Modeling "withdrawing just income generated"

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Post Reply
Topic Author
germark
Posts: 24
Joined: Thu Sep 06, 2018 2:18 pm

Modeling "withdrawing just income generated"

Post by germark » Fri Jan 04, 2019 6:37 pm

The most-cited model for withdrawing money from a portfolio seems to be the 4% rule / trinity study. That focused on how to convert a portfolio of stocks and bonds into an inflation-adjusted constant annual income each year for 30 years. It seems like it was geared towards retirees, who wanted to replace income from their job with income from a portfolio.

I'm in a different situation. I've received a windfall, and am still in my working years. I'm interested in modeling questions like "What happens if I limit distributions to the income the portfolio generates, and invest in X% total stock market and Y% total bond market? What is the range of values I can expect to receive in terms of income each year, and final amount of capital at the end of Z years?" Here "income" means the mix of dividends, interest and capital gains the mutual funds would generate.

I can't seem to find data on this, even though it seems like a natural question to me. Most calculations in this vein seem to be focused on the assumption of reinvested dividends, income and capital gains in a tax free environment.

If anyone could point me towards resources this would be much appreciated.

Thanks.

dziuniek
Posts: 523
Joined: Mon Jul 23, 2012 2:54 pm
Location: Corrupticut

Re: Modeling "withdrawing just income generated"

Post by dziuniek » Fri Jan 04, 2019 6:48 pm

Are you basically asking what your portfolio will be if you take the dividends and coupon payments out, say annually, and let the shares appreciate without selling any?

If so, dividends have been somewhere close to half of the total stock return. So HALF, for the stock part, is as good of an answer as any.

That being said, dividends are quite low historically, I think. So maybe the return will now come from share appreciation. I think some of this has to do with how dividends and capital gains have been taxes historically vs. now. Who knows about the future.

For bond funds, isn't the current SEC or Coupon Yield the best predictor of the future return? There's your answer.

Then again, we're supposedly in a rising rate environment, so maybe it's not a good idea to count on NAV going up.

Make a simple spreadsheet and play with the numbers. It isn't that hard. What it is, is uncertain.

Thesaints
Posts: 1718
Joined: Tue Jun 20, 2017 12:25 am

Re: Modeling "withdrawing just income generated"

Post by Thesaints » Fri Jan 04, 2019 6:49 pm

germark wrote:
Fri Jan 04, 2019 6:37 pm
The most-cited model for withdrawing money from a portfolio seems to be the 4% rule / trinity study. That focused on how to convert a portfolio of stocks and bonds into an inflation-adjusted constant annual income each year for 30 years. It seems like it was geared towards retirees, who wanted to replace income from their job with income from a portfolio.
Certainly the most cited. Unfortunately, not nearly as often understood.
I'm in a different situation. I've received a windfall, and am still in my working years. I'm interested in modeling questions like "What happens if I limit distributions to the income the portfolio generates, and invest in X% total stock market and Y% total bond market? What is the range of values I can expect to receive in terms of income each year, and final amount of capital at the end of Z years?" Here "income" means the mix of dividends, interest and capital gains the mutual funds would generate.

I can't seem to find data on this, even though it seems like a natural question to me. Most calculations in this vein seem to be focused on the assumption of reinvested dividends, income and capital gains in a tax free environment.

If anyone could point me towards resources this would be much appreciated.

Thanks.
A first issue is that distributions depends on the type of fund you have in mind. Some active funds can distribute 10% of more, while the underlying dividends were under 2%. Other funds with the same underlying assets will distribute maybe 1.5%.
"limiting distributions to the income the portfolio generates" is not a very clear concept.

Topic Author
germark
Posts: 24
Joined: Thu Sep 06, 2018 2:18 pm

Re: Modeling "withdrawing just income generated"

Post by germark » Fri Jan 04, 2019 6:55 pm

I'm in a different situation. I've received a windfall, and am still in my working years. I'm interested in modeling questions like "What happens if I limit distributions to the income the portfolio generates, and invest in X% total stock market and Y% total bond market? What is the range of values I can expect to receive in terms of income each year, and final amount of capital at the end of Z years?" Here "income" means the mix of dividends, interest and capital gains the mutual funds would generate.

