Author Topic: Unreliability of long run stock market returns  (Read 26079 times)

SeattleCPA

  • Handlebar Stache
  • *****
  • Posts: 2369
  • Age: 64
  • Location: Redmond, WA
    • Evergreen Small Business
Unreliability of long run stock market returns
« on: September 11, 2017, 06:57:06 AM »
Last week, in another thread, MDM and I got into what MDM described as a possible "violent agreement" about the goodness of the Bogleheads investment philosophy. In the end, all ended well, I think, when we realized we were making the same point.

But that brief discussion highlighted to me how hard it is to explain in words the point I was trying to make. And that point is this: There's a lot more variability in long run stock market returns than some people realize...

I'm going to blog a bit more on this subject this week, but I thought it'd be okay to post here a chart that shows in a picture the point I tried to express in words:



BTW, this chart is the standard "funnel chart" or "long-run stock market returns" chart you've seen a million times before if you noodle much with websites like www.portfoliocharts.com or www.portfoliovisualizer.com. Or if you've read books by John Bogle or Burton Malkiel.

Most of the time, when authors produce these charts, they talk about how the tail of the chart narrows over time. But something else is true, too, which is the tail never compresses to a single value.


Tyler

  • Handlebar Stache
  • *****
  • Posts: 1198
Re: Unreliability of long run stock market returns
« Reply #1 on: September 11, 2017, 10:40:20 AM »
Most of the time, when authors produce these charts, they talk about how the tail of the chart narrows over time. But something else is true, too, which is the tail never compresses to a single value.

Very true.

In addition, the tail on this type of chart is a little deceptive in how it obscures the effects of compound interest on your long-term account values.  Time perception is a tricky concept in investing, and this is a good example of when it helps to look at the numbers from a few different perspectives. 

SeattleCPA

  • Handlebar Stache
  • *****
  • Posts: 2369
  • Age: 64
  • Location: Redmond, WA
    • Evergreen Small Business
Re: Unreliability of long run stock market returns
« Reply #2 on: September 11, 2017, 10:56:15 AM »
Agreed, Tyler. Agreed.

BTW, another MMM member sent me a private message pointing out that it'd be nice to show a histogram of the returns, thereby visually depicting just how fat the tails are.

I think that's a lot of work going the cFIREsim route (which is what I did). Though maybe I'll try to do it. (BTW it's not that much work to use cFIREsim to get the values for a single scenario like 20 years as explained here .)

But two comments: First, I think the 20th percentile to 80th percentile runs roughly 4.4% to 7.2%... so the chance that someone would actually get 0% or 10% real return is infinitesimally small.

Nevertheless, as I think you Tyler are hinting, getting 4% or 5% annual return for 30 or 35 years rather than the 6% average makes a GIGANTIC difference in the outcome.

P.S. I think if you use the Portfolio Chart's rolling returns calculator available here, https://portfoliocharts.com/portfolio/rolling-returns/ , and then set the time frame to something like 30 years, you get a bar chart that gives you a good visual sense of the variability one can experience even if one invests for 30 years or whatever.

Mr. Green

  • Magnum Stache
  • ******
  • Posts: 4485
  • Age: 40
  • Location: Wilmington, NC
Re: Unreliability of long run stock market returns
« Reply #3 on: September 12, 2017, 07:22:55 AM »
Related to that funnel chart, I've always enjoyed the following chart that shows the cumulative average return in 10-year segments over a 30-year period. It's the standard 4% WR on $1 million, with Vanguard level (low) expense fees, using the CAGR from www.moneychimp.com/features/market_cagr.htm. It's interesting to see the occurrences where poor returns toward the end still can't overcome strong performance in the first 10 years, and also just how poor the performance can be in the first 10 years and still go the distance with less than stellar returns over the rest of the period. The individual breakdowns also allow someone to look at their personal performance as their retirement moves along and map it to the most similar historical scenario to see how that panned out. One can also add theoretical returns in the future for a partially finished 30-year (recent) period to see how the remaining unknown returns affect the outcome. It's fascinating to me to see how different sequences effect the ending value.
« Last Edit: September 12, 2017, 07:32:19 AM by Mr. Green »

SeattleCPA

  • Handlebar Stache
  • *****
  • Posts: 2369
  • Age: 64
  • Location: Redmond, WA
    • Evergreen Small Business
Re: Unreliability of long run stock market returns
« Reply #4 on: September 12, 2017, 10:00:24 AM »
Mr Green,

I agree that's a great table.

I share this comment, though, to move our collective knowledge forward.

This quotation from the referenced page, "... The most significant pattern is this: Over the very long run, the stock market has had an inflation-adjusted annualized return rate of between six and seven percent..." seems incomplete to me.

Absolutely no argument from me that the average is the average. And totally agree that's significant.

But I would also suggest that another significant pattern is the variability in the returns even over long accumulations.

I now have the blog post that talks in detail about this:

https://evergreensmallbusiness.com/long-run-stock-market-returns/

But one significant issue that gets missed (and I think is possibly missing from the table data ) is that people saving for 30 years don't actually get that 30 year period to average out to the 30-year average.

If they're saving for 30 years, they get 30 years for their first contribution, 29 years for the second contribution, and so on. In the end, it's sort of like they're really on average only investing for 15 or so years. (We all already know this of course.)

What is surprising to many of us--or least me--is that if one recalculates the returns using an annuity, one widens the range of outcomes by about 70%.

E.g., if the old range from worst to best measured maybe 3% to 8% with a balanced portfolio, the new range measures from 1% to 10%.

max9505672

  • Bristles
  • ***
  • Posts: 336
Re: Unreliability of long run stock market returns
« Reply #5 on: September 12, 2017, 10:25:46 AM »
This is new to me so I comment to follow.

So the difference between best and worst case would be which 30 year period you study?
« Last Edit: September 12, 2017, 10:40:03 AM by max9505672 »

DrF

  • Bristles
  • ***
  • Posts: 464
Re: Unreliability of long run stock market returns
« Reply #6 on: September 12, 2017, 11:14:01 AM »
Seattle, I see what you're getting at.

*The point* - Because each person starts investing at different dates with different amounts of money, their returns will be wildly different over the course of their saving timeframe. (We'll assume that everyone has the same asset allocation and identical funds to make this easier).

Sequence of return risk can bite you on the front end, as well as the rear!
https://earlyretirementnow.com/2017/05/17/the-ultimate-guide-to-safe-withdrawal-rates-part-14-sequence-of-return-risk/

The 3 groups below are buy and hold, a retiree who is drawing down their portfolio, and a saver adding to their portfolio.



Again, same 3 groups.



Summary, the sequence of stock market returns makes no difference if you buy and hold, while the retiree vs saver are inversely correlated with each other.

So... **Is there anything out there that can predict future returns** - Why yes, one widely used is CAPE.

https://earlyretirementnow.com/2017/08/30/the-ultimate-guide-to-safe-withdrawal-rates-part-18-flexibility-cape-based-rules/



If you use CAPE to guide how your portfolio/investments will do over the next 10 years you can more accurately gauge your expected return (on the money you have invested at the current moment).

FYI, the next 10 years don't look so pretty.

SeattleCPA

  • Handlebar Stache
  • *****
  • Posts: 2369
  • Age: 64
  • Location: Redmond, WA
    • Evergreen Small Business
Re: Unreliability of long run stock market returns
« Reply #7 on: September 12, 2017, 12:06:23 PM »
This is new to me so I comment to follow.

So the difference between best and worst case would be which 30 year period you study?

If you use cFIREsim and say 75% stocks and 25% bonds, assume an annuity (like regular savings into an IRA or 401(k)) and then accept cFIREsim's defaults, you're looking at a worst to best case range of basically 0% to 10%...

As I discuss in the blog post referenced earlier, this range is wider than people "think" for probably two big reasons:
1. They (we) miss fact that the "funnel" has always shown a gap of 3% or so... Maybe 4%...
2. If you use an annuity (annual savings) rather than a one-time deposit, you expand the range by 70% or so if you're going 75%/25%...

SeattleCPA

  • Handlebar Stache
  • *****
  • Posts: 2369
  • Age: 64
  • Location: Redmond, WA
    • Evergreen Small Business
Re: Unreliability of long run stock market returns
« Reply #8 on: September 12, 2017, 12:12:45 PM »
Seattle, I see what you're getting at.

