When do you determine your withdrawal amount after retiring?

[Winemaker]

My wife and I retired almost 9 and 10 years ago. While employed, we had ample funds to meet all expenses, charity, hobbies, and travel. We modeled our retirement income to match our "employed" income plus an inflation adjustment. We have never taken our 4% portfolio SW ever, we have had pension income, rental income, portfolio income, rental sales and 2 years of DW's SS, that has met our expenses. The portfolio income we "didn't take" has been noted and is there for possible future lumpy expenses. We have been blessed, but always afraid of rate of return sequence, but that train has left the station. We now just plan not to exceed the 3.8% NIIT limit.

 

[imjustawarrior]

I guess I determined our withdrawal rate while working. I tracked our expenses. I then made projections for spending in retirement. I ran the calculators and determined we could likely support our projected expenses.

I re-evaluate our expenses and portfolio every so often to see if we are still on track. If things started looking tight, we would spend less. If they are going well, we will spend more.

Retirement spending is not that much different from spending during accumulation. The source of the money has just changed.

 

[Koolau]

Did you spend way less than 4% in subsequent years? Any particular reason those first 4 years were extra spendy?

I spent less than 4% (of then current stash) in most subsequent years, but not a lot less (and occasionally more.)

In terms of what we had when we first retired, we NOW spend considerably more than 4% of that amount. As you know, there is a built in inflation factor in the so-called 4% rule (you are "allowed" to increase spending by the amount of the previous year's inflation.) We didn't actually calculate that and spend accordingly. We just spent what we needed to spend and it was more than 4% of the original stash. IOW if we wanted to spend it - we did!

So now, after almost 19 years, we do spend a LOT more than 4% of that original retirement stash figure. But it's not 4% of our current stash. SS for two has cut our need for taking 4% most years.

Why did we spend more than 4%?

When I retired, we began preparation to move to the Islands. Getting our old house ready for sale, funding kids college expenses, then moving, then rehabing our first dwelling in Paradise then rehabing our second dwelling on the other side of the Island, furnishing our townhouse and later our condo, buying two (used) cars for use in Paradise, all took WAY more funds than 4% of then current stash. We had sold our mainland house and spent all those proceeds and more on our moving "package" and rehabs, etc.

With all this happening at a time that included the "Great Recession," things looked just a bit dicey for a while, but as we all know, that same time frame began the greatest bull market in history, so it has w*rked out well for us. (SS waiting in the wings was great as we knew we could start anytime beginning at 62.)

My point in sharing all this is that people should not feel they need to slavishly follow the "4% rule" or spend their SWR as calculated by FIRECalc. Those are tools and not rules.

 

[Koolau]

I guess I determined our withdrawal rate while working. I tracked our expenses. I then made projections for spending in retirement. I ran the calculators and determined we could likely support our projected expenses.

I re-evaluate our expenses and portfolio every so often to see if we are still on track. If things started looking tight, we would spend less. If they are going well, we will spend more.

Retirement spending is not that much different from spending during accumulation. The source of the money has just changed.

We didn't have a good handle on what our spending would be in retirement as we had never lived anyplace but where we were born. We knew that it would be more expensive in the Islands, but only had a general idea from our various visits.

We had very few "shocks" as far as expenses in our new home land and a couple of pleasant surprises (state taxes/RE taxes were lower - way lower - than we had projected.)

All in all, once we got established in Paradise, things smoothed out and our spending was well within our control - AFTER our move and getting established. THAT was expensive!

 

[Midpack]

I guess I determined our withdrawal rate while working. I tracked our expenses. I then made projections for spending in retirement. I ran the calculators and determined we could likely support our projected expenses.

I re-evaluate our expenses and portfolio every so often to see if we are still on track. If things started looking tight, we would spend less. If they are going well, we will spend more.

Retirement spending is not that much different from spending during accumulation. The source of the money has just changed.

Nothing wrong with that methodology as long as your income sources keep pace with your (inflation related) spending increases. Predicting investment returns and inflation returns 20-30 years out isn't a no brainer if you're relying mostly/solely on investment returns. May be a different proposition for SIREs with COLAd income sources.

 

[teej1985]

Nothing wrong with that methodology as long as your income sources keep pace with your (inflation related) spending increases. Predicting investment returns and inflation returns 20-30 years out isn't a no brainer if you're relying mostly/solely on investment returns. May be a different proposition for SIREs with COLAd income sources.

