:Picking the starting date in 1999 , so that this includes
the 2001 crash, the 2003 crash and the 2008/2009 crash but does not
include the huge run-up in equities in the late 1990's certainly
makes equities look bad, but not by much.
Those are also for a lump sum invested just before the crashes, and do not reflect Dollar-Cost-Averaging with periodic contributions the way most people invest. I don't know of anyone who just invested a lump sum then never made any more contributions to their 401(k) or IRA for 15 years or more.
Looks like stocks are better at each point above.
... and they're better for the 15 years from 1999-2013 for dollar-cost-averaging instead of a single lump sum.
Yes, there will be times when bonds beat stocks, but I actually think that is a good thing.
If stocks were always better, nobody would ever buy bonds, and there would be no rebalancing opportunities.
Ged wrote:
Man I'm glad I didn't do a lump sum of all of my money in 1999.
Dollar-cost-averaging with a balance of stocks and bonds can be the best of both worlds. DCA did better than a lump sum for both those funds, and during that 15 years a good balanced fund far outperformed both of ‘em, especially with DCA.
chuppi wrote:
livesoft wrote:Well, if this doesn't prove that buy-and-hold is dead, I don't know what does. :twisted:
One needed to do something in those 15 years, didn't they?
I think the start of this period was the peak of the technology boom. So it looks really bad.
The only thing it I think it might prove is that investing nothing but a lump sum with no more contributions for 15 years is not a good way to invest in either stocks or bonds -- especially if you happen to invest that lump sum just before a series of really bad market crashes that include the second worst in history.
We contribute small chunks every month over a period of time. It is different from investing lump sum 15 years ago and seeing now that it has only grown 5% over these years. I think the numbers would be better for steady contribution over 15 years. I am hoping that someone here can break it down and crunch the numbers.
Bingo!

Got that data right here.


Using yearly data From Vanguard's own site, for a lump sum invested at the beginning of 1999 through 2013 my preliminary numbers show the average APY or CAGR (Compound Annual Growth Rate) for
  • VBMFX was 5.0%.
    VTSMX was 5.42%.
    Vanguard's Wellington (VWELX) balanced fund did much better, at 7.46%.
For dollar cost averaging by investing a fixed amount at the beginning of each year
  • VBMFX averaged 4.67%.
    VTSMX did 7.73% and
    VWELX averaged 8.2%.
To put that in more concrete numbers, a lump sum of $10,000 invested in 1999 in
  • VBMFX would be worth $20,798 at the end of 2013.
    VTSMX would be worth $22,082 and
    VWELX would be worth $29,427

Dollar-Cost-Averaging by Investing $1000 per year from 1999 through 2013:
  • VBMFX would give you $22,025
    You'd have $28,636 in VTSMX and
    $29,832 in VWELX.

Incidentally, I've lost track of the number of times I've seen pretty-well-thought-out and sometimes very complicated and pretty active portfolios and theories for balancing that turn out not to do have done as well as just buying Wellington (VWELX) and leaving it alone over long periods of time.

jimb
Last edited by jimb_fromATL on Sat Jul 26, 2014 2:17 pm, edited 1 time in total.

User avatar
Kevin M
Posts: 10211
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: Bonds beat stocks over the last 15 years

Post by Kevin M » Sat Jul 26, 2014 2:14 pm

Tamales wrote:Kevin, re: "...uncertainty of return over a long holding period. This is illustrated in this chart that I shared in another recent thread on the long-term risk of stocks..."

I might be off-point, but can't you see essentially the effect of increasing risk with time via the monte carlo simulator tool on portfolio visualizer, where the max and min lines are expanding in a funnel shape, outward?
Yes. I actually considered using a PV chart like this to demonstrate the point in the other thread, but thought John Norstad's annotated chart made the point better.
Tamales wrote:But also, you can affect the slopes and crossover points pretty significantly with different portfolio compositions, and arguably a portfolio which stretches the blue "min balance" line out as far as possible (which represents the "bad side" of risk) is one of the more desired attributes?
Well, you certainly can run the simulation for different asset classes and see that stocks have a much larger dispersion than bonds, and bonds more so than bills, and that also makes the point. I've seen charts before that show the different terminal wealth dispersions on a single chart, which also helps understand it.