*The point* - Because each person starts investing at different dates with different amounts of money, their returns will be wildly different over the course of their saving timeframe. (We'll assume that everyone has the same asset allocation and identical funds to make this easier).

Sequence of return risk can bite you on the front end, as well as the rear!
https://earlyretirementnow.com/2017/05/17/the-ultimate-guide-to-safe-withdrawal-rates-part-14-sequence-of-return-risk/

The 3 groups below are buy and hold, a retiree who is drawing down their portfolio, and a saver adding to their portfolio.



Again, same 3 groups.



Summary, the sequence of stock market returns makes no difference if you buy and hold, while the retiree vs saver are inversely correlated with each other.

So... **Is there anything out there that can predict future returns** - Why yes, one widely used is CAPE.

https://earlyretirementnow.com/2017/08/30/the-ultimate-guide-to-safe-withdrawal-rates-part-18-flexibility-cape-based-rules/



If you use CAPE to guide how your portfolio/investments will do over the next 10 years you can more accurately gauge your expected return (on the money you have invested at the current moment).

FYI, the next 10 years don't look so pretty.

Yes, exactly.

BTW, one of my favorite financial bloggers sent me a private message pointing to this very well written paper that covers the same ground with lots of more academic rigor than I did in my post:

http://www.norstad.org/finance/risk-and-time.html

I had not read this article when I wrote my blog post... but like many of you I've been alert to Paul Samuelson saying this over the years... and then I think I always notice when Larry Swedroe references this risk.

P.S. I will note that Norstad's suggested 'new and improved chart' to show this risk we're talking about also makes the mistake of using a one-time deposit rather than an annuity.

« Last Edit: September 12, 2017, 02:49:42 PM by SeattleCPA »

DrF

  • Bristles
  • ***
  • Posts: 464
Re: Unreliability of long run stock market returns
« Reply #9 on: September 12, 2017, 12:27:19 PM »
Here's my advice for everyone.

Go 90% the riskiest stock allocation you can find.

With the 10% you have left in your portfolio you will buy 2yr treasury futures contracts. (I haven't worked out the exact math, but you want your notional treasury futures to be ~8-10x your stock portfolio value).

Roll your treasury contracts every quarter and buy more if you fall below ~8-10x of your stock value.

Now a 50% stock drop turns into just 25%.

SeattleCPA

  • Handlebar Stache
  • *****
  • Posts: 2369
  • Age: 64
  • Location: Redmond, WA
    • Evergreen Small Business
Re: Unreliability of long run stock market returns
« Reply #10 on: September 12, 2017, 12:47:25 PM »
Here's my advice for everyone.

Go 90% the riskiest stock allocation you can find.

With the 10% you have left in your portfolio you will buy 2yr treasury futures contracts. (I haven't worked out the exact math, but you want your notional treasury futures to be ~8-10x your stock portfolio value).

Roll your treasury contracts every quarter and buy more if you fall below ~8-10x of your stock value.

Now a 50% stock drop turns into just 25%.

I'm thinking a little different... I like the idea of pairing the traditional plan "a" with a backup plan "b":

https://evergreensmallbusiness.com/retirement-plan-b-need-one/

But I acknowledge that this is really a lot about how one feels about the risk. Probably (possibly) many of us would agree that it's important to be sure one understands the risks as viscerally as one can.

aperture

  • Bristles
  • ***
  • Posts: 475
  • Location: Colorado
Re: Unreliability of long run stock market returns
« Reply #11 on: September 12, 2017, 12:48:45 PM »
PTF - on my lunch break - going to need way more time to read and understand (time may not be the only needed factor).  Thanks for the great discussion. -aperture

SeattleCPA

  • Handlebar Stache
  • *****
  • Posts: 2369
  • Age: 64
  • Location: Redmond, WA
    • Evergreen Small Business
Re: Unreliability of long run stock market returns
« Reply #12 on: September 12, 2017, 02:09:54 PM »
PTF - on my lunch break - going to need way more time to read and understand (time may not be the only needed factor).  Thanks for the great discussion. -aperture

Here's the best way I can think of for someone to understand this...

1. Go to cFIREsim.
2. Enter 0 into the Portfolio Value and Initial Yearly Spending boxes.
3. Scroll down to the Other Income area.
4. Enter the annual savings amount you'll do into the Other Income area's Amount box.
5. Enter 2017 into the Other Income area's Start Year box.
6. Enter 2046 into the Other Income area's End Year box.
7. Scroll back up to the top of the window and click the Run Simulation button. cFIREsim calculates away, draws a chart of outcomes and then gives you a series of ending values: average, median, st. dev. highest and lowest. E.g., if you calculate these values for the scenario where you save $5,500 a year for 30 years, you get these results:

Average:   $454,151   $0   $0
Median:   $459,246   $0   $0
St. Dev.:   $137,596   $0   $0
Highest:   $835,424   $0   $0
Lowest:   $170,973   $0   $0

You may not need to go any further... but if you want, you can take average, median,highest or lowest value and use it to calculate the return  (actually the internal rate of return) with Excel. Here's the formula you would use to calculate the internal rate of return of the best case scenario:

=RATE(30,-5500,,835424)

The function returns 10% (or if you want to be precise 9.5678%)

Here's the formula you would use to calculate the internal rate of return of the worst case scenario:

=RATE(30,-5500,,170973)

The function returns 0% (or if you want to be precise .244%)

Note: By changing the calculation inputs, you can probably tweak cFIREsim's numbers to get something pretty close to your actual plan.

DrF

  • Bristles
  • ***
  • Posts: 464
Re: Unreliability of long run stock market returns
« Reply #13 on: September 12, 2017, 02:14:38 PM »
I'm thinking a little different... I like the idea of pairing the traditional plan "a" with a backup plan "b":

https://evergreensmallbusiness.com/retirement-plan-b-need-one/

But I acknowledge that this is really a lot about how one feels about the risk. Probably (possibly) many of us would agree that it's important to be sure one understands the risks as viscerally as one can.

It's all about risk adjusted return right?

So, if you're comfortable with 100% equity, you are reducing your risk when you add 8-10x equity portion in short term bonds.

I saw this post and was bummed by the lazyness of it. Why would you leverage the equity without leveraging the bond portion? Look for my comment at the bottom of the article.

Basically, you need to model 100% equity with 900% short term bond. There's no good way to do this with any of the calculators, so you truncate it all the way down to 10% equity and 90% bond. Then you multiply all the findings by 10.
« Last Edit: September 12, 2017, 02:27:35 PM by DrF »

SeattleCPA

  • Handlebar Stache
  • *****
  • Posts: 2369
  • Age: 64
  • Location: Redmond, WA
    • Evergreen Small Business
Re: Unreliability of long run stock market returns
« Reply #14 on: September 12, 2017, 03:02:31 PM »

It's all about risk adjusted return right?


Gosh, we may be talking about different issues here. Sorry that I didn't realize that sooner.

The issue I'm trying to highlight is those funnel charts and popular thinking at places like Bogleheads say you bear less risk if you invest for a long time... like for decades. And that's not true.

The Nobel laureates say it's not true.

The funnel charts you can see in any book that talks about long run stock investments and risks show it (even though they sort of pretend it is true... see Bogle's books and Malkiel's...)

You can pretty easily use cFIREsim to prove with historical data we still bear massive risk even if you have a really long horizon. (See my earlier set of "how to" instructions.

DrF

  • Bristles
  • ***
  • Posts: 464
Re: Unreliability of long run stock market returns
« Reply #15 on: September 12, 2017, 03:17:31 PM »
We started off talking about the same thing. I agree that the "long return average" is highly dependent upon when you begin and how quickly you save.

Once we were in agreement, I switched things up on you and began talking about how to reduce risk and increase return.

secondcor521

  • Walrus Stache
  • *******
  • Posts: 5493
  • Age: 54
  • Location: Boise, Idaho
  • Big cattle, no hat.
    • Age of Eon - Overwatch player videos
Re: Unreliability of long run stock market returns
« Reply #16 on: September 12, 2017, 03:25:38 PM »
Average:   $454,151   $0   $0
Highest:   $835,424   $0   $0
Lowest:   $170,973   $0   $0

I think I get your point:  Some people get lucky saving up for retirement (if their accumulation years match up with good market returns - particularly their earlier years of savings).  Some people get unlucky saving up for retirement.  And I agree that this happens.