What is a SIRE?

 

[Al18]

FYI the OP hasn’t visited this forum since Saturday at 3:37 pm

 

[Koolau]

What is a SIRE?

Poll:Which are you/will you be? SIRE or FIRE?

SIRE (secure income, retire early). Consider yourself SIRE if more than half your retirement expenses are/will be covered by a pension (COLA or non), free or subsidized retiree health care, Social Security or other government and/or Corp benefits. Consider yourself FIRE if more than half...

www.early-retirement.org

secure income, retire early

 

[imjustawarrior]

Nothing wrong with that methodology as long as your income sources keep pace with your (inflation related) spending increases. Predicting investment returns and inflation returns 20-30 years out isn't a no brainer if you're relying mostly/solely on investment returns. May be a different proposition for SIREs with COLAd income sources.

We have a 75/25 AA. I would be surprised if it doesn't beat inflation. If 75/25 doesn't beat inflation, there is going to be a lot of whining on these retirement forums.

Outside our portfolio, our only income will be SS when that starts way down the road.

I don't directly make any predictions. I make my best guesses and see what the retirement calculators say. We are flexible, so we will adjust if need be. I would rather eat rice and beans and hike for entertainment than w*rk.

 

[ERD50]

.... Then recalculate on 12/31/24 using that day's portfolio value to get the spending for the year ahead. He could make that calculation again on 12/31/25. You may have heard this called the "retire again and again" strategy. It's a valid way to approach it, but, in my opinion, you have to be willing to go down as well as up.

That's not my understanding of the "retire again and again" strategy. You only go up! I know, sounds "too good to believe", but that data backs it up (again, assuming my understanding is correct, but I feel pretty confident in this).

Remember, the typical initial SWR calculation says you survive 95% of historical cycles with X% initial spending (inflation adjusted each year). If you run it again, say 5 years later, and your inflation adjusted stash has grown, it will show you can spend more (even w/o accounting for knocking 5 years off your expected lifetime).

And if your stash shrunk, you don't need to lower spending - that was already accounted for in the initial calculation - you survive the worst of the worst (well, 95% of the time - which is why I like to demonstrate this with a truly SWR that survives 100% of history, ~ 3.3%, it just makes the discussion simpler).

The model can only work with its historical database, and sure, the future could be worse than the historical worst, but that's outside of what we can glean from these types of models.

But yes, with the "retire again and again" strategy, you can increase spending if the model says so, but you never need to decrease spending - that was already 'baked in the cake'.

 

[ERD50]

Another way to illustrate the above in maybe simpler terms:

Run FIRECalc with $33,000 spend and $1M portfolio (3.3% initial WR), for 30 years. Result, 100% success.

Now change the "Full Years" to 5, which makes it easier to see what happens after 5 years (the lines blur together and the scales are hard to read in the 30 year simulation at 5 years). One run was at $462,156 at the 5 year mark (down ~ 54% !) - but the important point is that this same run succeeded for the 30 year period, without cutting inflation adjusted spending.

So you see, you don't cut spending on a downturn - that was accounted for in the initial calculation.

edit/add: A bit more playing, and it looks like 9 years in is the low point @ $418,808, down from $1M initial. And... it survives. That would make the WR at year nine a whopping 7.88%, from the initial 3.30%. But that's what it's all about - the initial % accounts for the drops w/o adjusting spending.

 

[Gumby]

That's not my understanding of the "retire again and again" strategy. You only go up! I know, sounds "too good to believe", but that data backs it up (again, assuming my understanding is correct, but I feel pretty confident in this).

Remember, the typical initial SWR calculation says you survive 95% of historical cycles with X% initial spending (inflation adjusted each year). If you run it again, say 5 years later, and your inflation adjusted stash has grown, it will show you can spend more (even w/o accounting for knocking 5 years off your expected lifetime).

And if your stash shrunk, you don't need to lower spending - that was already accounted for in the initial calculation - you survive the worst of the worst (well, 95% of the time - which is why I like to demonstrate this with a truly SWR that survives 100% of history, ~ 3.3%, it just makes the discussion simpler).

The model can only work with its historical database, and sure, the future could be worse than the historical worst, but that's outside of what we can glean from these types of models.

But yes, with the "retire again and again" strategy, you van increase spending if the model says so, but you never need to decrease spending - that was already 'baked in the cake'.

