but they gloss over the fact that simply having a mortgage increases sequence of returns risk.

No it doesn't. SORR is the risk that more unfavorable returns occur earlier in retirement. Having a mortgage does not affect market returns, so you must be talking about some other type of risk.

You are technically correct.

This is what I think is being considered by the OP. I'm sure they'll correct me if I'm wrong!

Having a mortgage increases your minimum monthly costs. It's typically one of the bigger monthly costs. A higher minimum monthly cost means it's harder to cut costs, and therefore it's more likely you'll need to withdraw while the market is low. So it didn't increase the odds of having a SORR problem, it just increased the odds of that problem causing financial issues.

He is technically correct... the best kind of correct. But you also have to factor in paying off a house vs having more liquid cash in investments/emergency funds etc. which may be more beneficial in case of "financial issues" than a paid off house.

This issue was definitively answered by the following:

https://earlyretirementnow.com/2017/10/11/the-ultimate-guide-to-safe-withdrawal-rates-part-21-mortgage-in-retirement/

Actually, after thinking about this issue a lot I think this is one of the few ERN analyses that I disagree with.

Karsten's analysis assumes a fixed 2% inflation rate, so the mortgage payment effectively decreases by 2% every year. But some of the greatest SORR eras in history were driven by high inflation, so this one assumption makes his conclusion unnecessarily conservative. I suspect that his calculations being done in CPI-adjusted terms, rather than nominal terms, made modeling historical inflation difficult, so he simply made a convenient assumption to make things easier to model.

Try this: use cFireSIM to model paying down a mortgage.

Take a $1M portfolio, and a $1M mortgage, excluding taxes and insurance. That works out to $57,000 per year at 4%. Select the option for withdrawals to not be CPI adjusted.

CFireSIM suggests that the above scenario fails 4% of the time over 30 years - roughly the same as the "4% rule". Ergo, funding a 30 year fixed rate loan at 4% is as risky as the "4% Rule", despite the actual withdrawal rate being 5.7% at the beginning. The non-CPI adjusted payments make up the difference.

So in contrast to the ERN analysis, I'd deduce that if you're comfortable with the failure rates of the "4% Rule" then you should be entirely comfortable funding a 4% or below mortgage off a 100% stock portfolio. Adjust to suit your risk tolerance as appropriate.

A 3% mortgage fails 2% of the time. If your loan term is shorter than 30 years it's even lower. Of course the lowest possible risk is always paying off the loan, but at some point I start worrying more about optimizing for the median end of life portfolio value than the worst case portfolio failure risk.

Cliff notes: if you have a cheap mortgage (4% or lower) then don't worry about paying it down - you just need a portfolio big enough to live off + enough to cover the mortgage balance.

Because you show 4% failure at 4% rate and 2% failure at 3% rate, then the 20% failure at 5% probably should be shown as well in the spirit of showing both sides of the coin. Rates sub-4% are unlikely going forward and if the 30-year rate does go below that then there other economic issues at hand. So yes, i agree that if there is a sub-4% mortgage then paying it off doesn't make sense and is a positive force on the SWR analysis. Unfortunately I already had a paid off house and didn't see the need or hassle to take out a mortgage.