Matter/scatter, Part 2: Bond fund choices

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Post Reply
User avatar
Topic Author
nisiprius
Advisory Board
Posts: 42242
Joined: Thu Jul 26, 2007 9:33 am
Location: The terrestrial, globular, planetary hunk of matter, flattened at the poles, is my abode.--O. Henry

Matter/scatter, Part 2: Bond fund choices

Post by nisiprius »

Part 1: Introduction
Brief notes: 1) In these Monte Carlo simulations in which the return for a given month is varied by randomly choosing the actual historical return or the actual historic return for an adjacent month. 2) They show the final value from $100 monthly contributions made over the whole time period. 4) Green and red crosses mark actual historical values. 5) Green and red bars show the 10% percentile, median, and 90% percentile of the range of outcomes. 5) The actual data source is PortfolioVisualizer.com, Backtest Portfolio, Monthly Returns; this is used as calculation input; and no PV content or numeric values are directly reproduced.


One of the more dramatic examples of active having outperformed passive: the PIMCO Total Return Fund, PTTRX, versus the Vanguard Total Bond Market Index Fund, VBMFX. PTTRX was long managed by the "bond king," Bill Gross, who developed a near-cult following.

Image

There is almost no overlap in the range of outcomes. The 10th-percentile return of PTTRX is almost as large as the 90th-percentile return of VBMFX. The choice between the two looks hugely important, and would be if this were our only investment.

But few of us are in 100% bonds. What happens when we put the bond choice into the context of a whole portfolio? Here, we compare 60/40 portfolios in which the 60% stock allocation is the Vanguard 500 Index Fund, and the 40% bond allocation is either VBMFX or PTTRX.

Image

Ever since 2009 or so it has been continuously suggested that investors "shorten up duration" because of rising interest rates. Here is a comparison of 60/40 portfolios in which the bond component is intermediate-term and short-term. In order to get the longest possible time period, I've used VFINX, 500 Index for stocks, and either VFITX, the Vanguard Intermediate-Term Treasury Fund, or VFISX, the Vanguard Short-Term Treasury fund, for bonds.

Image

Taking "shortening duration" to the logical conclusion, do we even need a bond fund at all? What happens if we use a 60/40 allocation in which the 40% allocation is to cash (as represented by PortfolioVisualizer's CASHX asset, Treasury bills).

Image

Returning to VBMFX as our baseline, it is sometime suggested that a long-term bond fund, such as the Vanguard Long-Term Treasury Fund, VUSTX, should be superior to Total Bond. Both have had almost zero correlation with stocks, but because VUSTX is more volatile, the diversification benefits of that low correlation should be higher. Here's how it looks in practice:

Image

Some authorities object to Total Bond because of claimed issues with the (government-guaranteed) mortgage-backed securities it contains. Although they have excellent credit quality, they are said to have high convexity due to the fact that mortgage holders can prepay at will. Therefore, some people prefer to substitute the Vanguard Intermediate-Term Bond Index Fund, VBIIX, for VBMFX:

Image

Since we've made a point of excluding mortgage-backed securities in our last comparison, let's give equal time and compare VBMFX to the Vanguard GNMA fund, VFIIX, which is nothing but mortgage-backed securities:

Image

It is sometimes suggested that adding bonds that are below investment grade--"high yield" corporate bonds, aka "junk bonds"--can improve a portfolio. Vanguard's "high yield" corporate bond fund, VWEHX, is older than Total Bond, and I wanted to include all the data. So this time, I used a baseline 60/40 portfolio of 60% VFINX (Vanguard 500 Index) and 40% Vanguard VFITX (Vanguard Intermediate-Term Treasury). For the comparison, I swapped in VWEHX for 15% of the allocation: 60% VFINX, 25% VFITX, and 15% VWEHX.

Image
Annual income twenty pounds, annual expenditure nineteen nineteen and six, result happiness; Annual income twenty pounds, annual expenditure twenty pounds ought and six, result misery.
Lastrun
Posts: 331
Joined: Wed May 03, 2017 6:46 pm

Re: Matter/scatter, Part 2: Bond fund choices

Post by Lastrun »

Thanks so much for these. Query: I am strugglng as to why the SD for cashx is so high in your extreme shortening example?
User avatar
Topic Author
nisiprius
Advisory Board
Posts: 42242
Joined: Thu Jul 26, 2007 9:33 am
Location: The terrestrial, globular, planetary hunk of matter, flattened at the poles, is my abode.--O. Henry

Re: Matter/scatter, Part 2: Bond fund choices

Post by nisiprius »

Lastrun wrote: Wed Aug 01, 2018 6:16 am Thanks so much for these. Query: I am strugglng as to why the SD for cashx is so high in your extreme shortening example?
Do you mean, "why didn't CASHX visibly reduce the standard deviation?" Remember, my illustrations are in the context of 60/40 portfolios with 60% stocks, and I think the answer can be found by looking a "straight" chart:
Source
Image
Blue, stocks. Orange, short-term bonds. Green, money market, AKA cash.

Looking at 60% VFINX + 40% VFISX versus 60% VFINX + 40% cash, For a rough approximation, you can say that the portfolio is 60% stocks and 40% "something that doesn't fluctuate at all." Therefore, in both cases, you'd expect the standard deviation of the portfolio simply to be 60% of the standard deviation of stocks, for either VFISX or cash. Now, you can say that the fluctuations in VFISX ought to add something to the total portfolio fluctuation, but the point is that they are so small, and of course you are only seeing 40% of whatever fluctuation there is.

Now, the correct but often-overemphasized theory is that because bond fluctuations have very low correlation with stock fluctuations, they don't actually add much to the portfolio's total fluctuation, and in fact they can (in theory) actually reduce it in comparison to cash. But in practice the point is that by the time you get down to short-term bonds, the fluctuations are so small that it hardly matters whether or not they are in sync with stocks or not, they are effectively almost zero. So the difference in standard deviation between a portfolio with 40% short-term bonds and 40% cash is going to be almost invisible.
Annual income twenty pounds, annual expenditure nineteen nineteen and six, result happiness; Annual income twenty pounds, annual expenditure twenty pounds ought and six, result misery.
User avatar
dwickenh
Posts: 2021
Joined: Sun Jan 04, 2015 9:45 pm
Location: Illinois

Re: Matter/scatter, Part 2: Bond fund choices

Post by dwickenh »

Thanks nispirius, a great display for the brain. You do a great job of presenting the facts without opinion getting in the way.

Dan
The market is the most efficient mechanism anywhere in the world for transferring wealth from impatient people to patient people.” | — Warren Buffett
Lastrun
Posts: 331
Joined: Wed May 03, 2017 6:46 pm

Re: Matter/scatter, Part 2: Bond fund choices

Post by Lastrun »

nisiprius wrote: Wed Aug 01, 2018 7:54 am Do you mean, "why didn't CASHX visibly reduce the standard deviation?" Remember, my illustrations are in the context of 60/40 portfolios with 60% stocks, and I think the answer can be found by looking a "straight" chart:
Yep, now I see. Same issue (although inverse) with PTTRX.
Post Reply