I recall that we had this discussion a very long time ago and I understand the argument for "only go up", which you have laid out well. But think on this:

Suppose you hold only the S&P 500. You retire on Nov 1, 1966, when the S&P 500 was 80.99 and calculate your SWR. Based on the original 4% rule, you would take that amount until Nov 1, 1967 and then adjust for inflation. In Nov 1966, the CPI-U was 32.9. In November 1967, it was 33.8. So inflation was 2.7% over the year and you increase your draw by that amount. As you know, using FIRECalc, you barely survive a 30 year retirement because the market is soon to crash and inflation will run rampant in the 1970s.

But you desire to "retire again" on Nov 1, 1967, when the S&P was 92.66 (14.4% higher), so you increase your draw by that much instead. It is difficult to see how someone who started with the same portfolio in Nov 1966, but after one year jacked his withdrawal up by over 5 times the rate of inflation and then never lowered it when the market tanked but continued to inflation adjust per the original 4% rule could come out ahead of or even equal to the guy who took the same draw from 11/66 to 11/67, then increased by only 1/5th as much in Nov 1967 and continued to inflation adjust per the original 4% rule.

You have the same portfolio dollars on 11/1/67 and the "retire again" you is spending more of them.

some data

S&P 500 Historical Prices by Month - Multpl

S&P 500 Historical Prices table by month, historic, and current data. Current S&P 500 Historical Prices is 6,063.40, a change of +11.18 from previous market close.

www.multpl.com

Consumer Price Index Data from 1913 to 2024

The Consumer Price Index (CPI-U) is provided by the U.S. Department of Labor Bureau of Labor Statistic and is used to measure inflation. The CPI is calculated and released monthly and powers this site's always-current Inflation Calculator, among other things. The government agency last updated...

www.usinflationcalculator.com

 

[audreyh1]

That's not my understanding of the "retire again and again" strategy. You only go up! I know, sounds "too good to believe", but that data backs it up (again, assuming my understanding is correct, but I feel pretty confident in this).

Remember, the typical initial SWR calculation says you survive 95% of historical cycles with X% initial spending (inflation adjusted each year). If you run it again, say 5 years later, and your inflation adjusted stash has grown, it will show you can spend more (even w/o accounting for knocking 5 years off your expected lifetime).

And if your stash shrunk, you don't need to lower spending - that was already accounted for in the initial calculation - you survive the worst of the worst (well, 95% of the time - which is why I like to demonstrate this with a truly SWR that survives 100% of history, ~ 3.3%, it just makes the discussion simpler).

The model can only work with its historical database, and sure, the future could be worse than the historical worst, but that's outside of what we can glean from these types of models.

But yes, with the "retire again and again" strategy, you can increase spending if the model says so, but you never need to decrease spending - that was already 'baked in the cake'.

That is correct. Retire again and again means you only go up. You basically work yourself closer to a worse case scenario, but only if your initial growth is good.

Adjusting to your portfolio value every year, up or down is not the same thing at all.

 

[SecondCor521]

I recall that we had this discussion a very long time ago and I understand the argument for "only go up", which you have laid out well.

Suppose you hold only the S&P 500. You retire on Nov 1, 1966, when the S&P 500 was 80.99 and calculate your SWR. Based on the original 4% rule, you would take that amount until Nov 1, 1967 and then adjust for inflation. In Nov 1966, the CPI-U was 32.9. In November 1967, it was 33.8. So inflation was 2.7% over the year and you increase your draw by that amount. As you know, using FIRECalc, you barely survive a 30 year retirement because the market is soon to crash and inflation will run rampant in the 1970s.

But you desire to "retire again" on Nov 1, 1967, when the S&P was 92.66 (14.4% higher), so you increase your draw by that much instead. It is difficult to see how someone who started with the same portfolio in Nov 1966, but in one year jacked his withdrawal up by over 5 times the rate of inflation and then never lowered it when the market tanked but continued to inflation adjust per the original 4% rule could come out ahead of the guy who took the same draw from 11/66 to 11/67, then increased by only 1/5th as much in Nov 1967 and continued to inflation adjust per the original 4% rule.

Both 1966 and 1967 are "bad start years" and are part of the 5% of the time when the 4% rule fails, at least in FIREcalc. So both of your hypothetical retirees - the normal 4%er and the "retire again and again"er are doomed.

Both 1966 and 1967 succeed if you start with a 3.3% WR in FIREcalc.