I can't seem to find data on this, even though it seems like a natural question to me. Most calculations in this vein seem to be focused on the assumption of reinvested dividends, income and capital gains in a tax free environment.

If anyone could point me towards resources this would be much appreciated.

Thanks.
A first issue is that distributions depends on the type of fund you have in mind. Some active funds can distribute 10% of more, while the underlying dividends were under 2%. Other funds with the same underlying assets will distribute maybe 1.5%.
"limiting distributions to the income the portfolio generates" is not a very clear concept.
[/quote]

Thank you for answering. I only do passive investing and currently use the 3-fund portfolio. So I was planning on investing in Vanguard's Total Stock Market, Total Bond Market and Total International. I probably should have clarified that. Each quarter the funds distribute "something", which can either be reinvested or taken in cash. That "something" is what I was referring to.

Does that clarify what I meant?

DouroBound
Posts: 48
Joined: Mon Mar 12, 2018 7:55 pm

Re: Modeling "withdrawing just income generated"

Post by DouroBound » Fri Jan 04, 2019 6:58 pm

germark wrote:
Fri Jan 04, 2019 6:37 pm
The most-cited model for withdrawing money from a portfolio seems to be the 4% rule / trinity study. That focused on how to convert a portfolio of stocks and bonds into an inflation-adjusted constant annual income each year for 30 years. It seems like it was geared towards retirees, who wanted to replace income from their job with income from a portfolio.

I'm in a different situation. I've received a windfall, and am still in my working years. I'm interested in modeling questions like "What happens if I limit distributions to the income the portfolio generates, and invest in X% total stock market and Y% total bond market? What is the range of values I can expect to receive in terms of income each year, and final amount of capital at the end of Z years?" Here "income" means the mix of dividends, interest and capital gains the mutual funds would generate.

Thanks.
Unless I am misunderstanding you, it sounds like the situation you describe is something like this: You have inherited a portfolio that consists of mutual funds, and you wish to take the income generated by that portfolio and redirect it to another portfolio of index funds. Is that right? If so, it sounds to me like you may be over-complicating the analysis. If you think of the inherited portfolio as part of your overall portfolio (and I don't see why you wouldn't, unless there is some special circumstance not described here), and you plan to reinvest and not to spend the distributions, then I would think you just need to look at any of the numerous models of portfolio growth (which will include reinvestment of dividends, interest, etc.) to get a sense of the range of final values.

Topic Author
germark
Posts: 24
Joined: Thu Sep 06, 2018 2:18 pm

Re: Modeling "withdrawing just income generated"

Post by germark » Fri Jan 04, 2019 7:08 pm

dziuniek wrote:
Fri Jan 04, 2019 6:48 pm
Are you basically asking what your portfolio will be if you take the dividends and coupon payments out, say annually, and let the shares appreciate without selling any?
Yes, that's exactly what I'm asking about.

It would be nice to answer questions such as "after 10 years of following this strategy, how does the principle tend to do with inflation?" It would also be nice to plug in starting numbers and see what annual distributions tend to be, how they compare with inflation over time, and so on.
dziuniek wrote:
Fri Jan 04, 2019 6:48 pm
Make a simple spreadsheet and play with the numbers. It isn't that hard. What it is, is uncertain.
To do it myself I'd need the real historical data (including inflation), which I don't have, and don't know how to get. I'm also concerned whether I'm really the best person to be performing the analysis. Maybe I am, but at this point reading at book or paper seems more appealing to me.

Topic Author
germark
Posts: 24
Joined: Thu Sep 06, 2018 2:18 pm

Re: Modeling "withdrawing just income generated"

Post by germark » Fri Jan 04, 2019 7:12 pm

DouroBound wrote:
Fri Jan 04, 2019 6:58 pm
germark wrote:
Fri Jan 04, 2019 6:37 pm
The most-cited model for withdrawing money from a portfolio seems to be the 4% rule / trinity study. That focused on how to convert a portfolio of stocks and bonds into an inflation-adjusted constant annual income each year for 30 years. It seems like it was geared towards retirees, who wanted to replace income from their job with income from a portfolio.

