Kevin M wrote:
As mentioned earlier, the data is from the Simba backest spreadsheet. I believe that the ITT data prior to availability of the Vanguard ITT fund is based on longinvest's 10-4 year rolling bond ladder simulation, so perhaps longinvest can weigh in and say if the extra two years of shorter-maturity bonds in the 10-2 year ladder is enough to explain the difference in performance. Another factor is that the Simba spreadsheet applies an expense ratio to better simulate a fund, but I wouldn't think this is enough to explain the difference. At any rate, this is most definitely not data for a 20-year Treasury rolled annually, but for a bond ladder that is intended to be representative of the Vanguard ITT fund.
Here's the reason for the difference.
My chart's legend was incorrect, for bond returns. I edited my posts to clarify it. The VPW spreadsheet (the source of my data) always uses the most reliable source of returns to approximate TBM and TSM, for each historical period. In pre-1972 years it uses FRED and Shiller derived
returns, and starting in 1972 it uses Simba's TBM and TSM returns.
Specifically, the VPW spreadsheet uses Bond Fund 10-2 from 1966 to 1971 (inclusively), then uses Simba's TBM returns from 1972 to 1981 (incusively), which happen to be significantly higher than the returns of Bond Fund 10-2 and 10-4, possibly (?) because its inclusion of corprorate bonds, I don't know; this continues to intrigue me. It does not subtract an expense ratio from Bond Fund 10-2, but an expense ratio is taken out (in Simba's) from TBM's returns. It also uses Shiller's S&P500 returns from 1966 to 1971, then Simba's TSM returns from 1972 to 1981. It does not subtract an expense ratio from Shiller's S&P500, but an expense ratio is taken out (in Simba's) from TSM's returns.
Kevin M wrote:Minor technical note: The values on the horizontal axis are year-end, which is consistent with the convention used in the Simba spreadsheet and PortfolioVisualizer. So if you specify start year as 1967 and end year as 1981, you will get returns for 12/31/1966-12/31/1981. I believe longinvest's chart is showing year-start on the horizontal axis.
Yes, in my charts, years indicate the start of year (or end of day on the preceding December 31st). This way the line between 1966 and 1967 represents the unrolling of year 1966, which seems pretty intuitive to me. This would allow me to hypothetically add monthly returns without having to slide anything. Marking the last day of the year on the chart seems wrong, to me. Maybe that's an improvement suggestion I should make to Simba's maintainers.
Kevin M wrote:EDIT: However, it is true that returns lag changes in yield proportionate to duration, since the longer the duration the longer it takes for higher/lower yields to compensate for decrease/increase in bond price.
Here's my intuition. Over the long term, the total return of a ladder is the average yield of its bonds at the time they were bought. If I invest into a ladder of bonds bought, respectively, with yields of 3%, 4%, 3%, 5%, and 5%, holding them to maturity, my return should be near the 4% average. Of course, there will be a lag, because new bonds continue to be acquired at different yields. But, it cannot be otherwise, mathematically. If, at the same time, inflation was running 3%, 4%, 3%, 5%, and 5%, then the bond fund will necessarily keep pace with inflation, even though it will lag behind. A full recovery will eventually happen as long as yields (on the long end of the ladder) reflect a small bonus on top of inflation, which has been the case pretty much all the time, except in the 1940s-1950s period. With yields, on the long side of the ladder, reflecting a small bonus above inflation*, the ladder is likely to recover from a one-time increase in inflation (with an accompanying increase in yields) within a time more or less equal to its duration, the point of indifference (adjusted based on the size of the bonus).
* This is like having yields of 3.5%, 4.5%, 3.5%, 5.5%, and 5.5% at the time of purchase, in my example.
Nominal bonds are very vulnerable to inflation risk, as we learned in the 1940s-1950s. There exist TIPS which are not vulnerable to inflation.
Bogleheads investment philosophy |
Lifelong Portfolio: 25% each of (domestic/international) stocks / (nominal/inflation-indexed) long-term domestic bonds |