Yes, I assumed 20 years would be closer to the typical user's spending duration (which will be a bit less than half of years remaining) versus 10 years. The reason we went with 10 year for break even inflation (BEI) was ready data availability. The 10 year BEI series is the longest term with daily data that we could get from the St Louis Fed. The 20 and 30 year BEI are updated only monthly.Ben, question on assumption time periods for the power users lurking here. Am curious the thinking behind using 20-yr rate for bonds (vs 10 or 30) while what looks like 10-yr breakeven for inflation. I'm guessing bond rate is an attempt to match a "typical" retirement?
For the expected stock return regressions, restricting the data to post-1950 (roughly the second half of the data series) was based on the possibility that the more recent data is more applicable to present day conditions. 1871 was a long time ago and maybe it was a different world and that data is less applicable. So I ran an extra set of regressions on the post-1950 data and included those estimates in the averages used for the presets. This gives the recent data more weight. This doesn't relate to the duration of the stocks. It's just based on the idea that the relationship between these variables today is more similar to the post-1950 period than pre-1950.Likewise the 1950 cutoff option on stocks.
I'm not sure if the stock expected return has a forward period associated with it like for bonds and inflation (or somehow embedded in the regression methodology), or is to be thought of more as "indefinite." thank you
Stocks should have a duration based on the timing of earnings, but it's hard to estimate and match to spending. I don't think it's practical to do much with stock duration when it comes to investing.
Statistics: Posted by Ben Mathew — Wed May 22, 2024 12:19 am — Replies 764 — Views 219380