So what?  What can the average FIRE-aspiring Joe do in response to this reality, since market returns and when in history Joe happened to live are both out of his control?

In my case, my retirement-savings years were from about 1993 to 2016.  Back in the late 90's, I started doing a spreadsheet to predict my retirement date.  Over the years, I updated it with actual account balances.  No matter what I did, I really couldn't get the spreadsheet to predict a date much past age 48, and it predicted age 47 quite often.  I retired at age 46 about 2 years ago and thought that my technique was effective, but your point seems to imply I just got lucky in that my actual ROI matched my predicted ROI (and I predicted market averages for everything - market returns, inflation, etc.) pretty closely.

I suspect part of your reply is for Joe to have a plan B, which I agree may be reasonable depending on what that plan B is.  I visited your website but didn't see the article on that - it may not be posted yet.

SeattleCPA

  • Handlebar Stache
  • *****
  • Posts: 2369
  • Age: 64
  • Location: Redmond, WA
    • Evergreen Small Business
Re: Unreliability of long run stock market returns
« Reply #17 on: September 12, 2017, 03:30:50 PM »
We started off talking about the same thing. I agree that the "long return average" is highly dependent upon when you begin and how quickly you save.

Once we were in agreement, I switched things up on you and began talking about how to reduce risk and increase return.

Okay, maybe i'm not supposed to be laughing at this. But i am. :-)

I don't want to start a donnybrook, but I think the risk reduction stuff quickly steers us off into an MPT discussion...which I'm happy to have. But you need to give me a heads-up that the game has changed...

DrF

  • Bristles
  • ***
  • Posts: 464
Re: Unreliability of long run stock market returns
« Reply #18 on: September 12, 2017, 03:36:33 PM »
My apologies for hijacking the thread. Once I was clear on what you were saying, I was ready to move on to the topic on the top of my mind.

SeattleCPA

  • Handlebar Stache
  • *****
  • Posts: 2369
  • Age: 64
  • Location: Redmond, WA
    • Evergreen Small Business
Re: Unreliability of long run stock market returns
« Reply #19 on: September 12, 2017, 03:54:57 PM »
I think I get your point:  Some people get lucky saving up for retirement (if their accumulation years match up with good market returns - particularly their earlier years of savings).  Some people get unlucky saving up for retirement.  And I agree that this happens.

Exactly. We are on the same page.

So what?  What can the average FIRE-aspiring Joe do in response to this reality, since market returns and when in history Joe happened to live are both out of his control?

Well first, if you're already FIRE, you're actually past this risk I'm wringing my hands about... you've made your way through the rapids. The risks once someone is FI or FIRE is all about the safe withdrawal stuff. Or at least that's the way it seems to me.

What I think we (you, me, the others who understand this) can do is spotlight the reality that we were greatly assisted by a good patch of investment history.

And then here's the reality for people who still bear this risk (so people still working toward FI...) I think people ought to plan for variability in outcomes, maybe save a little more, maybe keep open the option of working a little longer, etc.

Finally, if this reality is totally new to someone, it seems possible that someone wants to think further about the risk baked into the portfolio. Because if this stuff is totally new, maybe someone bears more risk than they realize.

In my case, my retirement-savings years were from about 1993 to 2016.  Back in the late 90's, I started doing a spreadsheet to predict my retirement date.  Over the years, I updated it with actual account balances.  No matter what I did, I really couldn't get the spreadsheet to predict a date much past age 48, and it predicted age 47 quite often.  I retired at age 46 about 2 years ago and thought that my technique was effective, but your point seems to imply I just got lucky in that my actual ROI matched my predicted ROI (and I predicted market averages for everything - market returns, inflation, etc.) pretty closely.

I think that you did get lucky. And just to say it , I think I did too. (I started investing in the early 80s using a SEP-IRA.)

But you were also tactically smart because you were continually double-checking the math. That's probably something else someone should do...

BTW, you might find it interesting to visit www.portfoliocharts and find a heat map of the portfolio closest to what you used. When I do this, I see that I basically accidentally used a portfolio that's given me nearly the long-run average exactly... I.e., I got (and I bet you might have too) the average return or a little better. Which is kind of a coincidence that hides the variability.

I suspect part of your reply is for Joe to have a plan B, which I agree may be reasonable depending on what that plan B is.  I visited your website but didn't see the article on that - it may not be posted yet.

Sorry... it starts here:
https://evergreensmallbusiness.com/retirement-plan-b-need-one/

I tried to give some step-by-step instructions for popular online calculators here: FIRECalc, cFIREsim and PortfolioVisualizer so someone can better understand a likely range of outcomes.

And then I pulled together the tactical options I thought probably made sense to consider here:

https://evergreensmallbusiness.com/retirement-plan-b-tips/


SeattleCPA

  • Handlebar Stache
  • *****
  • Posts: 2369
  • Age: 64
  • Location: Redmond, WA
    • Evergreen Small Business
Re: Unreliability of long run stock market returns
« Reply #20 on: September 12, 2017, 03:56:11 PM »
My apologies for hijacking the thread. Once I was clear on what you were saying, I was ready to move on to the topic on the top of my mind.

I don't think you have any reason to apologize! Aren't these threads basically communal property? :-)

secondcor521

  • Walrus Stache
  • *******
  • Posts: 5493
  • Age: 54
  • Location: Boise, Idaho
  • Big cattle, no hat.
    • Age of Eon - Overwatch player videos
Re: Unreliability of long run stock market returns
« Reply #21 on: September 12, 2017, 04:28:18 PM »
...

Thanks for the detailed reply.

I agree I am past this for myself, but I do have three children who are 22, 17, and 15, and I'd like to advise them as best I can.  One general piece of advice I tend to give them (and I think it's in your "Plan B tips" article) is to just save more - say 20%+.  In my case, I was saving over half of a good salary from 2009 to 2016, which obviously helped me a great deal.  I also talk with them about asset allocation and all the other personal financial stuff.

I think another thing that could have happened with me is that as I was watching my prediction I was probably subconsciously working to make it reality.  So if the market was dropping, I probably reflexively increased my savings rate to compensate.  Or I worked to reduce my spending so I would hit that 4% SWR at age 47.

(My portfolio was very roughly 100% S&P500 low-cost index funds the entire time, with some NQSOs and some section 1202 small-business stock thrown in at random times.  I think my spreadsheet assumed 11% ROI and 3% inflation of both wages and prices.  In reality I suspect that the ROI was lower, the wage increases were higher, and inflation was lower, so I did do better than average probably.)

Your last link, with your tactical tips, was what I was interested in with my original question.  Thanks!

max9505672

  • Bristles
  • ***
  • Posts: 336
Re: Unreliability of long run stock market returns
« Reply #22 on: September 12, 2017, 07:39:16 PM »
Average:   $454,151   $0   $0
Median:   $459,246   $0   $0
St. Dev.:   $137,596   $0   $0
Highest:   $835,424   $0   $0
Lowest:   $170,973   $0   $0
Man, I am grateful to have come across this thread. Not many people here talk about this and only use the 6-7%/year return as being guaranteed.

So, I've run different scenarios on cFIREsim and yes, those lowest ending values seem pretty scary and are a possibility. But I would be curious to see the probabilities of each cases. You already mentioned the 20th percentile to 80th percentile runs roughly 4.4% to 7.2%, how did you calculate it? Are there any calculations/tables out there with ''realistic'' ROI percentage range depending on the length of the period invested, something like a normal distribution?

Finally, from what I understand, the intend of this thread is to inform people that the traditional expected 6% to 7% shouldn't be considered as a fix variable, and that, as secondcor521 mentioned, you can either get (very) lucky or get (very) unlucky. But at the end of the day, there's nothing you can really do about it. Yes you can (and should) have plan B's, but the general investment strategy popular with this community, i.e. investing as much as possible until you reach you desired ''FI'' or ''FIRE'' or ''whatever other goal ''amount without trying to time the market, will be the same, right?

