Author Topic: New withdrawal method: Retire Again & Again  (Read 10630 times)

Siamond

  • 5 O'Clock Shadow
  • *
  • Posts: 39
New withdrawal method: Retire Again & Again
« on: November 16, 2013, 12:14:14 PM »
In the past year, I've been on a quest to better understand various approaches to withdrawal methods (post retirement). On the way, I came up with a simple method I've not seen described elsewhere. I do NOT believe this is a one-size-fits-all, but still, this is an idea that could work for some, so let me share it... All my numbers below are in 'real' dollars. And my references to 4% or 3% apply to the US, things are different other countries.

Ok, say you save a given stash. Conventional wisdom say that you can spend 4% of it every year (inflation adjusted) post retirement, and you should probably make it. Now if person A has $1M in 2013 and retires, person A gets $40K to spend per year. Now person B has the same stash and keep working one more year, and there is a stock market crash in 2014, the portfolio is now $750K, and Person B retires for a paltry $30K. But wait, person A went through the same crash, and still it's ok for person A to take $40k in 2014? Huh? Let's keep going. In 2015, the market recovers, and the portfolio of both guys went up to $1M. But Person A gets $40K while Person B gets $30K. Are you kidding me?

Fact is the 4% method has a very fundamental flaw, being overly reliant on the starting point. So what if we all try to be Person A (the lucky guy who retired close to a market high, hence his 4% is higher than others!).

Basic idea is simple. Do retire when you feel like it. Take 4% of your stash as annual spend. Next year, look at your portfolio, do the math again (current portfolio times 4%). If this is better than your current withdrawal, well, RETIRE AGAIN, take 4% of your current portfolio, i.e. you got a raise! Now Person B will catch up with Person A. Rinse and repeat every year.

Does this work? Not exactly like that. I used the word "probably" earlier. Fact is 4% can fail if you happen to retire at the very exact craziest peak (1999, anyone?). So if you follow what I just suggested, you might significantly increase the probability of stepping in this failure zone (which existence is another reason to dislike the 4% starting point randomness).

Easy fix. Start at 3.2% or even 3%. Now that is super safe. Too safe? Too little to start with? Well, remember, you will get raises along the way (unless you started at a crazy high peak already). Until you settle to a plateau, which will remain a) very safe b) reasonably optimal c) fairly independent from your starting point.

Want to play with this idea? Well, take a look at http://www.cfiresim.com/input.php and select the appropriate spending method. Thanks Bo, for implementing my idea.

Siamond

  • 5 O'Clock Shadow
  • *
  • Posts: 39
Re: New withdrawal method: Retire Again & Again
« Reply #1 on: November 16, 2013, 12:17:10 PM »
Now what are the pros & cons of this idea?

What’s to like?
-   This is very simple
-   This fixes major flaws in the 4% SWR method
-   You only get 'raises', no spending cuts, no complex dual-budgeting
-   If you’re disciplined enough to start at 3% or so, you are fully safe, and you no longer make a bet on your starting point
-   Being safe doesn’t preclude to reap the benefits of large market upswings (backtest a scenario starting in 1975, and you’ll see what I mean)
-   The method auto-adjusts itself if you receive & invest a large lump sum (e.g. inheritance)

What’s NOT to like:
-   The future may NOT rhyme with the past, so if the market goes in an unprecedented & very sustained drought, the method MAY fail
- You'll probably start at a lower point (e.g. 3.2%) than what you've been told (e.g. 4%)
-   Depending on your starting point, it may take quite some time before you reach a plateau, and you may not like to spend more money when you’re older and less money beforehand… Then you need another withdrawal method.
-   Any recurring side revenue (part-time work, social security, pension, etc) has to be accounted for and tracked separately. You can’t view such revenue as a recurring portfolio increase, this breaks the logic of the method.
« Last Edit: November 16, 2013, 12:23:59 PM by Siamond »

arebelspy

  • Administrator
  • Senior Mustachian
  • *****
  • Posts: 28444
  • Age: -997
  • Location: Seattle, WA
Re: New withdrawal method: Retire Again & Again
« Reply #2 on: November 16, 2013, 12:44:44 PM »
I guess I'm not seeing how this is that different than many of the "recalculate annually" ideas, except that you don't decrease your spending in bad times (one of the biggest things that will lead to failure).  (Unlike, say, Pfau's recent glide path with increasing equity research which was interesting.)

You seem to basically be advocating for a plan that only increases one's WR, without ever decreasing.  That seems awfully risky to me.  So to compensate, you have someone start with a lower SWR.

Why not just go with a normal SWR and not do all the funky increase if X but not if Y shenanigans?

