This is true. But I suppose the argument suggests that even a year makes a big difference. If a retiree quit at YE 2020, their portfolio might have grown by nearly 30% in 2021. Alternatively, someone retiring exactly one year later (YE 2021) has seen their portfolio eroded by roughly 25% (not including spending) right off the bat. That’s a very different sequence of returns in only a single year. There are even better examples.

Returns from CRE syndications can be lumpy as properties are sold. And you don’t have control as a limited partner. As a general rule, I will likely assume the syndication returns are similar to stock index etfs over the long haul and when large returns are distributed, park the excess as cash to spend down rather than sell equities at whatever the CAPE rule calculates. At some point, cash surplus will deplete back to normal and I will resume selling equities per the CAPE rule. Or I may buy new CRE deals and sell equities at the CAPE rule level to maintain my diversification. I still need to think about this some more when the time comes.

The issue, as I currently understand it, with the 66ish 30 year periods is that they all share a lot of the data. For example, the 1926 and 1927 cohort will share 29 out of 30 years. This is very different from having 66 cohorts that all share 0 years of data.

Interesting. Great thoughts. I guess I’m not clear on the significance of whether the 30-year periods were overlapping or not. Would we not want to model each cohort and the real returns observed no matter which year of initial retirement? If that’s the case, starting at 1926 we have 66-ish 30-year periods to analyze. The issue becomes modeling the long horizons, of say 60 years. Even the dreaded 1966 scenario is still four years out from a 60-year look.

I’m certainly glad that we agree that there is probably too much hand-wringing on the specifics of withdrawal rates. I think it’s a useful exercise in the theoretical case, but I doubt there will ever be a true retiree with a 30+ year retirement horizon living off only investment income. A die-hard 4% rule adherent will be just fine if they have real spending flexibility and even modest streams of income. Even more likely, especially if young, they’ll be back involved in the world and probably making real wages in 5-10 years, max.

That said, having a low withdrawal rate certainly gives me the comfort to know I can tinker for a while!

I would disagree with this for the following reasons: We basically have 3ish 30 year non-overlapping samples in the historical data (post 1926). We don’t really know what the true equity risk premium was in the past. It could easily be 2-3% points higher or lower than what has been observed. We also don’t know if the equity risk premium should remain static over time. Perhaps, going forward the equity risk premium is 3-4% instead of the observed 6-7%? Maybe it was 3-4% all along and the observed outcome in the US was lucky? Maybe there is less serial correlation in market returns going forward than has been observed in the historical data? The same also applies to the returns from fixed income.

The uncertainty around these inputs creates a lot of possible error in the modeling of SWRs. It’s undeniable that small differences in initial spend have a massive impact on outcomes. The question I have is can we calculate them with enough precision to be actionable today? It certainly seems reasonable to spend a bit less in drawdowns and perhaps a bit more after years of good returns. Determining exactly how much is really tricky though!

I thought your simplified advice and the 4 things to survive a bear market were great. I think it will be much fruitful to focus on these things than calculating withdrawal rates with a high level of precision. Don’t retire in particular is incredibly under rated!

The unfortunate reality is there is simply a tremendous amount of uncertainty in determining what amount to spend from a portfolio. The best historical simulation or Monte Carlo model cannot eliminate that uncertainty. It’s just the nature of the beast.

Thanks for this discussion, Kyle.

As far as precision goes, what’s the downside? I think (hope) we can agree that precision to the tenth decimal place is certainly significant (i.e. big difference between even 4% and 3.9%). I’ve seen people shrug off the difference between 3% and 4% as “only 1%,” which means we have a high-school-level math problem on our hands. Yikes! There is certainly a degree of conservatism here when considering the difference between, say, 3.3% and 3.25%. But again, I guess I don’t see the downside to this degree of precision. How is it negatively impacting our lives? In terms of what is and is not knowable, we can use past market performance and a range of portfolio asset allocation and retirement windows to assess appropriate withdrawal rates with a high degree of precision. That’s me speaking for ERN anyway, who does the real heavy lifting on the modeling. The future is, of course, unknowable.

Regarding CAPE in international stocks: I’d be weary of putting a high degree of confidence in the accounting standards of emerging or even some major international markets (cough…China…cough).

Finally, I’m confused by this statement: “History has shown that successfully implementing a tactical approach has a low probability of success and not something I think most would recommend.”

I can’t help but be skeptical about calculating SWRs to two decimal places. To me, it feels like an exercise in false precision. Of course precision IS important with SWRs as Big Ern points out, but I don’t think is something that is knowable. I’d guess that a 95% confidence interval on whatever SWR we calculate is easily +/- 1.5-2%. Furthermore, if we could be precise with our ability to calculate a CAPE based withdrawal rate it is logical that increasing an investors allocation to international stocks with their lower CAPE ratio would improve SWRs. History has shown that successfully implementing a tactical approach has a low probability of success and not something I think most would recommend.

Thanks for this comment, Phillip. I guess RE can act as a sort of de-facto bond, but I’d probably just treat those proceeds as normal income and subtract that from what I’d need to withdraw on other investments. Is that your plan?

Thanks, Craig. In truth, I still believe anyone saving anywhere approaching 25x their appending will be light years ahead of most. The technical details really just concern those truly interested in retirement.