I feel like I have a lot more reading to do on this (thanks for the multiple links you posted), but do some poeple diverge from this strategy in order to better control the variability?

max9505672

  • Bristles
  • ***
  • Posts: 336
Re: Unreliability of long run stock market returns
« Reply #23 on: September 12, 2017, 07:47:32 PM »


If you use CAPE to guide how your portfolio/investments will do over the next 10 years you can more accurately gauge your expected return (on the money you have invested at the current moment).

FYI, the next 10 years don't look so pretty.
Any existing CAPE calculator for investments available?

SeattleCPA

  • Handlebar Stache
  • *****
  • Posts: 2369
  • Age: 64
  • Location: Redmond, WA
    • Evergreen Small Business
Re: Unreliability of long run stock market returns
« Reply #24 on: September 12, 2017, 08:17:50 PM »
But I would be curious to see the probabilities of each cases. You already mentioned the 20th percentile to 80th percentile runs roughly 4.4% to 7.2%, how did you calculate it?

This will give you a little bit more granular understanding (see especially the little table about a third of the way down the post):

https://evergreensmallbusiness.com/average-retirement-savings-return/

I've also got some tables of probabilities I did for the blog posts I've been doing the last few weeks. I'll see if I can put together some that's useful so people have a better understanding of the distribution of returns.

Finally, from what I understand, the intend of this thread is to inform people that the traditional expected 6% to 7% shouldn't be considered as a fix variable, and that, as secondcor521 mentioned, you can either get (very) lucky or get (very) unlucky. But at the end of the day, there's nothing you can really do about it. Yes you can (and should) have plan B's, but the general investment strategy popular with this community, i.e. investing as much as possible until you reach you desired ''FI'' or ''FIRE'' or ''whatever other goal ''amount without trying to time the market, will be the same, right?

I think that's right. An FI or FIRE strategy doesn't change. But your tactics may need to be changed. I.e., you may need to save more or be able to work less, may need to work longer or be able to work shorter, etc.


I feel like I have a lot more reading to do on this (thanks for the multiple links you posted), but do some people diverge from this strategy in order to better control the variability?

Sadly, I think most people haven't even recognized this "variability" issue. You are way ahead if you know it's not exactly a lock.


gerardc

  • Pencil Stache
  • ****
  • Posts: 767
  • Age: 40
  • Location: SF bay area
Re: Unreliability of long run stock market returns
« Reply #25 on: September 12, 2017, 08:57:28 PM »
Most of the time, when authors produce these charts, they talk about how the tail of the chart narrows over time. But something else is true, too, which is the tail never compresses to a single value.

Are you talking about the tail after a predefined period, e.g. 30 years? If so, this is trivially true. If not, please provide proof. I seriously doubt that's the case.

max9505672

  • Bristles
  • ***
  • Posts: 336
Re: Unreliability of long run stock market returns
« Reply #26 on: September 13, 2017, 05:49:51 AM »
But I would be curious to see the probabilities of each cases. You already mentioned the 20th percentile to 80th percentile runs roughly 4.4% to 7.2%, how did you calculate it?

This will give you a little bit more granular understanding (see especially the little table about a third of the way down the post):

https://evergreensmallbusiness.com/average-retirement-savings-return/

I've also got some tables of probabilities I did for the blog posts I've been doing the last few weeks. I'll see if I can put together some that's useful so people have a better understanding of the distribution of returns.
Thanks for the link, very helpful.

I would also love to see those tables of probabilities.

P.S. I've tried to find the thread you were talking about in your first post but couldn't: ''Last week, in another thread, MDM and I got into what MDM described as a possible "violent agreement"...'' Could you please link it here?

SeattleCPA

  • Handlebar Stache
  • *****
  • Posts: 2369
  • Age: 64
  • Location: Redmond, WA
    • Evergreen Small Business
Re: Unreliability of long run stock market returns
« Reply #27 on: September 13, 2017, 06:39:44 AM »
Nice thread.  Be prepared for reactionary responses, though this is probably over the heads of most on this forum.  Of note is the funnel is backwards looking and does not truely represent the actual possible range outcomes (either positive or negative...see Man in the High Tower).

The MMM community would provide a great service to individual investors if it led a discussion on this topic. Most people, I worry, believe a myth... the myth that returns will even out over time and allow the individual investor to earn the average...

My personal opinion is that there are to many fundamental disruptive forces to the global economy and capital markets to rely solely on historic financial market data to predict future returns.  We all trust the system, but some elements that historically drove demand for investor capital are eroding (lower financial barriers to the creation of a global enterprise, being one).  I am none the less optimistic, the world having some many fundamental wealth creation engines under construction.

I agree we can't reflexively extrapolate. But we probably would also agree history gives no indication that we can count on the sort of stability and predictably that means every individual saving for retirement, in the end, gets the same average return. Or nearly the same average return.

SeattleCPA

  • Handlebar Stache
  • *****
  • Posts: 2369
  • Age: 64
  • Location: Redmond, WA
    • Evergreen Small Business
Re: Unreliability of long run stock market returns
« Reply #28 on: September 13, 2017, 06:49:57 AM »
... I am past this for myself, but I do have three children who are 22, 17, and 15, and I'd like to advise them as best I can.  One general piece of advice I tend to give them (and I think it's in your "Plan B tips" article) is to just save more - say 20%+.  In my case, I was saving over half of a good salary from 2009 to 2016, which obviously helped me a great deal.  I also talk with them about asset allocation and all the other personal financial stuff.

I'm the dad to two growth daughters, ages 25 and 26, so I think about this a lot too.

Agree with your thoughts above... and as a parent, I propose these three additional notions:
1. It would not be surprising if our kids don't have the investment good luck we've had, so I want to careful I'm not intentionally or unintentionally setting unrealistic expectations.
2. I want to, if at all possible, leave some meaningful financial legacy for them if I can. E.g., I'm still working (with one of my daughters in fact) but if I need to dial down my consumption in retirement a bit to increase chances there's a nice nest egg for them, I want to try and do that.
3. Our kids can adequately deal with a below average patch of financial history if they save more... which probably means a good job or income production activity (so higher earnings and a longer runway) becomes something to think about.

ooeei

  • Handlebar Stache
  • *****
  • Posts: 1142
Re: Unreliability of long run stock market returns
« Reply #29 on: September 13, 2017, 06:53:55 AM »

The MMM community would provide a great service to individual investors if it led a discussion on this topic. Most people, I worry, believe a myth... the myth that returns will even out over time and allow the individual investor to earn the average...


I'd like to see how your chart changes if you add in a constant investment metric. During the accumulation phase when you're constantly investing in the market, I suspect the lines do get closer together. They still won't touch, but it will help avoid the worst (and best) case scenarios in a given time period.

This doesn't help much during the draw down phase, but I think most people do actually know that's somewhat up to luck. Usually the accumulation phase is where people "assume the average."

Related to that funnel chart, I've always enjoyed the following chart that shows the cumulative average return in 10-year segments over a 30-year period. It's the standard 4% WR on $1 million, with Vanguard level (low) expense fees, using the CAGR from www.moneychimp.com/features/market_cagr.htm. It's interesting to see the occurrences where poor returns toward the end still can't overcome strong performance in the first 10 years, and also just how poor the performance can be in the first 10 years and still go the distance with less than stellar returns over the rest of the period. The individual breakdowns also allow someone to look at their personal performance as their retirement moves along and map it to the most similar historical scenario to see how that panned out. One can also add theoretical returns in the future for a partially finished 30-year (recent) period to see how the remaining unknown returns affect the outcome. It's fascinating to me to see how different sequences effect the ending value.

I'm confused as to how this has a different result than the data that led to the 4% rule in the first place. It's my understanding 4% always worked for 30 years (even though you might be broke at the end of 30 years). Are you using 100% stocks?


SeattleCPA

  • Handlebar Stache
  • *****
  • Posts: 2369
  • Age: 64
  • Location: Redmond, WA
    • Evergreen Small Business
Re: Unreliability of long run stock market returns
« Reply #30 on: September 13, 2017, 07:17:53 AM »
Most of the time, when authors produce these charts, they talk about how the tail of the chart narrows over time. But something else is true, too, which is the tail never compresses to a single value.

Are you talking about the tail after a predefined period, e.g. 30 years? If so, this is trivially true. If not, please provide proof. I seriously doubt that's the case.