I'm in a different situation. I've received a windfall, and am still in my working years. I'm interested in modeling questions like "What happens if I limit distributions to the income the portfolio generates, and invest in X% total stock market and Y% total bond market? What is the range of values I can expect to receive in terms of income each year, and final amount of capital at the end of Z years?" Here "income" means the mix of dividends, interest and capital gains the mutual funds would generate.

Thanks.
Unless I am misunderstanding you, it sounds like the situation you describe is something like this: You have inherited a portfolio that consists of mutual funds, and you wish to take the income generated by that portfolio and redirect it to another portfolio of index funds. Is that right? If so, it sounds to me like you may be over-complicating the analysis. If you think of the inherited portfolio as part of your overall portfolio (and I don't see why you wouldn't, unless there is some special circumstance not described here), and you plan to reinvest and not to spend the distributions, then I would think you just need to look at any of the numerous models of portfolio growth (which will include reinvestment of dividends, interest, etc.) to get a sense of the range of final values.
Sorry, I wasn't clear. I received cash.

The question for me is what to do with the cash, how to balance / model consumption vs. investment and so on.
Last edited by germark on Fri Jan 04, 2019 7:50 pm, edited 1 time in total.

Thesaints
Posts: 1718
Joined: Tue Jun 20, 2017 12:25 am

Re: Modeling "withdrawing just income generated"

Post by Thesaints » Fri Jan 04, 2019 7:45 pm

germark wrote:
Fri Jan 04, 2019 6:55 pm
Thank you for answering. I only do passive investing and currently use the 3-fund portfolio. So I was planning on investing in Vanguard's Total Stock Market, Total Bond Market and Total International. I probably should have clarified that. Each quarter the funds distribute "something", which can either be reinvested or taken in cash. That "something" is what I was referring to.

Does that clarify what I meant?
So, essentially you invest your capital distributed over those three funds in percentages TBD and only spend distributions.
You want to know how much you can spend every year and how much your invested capital grows (or declines) over the years.
There is no answer that can be useful to you, since future market returns, dividend yields and interest rates cannot be forecasted with much accuracy.

Dividend yields have changed over time and of course interest rates have changed even more.
What we can say is that the bond fund will tend to pay you everything in the form of dividends over the year and maintain its nominal value.
For the stock funds it is a crapshoot. These days yields are around 2% and returns over long periods have been ~10% (but you have to subtract the dividend yield, which has not been constant over long periods). Maybe yields won't change much, but there is no assurance that over the next, let's say, 20 years your stock component will appreciate 8%/yr. nominal.
Not knowing what each fund will return and pay out it is not possible to say what a combination of them will do.
But maybe I still have not fully understood your question..

dziuniek
Posts: 523
Joined: Mon Jul 23, 2012 2:54 pm
Location: Corrupticut

Re: Modeling "withdrawing just income generated"

Post by dziuniek » Fri Jan 04, 2019 10:36 pm

Search for SIMBA BACKTESTING SPREADSHEET - there's a new, new version which has a lot of what you're looking for - historical data.

You can see the return of different asset classes using any asset mix you want.

Run it for US stocks, Intl stocks, as well as bonds.

This is probably the most helpful thing you'll be able to find.

You can mess with the data so you can account for dividends there maybe?

venkman
Posts: 792
Joined: Tue Mar 14, 2017 10:33 pm

Re: Modeling "withdrawing just income generated"

Post by venkman » Fri Jan 04, 2019 10:48 pm

Go to the Backtest Portfolio section of Portfolio Visualizer.

https://www.portfoliovisualizer.com/backtest-portfolio

Change the "Display Income" option to Yes. That will bring up an option right below it for "Reinvest Dividends." Select No for that. Then just enter whatever funds you want to backtest (use VTSAX or VFIAX as a proxy for US stocks and VBTLX for US bonds). The results will show how much you would've gotten in income and how much your initial principal would've grown.