All in all, I agree with the idea of recalculating income and having safety margins to be able to be flexible with spending, but this whole method seems a bit like the tail wagging the dog, IMO.
I am a former teacher who accumulated a bunch of real estate, retired at 29, spent some time traveling the world full time and am now settled with three kids.
If you want to know more about me, this Business Insider profile tells the story pretty well.
I (rarely) blog at AdventuringAlong.com. Check out the Now page to see what I'm up to currently.

AJDZee

  • Bristles
  • ***
  • Posts: 310
Re: New withdrawal method: Retire Again & Again
« Reply #3 on: November 16, 2013, 03:08:58 PM »

You seem to basically be advocating for a plan that only increases one's WR, without ever decreasing.  That seems awfully risky to me.  So to compensate, you have someone start with a lower SWR.

Yeah, my thoughts exactly when reading above...

Taking out 4% WR when you have a bad year still fits into the SWR model only if you don't take more than you were taking before (indexed to inflation) when you have a great year - i.e. your 'raises' you mention above. That's where it evens out, and that's where the premise of using 'average market return' assumptions come into your retirement planning.

The SWR studies don't take into account someone who's adjusting their withdrawal based on what kind of year their investments had - they are blindly taking out their same income every year - and probability for success is still safe.

Siamond

  • 5 O'Clock Shadow
  • *
  • Posts: 39
Re: New withdrawal method: Retire Again & Again
« Reply #4 on: November 17, 2013, 01:33:34 PM »
I guess I'm not seeing how this is that different than many of the "recalculate annually" ideas, except that you don't decrease your spending in bad times (one of the biggest things that will lead to failure).  (Unlike, say, Pfau's recent glide path with increasing equity research which was interesting.)

You seem to basically be advocating for a plan that only increases one's WR, without ever decreasing.  That seems awfully risky to me.  So to compensate, you have someone start with a lower SWR.

I never said to change your SWR. On the contrary, I suggested to stick to a conservative one, but recalibrate your spend based on your current portfolio value. Way different. As to the risk, if you start at a proper point, the risk is actually significantly lower than the regular 4% method. Just run some backtesting and you'll see the point.

A very simple run with cFIREsim with $1M/$40K (a 4% SWR method) leads to 5% failures, and a median spend of $40k (of course!). It's actually worse than that because many people tend to retire when the market is high, which makes the 4% regular method significantly more risky.

While a simple run of $1M/$32K (with RA&A at 3.2%) leads of 0% failures, and a median spend of $46k (could be significantly more if you're a tad lucky with your time period; could also be less, but then you'll be happy to have done so, as otherwise, the portfolio would have tanked).

As to decreasing your spending in bad times, I do agree this is probably a good idea, but this is actually much harder to put in practice than one might think. When? By how much? For how long? Personally, I like Guyton-Klinger for such an adaptive method, but then it's another thread of discussion... I was just explaining one very simple method which works without such complexity...

Quote
Why not just go with a normal SWR and not do all the funky increase if X but not if Y shenanigans?

For the reasons I explained. The randomness of the starting point has really bad side-effects... Read my first post again, maybe?

PS. I am not advocating anything. No withdrawal method is perfect, far from it. I'm just sharing one possible simple approach, that's all...
« Last Edit: November 17, 2013, 01:36:35 PM by Siamond »

arebelspy

  • Administrator
  • Senior Mustachian
  • *****
  • Posts: 28444
  • Age: -997
  • Location: Seattle, WA
Re: New withdrawal method: Retire Again & Again
« Reply #5 on: November 17, 2013, 01:35:09 PM »
If it works for you, go for it.  :)

Here's my main point:
You're doing funky stuff to increase spending but never decrease.  That, in general, just adds risk.

Further, you know that it's riskier, so you advocate starting with a lower SWR so that you can do this method.  Why not just have a normal SWR and not have to?

In other words, you say:
Quote
Why not just go with a normal SWR and not do all the funky increase if X but not if Y shenanigans?

For the reasons I explained. The randomness of the starting point has really bad side-effects... Read my first post again, maybe?

Sure, there's randomness of a starting point. That's why I advocate for flexible withdrawals, rather than rigid.

But I don't see that your method solves that, and it introduces issues of its own.

You specifically increase spending, never decrease it, and start with a lower SWR to compensate for that.  I say keep it simple and just do a more normal SWR with the standard CPI spending increase.

Running a simple scenario:
Load up cFIREsim.  Leave everything default, and run it.  40k spending = 94.7% success rate (6 failures)
Leave everything default and put in Retire Again and Again method.  Drops to a 90.3% success rate (11 failures, almost double).

So your method fails more often, leaving one to have to lower their SWR to compensate (i.e. spend less money).  I'd rather be able to spend more with the standard SWR, and then have the higher success rate.  Further, it's a lot simpler.