Gerardc,

I may not understand your comment, but I'm thinking you're asking for proof there's tons more variability in outcomes than many people commonly believe. So I'm going to address that issue... try to "provide proof"... I'm also going to break this response out into sections so people can choose a method of proof that meshes with the way they like to process information.

For People Who Know How to Use cFIREsim and Microsoft Excel:

In message response #12 above, I provide the steps for using the cFIREsim online calculator to calculate the average, worst and best case returns historically for a given retirement savings plan. I also describe how to use Microsoft Excel to quickly calculate the average annual returns for the investor who experienced the worst case scenario, the average scenario, and the best case scenario.

Going through the steps of using cFIREsim takes less than a minute (I just timed myself)...  And this exercise will provide proof.

For People With Good Spatial Reasoning and Chart Reading Skills


Another pretty easy way to prove this to yourself is to look at any--any--funnel chart you see in books or on websites talking about long return returns.

What you will see is that the worst case and best case scenario returns never converge to a single percentage. The charts always show a gap between the worst and best case. Always, always the charts show this.

The gap varies. Portfolio Charts, it looks like, shows the range of possible average 30-year returns as maybe 5%-ish to 8%-ish for the portfolio I use.

In John Bogle's books and Burton Malkiel's books, the chart shows, roughly, the same sort of 3%-ish to 4%-ish gap.

That gap is what we're talking about here. What I'm saying is that gap exists. That gap is the proof.

And two things to note: First, if the average return of a portfolio is 6% but you or I earn 5% or 4%, that 1% or 2% shortfall over thirty years makes a huge difference.

Second, note that you shouldn't look just at the point on the chart that displays the range from best to worst at the 30-year-marker... some of your or my contributions to an IRA or 401(k) will get 30 years of returns... but some of the money we contribute will only be "in" the market for 2 years or 5 years or 10 years. So look at the range of outcomes at those points in the chart too.

Looking at the funnel charts you see with this mindset will provide the proof every time you see them from now on.

For People Who Just Want to Read a Quick Little Narrative


I did the very best job I could do to explain this in two posts at my blog: Unreliability of long return stock market returns and Retirement Plan B: Why You Need One...

If someone doesn't want to use cFIREsim or doesn't want to try and read the funnel charts, they might want to look at those two blog posts.

BTW, as part of the little "retirement plan b" blog post series I did, I also provides step-by-step instructions to make many of the sorts of calculations one needs to make to see this variability.

SeattleCPA

  • Handlebar Stache
  • *****
  • Posts: 2369
  • Age: 64
  • Location: Redmond, WA
    • Evergreen Small Business
Re: Unreliability of long run stock market returns
« Reply #31 on: September 13, 2017, 07:29:08 AM »
I'd like to see how your chart changes if you add in a constant investment metric. During the accumulation phase when you're constantly investing in the market, I suspect the lines do get closer together. They still won't touch, but it will help avoid the worst (and best) case scenarios in a given time period.

This doesn't help much during the draw down phase, but I think most people do actually know that's somewhat up to luck. Usually the accumulation phase is where people "assume the average."

I'm confused as to how this has a different result than the data that led to the 4% rule in the first place. It's my understanding 4% always worked for 30 years (even though you might be broke at the end of 30 years). Are you using 100% stocks?

The thing about "assuming the average" is the problem. That's what's really shaky as an assumption.

And this important point. On my funnel chart (see message #1), my lines are further apart because I'm assuming you don't get the 30-year averaging out period for any money other than the contribution to an IRA or 401(k) you may at the start of a 30-year accumulation.

For the second contribution in a 30-year accumulation, the investor only gets 29 years for any "averaging out" or "evening out" to occur. For the third contribution, only 28 years, and so on.

Because this subject is such a paradigm shift, let me again make this point: If you redraw a funnel chart to show the range of average annual returns on a 30-year accumulation with annual savings (as I did in the original post in this thread) the tail fattens up... it doesn't get thinner.





SeattleCPA

  • Handlebar Stache
  • *****
  • Posts: 2369
  • Age: 64
  • Location: Redmond, WA
    • Evergreen Small Business
Re: Unreliability of long run stock market returns
« Reply #32 on: September 13, 2017, 07:52:29 AM »
I would also love to see those tables of probabilities.

So I copied the info below from the Excel workbook I used for my blog post on using cFIREsim to make retirement plan b calculations. (That blog post, BTW, describes how to get the numbers and download the full spreadsheet with all the data.)

These numbers show what $5500 a year into an IRA grows into after 35 years using the cFIREsim defaults:

Average   $614,262
Median   $620,358
Min   $231,314
Max   $1,007,843
StDev   $179,489
AvgDev   $152,747

Here are the percentile ranks... I.e.,the 10th percentile is $371,485.  The 50th percentile or median is $620,358.
0.1   $371,485.40
0.2   $437,550.60
0.3   $511,969.90
0.4   $548,222.20
0.5   $620,358.00
0.6   $674,830.20
0.7   $730,134.60
0.8   $800,482.00
0.9   $832,878.80
0.99   $966,887.86

I would think a pragmatic way to look at the above data is to say (a) quite a bit of variability in outcomes... and then (b) it's probably good idea to plan for a range of outcomes from maybe 30th percentile to 70th percentile... or from the 20th to 80th percentile.

BTW if someone here personally knows Bo the creator of cFIREsim and gets his explicit permission, I'm happy to attach the full Excel spreadsheet (with all the cFIREsim data and Bo's calculations)...


P.S. I've tried to find the thread you were talking about in your first post but couldn't: ''Last week, in another thread, MDM and I got into what MDM described as a possible "violent agreement"...'' Could you please link it here?

Here is link:

https://forum.mrmoneymustache.com/investor-alley/brand-new-help-us-make-smart-investment-choices/

What started the possible "violent agreement" was a comment I made about the Bogleheads investment philosophy having some flaws--including (in my opinion) a blindness to this issue of variability in outcomes.

Just to say this, past and current thought leaders in the Bogleheads community like Larry Swedroe and Harry Sit (aka "the finance buff") absolutely get this. But lots and and lots of Bogleheads, in my opinion, don't.

CorpRaider

  • Bristles
  • ***
  • Posts: 442
    • The Corpraider Blog
Re: Unreliability of long run stock market returns
« Reply #33 on: September 13, 2017, 08:15:53 AM »
If you use CAPE to guide how your portfolio/investments will do over the next 10 years you can more accurately gauge your expected return (on the money you have invested at the current moment).

FYI, the next 10 years don't look so pretty.
Any existing CAPE calculator for investments available?

Research Affiliates has an asset allocation tool/function/thing that has CAPEs for various markets.

PizzaSteve

  • Pencil Stache
  • ****
  • Posts: 501
Re: Unreliability of long run stock market returns
« Reply #34 on: September 13, 2017, 08:18:43 AM »
Nice thread.  Be prepared for reactionary responses, though this is probably over the heads of most on this forum.  Of note is the funnel is backwards looking and does not truely represent the actual possible range outcomes (either positive or negative...see Man in the High Tower).

The MMM community would provide a great service to individual investors if it led a discussion on this topic. Most people, I worry, believe a myth... the myth that returns will even out over time and allow the individual investor to earn the average...

My personal opinion is that there are to many fundamental disruptive forces to the global economy and capital markets to rely solely on historic financial market data to predict future returns.  We all trust the system, but some elements that historically drove demand for investor capital are eroding (lower financial barriers to the creation of a global enterprise, being one).  I am none the less optimistic, the world having some many fundamental wealth creation engines under construction.

I agree we can't reflexively extrapolate. But we probably would also agree history gives no indication that we can count on the sort of stability and predictably that means every individual saving for retirement, in the end, gets the same average return. Or nearly the same average return.
Agreed.  Don't want to take it off topic, but in a prior thread I pointed out the idea that investments in personal resiliency, especially for the young (e.g. start a business, education to grow skills, spending flexibility) are the best `ROI' strategies to handle the outlier scenarios.  Overthinking asset allocations has diminishing returns, so i think we totally agree.

What i hope our young forum minds absorb is the notion that some variability of outcomes should be expected.  I dont agree that CAPE or other fairly simple metrics can predict the direction of variability (e.g. we are certain for lesser outcomes).  That is why i bring up the disruptive forces.