Topic Author
germark
Posts: 24
Joined: Thu Sep 06, 2018 2:18 pm

Re: Modeling "withdrawing just income generated"

Post by germark » Fri Jan 04, 2019 11:07 pm

venkman wrote:
Fri Jan 04, 2019 10:48 pm
Go to the Backtest Portfolio section of Portfolio Visualizer.

https://www.portfoliovisualizer.com/backtest-portfolio

Change the "Display Income" option to Yes. That will bring up an option right below it for "Reinvest Dividends." Select No for that. Then just enter whatever funds you want to backtest (use VTSAX or VFIAX as a proxy for US stocks and VBTLX for US bonds). The results will show how much you would've gotten in income and how much your initial principal would've grown.
Hey - thanks so much for this link. At first glance it is exactly what I was looking for :happy

dbr
Posts: 27450
Joined: Sun Mar 04, 2007 9:50 am

Re: Modeling "withdrawing just income generated"

Post by dbr » Sat Jan 05, 2019 10:01 am

germark wrote:
Fri Jan 04, 2019 11:07 pm
venkman wrote:
Fri Jan 04, 2019 10:48 pm
Go to the Backtest Portfolio section of Portfolio Visualizer.

https://www.portfoliovisualizer.com/backtest-portfolio

Change the "Display Income" option to Yes. That will bring up an option right below it for "Reinvest Dividends." Select No for that. Then just enter whatever funds you want to backtest (use VTSAX or VFIAX as a proxy for US stocks and VBTLX for US bonds). The results will show how much you would've gotten in income and how much your initial principal would've grown.
Hey - thanks so much for this link. At first glance it is exactly what I was looking for :happy
I wonder how that calculation handles capital gains distributions. It would be possible to do that for individual funds if the data has been tabulated as such. For a person interested in tax performance ST and LT would have to be separated. For the case of performance of asset classes I doubt there is meaningful data for capital gains distributions as such.

jcar
Posts: 173
Joined: Fri Jul 03, 2015 9:41 pm
Location: NC

Re: Modeling "withdrawing just income generated"

Post by jcar » Sat Jan 05, 2019 10:32 am

As far as a good tool for tracking dividends, distributions, cap gains wouldn't looking at a monthly brokerage statement like at Vanguard do the job. Under my statement they list estimated annual income for each fund or security. My rep at VG tells me that this is a reverse look only so would not account for an announced dividend increase until it is actually paid. Still a very good estimate as I track quarterly and there estimate is always within a few cents.

dbr
Posts: 27450
Joined: Sun Mar 04, 2007 9:50 am

Re: Modeling "withdrawing just income generated"

Post by dbr » Sat Jan 05, 2019 10:36 am

jcar wrote:
Sat Jan 05, 2019 10:32 am
As far as a good tool for tracking dividends, distributions, cap gains wouldn't looking at a monthly brokerage statement like at Vanguard do the job. Under my statement they list estimated annual income for each fund or security. My rep at VG tells me that this is a reverse look only so would not account for an announced dividend increase until it is actually paid. Still a very good estimate as I track quarterly and there estimate is always within a few cents.
The OP appears to want these numbers estimated 10, 20 or 30 years into the future, or more.

A backtest using PortfolioVisualizer more or less assumes the next several decades will look like the last several decades, but at least there is some sample over time there. The tougher one is getting any idea how much any particular fund is going to distribute in capital gains distributions.

Topic Author
germark
Posts: 24
Joined: Thu Sep 06, 2018 2:18 pm

Re: Modeling "withdrawing just income generated"

Post by germark » Sat Jan 05, 2019 11:30 am

venkman wrote:
Fri Jan 04, 2019 10:48 pm
Go to the Backtest Portfolio section of Portfolio Visualizer.

https://www.portfoliovisualizer.com/backtest-portfolio

Change the "Display Income" option to Yes. That will bring up an option right below it for "Reinvest Dividends." Select No for that. Then just enter whatever funds you want to backtest (use VTSAX or VFIAX as a proxy for US stocks and VBTLX for US bonds). The results will show how much you would've gotten in income and how much your initial principal would've grown.
Hey I just wanted to let you know that I spent more time with this tool and it is exactly what I was looking for (assuming that the results that it gives are correct!)

I put in a $1M portfolio with 50% stocks (VTSMX) and 50% bonds (VBTIX). The data goes back to 1996. I was completely shocked by the results.

For me the real question was what has happened, historically, to the principal when dividends etc. aren't reinvested. In particular, how has principal fared against inflation?