I'm just not seeing the advantage to your scenario.
« Last Edit: November 17, 2013, 01:54:01 PM by arebelspy »
I am a former teacher who accumulated a bunch of real estate, retired at 29, spent some time traveling the world full time and am now settled with three kids.
If you want to know more about me, this Business Insider profile tells the story pretty well.
I (rarely) blog at AdventuringAlong.com. Check out the Now page to see what I'm up to currently.

Siamond

  • 5 O'Clock Shadow
  • *
  • Posts: 39
Re: New withdrawal method: Retire Again & Again
« Reply #6 on: November 17, 2013, 02:12:02 PM »
Thanks for providing more details about your feedback, appreciated.

Yes, I know it doesn't work well when starting at 4%, I made it clear in my 1st post, I believe.

I think I addressed your points when I stated the following.

Quote
A very simple run with cFIREsim with $1M/$40K (a 4% SWR method) leads to 5% failures, and a median spend of $40k (of course!). It's actually worse than that because many people tend to retire when the market is high, which makes the 4% regular method significantly more risky.

While a simple run of $1M/$32K (with RA&A at 3.2%) leads of 0% failures, and a median spend of $46k (could be significantly more if you're a tad lucky with your time period; could also be less, but then you'll be happy to have done so, as otherwise, the portfolio would have tanked).

So you fear more risk? Nope, this significantly REDUCES risk (not strictly to zero, as the future might be less rosy than the past, but much better than the classic method - notably with the bias of often retiring at a high portfolio point).

And you'd like to spend more? Well, this is exactly what happens (cf. the higher median spend), and maybe more important, all the cases where things get rosier than that. It's just that the spend increase only happens when it is safe to do so.

Overall, this is safer and more adaptive (and definitely less 'random') than the classic method, while staying quite true to its core principle.

This also sounds especially suitable for early retirees, you take your 3.2%, you take your 'raises' when they come, and while you're still relatively young, you'll probably develop some side revenues to beef up your spending level.

Look, this is indeed a very coarse and simple method. Definitely not perfect (this is why I included Pros & Cons). I would guess that you are a supporter of more adaptive methods (like Hebeler or Guyton-Klinger). Full disclosure: me too! But they are harder to implement and to understand, and they have their own load of issues. So... again, I was just sharing one simple idea which works...

Thanks for the feedback.

arebelspy

  • Administrator
  • Senior Mustachian
  • *****
  • Posts: 28444
  • Age: -997
  • Location: Seattle, WA
Re: New withdrawal method: Retire Again & Again
« Reply #7 on: November 17, 2013, 02:24:50 PM »
Under what basis do you claim it reduces risk?

In every single scenario I run, Retire Again and Again has a higher failure rate.

It only reduces risk if you drop your initial WR, but then you aren't comparing apples to apples.  Of course random.retirement.method.X.Y.Z can beat the standard method when employing a 3.2% initial WR compared to a 4% SWR.
I am a former teacher who accumulated a bunch of real estate, retired at 29, spent some time traveling the world full time and am now settled with three kids.
If you want to know more about me, this Business Insider profile tells the story pretty well.
I (rarely) blog at AdventuringAlong.com. Check out the Now page to see what I'm up to currently.

Siamond

  • 5 O'Clock Shadow
  • *
  • Posts: 39
Re: New withdrawal method: Retire Again & Again
« Reply #8 on: November 17, 2013, 02:40:56 PM »
Yes, true, one should NOT compare classic-SWR and RA&A with the same starting point. What you need to compare is the full risk *AND* reward (i.e. spend) trajectory when using sensible entry parameters.

Ok, so maybe I should have said improve the risk/reward balance? This is really the point, not risk in absolute, you're right.

People go to 4% because they want a better 'reward' (spend), and so they take this 5% failure risk (or more). And then they often fall in two very unfortunate scenarios, either we fall in the failure zone, or their portfolio starts increasing a lot, and they still don't get more reward. That is a very poor risk/reward balance if you ask me.

What I am suggesting strongly minimizes the failure zone, while improving the rewards WHEN IT IS SAFE TO DO (which happens more often than not). That's all.

arebelspy

  • Administrator
  • Senior Mustachian
  • *****
  • Posts: 28444
  • Age: -997
  • Location: Seattle, WA
Re: New withdrawal method: Retire Again & Again
« Reply #9 on: November 17, 2013, 03:05:22 PM »
Hmm, okay.  I'll have to noodle over it under that context.
I am a former teacher who accumulated a bunch of real estate, retired at 29, spent some time traveling the world full time and am now settled with three kids.
If you want to know more about me, this Business Insider profile tells the story pretty well.
I (rarely) blog at AdventuringAlong.com. Check out the Now page to see what I'm up to currently.

 

Wow, a phone plan for fifteen bucks!