As to using MC to decide on portfolio construction, I think it's a useful tool to get a sense of different possible outcomes--very educational--but it's very sensitive to the inputs, so I'd be careful about relying on it too much.

Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)

User avatar
Robert T
Posts: 2534
Joined: Tue Feb 27, 2007 9:40 pm
Location: 1, 0.2, 0.4, 0.5
Contact:

Re: Bonds beat stocks over the last 15 years

Post by Robert T » Sat Jul 26, 2014 2:37 pm

ginmqi wrote:I've come across that piece before and glad I am revisiting it. So if it's true that there is in fact no time diversification, and in fact this is one of the most pervasive things being told across all forms of media: that young investors should be more into equity than elderly investors.

If that is actually a fallacy, as Norstad argues. Then what is the practical, real-world decision that would impact actual real world portfolios? Should we go 50/50 then?
The evidence presented by those who argue that there is time diversification in equity returns is the narrowing gap in the standard deviation of returns for longer horizon investments eg. the standard deviation of annualized returns for 15 years is lower than for 5 years. However what those who are say there is no time diversification show that even with lower standard deviations the actual cumulative dollar return variances goes up over time e.g. even a 1% difference in return over 40 years results in a larger difference in final portfolio $ amounts.

For me the practical, real-world decisions is; (i) recognize the likely dispersion in portfolio dollar values relative to your goals; (ii) use modest expected returns in calculations; (iii) leave room for adjustment in planning. Here is a practical example:

The chart below is the glide part of a portfolio with an expected return of 7.5 percent per year (similar to my own when set up - 75:25 stock:bond portfolio with small cap and value tilt). The dotted lines are derived from Bodie's formula - similar to that used in the Norstad article - except the chart adds in a constant annual savings which dampens the cumulative dispersion effect. The orange line tracks actual performance to date (which I update annually). So not too far off so far. Would suggest having some flexibility at toward end (e.g. possibility of working longer/saving more if falling to far off track.

Image
As a young investor who currently subscribes to the mantra of being aggressive in equity when I am young...this is quite concerning. Should I change my IPS and dramatically reduce my equity???
Personally I don't think your IPS, if well thought out, needs to changes based on this thread. I am not changing mine.

One of the challenges in the interpretation of the Fama-French data is their use of only US data. If you add performance of other countries around the world the story for bonds becomes a bit more complicated. If you look at the Credit Suisse Global Investment Return Yearbook 2013 (linked below)

From 1900 - 2012 (over a 100+ year investment horizon):
I would highly recommend reading Bill Bernstein's Deep Risk, and Ages of Investors books. They may be helpful as you chart you course.

Best,

Robert
.

User avatar
packer16
Posts: 1045
Joined: Sat Jan 04, 2014 2:28 pm

Re: Bonds beat stocks over the last 15 years

Post by packer16 » Sat Jul 26, 2014 3:02 pm

I am skeptical of the Norstrum article as it uses a theoretical model (random walk) and not real data to make its point. When the real data is shown to be different, the data is ignored and the model trotted out. This is physics envy to the nth degree. Most practitioners and some academics know what he is peddling is not reflected in the historical data but only in his model. Who is right? I will go for the real data versus a random walk model that explains very little in terms of variations in stock prices. You can be skeptical of the past data but I think you should be more skeptical of model that has clear flaws in terms of explaining equity returns. Even Norstrum states that a random walk is at best it is a rough approximation of reality. So do you want to rely in historic data or what Norstrum describes at best a rough approximation of reality?

Packer
Buy cheap and something good might happen

edge
Posts: 3333
Joined: Mon Feb 19, 2007 7:44 pm
Location: Great Falls VA

Re: Bonds beat stocks over the last 15 years

Post by edge » Sat Jul 26, 2014 3:36 pm

Nothing to see here. At least both were above 5 percent. Anyone who didn't know this was possible and in fact very likely given the sequence of returns in the 90s should not be in stocks.

User avatar
Kevin M
Posts: 10211
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: Bonds beat stocks over the last 15 years

Post by Kevin M » Sat Jul 26, 2014 6:39 pm

packer16 wrote:I am skeptical of the Norstrum article as it uses a theoretical model (random walk) and not real data to make its point.
Skepticism is good, but the article does much more than just rely on the random walk model. It addresses the issue from several points of view, including discussions of several historical examples. The Bodie option pricing argument also has nothing to do with the random walk model.