I was shocked that even with 50% bonds and not reinvesting distributions, the principal beat inflation!

For those with an interest, in this backtest income started and ended at roughly the same place ($54k). Its lowest value was $34k, although several years it dipped below $40k.

msk
Posts: 995
Joined: Mon Aug 15, 2016 10:40 am

Re: Modeling "withdrawing just income generated"

Post by msk » Sat Jan 05, 2019 11:59 am

I did some research myself into this using Shiller's data reconstructed back to 1871. 100% stocks. My conclusion agrees with Monte Carlo simulations.

Never withdraw more than 5% of portfolio value in any year and if the market behaves somewhat like it has since 1871, world wars and the Great Depression, etc., both your withdrawals and the balance portfolio at the end of each year will, on average, keep pace with inflation forever.

Bonds: Monte Carlo simulations show that you should never exceed 2.5% p.a.

Mix stocks and bonds to taste but personally, for the really long term I stick to 100% stocks. Be wary as to what you call "income". Dividends, the stuff the ETFs pay out are, frankly, unimportant. Berkshire Hathaway never pays dividends and has been a super performing long term stock. Would be silly to say it never generates "income"! I would be more tempted to define income as portfolio value including dividends and interest at end year less value at the beginning of the year. To smooth out large swings, do it over 3 years; e.g. portfolio value at end 2018 less at 1/1/2016 divided by 3. Repeat each year.

Topic Author
germark
Posts: 24
Joined: Thu Sep 06, 2018 2:18 pm

Re: Modeling "withdrawing just income generated"

Post by germark » Sat Jan 05, 2019 5:21 pm

germark wrote:
Sat Jan 05, 2019 11:30 am
venkman wrote:
Fri Jan 04, 2019 10:48 pm
Go to the Backtest Portfolio section of Portfolio Visualizer.

https://www.portfoliovisualizer.com/backtest-portfolio

Change the "Display Income" option to Yes. That will bring up an option right below it for "Reinvest Dividends." Select No for that. Then just enter whatever funds you want to backtest (use VTSAX or VFIAX as a proxy for US stocks and VBTLX for US bonds). The results will show how much you would've gotten in income and how much your initial principal would've grown.
Hey I just wanted to let you know that I spent more time with this tool and it is exactly what I was looking for (assuming that the results that it gives are correct!)

I put in a $1M portfolio with 50% stocks (VTSMX) and 50% bonds (VBTIX). The data goes back to 1996. I was completely shocked by the results.

For me the real question was what has happened, historically, to the principal when dividends etc. aren't reinvested. In particular, how has principal fared against inflation?

I was shocked that even with 50% bonds and not reinvesting distributions, the principal beat inflation!

For those with an interest, in this backtest income started and ended at roughly the same place ($54k). Its lowest value was $34k, although several years it dipped below $40k.
Continuing with the analysis, it's amazing that even though principal increase, annual income actually trends down over time. And these numbers are in nominal numbers, which means that inflation-adjusted it's even worse. The worst year for income was 2003. Off-hand I can't recall any historic events then (it seems between dot-com bust and credit crisis).

I have to say, though, I wonder about the accuracy of this data. It looks like a free tool with no individual or corporation taking responsibility for its accuracy. Does anyone know of similar pay tools that stand by the accuracy of their data and analyses?

jcar
Posts: 173
Joined: Fri Jul 03, 2015 9:41 pm
Location: NC

Re: Modeling "withdrawing just income generated"

Post by jcar » Sat Jan 05, 2019 10:36 pm

dbr wrote:
Sat Jan 05, 2019 10:36 am
jcar wrote:
Sat Jan 05, 2019 10:32 am
As far as a good tool for tracking dividends, distributions, cap gains wouldn't looking at a monthly brokerage statement like at Vanguard do the job. Under my statement they list estimated annual income for each fund or security. My rep at VG tells me that this is a reverse look only so would not account for an announced dividend increase until it is actually paid. Still a very good estimate as I track quarterly and there estimate is always within a few cents.
The OP appears to want these numbers estimated 10, 20 or 30 years into the future, or more.

