95% Rule and Variable spending - is this correct?

[ERD50]

Not making sense to me.

I use the defaults, then change the 'Spending Models' to Percentage of Remaining Portfolio: , then enter '95'.

The results provide a graph of the variable spending. In a few short years, it dips from $30,000 to near the blue 'two-person-poverty' line of $13,167. It's a bit tough to interpolate the numbers, but one line at 4 years is nearing $15,000 (much closer to the $13,167 line than the $20,394 line) ; at 7 years very close to the blue line, and below it at 14 years.

This does not add up for me.

Four years of 95% reductions in spending would get you to 81.45% (.95^4) of $30,000 = $24,435. So how can the line dip so low?


13167/30000 ~ 44%. You need to go 16 years to get that reduction at 95%. Yet, the spending lines are coming very close by year 7, and below the line by year 14.

Is this a bug, or am I confused (or both?).

-ERD50

According to the rule, each year's withdrawal is the greater of 95% of last year's withdrawal or 4% of the current portfolio as you started with.

[ESRBob]

Erd50


I got the same results, and also find them puzzling.

As you say the 95% Rule should limit the speed of decline in the lower (spending) graph even if every single year were a loser to 95% of the year prior. I also had a few rogue lines falling faster. It could be something to do with high inflation and negative returns in a handful of years driving the real value of the withdrawal down close to the poverty line. That is the only thing I can think of to explain what you're seeing. Market returns shouldn't drive those minimum spending levels down in nominal dollars any faster than .95^n * (Original Spending) where n is the number of years in retirement.

Anyone see something we're missing?

[ERD50]

OK thanks, at least I'm not misinterpreting things.

After I posted, I was thinking that maybe it is taking the first 95% calculation using the *EOY* portfolio, rather than the starting portfolio. The 'worst case' lines are going to be series that started with a really bad year. So, if we had a 30% drop in the market the first year (70% of portfolio left), and then it took .95 x .7 = .665, that would already drop you from $30,000 starting spend to about $20,000 - and you would have a series of .95's after that.

Just a guess at what the bug might be, but it would probably give the numbers we are seeing. Picking the wrong start/end point is a common and easily overlooked program bug (or so I'm told ).

-ERD50

[ESRBob]

Yeah, the 95% needs to be 95% of last year's withdrawal, which would of course be based on last year's Start of Year portfolio value, not end of Year.

I was a coder for several years early on in my career, so I understand the thought process you're going through with developing ideas for where the bug might be. I actually enjoyed bug hunting -- not every programmer does-- I found it intellectually satisfying, like solving a puzzle. There's an art to figuring out which variables to print out at what point in the logic loops so you can trace through what values are there, what they should be and just chop in tighter and tighter until you've nailed the little sucker. Fun. But I left all that to go to business school -- more pay and bigger more abstract problems that didn't lend themselves to such nice clear wins. But did lead to a portfolio big enough to FIRE on!