Other simple factors that may push returns below the average (examples only):

* Taxes.  Most assume taxes will be the same.  This could be a shaky assumption if Congress eyes fat 401k balances as a source of potential revenue (at least for US MMMers).  Tax law changes can impact capital gains rates (net from taxible) and net withdrawable capital (net from tax diferred)  pushing realized returns down.

* Inflation.  Significant inflation could impact real returns vs nominal returns in a way that challenges a MMM like plan.  None of us know the probability of hyperinflation and equities are probably the best defense anyway, but still inflation impacts could lower our net returns or require increased spending, impacting stash cushion calculations.  Especially true if price increases are asymmetric towards certain goods we all need, like food staples and healthcare.

So, independent of the funnel factors, other variables are also important to take into account (no response needed, dont want too derail thread). Note these could improve or reduce success chances, as taxes or inflation move both positive and negative directions.  They likely have their own funnel charts of probably, though the data to make them would be spotty (e.g. avg tax rates, imoact of real vs nominal returns).  Firecalc may or may not be able to somewhat incorporate them via assumptions.
« Last Edit: September 13, 2017, 08:40:37 AM by PizzaSteve »

SeattleCPA

  • Handlebar Stache
  • *****
  • Posts: 2369
  • Age: 64
  • Location: Redmond, WA
    • Evergreen Small Business
Re: Unreliability of long run stock market returns
« Reply #35 on: September 13, 2017, 08:34:31 AM »
Nice thread.  Be prepared for reactionary responses, though this is probably over the heads of most on this forum.  Of note is the funnel is backwards looking and does not truely represent the actual possible range outcomes (either positive or negative...see Man in the High Tower).

The MMM community would provide a great service to individual investors if it led a discussion on this topic. Most people, I worry, believe a myth... the myth that returns will even out over time and allow the individual investor to earn the average...

My personal opinion is that there are to many fundamental disruptive forces to the global economy and capital markets to rely solely on historic financial market data to predict future returns.  We all trust the system, but some elements that historically drove demand for investor capital are eroding (lower financial barriers to the creation of a global enterprise, being one).  I am none the less optimistic, the world having some many fundamental wealth creation engines under construction.

I agree we can't reflexively extrapolate. But we probably would also agree history gives no indication that we can count on the sort of stability and predictably that means every individual saving for retirement, in the end, gets the same average return. Or nearly the same average return.
Agreed.  Don't want to take it off topic, but in a prior thread I pointed out the idea that investments in personal resiliency, especially for the young (e.g. start a business, education to grow skills, spending flexibility) are the best `ROI' strategies to handle the outlier scenarios.  Overthinking asset allocations has diminishing returns, so i think we totally agree.

What i hope our young forum minds absorb is the notion that some variability of outcomes should be expected.  I dont agree that CAPE or other fairly simple metrics can predict the direction of variability (e.g. we are certain for lesser outcomes).  That is why i bring up the disruptive forces.

Agree. Totally agree. Thanks for contributing to this thread. :-)

ooeei

  • Handlebar Stache
  • *****
  • Posts: 1142
Re: Unreliability of long run stock market returns
« Reply #36 on: September 13, 2017, 08:48:43 AM »
I'd like to see how your chart changes if you add in a constant investment metric. During the accumulation phase when you're constantly investing in the market, I suspect the lines do get closer together. They still won't touch, but it will help avoid the worst (and best) case scenarios in a given time period.

This doesn't help much during the draw down phase, but I think most people do actually know that's somewhat up to luck. Usually the accumulation phase is where people "assume the average."

I'm confused as to how this has a different result than the data that led to the 4% rule in the first place. It's my understanding 4% always worked for 30 years (even though you might be broke at the end of 30 years). Are you using 100% stocks?

The thing about "assuming the average" is the problem. That's what's really shaky as an assumption.

And this important point. On my funnel chart (see message #1), my lines are further apart because I'm assuming you don't get the 30-year averaging out period for any money other than the contribution to an IRA or 401(k) you may at the start of a 30-year accumulation.

For the second contribution in a 30-year accumulation, the investor only gets 29 years for any "averaging out" or "evening out" to occur. For the third contribution, only 28 years, and so on.

Because this subject is such a paradigm shift, let me again make this point: If you redraw a funnel chart to show the range of average annual returns on a 30-year accumulation with annual savings (as I did in the original post in this thread) the tail fattens up... it doesn't get thinner.

Ah, I somehow missed that your chart assumed annual contributions. That is a pretty shocking graph in that case, and my whole point was already taken into account. How do they calculate the average return for that graph?

I find that the numbers get tricky the more in depth you go. For example, average yearly return when you're adding in contributions/withdrawals can be misleading. For example in year 1 maybe you have a 50% loss, but you only have $10k in the account. If the next 28 years you have 0% returns, and in year 30 you have a 50% gain. Your average yearly return over that time is 0%, but you still end up with $442,500 with only $300,000 invested. That's a gain of 47.5% over the money you invested, but your average yearly return can still be said to be 0%.

Flip the returns around and you end up with $300,000 invested, but only $155,000 when all is said and done. A loss of around 48%, with the average return still 0%.

Obviously this is an extreme example, but it goes to show that averaging out returns each year doesn't tell the whole story when we're adding in contributions/withdrawals. When you're adding constantly, the later years have the largest effect. When you're withdrawing constantly, the early years have the largest effect. Interestingly enough these two time frames meet shortly before and after retirement. This means the 5-10 years before and after your retirement are the most important as far as investment returns go.
« Last Edit: September 13, 2017, 08:50:42 AM by ooeei »

SeattleCPA

  • Handlebar Stache
  • *****
  • Posts: 2369
  • Age: 64
  • Location: Redmond, WA
    • Evergreen Small Business
Re: Unreliability of long run stock market returns
« Reply #37 on: September 13, 2017, 10:26:26 AM »
I'd like to see how your chart changes if you add in a constant investment metric. During the accumulation phase when you're constantly investing in the market, I suspect the lines do get closer together. They still won't touch, but it will help avoid the worst (and best) case scenarios in a given time period.

This doesn't help much during the draw down phase, but I think most people do actually know that's somewhat up to luck. Usually the accumulation phase is where people "assume the average."

I'm confused as to how this has a different result than the data that led to the 4% rule in the first place. It's my understanding 4% always worked for 30 years (even though you might be broke at the end of 30 years). Are you using 100% stocks?

The thing about "assuming the average" is the problem. That's what's really shaky as an assumption.

And this important point. On my funnel chart (see message #1), my lines are further apart because I'm assuming you don't get the 30-year averaging out period for any money other than the contribution to an IRA or 401(k) you may at the start of a 30-year accumulation.

For the second contribution in a 30-year accumulation, the investor only gets 29 years for any "averaging out" or "evening out" to occur. For the third contribution, only 28 years, and so on.

Because this subject is such a paradigm shift, let me again make this point: If you redraw a funnel chart to show the range of average annual returns on a 30-year accumulation with annual savings (as I did in the original post in this thread) the tail fattens up... it doesn't get thinner.

Ah, I somehow missed that your chart assumed annual contributions. That is a pretty shocking graph in that case, and my whole point was already taken into account. How do they calculate the average return for that graph?

I find that the numbers get tricky the more in depth you go. For example, average yearly return when you're adding in contributions/withdrawals can be misleading. For example in year 1 maybe you have a 50% loss, but you only have $10k in the account. If the next 28 years you have 0% returns, and in year 30 you have a 50% gain. Your average yearly return over that time is 0%, but you still end up with $442,500 with only $300,000 invested. That's a gain of 47.5% over the money you invested, but your average yearly return can still be said to be 0%.

Flip the returns around and you end up with $300,000 invested, but only $155,000 when all is said and done. A loss of around 48%, with the average return still 0%.

Obviously this is an extreme example, but it goes to show that averaging out returns each year doesn't tell the whole story when we're adding in contributions/withdrawals. When you're adding constantly, the later years have the largest effect. When you're withdrawing constantly, the early years have the largest effect. Interestingly enough these two time frames meet shortly before and after retirement. This means the 5-10 years before and after your retirement are the most important as far as investment returns go.

My message #12 above tells you how to use cFIREsim to make the calculations for a situation like the one you probably "personally" experienced. Seriously, try it. Going through the steps takes about a minute. If you want to fiddle with Excel to get the your internal rate of return (which takes into account all the timing issues you mention), that's another minute. Two minutes in total. Tops. And then you'll know the historically possible range of numbers for your situation.