A backtest using PortfolioVisualizer more or less assumes the next several decades will look like the last several decades, but at least there is some sample over time there. The tougher one is getting any idea how much any particular fund is going to distribute in capital gains distributions.
I was aware of visualizer but this would not be reliable at all. I'm retired recently so I'm focused on what to expect more near term. I'm with you about cap gains. At best even 5 years out is a crap shoot. Always great comments and insights on this site.

venkman
Posts: 792
Joined: Tue Mar 14, 2017 10:33 pm

Re: Modeling "withdrawing just income generated"

Post by venkman » Sat Jan 05, 2019 10:58 pm

germark wrote:
Sat Jan 05, 2019 5:21 pm
Continuing with the analysis, it's amazing that even though principal increase, annual income actually trends down over time. And these numbers are in nominal numbers, which means that inflation-adjusted it's even worse. The worst year for income was 2003. Off-hand I can't recall any historic events then (it seems between dot-com bust and credit crisis).
In 2003, stocks were still recovering from the 2000-2002 bear market, at the same time bond yields were starting to significantly decline.

I have to say, though, I wonder about the accuracy of this data. It looks like a free tool with no individual or corporation taking responsibility for its accuracy. Does anyone know of similar pay tools that stand by the accuracy of their data and analyses?
I'm 99% sure that PV was created and is maintained by someone affiliated with the BH forums, though I can't remember who. If you look at the FAQ section, it lists the sources it uses for the data.

AlohaJoe
Posts: 4048
Joined: Mon Nov 26, 2007 2:00 pm
Location: Saigon, Vietnam

Re: Modeling "withdrawing just income generated"

Post by AlohaJoe » Sat Jan 05, 2019 11:14 pm

germark wrote:
Fri Jan 04, 2019 6:37 pm
I'm interested in modeling questions like "What happens if I limit distributions to the income the portfolio generates, and invest in X% total stock market and Y% total bond market? What is the range of values I can expect to receive in terms of income each year, and final amount of capital at the end of Z years?"
Image

You can see that a bond ladder that only spends income will vary dramatically. Variability was as high as 36% over 30-year periods.

Image

The same is true when you add dividends to the mix. See how it can go from under $6,000 to over $10,000 back down to $4,000.
First, yields change over time, leading to inconsistent spending for yield-harvesting investors. Second, using a yield portfolio forces an inefficient tradeoff between retirement spending and estate goals. Third, drawing down an efficient portfolio with the same risk as a portfolio emphasizing yield has higher expected principal accumulation, and fourth, drawing down a portfolio is more tax-efficient than using yield to generate retirement income.
From "Is a Portfolio Built to Produce Yield a Sensible Retirement Income Portfolio?" by Sam Pittman

Topic Author
germark
Posts: 24
Joined: Thu Sep 06, 2018 2:18 pm

Re: Modeling "withdrawing just income generated"

Post by germark » Sat Jan 05, 2019 11:46 pm

AlohaJoe wrote:
Sat Jan 05, 2019 11:14 pm
...

From "Is a Portfolio Built to Produce Yield a Sensible Retirement Income Portfolio?" by Sam Pittman
Thank you - reading now!

msk
Posts: 995
Joined: Mon Aug 15, 2016 10:40 am

Re: Modeling "withdrawing just income generated"

Post by msk » Sun Jan 06, 2019 3:13 am

I feel that focusing on cash payouts, i.e. dividends or interest, is a dangerous path. E.g. recently I bought some shares in an IPO of a small utility. The prospectus promises a dividend yield of 8%. Looks highly attractive! But digging deeper also brings out the planned management of the company is to continue paying the same annual dividend until the end of the utility's 20-year contract, i.e. dividend is constant in nominal terms (implies declining in real terms) and of course the share price will decline if interest rates rise, etc. A high dividend yield is no panacea. I would expect that a company paying nil dividends may well do much better as an investment, e.g. Berkshire Hathaway. But the stock market is not dumb either. It will price each share as deserved. Inflation matters when you talk of 30+ year horizons!

Topic Author
germark
Posts: 24
Joined: Thu Sep 06, 2018 2:18 pm

Re: Modeling "withdrawing just income generated"

Post by germark » Sun Jan 06, 2019 12:58 pm

I can't help but feel that people here have an odd reaction to focusing on mutual fund distributions / portfolio yield.

