long term a whole world market 100% equities portfolio has returned about 7% annually in inflation-adjusted dollars. In any given 25 year period you could easily see 4% real return (or perhaps a bit worse) or a 10% real return (probably not much better). That's where bonds are so useful, particularly long treasuries. You want bonds to even out the ups and downs of the market, and they do that because many bonds are anti-correlated to the equities markets. Long treasuries are the most anti-correlated to the stock market, more than medium and short term. In terms of correlations with the broad US stock market it probably goes
long treasuries
medium treasuries
gold (???, could be higher or lower, dunno)
short term gov't bonds
cash (neutral)
high quality corporate debt
junk bonds
REITS
Foreign equities
SPY
So while long treasuries are the most risky government bond to hold that is actually a feature rather than a bug when held with equities. When equities tank the long government bond tends to spike, and it will spike more dramatically than shorter term government bonds, exactly because they are more sensitive to interest rate changes. And it avoids the problem with total bond market funds -- something like a third of BND is corporate debt, which will crash right along with equities, just not as hard. Look at the performance in 2008 and 2009 of BND vs TLT and tell me which you'd rather have as 20% of your portfolio. Then look at a corporate investment grade ETF over the same period and wonder why the hell you'd want to include that in your portfolio as a hedge against equities crashing.
So the question really is, does it make sense to use the median or average case to plan for your retirement? I would submit that it does not. You should plan for your retirement based off the tenth or twentieth percentile outcomes. And you can do that a couple of ways -- either go 100% equities and just plan on gamble, saving as much as you can and working as long as you have to. That would tend to minimize the average number of years you have to work, but there would be outcomes that would force you to work much longer than you might like. That's what I'm doing (and writing this I'm beginning to reconsider). Or you can buy 20-40% bonds and reduce both the median rate of return and increase the 10th percentile rate of return, which is probably the more important metric. For your planning you should be much more concerned with your 10th percentile rate of return than your median. But for each 10% of bonds you add to your portfolio you're probably sacrificing half a percent return and increasing the 10th percentile rate by a half percent.
Here's a great resource for gaming all this out:
https://www.portfoliovisualizer.com/
Last edited by SenorKeeed; 02-18-2019 at 03:31 PM.