Regarding the shortcomings of the random walk model, it actually is acknowledged and addressed.
There's one problem with this chart. It involves a phenomenon called "reversion to mean." Some (but not all) academics and other experts believe that over long periods of time financial markets which have done better than usual in the past tend to do worse than usual in the future, and vice-versa. The effect of this phenomenon on the pure random walk model we've used to draw the chart is to decrease somewhat the standard deviations at longer time horizons. The net result is that the dramatic widening of the spread of possible outcomes shown in the chart is not as pronounced. The +1 standard deviation ending values (the tops of the bars) come down quite a bit, and the -1 standard deviation ending values come up a little bit. The phenomenon is not, however, anywhere near so pronounced as to actually make the +1 and -1 standard deviation curves get closer together over time. The basic conclusion that the uncertainty of the ending values increases with time does not change.


Furthermore, if you look at John's more complex article on the random walk, he discusses the weaknesses of the model. One of the weaknesses is that it actually underestimates long-term risk relative to some other models:
These kinds of distributions also have the “fat tails” that we notice in the historical data – very large losses and very large gains are more likely than in the lognormal model. Investing in volatile assets like stocks is even more risky in these models than in the lognormal model.
packer16 wrote:So do you want to rely in historic data or what Norstrum describes at best a rough approximation of reality?
One of the problems with relying only on historical data is that we simply don't have enough independent, long-term periods to make any statistically justified conclusions based only on observations. For example, in one hundred years there are only four independent 25-year periods, so we can't used direct observations to show a reasonable estimate of dispersion of 25-year returns. So to come to any conclusions at all on the variance of long-term returns, even tentative ones, some sort of theoretical, statistical model must be used.

Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)

User avatar
jimb_fromATL
Posts: 2278
Joined: Sun Nov 10, 2013 12:00 pm
Location: Atlanta area & Piedmont Triad NC and Interstate 85 in between.

Re: Bonds beat stocks over the last 15 years

Post by jimb_fromATL » Sat Jul 26, 2014 6:51 pm

Kevin M wrote: One of the problems with relying only on historical data is that we simply don't have enough independent, long-term periods to make any statistically justified conclusions based only on observations. For example, in one hundred years there are only four independent 25-year periods, so we can't used direct observations to show a reasonable estimate of dispersion of 25-year returns. So to come to any conclusions at all on the variance of long-term returns, even tentative ones, some sort of theoretical, statistical model must be used.

Kevin
In a hundred year period, after the first 25 years there are 76 different 25 year periods, each with a different start and end date. Some started just before and some just after all of the highs and lows; they encompass long periods of decline, flat performaince, and increases; and the last hundred years include the best and worst years and decades of applicable modern history. If you use monthly data, multiply the number of rolling periods by 12. That's a pretty big sample.

jimb

Jack
Posts: 3254
Joined: Tue Feb 27, 2007 2:24 am

Re: Bonds beat stocks over the last 15 years

Post by Jack » Sat Jul 26, 2014 7:10 pm

jimb_fromATL wrote:
Kevin M wrote: One of the problems with relying only on historical data is that we simply don't have enough independent, long-term periods to make any statistically justified conclusions based only on observations. For example, in one hundred years there are only four independent 25-year periods, so we can't used direct observations to show a reasonable estimate of dispersion of 25-year returns. So to come to any conclusions at all on the variance of long-term returns, even tentative ones, some sort of theoretical, statistical model must be used.

Kevin
In a hundred year period, after the first 25 years there are 76 different 25 year periods, each with a different start and end date. Some started just before and some just after all of the highs and lows; they encompass long periods of decline, flat performaince, and increases; and the last hundred years include the best and worst years and decades of applicable modern history. If you use monthly data, multiply the number of rolling periods by 12. That's a pretty big sample.

jimb
Unfortunately that's not the way statistics work. Your 76 "different" 25-year periods actually consist of 96% of the previous sample. They aren't that different at all.

By your reasoning, all you need is one day of data broken up into 25,000 1-second periods. "That's a pretty big sample!"