And you're right. This is shocking. Lots of people--me included for a long while--take a shortcut in their analysis and say, "oh well, things will average out in the end..." But that's not really rigorous thinking.

And then this surprising reality... because we've had some really good investment returns over the last three decades or so, lots of people who are already retired now actually did end up with average or above average returns.

It's a bit like they walked through a minefield without realizing it... and now think there wasn't really a minefield.

max9505672

  • Bristles
  • ***
  • Posts: 336
Re: Unreliability of long run stock market returns
« Reply #38 on: September 13, 2017, 10:50:31 AM »
And this important point. On my funnel chart (see message #1), my lines are further apart because I'm assuming you don't get the 30-year averaging out period for any money other than the contribution to an IRA or 401(k) you may at the start of a 30-year accumulation.

For the second contribution in a 30-year accumulation, the investor only gets 29 years for any "averaging out" or "evening out" to occur. For the third contribution, only 28 years, and so on.

Because this subject is such a paradigm shift, let me again make this point: If you redraw a funnel chart to show the range of average annual returns on a 30-year accumulation with annual savings (as I did in the original post in this thread) the tail fattens up... it doesn't get thinner.
***Edit : I haven't had the time to look in details at the last links you quoted me, feel free to not answer if the answer's in your previous post***

All the examples you have been giving up to now are great for people who haven't actually started investing yet, our young kids for example.

But I think most of us here has already began investing. Basically, what is interesting to them is a similar chart but taking into account the actual investments they have (Portfolio value isn't 0$) and the retirement date they are aiming for as of today and considering annual savings are probably going up over the years.

Is this something easily reproducible with cFIREsim? What would be the impact of already having a positive portfolio value and increasing investments on the variability?
« Last Edit: September 13, 2017, 10:55:10 AM by max9505672 »

SeattleCPA

  • Handlebar Stache
  • *****
  • Posts: 2369
  • Age: 64
  • Location: Redmond, WA
    • Evergreen Small Business
Re: Unreliability of long run stock market returns
« Reply #39 on: September 13, 2017, 11:00:58 AM »

All the examples you have been giving up to now are great for people who haven't actually started investing yet, our young kids for example.

But I think most of us here has already began investing. Basically, what is interesting to them is a similar chart but taking into account the actual investments they have (Portfolio value isn't 0$) and the retirement date they are aiming for as of today and considering annual savings are probably going up over the years.

Is this something easily reproducible with cFIREsim? What would be the impact of already having a positive portfolio value and increasing investments on the variability?

Yes. Here's how to do it...

1. Go to cFIREsim.
2. Enter your current retirement savings into the Portfolio Value and 0 Initial Yearly Spending boxes.
3. Scroll down to the Other Income area.
4. Enter the annual savings amount you're adding to your portfolio into the Other Income area's Amount box. (E.g., if you're saving $10,000 a year, enter 10000.)
5. Enter 2017 into the Other Income area's Start Year box.
6. Enter your retirement year into the Other Income area's End Year box. (E.g. if you plan on retiring in 2030, enter 2030.)
7. Scroll back up to the top of the window and click the Run Simulation button. cFIREsim calculates away, draws a chart of outcomes and then gives you a series of ending values: average, median, st. dev. highest and lowest.

E.g., if you calculate these values for the scenario where you save $5,500 a year for 30 years, you get the results shown below, but obviously you will get different results using a different set of inputs:

Average:   $454,151   $0   $0
Median:   $459,246   $0   $0
St. Dev.:   $137,596   $0   $0
Highest:   $835,424   $0   $0
Lowest:   $170,973   $0   $0

Once you have your own real data that sort of looks like this, you can take average, median,highest or lowest value and use it to calculate the return  (actually the internal rate of return) with Excel.

Here's the formula you would use to calculate the internal rate of return for a $5500 a year savings amount where you still have 30 years of work left and you end up with that best case scenario of $835,424:

=RATE(30,-5500,,835424)

The function returns 10% (or if you want to be precise 9.5678%)

Here's the formula you would use to calculate the internal rate of return of the worst case scenario:

=RATE(30,-5500,,170973)

The function returns 0% (or if you want to be precise .244%)

« Last Edit: September 13, 2017, 11:03:27 AM by SeattleCPA »

ooeei

  • Handlebar Stache
  • *****
  • Posts: 1142
Re: Unreliability of long run stock market returns
« Reply #40 on: September 13, 2017, 12:24:31 PM »
My message #12 above tells you how to use cFIREsim to make the calculations for a situation like the one you probably "personally" experienced. Seriously, try it. Going through the steps takes about a minute. If you want to fiddle with Excel to get the your internal rate of return (which takes into account all the timing issues you mention), that's another minute. Two minutes in total. Tops. And then you'll know the historically possible range of numbers for your situation.

And you're right. This is shocking. Lots of people--me included for a long while--take a shortcut in their analysis and say, "oh well, things will average out in the end..." But that's not really rigorous thinking.

And then this surprising reality... because we've had some really good investment returns over the last three decades or so, lots of people who are already retired now actually did end up with average or above average returns.

It's a bit like they walked through a minefield without realizing it... and now think there wasn't really a minefield.

So while it is somewhat surprising, one thing I notice when looking at the data as you laid out (very helpful btw) is that usually in these "worst case" scenarios, just waiting a few more years boost returns significantly. That is to say, after each dip there is an associated rise. The real world "use" of this is pretty obvious, if your investments crash right around retirement time, ride it out a bit longer. While this is sort of "market timing" and assumes you can determine where the bottom is, which you can't, it is pragmatic and obvious advice.

For example, if in your example you retired in 1981 (starting 1951) you'd have $218,000 (1.9% return per excel). If you hold out for 4 more years to 1985 you end up with $371,000 (3.8% return per excel).  While there isn't a surefire way to predict exactly when the bottom will happen, waiting and contributing for another 5 years significantly helps your chances, and in this case doubles your average return.

With that being said, it is a good reminder that investment returns are largely out of your control, and luck plays a significant factor in them. All the more reason to be flexible and prepared. With early retirees the effect is even more pronounced, but so is the flexibility.

SeattleCPA

  • Handlebar Stache
  • *****
  • Posts: 2369
  • Age: 64
  • Location: Redmond, WA
    • Evergreen Small Business
Re: Unreliability of long run stock market returns
« Reply #41 on: September 13, 2017, 12:43:56 PM »

Average:   $454,151   $0   $0
Median:   $459,246   $0   $0
St. Dev.:   $137,596   $0   $0
Highest:   $835,424   $0   $0
Lowest:   $170,973   $0   $0

Very worrisome numbers - though the worst case scenario is low probability and (what I didn't realize until I checked Cfiresim myself) is that these are inflation adjusted "real dollars".  So, even in the worst possible case 1% probability you are still doing better to be invested in equities (over this time horizon) vs just stuffing your money in a savings account or CDs based on the rates available today.  Adjusting the expense ratios to reflect my 401K (TSP) yields me about $10,000 more in the worst case scenario.  I'd be sad if I had these results, but I'd know it while working and adjust by either saving more or working longer (to bulk up my pension and savings).

Just so there's no misunderstanding, I don't think any of this variability means we back away from an equity-heavy portfolio. I have not, I confess, done the numbers with a bonds-heavy portfolio. But surely that emphasis would only make this all look worse.

Apologies to anybody who's heard me say this too many times now... but I think the big and small big lessons here are:

1. (the BIG lesson) Run the numbers so you understand the variability in outcomes you could experience...
2. (the SMALLER lesson?) If this variability is totally out of left field for someone, that may signal person wants to revisit their asset allocation formula and perhaps dial down risk.

SeattleCPA

  • Handlebar Stache
  • *****
  • Posts: 2369
  • Age: 64
  • Location: Redmond, WA
    • Evergreen Small Business
Re: Unreliability of long run stock market returns
« Reply #42 on: September 13, 2017, 12:47:25 PM »
My message #12 above tells you how to use cFIREsim to make the calculations for a situation like the one you probably "personally" experienced. Seriously, try it. Going through the steps takes about a minute. If you want to fiddle with Excel to get the your internal rate of return (which takes into account all the timing issues you mention), that's another minute. Two minutes in total. Tops. And then you'll know the historically possible range of numbers for your situation.