FIREcalc also starts on 1/1 of every year. I know that folks have extended the research to monthly starts, but I don't recall the results of that stuff other than it was approximately the same.

One known side effect of the "retire again and again" model is that for WRs that are not 100% safe, you do increase SORR risk. And in any case, retiring again and again does reduce safety margin in general and leaves you with a smaller average terminal balance. This latter point can be viewed either as a bug or a feature.

If you picked two adjacent years that were "success" years and use a 100% historically safe WR (like 3.3%), then you would see that someone who "retired again and again" would still succeed but would have less money left over, on average, at the end.

The original research on this, back when it was termed "POPR" or payout period reset, is at https://retireearlyhomepage.com/popr and may be of interest.

 

[Gumby]

Both 1966 and 1967 are "bad start years" and are part of the 5% of the time when the 4% rule fails, at least in FIREcalc. So both of your hypothetical retirees - the normal 4%er and the "retire again and again"er are doomed.

Both 1966 and 1967 succeed if you start with a 3.3% WR in FIREcalc.
......

Based on what I set forth earlier, there mathematically must be a withdrawal rate somewhere above 3.3% where the guy who retired on 11/1/66 and adhered to the original 4% rule survives but the guy who "retired again" on 11/1/67 does not, just because the second guy is spending more. I don't know precisely what that WR is, but it must exist.

 

[SecondCor521]

Based on what I set forth earlier, there mathematically must be a withdrawal rate somewhere above 3.3% where the guy who retired on 11/1/66 and adhered to the original 4% rule survives but the guy who "retired again" on 11/1/67 does not, just because the second guy is spending more. I don't know precisely what that WR is, but it must exist.

Yep, you're right. But I think that your statements are true for any set of years, not just 66/67. The exact percentage might vary based on starting years, but there will always be some rate that is safe under original 4% rules but not under POPR. "Retiring again" both (a) allows you to draw more, and (b) increases the risk of running out of money. Whether that risk is fatal to one's plan is always the next question.

And there's probably how we're hardwired. Some of us would prefer safety of more dollars in the bank, some of us would prefer that European river cruise with the balcony cabin. We might change our mind over time, too.

It's the age old question of FIRE: How much can I get away with? It's essentially the question that OP is asking at the beginning of this thread. It's also probably the question motivating ERN's gazillion part blog series on a safe WR (The Safe Withdrawal Rate Series - Early Retirement Now). Probably also the question behind a large percentage of the threads here.

 

[ERD50]

I recall that we had this discussion a very long time ago and I understand the argument for "only go up", which you have laid out well. But think on this:

Suppose you hold only the S&P 500. You retire on Nov 1, 1966, when the S&P 500 was 80.99 and calculate your SWR. Based on the original 4% rule, you would take that amount until Nov 1, 1967 and then adjust for inflation. ...

There are a lot of problems with your example, some have been pointed out. I'll detail them here, but that's OK, we can still have the discussion, it's the concept that is important.

As pointed out, to make this easier to discuss, you really need to start with a historically 100% safe WR. That's a problem with your example (4% has ~ 5% failures - the 'again and again' method *will* increase failures with this data set), in addition, your S&P numbers don't include divs, and I don't think you accounted for the previous year WD from the portfolio, so it all gets kinda mushy, number-wise.

But that's OK, let's discuss the concept. Start with historically 100% safe Withdraw % (say 3.3%). Yes, at any point along the way, if the portfolio has grown such that a 3.3% withdraw $ amount is greater than the original calculated 3.3% inflation adjusted withdraw $ amount, you can take the larger $ amount, and inflation adjust that going forward.

I'll try to come back later, and see if I can demonstrate this in that other calculator, it lets you (or used to) select specific years to display.

Here's a key point, I think. When I looked at retire again and again, and reviewed the data, what I came to realize
is that (within this data set), it appears to be the best generalized approach to maximize the amount you can withdraw/spend. Because it is always increasing (when conditions allow it) the withdraw amount within a current 100% safe amount, it minimizes the area under the curve. And those worst case years will never see an increase under the rules. Your example seems to counter that (but the numbers might not be appropriate), so I'll take a closer look later, but from what I recall, that doesn't happen, the worst case years only saw drops in the portfolio, with no opportunities to raise the WD amount under the RA&G rules.

 

[ERD50]

I'll also address this for clarity:

Does the retire again and again approach reduce your margin of safety in the event we have a 'worse than the worst in history' event?