User avatar
Kevin M
Posts: 10211
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: Bonds beat stocks over the last 15 years

Post by Kevin M » Sat Jul 26, 2014 7:17 pm

jimb_fromATL wrote: In a hundred year period, after the first 25 years there are 76 different 25 year periods, each with a different start and end date.
One thing I've learned from forum members who have strong statistics backgrounds is that rolling periods, which is what you're discussing, have high correlation to each other. Therefore looking at many rolling periods adds little if any information relative to looking at the independent (non-overlapping) periods. For a detailed explanation of this, see forum member RodC's paper, Indpendence of rolling periods.
jimb_fromATL wrote:If you use monthly data, multiply the number of rolling periods by 12. That's a pretty big sample.
No help at all. The variance of monthly returns tells us nothing about the observable variance of longer-term returns. To use shorter-term periods to estimate variance of longer-term period returns you must use a statistical model. With the most common model you multiply the standard deviation of the shorter-time period by the square root of the number of such periods contained in the longer time period.

So to estimate the standard deviation of annual returns from monthly data you would multiply the monthly standard deviation by sqrt(12). Similarly, to estimate the standard deviation of cumulative 15-year returns from annual data, you'd multiply the annual standard deviation by sqrt(15).

Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)

User avatar
packer16
Posts: 1045
Joined: Sat Jan 04, 2014 2:28 pm

Re: Bonds beat stocks over the last 15 years

Post by packer16 » Sat Jul 26, 2014 9:27 pm

Kevin M wrote:
packer16 wrote:I am skeptical of the Norstrum article as it uses a theoretical model (random walk) and not real data to make its point.
Skepticism is good, but the article does much more than just rely on the random walk model. It addresses the issue from several points of view, including discussions of several historical examples. The Bodie option pricing argument also has nothing to do with the random walk model.

Regarding the shortcomings of the random walk model, it actually is acknowledged and addressed.
There's one problem with this chart. It involves a phenomenon called "reversion to mean." Some (but not all) academics and other experts believe that over long periods of time financial markets which have done better than usual in the past tend to do worse than usual in the future, and vice-versa. The effect of this phenomenon on the pure random walk model we've used to draw the chart is to decrease somewhat the standard deviations at longer time horizons. The net result is that the dramatic widening of the spread of possible outcomes shown in the chart is not as pronounced. The +1 standard deviation ending values (the tops of the bars) come down quite a bit, and the -1 standard deviation ending values come up a little bit. The phenomenon is not, however, anywhere near so pronounced as to actually make the +1 and -1 standard deviation curves get closer together over time. The basic conclusion that the uncertainty of the ending values increases with time does not change.


Furthermore, if you look at John's more complex article on the random walk, he discusses the weaknesses of the model. One of the weaknesses is that it actually underestimates long-term risk relative to some other models:
These kinds of distributions also have the “fat tails” that we notice in the historical data – very large losses and very large gains are more likely than in the lognormal model. Investing in volatile assets like stocks is even more risky in these models than in the lognormal model.
packer16 wrote:So do you want to rely in historic data or what Norstrum describes at best a rough approximation of reality?
One of the problems with relying only on historical data is that we simply don't have enough independent, long-term periods to make any statistically justified conclusions based only on observations. For example, in one hundred years there are only four independent 25-year periods, so we can't used direct observations to show a reasonable estimate of dispersion of 25-year returns. So to come to any conclusions at all on the variance of long-term returns, even tentative ones, some sort of theoretical, statistical model must be used.

Kevin
I think you are looking at the markets as a science where the market is much more psychology and economics with a dash of science. The tools of science are limited in there applicability to markets. You are not going to find many hypothesis test results with high confidence intervals. If you run betas you see very low R squares for example. Most of the finance academics have a hard science or engineering background so they apply the tools used to study nature to markets implicitly assuming that markets are like nature, which they are not because they are run by emotional people. However, the larger markets will tend to show more efficiency while the smaller ones not as much. The past is important because it is the only evidence we have of how psychology effects the markets, there is no theory for this. In my mind it is much more important to study and understand how markets work versus how the should in theory work because in practice there are very few fully efficient markets.

You can also look at the underlying economics of markets to get a feel for long-term returns. Markets that have a long Anglo-Dutch tradition do have higher returns to shareholders because other market have more intervention either via taxes or direct gov't intervention and the intervention can be unexpected.