And you're right. This is shocking. Lots of people--me included for a long while--take a shortcut in their analysis and say, "oh well, things will average out in the end..." But that's not really rigorous thinking.

And then this surprising reality... because we've had some really good investment returns over the last three decades or so, lots of people who are already retired now actually did end up with average or above average returns.

It's a bit like they walked through a minefield without realizing it... and now think there wasn't really a minefield.

So while it is somewhat surprising, one thing I notice when looking at the data as you laid out (very helpful btw) is that usually in these "worst case" scenarios, just waiting a few more years boost returns significantly. That is to say, after each dip there is an associated rise. The real world "use" of this is pretty obvious, if your investments crash right around retirement time, ride it out a bit longer. While this is sort of "market timing" and assumes you can determine where the bottom is, which you can't, it is pragmatic and obvious advice.

For example, if in your example you retired in 1981 (starting 1951) you'd have $218,000 (1.9% return per excel). If you hold out for 4 more years to 1985 you end up with $371,000 (3.8% return per excel).  While there isn't a surefire way to predict exactly when the bottom will happen, waiting and contributing for another 5 years significantly helps your chances, and in this case doubles your average return.

With that being said, it is a good reminder that investment returns are largely out of your control, and luck plays a significant factor in them. All the more reason to be flexible and prepared. With early retirees the effect is even more pronounced, but so is the flexibility.

I totally agree. And that sort of actionable insight is exactly the sort of "takeaway" I would hope someone gets from our discussion of this.

Telecaster

  • Magnum Stache
  • ******
  • Posts: 3543
  • Location: Seattle, WA
Re: Unreliability of long run stock market returns
« Reply #43 on: September 13, 2017, 02:12:58 PM »

Well first, if you're already FIRE, you're actually past this risk I'm wringing my hands about... you've made your way through the rapids. The risks once someone is FI or FIRE is all about the safe withdrawal stuff. Or at least that's the way it seems to me.

What I think we (you, me, the others who understand this) can do is spotlight the reality that we were greatly assisted by a good patch of investment history.

And then here's the reality for people who still bear this risk (so people still working toward FI...) I think people ought to plan for variability in outcomes, maybe save a little more, maybe keep open the option of working a little longer, etc.

Interesting discussion (seriously), but is any of this actionable?   Let's say you were planning on retiring in say, 2025 and you happen to be the poor slob who gets the low end of the return sequence.  Now lets say it is 2023 and you realize you're not going to meet your target number.  Your choices at that point are 1) working longer, or 2) taking out less money in retirement. 

DavidAnnArbor

  • Handlebar Stache
  • *****
  • Posts: 2266
  • Age: 58
  • Location: Ann Arbor, Michigan
Re: Unreliability of long run stock market returns
« Reply #44 on: September 13, 2017, 02:19:41 PM »
When I simply put numbers in cfiresim (the way I always did) to look at %chance of FIRE, I saw such variability in outcomes.

Isn't the unreliability a given ?  1966-1983 was a pretty bad period for stocks.

talltexan

  • Walrus Stache
  • *******
  • Posts: 5344
Re: Unreliability of long run stock market returns
« Reply #45 on: September 13, 2017, 03:15:03 PM »
ptf

SeattleCPA

  • Handlebar Stache
  • *****
  • Posts: 2369
  • Age: 64
  • Location: Redmond, WA
    • Evergreen Small Business
Re: Unreliability of long run stock market returns
« Reply #46 on: September 13, 2017, 03:17:42 PM »

Well first, if you're already FIRE, you're actually past this risk I'm wringing my hands about... you've made your way through the rapids. The risks once someone is FI or FIRE is all about the safe withdrawal stuff. Or at least that's the way it seems to me.

What I think we (you, me, the others who understand this) can do is spotlight the reality that we were greatly assisted by a good patch of investment history.

And then here's the reality for people who still bear this risk (so people still working toward FI...) I think people ought to plan for variability in outcomes, maybe save a little more, maybe keep open the option of working a little longer, etc.

Interesting discussion (seriously), but is any of this actionable?   Let's say you were planning on retiring in say, 2025 and you happen to be the poor slob who gets the low end of the return sequence.  Now lets say it is 2023 and you realize you're not going to meet your target number.  Your choices at that point are 1) working longer, or 2) taking out less money in retirement.

Fair question and if I can take liberties, I'd like to rephrase the question like this, "Other than the obvious steps of deciding two years before retirement that you need to work longer or that you'll be spending less, are there other, earlier actions someone can take to react to the variability?"

Okay, I think one can answer this question more robustly by considering the variability we're discussing. For example, if someone is using the average, they should understand from the get go that there's maybe a 50% chance their plan will need to change up (or down) and maybe a lot.  Or maybe another way to say the same sort of thing is there's maybe a 30% chance their plan will need to change massively. I think that's an actionable insight--especially if you have the insight early. It'll probably make someone think differently about those trade-offs we all make when it comes to planning how much we save and for how long we work.

Another actionable insight: If this variability is "new" to someone's financial thinking and strategic planning, I wonder if someone doesn't want to maybe use a different asset allocation... maybe one with less risk and less return but possibly a narrower range of likely outcomes... or maybe one with more risk but potentially more return...
« Last Edit: September 13, 2017, 03:23:25 PM by SeattleCPA »

DrF

  • Bristles
  • ***
  • Posts: 464
Re: Unreliability of long run stock market returns
« Reply #47 on: September 13, 2017, 03:21:21 PM »
Seattle and I are already on the same page, but I thought this was a better visual to aid everyone in seeing why this matters.

Based on prior returns, if one wants to have enough to retire in 30 years you would need to save 22% of your income. It will take you 30 years to save enough based on the worst case returns in the portfoliocharts database. If you happen to start saving in the best year possible, it would only take you ~20 years. A whole 10 year spread from best to worst.



But, we're mustachians right? So, if you save more of your income you are less exposed to this front end sequence of return risk. Same AA, but vastly different results when you save more of your income. By saving 75% of your income, you're guaranteed to have enough in 10 years (based on the portfoliocharts data). Best case scenario you would have enough in just 4-5 years. This is a spread of only 5-6 years, which reduces your exposure to sequence of return risk.


SeattleCPA

  • Handlebar Stache
  • *****
  • Posts: 2369
  • Age: 64
  • Location: Redmond, WA
    • Evergreen Small Business
Re: Unreliability of long run stock market returns
« Reply #48 on: September 13, 2017, 03:22:23 PM »
When I simply put numbers in cfiresim (the way I always did) to look at %chance of FIRE, I saw such variability in outcomes.

Isn't the unreliability a given ?  1966-1983 was a pretty bad period for stocks.

I think it is a given. And, David, if you've always "gotten" that you are way ahead of many who mistakenly believe that one can use "time" to diversify away a bunch of the risk! (Good job!)

Some people intuitively get this. And some people don't.

SeattleCPA

  • Handlebar Stache
  • *****
  • Posts: 2369
  • Age: 64
  • Location: Redmond, WA
    • Evergreen Small Business
Re: Unreliability of long run stock market returns
« Reply #49 on: September 13, 2017, 03:38:10 PM »
I think I agree with DrF.

Two tangential comments... One, there are great resources at the www.portfoliocharts.com website. I love it! But as an author I'm always alert about not stepping over the fair use boundary and violating an author's copyrights. (I.e., Tyler's copyrights.) He can share his work. But we should be careful about showing his charts, I think.

Second comment:I know I said earlier in this thread that what I'm talking about here is sequence of returns risk. But after more thought, I realize it's not really.

Don't we think that sequence of returns risk refers to situations like this:

Guy #1 earns 4% for the first 15 years of retirement and then 8% for the second 15 years.

Guy #2 earns 8% for the first 15 years of retirement and then 4% for the second 15 years.

Both of these guys earn about a 6% IRR or actually 5.9811%. And sequence of returns risk refers to weirdness that even though they get the exact same average annual return, the first guy gets limited in what he can draw because he gets the bad returns early.

The variability in outcomes issue is that even if the average run 6%, lots of people will get 4% or 5%... lots will actually get 7% or 8%... and some outliers will for good or bad get even more extreme results.