Yes, of course it reduces your margin of safety, there is no free lunch.

But it shows that it can be done within the data set we have. The 100% historically safe covers the worst of the worst, so clearly, all other years could have w/d more and still be 100% safe. Th RA&A method shows how to take advantage of those times, rather than slavishly sticking to the original rate + inflation.

 

[ERD50]

OK, here you go:

https://www.********.com/5adfc5bc-f1aa-4ab8-99c1-9adabcd2925c

I first ran the investigate for 100% success, and got $35,863 (all defaults, $1M starting portfolio), so 3.5863% initial SWR. The failed year was 1966 if I boosted that to $35,900.

Then I ran it (w/o 'investigate') with $35,862, ($1 less, to avoid rounding errors - maybe didn't need to), for years 1964~1968 to make it easier to view the graph.

Look at the graph, the inflation adjusted portfolio never exceeds the original $1M, so there is no opportunity to raise the spend, which makes sense. You can't take more from a year that's on the brink of failure, and conversely, of course you can take more from years that end up with a balance at the end.

The RA&G method optimizes the amount you can WD/spend, and reduces the amount 'left on the table', within the data set provided. How can it be otherwise?

Another 'test' would be to take a different year that does see a significant increase, and run with that increased WD amount starting that year (for the remaining years though - if that was year 5 of 30, you need to cut the time frame to the remaining 25 years).

 

[Gumby]

I gave you the source of the numbers, so you can check whether they are correct. If it matters to the example (although I don't think it does) you could shift to Jan 1, 1967 and Jan 1, 1968. But I am confident that, regardless of year, at some withdrawal rate, a guy who used the "retire again" methodology and increased spending that first year by 5 times as much as inflation and then follows the standard yearly inflation adjustment will fail, when he would not have failed if he had simply stuck with the original "retire and increase for inflation" methodology. Simply put, if you spend more money from the same portfolio, you will have a higher risk of failure, no matter what happens. That is especially the case if you make a larger than inflation increase early in retirement.

It is not substantially different than what I call the Sequence of Inflation Risk. Specifically, if you have high inflation immediately after you retire, your risk of failure increases. It is the same theory as the traditional Sequence of Return Risk we always consider. And if you have both poor returns and high inflation early in your retirement, as the late 60' retirees did, you are especially at risk.

In my hierarchy of goals, not failing is more important than increased spending, which is why, again in my opinion, if you use the retire again method, you should be willing to go down as well as up. I suppose this is like spending a constant 4% of the portfolio balance every year. You are guaranteed never to hit zero balance , but you have to be willing to reduce spending as necessary.

 

[ERD50]

wooops, I forgot the site here blocks that other site. It was FI Calc that I meant to post, I'll try re-doing those numbers there.

 

[ERD50]

OK, I was able to replicate this in FI Calc (the 'share link seems a little flaky, so just enter as I did if you want to test it, my only change was $ amount WD and 100% equities).

To better match Gumby's S&P, I switched to 100% equities, and $36,830 gave me 100% success. Again, the near-failing 1966 shows the inflation adjusted portfolio never went above the starting value, so there was no opportunity to give yourself a raise.

Maybe later, I can try Gumby's approach in a spreadsheet, including divs and withdrawals, but unless both of those calculators are doing it wrong, I don't expect a difference.

 

[Gumby]

OK, here you go:

https://www.********.com/5adfc5bc-f1aa-4ab8-99c1-9adabcd2925c

I first ran the investigate for 100% success, and got $35,863 (all defaults, $1M starting portfolio), so 3.5863% initial SWR. The failed year was 1966 if I boosted that to $35,900.

Then I ran it (w/o 'investigate') with $35,862, ($1 less, to avoid rounding errors - maybe didn't need to), for years 1964~1968 to make it easier to view the graph.

Look at the graph, the inflation adjusted portfolio never exceeds the original $1M, so there is no opportunity to raise the spend, which makes sense. You can't take more from a year that's on the brink of failure, and conversely, of course you can take more from years that end up with a balance at the end.

The RA&G method optimizes the amount you can WD/spend, and reduces the amount 'left on the table', within the data set provided. How can it be otherwise?

Another 'test' would be to take a different year that does see a significant increase, and run with that increased WD amount starting that year (for the remaining years though - if that was year 5 of 30, you need to cut the time frame to the remaining 25 years).