Packer
Buy cheap and something good might happen

User avatar
grayfox
Posts: 4936
Joined: Sat Sep 15, 2007 4:30 am

Re: Bonds beat stocks over the last 15 years

Post by grayfox » Sun Jul 27, 2014 12:49 am

ginmqi wrote:
I've come across that piece before and glad I am revisiting it. So if it's true that there is in fact no time diversification, and in fact this is one of the most pervasive things being told across all forms of media: that young investors should be more into equity than elderly investors.

If that is actually a fallacy, as Norstad argues. Then what is the practical, real-world decision that would impact actual real world portfolios? Should we go 50/50 then?

<snip>

As a young investor who currently subscribes to the mantra of being aggressive in equity when I am young...this is quite concerning. Should I change my IPS and dramatically reduce my equity???
Consider this chart from a previous post.

Image

Suppose you plan to retire at age 67 when your human capital is fully expended. For retirement income to pay your monthly bills for the rest of your life, you plan to liquidate your portfolio at age 67 and purchase an inflation-adjusted annuity.

:arrow: To state the obvious, at age 67, you must be 100% cash immediately before purchasing the annuity.
:idea: But 30 and 20 years before, it would make sense to have 100% stocks, since there is 90-99 percent chance stocks will beat bonds.

The start and end points are known. The middle is the tricky part.
:?: The question is: What mix of assets to hold between 20 years and 0 years before retirement?

By 10 years before retirement, there is a 20% chance bonds will beat stocks. 100% stock probably no longer makes sense.
5 years before retirement, there is a 30% chance bonds beat stocks.
1 year before retirement there is a 40% chance bonds beat stocks. Mostly 1-year Treasury now makes more sense.

Somehow, between 20 years and 0 years before retirement, you want to go from 100% stock to 100% cash.

:?: I don't know what the optimum way to make the switch from stocks to bonds to cash. Maybe some kind of glide path where you gradually add Treasury bonds?

For example, maybe 20 years before retirement you sell 5% of the stock and buy 20-year Treasuries that will mature at age 67 when you actually need the cash to purchase the annuity.
19 years before retirement, sell another fraction of the stock and buy 19-year Treasury.
Continue every year selling some stock and buying Treasuries that mature at age 67.
Then at age 67 you will have the cash to buy the inflation-indexed annuity.

I don't think there is one universally correct answer to this for all situations.

User avatar
grayfox
Posts: 4936
Joined: Sat Sep 15, 2007 4:30 am

Re: Bonds beat stocks over the last 15 years

Post by grayfox » Sun Jul 27, 2014 1:43 am

Kevin M wrote:
berntson wrote: My claim was that most of the extra volatility over long time periods comes from upward volatility. I'm sure there's still downward volatility, but I'm not convinced that it's less than the volatility for nominal bonds. Maybe inflation adjusted bonds.
I wanted to get back to this point. My concern is not so much volatility, which I think brings to mind uncertainty of returns over shorter periods, like a month or a year, but uncertainty of return over a long holding period. This is illustrated in this chart that I shared in another recent thread on the long-term risk of stocks. It is from this excellent paper by forum member John Norstad: Risk and Time, which I think would be well worth reading.

Image

Kevin
You can see from this chart that the dispersion of outcomes increases with longer holding periods.

But in this thread, we looked at worst outcome over longer holdings periods.

Cruncher presented empirical data that shows worst outcomes improve with longer holding periods. After about 20 years, the worst outcome was still a positive return.

Image

Grayfox presented theoretical data that shows Value-At-Risk gets smaller with longer holding periods. After about 20 years, the chance of a loss gets smaller and smaller.

Image

The explanation is that you have variable returns, but the mean is positive. In the short term, the variability can lead to big losses. But over time, the positive mean swamps the variability. Read the thread.

:arrow: So, does risk with stocks increase or decrease over time? It depends on how you measure risk.
If you use Standard Deviation to measure risk, the risk increases with time.
If you use worst case outcome or Value-At-Risk or the chance of showing a loss, the risk decreases with time.

bs1
Posts: 22
Joined: Sun Jun 24, 2012 7:10 pm

Re: Bonds beat stocks over the last 15 years

Post by bs1 » Sun Jul 27, 2014 6:30 am

Last edited by bs1 on Wed Jul 30, 2014 7:20 pm, edited 1 time in total.

Tamales
Posts: 1297
Joined: Sat Jul 05, 2014 10:47 am

Re: Bonds beat stocks over the last 15 years

Post by Tamales » Sun Jul 27, 2014 11:40 am

Thanks for those. The 2nd link, "In Investing, It’s When You Start And When You Finish" contains a very interesting graphic on how different start and end dates affect results, and the source for that: http://www.crestmontresearch.com/ has a wealth of interesting things on their web page. An updated version of their start/finish date matrix can be found here: http://www.crestmontresearch.com/stock-matrix-options/

Has anyone read either of their books, "Unexpected Returns" and "Probable Outcomes?"

berntson
Posts: 1366
Joined: Mon Oct 29, 2012 12:10 pm

Re: Bonds beat stocks over the last 15 years

Post by berntson » Sun Jul 27, 2014 12:56 pm

grayfox wrote: If you use Standard Deviation to measure risk, the risk increases with time.
If you use worst case outcome or Value-At-Risk or the chance of showing a loss, the risk decreases with time.
I agree. No simple measure of risk corresponds exactly to how real investors think about risk in part because we each think about risk in different ways. But for myself, I think more about value at risk or the chance of showing a loss (against inflation) than about standard deviation. I don't care much about the spread of possible returns. I care about the likelyhood of failing to meet my goals. Over longer time periods, the chance of failing to reach my goals decreases. So risk decreases too.

MapleHermit
Posts: 68
Joined: Sat Oct 26, 2013 2:37 am

Re: Bonds beat stocks over the last 15 years

Post by MapleHermit » Sun Jul 27, 2014 4:00 pm

In the long term, it doesn't make sense for bonds to beat stocks.

User avatar
grayfox
Posts: 4936
Joined: Sat Sep 15, 2007 4:30 am

Re: Bonds beat stocks over the last 15 years

Post by grayfox » Mon Jul 28, 2014 12:24 am

MapleHermit wrote:In the long term, it doesn't make sense for bonds to beat stocks.
Nevertheless it can happen.

Think of a simpler case that has a definite endpoint: 10-year default-free Treasury bond vs. 10-year A Corporate Bond.
They both pay annual interest and then pay back $1000 at the end of 10 years.

The Treasury bond pays $25 per year interest. (2.5% yield)
(Cashflow = -1000+25+25+25+25+25+25+25+25+25+25+1000) IRR= 2.50% :|

The A Corporate pays $34.30 per year interest (3.43% yield)
(Cashflow = -1000+34.30+34.30+34.30+34.30+34.30+34.30+34.30+34.30+34.30+34.30+1000) IRR=3.43% :happy

The A should beat the Treasury bond, and most of the time it will.
But every once in a while the A defaults and only pays back, say, $750.
(Cashflow = -1000+34.30+34.30+34.30+34.30+34.30+34.30+34.30+34.30+34.30+34.30+750) IRR=1.05% :(

You get a shot at the higher reward (3.43 vs. 2.5) with some chance of it being lower (1.05 vs 2.5).
This is a calculated risk, but sometimes the lower risk asset ends up with higher return.

Another point I will make is about expected return, E(return).
E(return) of the Treasury is the yield 2.50% because there is only one outcome, 2.5%.

But E(return) of the A-Corporate is not the yield 3.43% because that is not the only outcome.
Suppose there is a 5% chance of default and getting only 1.05% return. E(return) = .95*3.43 + .05*1.05 = 3.31% which is less than the yield.
It's still higher than the 2.50% expected return of the T, but more uncertain in the outcome.

This is the basis of investing. You must demand a higher expected return when the outcome is less certain and can go against you.

Browser
Posts: 4857
Joined: Wed Sep 05, 2012 4:54 pm

Re: Bonds beat stocks over the last 15 years

Post by Browser » Mon Jul 28, 2014 9:14 am

My extensive research has shown that when stocks outperform bonds, it is because there is positive "equity twinkle dust". Sometimes the twinkle dust does not show up, in which case bonds outperform stocks. I figure that one imaginary ex post explanation is as good as any other.
We don't know where we are, or where we're going -- but we're making good time.

Post Reply