Although I cannot see your graph, I don't doubt that it shows as you describe. What I have been trying to say (perhaps unclearly) is the following hypothetical comparing traditional method versus retire again after 1 year and then use traditional method for subsequent years:

Traditional from day one method

Beginning portfolio = 100,000
Retirement time = t0
Initial Draw = 4,000 (draw on Jan 1)
Inflation rate t0 > t1 = 3%
Portfolio return t0 > t1 = 15%
Portfolio value at t1 = (100,000 - 4,000) x 1.15 = 110,400
Year 2 draw = 4000 x 1.03 = 4120 (draw on Jan 1)
Inflation rate t1 > t2 = 4%
Portfolio return t1 >t2 = -5%
Portfolio value at t2 = (110,400 - 4120) x .95 = 100,966
Year 3 draw = 4120 x 1.04 = 4284.80

Retire again method

Beginning portfolio = 100,000
Retirement time = t0
Initial Draw = 4,000 (draw on Jan 1)
Inflation rate t0 > t1 = 3%
Portfolio return t0 > t1 = 15%
Portfolio value at t1 = (100,000 - 4,000) x 1.15 = 110,400
Year 2 draw = 4000 x 1.15 = 4600 (draw on Jan 1)
Inflation rate t1 > t2 = 4%
Portfolio return t1 >t2 = -5%
Portfolio value at t2 = (110,400 - 4600) x .95 = 100,510
Year 3 draw = 4600 x 1.04 = 4784.00


So we can see that at the end of Year 2, the retire again method already has a smaller portfolio than the traditional from day one method and the gap will only grow in subsequent years, because the draw will always be larger and the smaller portfolio will not gain as much in up years. It is a mathematical certainty that the retire again method will fail before the traditional from day one method for the exact same starting year.

 

[SecondCor521]

Simply put, if you spend more money from the same portfolio, you will have a higher risk of failure, no matter what happens.

Yup. I think ERD50 and I have agreed with you on this point at least several times.

In my hierarchy of goals, not failing is more important than increased spending

I'm a "safety first" person also. Most people here probably are. And if we're talking about an "unlucky" 1966 retiree, we might be averse to RA&A / POPR.

But let's go to the other end of the spectrum in terms of examples, just as a thought experiment. Suppose instead of that one year of good performance in 1966, we have a number of years of good performance (as we have had over the past decade or so). You wake up one fine January 1st and discover that, after a decade of good performance, your 4%-initial-adjusted-for-inflation amount is now 2% of your current portfolio. This is because that good market performance has outstripped inflation by quite a bit over that decade.

At some point, wouldn't you agree, the "lucky" retiree would maybe decide that they'd still be "safe enough" if they raised their WR%? They could compromise and raise it to 3% of their current portfolio and then resume the inflation adjustments. They'd get to spend 50% more (European river cruise anyone) and still be "very safe".

The problem of course with "safe" is that we're discussing the odds of running out of money in the future. Which gets into one's assumptions about future history being better or worse than the past.

We can also have different opinions about how "safe" is safe "enough". You might be comfortable at 3%, I might be comfortable only at 1%. YMMV as they say.

Some people just don't want to spend money on anything, and therefore wouldn't feel the urge to splurge just because they could. Maybe they've been brainwashed into frugality by their parents, or maybe they Fat FIREd or overshot the mark to begin with and already spend all they want or need to. But there are probably many who have something they want to spend more on (like this thread OP).

 

[ERD50]

... Simply put, if you spend more money from the same portfolio, you will have a higher risk of failure, no matter what happens. ...

True, but as I've been pointing out, the RA&A approach will not increase spending on those paths that are in danger of failure (unless those calculators are making mistakes). So the near-failing paths are not affected.

That's the 'beauty' of the approach, it preserves the safety of the near-failures, and allows you to take from the better paths, leaving less on the table at the end. Sure, less on the table means a higher risk of failure if the future is worse than the worst of the past, but we are already playing that game.

I'm not trying to convince anyone to follow this. All I'm saying is that I believe it holds up for the data-set we have. And people should see that there is some comfort factor and leeway if their inflation adjusted portfolio has grown since they retired.

And it partially explains that so-called 'paradox' of two retirees starting their retirement two years apart, and the market has changed, so one can WD more than the other (well, no, they both can). I say 'partially explains, because there is another flaw in that scenario - if they both had the the ame amount two years ago, and the market goes up, the later retiree now has a higher portfolio, but they make them the same, so it's off anyhoe, but that's another story.