Assume a hypothetical 30 year old windfall recipient ("John") received a $1M check 12 months ago and invested it 50% in total bond and 50% in total stock market. This is, I gather, a very "boglehead" thing to do with a windfall.

Looking online, the trailing 12 month yield on these investments was 2.67% and 1.92%. I believe this means that the income generated from the portfolio over this period would have been roughly $13k + $10k = $23k. Is that correct?

Unless I'm mistaken, the IRS would consider that $23k to be taxable, regardless of whether or not John had actually reinvested the distributions. So if he had reinvested them, he would have to go elsewhere (e.g. employment income) in order to pay the tax liability. Is this correct?

John's situation is close enough to my own to make the general point. It appears that blindly reinvesting distributions creates a tax liability, which becomes non-trivial as taxable portfolios grow. So it appears that in situations like this, even if you create a "boglehead" portfolio, there are still legitimate questions such as
-what happens to a portfolio vis-a-vis inflation if you choose not to reinvest distributions
-how the distributions have historically behaved over time

This was the point of my original question.

For some reason, when discussing mutual fund yield, people here seem to immediately jump to a conclusion that the poster is interested in rejiggering their entire portfolio so as to maximize yield. But that's what not what I've been talking about at all.

dbr
Posts: 27450
Joined: Sun Mar 04, 2007 9:50 am

Re: Modeling "withdrawing just income generated"

Post by dbr » Sun Jan 06, 2019 1:51 pm

germark wrote:
Sun Jan 06, 2019 12:58 pm
I can't help but feel that people here have an odd reaction to focusing on mutual fund distributions / portfolio yield.

That is because you have never made clear your purpose in creating this model, or forecast, if you will.


Assume a hypothetical 30 year old windfall recipient ("John") received a $1M check 12 months ago and invested it 50% in total bond and 50% in total stock market. This is, I gather, a very "boglehead" thing to do with a windfall.

Looking online, the trailing 12 month yield on these investments was 2.67% and 1.92%. I believe this means that the income generated from the portfolio over this period would have been roughly $13k + $10k = $23k. Is that correct?

Yes, that would be the dividends paid. There may also have been capital gains distributions. We don't know your purpose so we don't know if we should count those if they exist, but I think you are saying yes. I haven't looked up what distributions might or might not have been made last year (2018).

Unless I'm mistaken, the IRS would consider that $23k to be taxable, regardless of whether or not John had actually reinvested the distributions. So if he had reinvested them, he would have to go elsewhere (e.g. employment income) in order to pay the tax liability. Is this correct?

At this point it seems your question is a tax question. Is it?

If the account is taxable account, yes. However the tax liability is much less than 23K and could even be zero. But, yes, you have to pay the tax liability from something. Money is fungible so saying where you paid the tax liability from is somewhat meaningless, but it could be a person would reinvest the income less the additional taxes above what the person already has to pay.


John's situation is close enough to my own to make the general point. It appears that blindly reinvesting distributions creates a tax liability, which becomes non-trivial as taxable portfolios grow. So it appears that in situations like this, even if you create a "boglehead" portfolio, there are still legitimate questions such as
-what happens to a portfolio vis-a-vis inflation if you choose not to reinvest distributions
-how the distributions have historically behaved over time

Now things seem confused. Blindly reinvesting distributions does not immediately create any additional tax liability that year. Naturally as a taxable portfolio grows it will possibly distribute more and more taxable income and also accrue unrealized gain that may eventually be realized and taxed. Most people consider this a good outcome. The question for you is what is the alternative to reinvesting the income. You could spend it now. In that case your wealth would not grow as much and future income creating tax costs would not be as much. You could eliminate all your tax problems by simply cashing in and spending the inheritance. Depending on how the money was inherited there would be a basis step up and little tax cost to liquidating everything as soon as possible.

This was the point of my original question.

The point of the original question is not clear at all, at least to me.


For some reason, when discussing mutual fund yield, people here seem to immediately jump to a conclusion that the poster is interested in rejiggering their entire portfolio so as to maximize yield. But that's what not what I've been talking about at all.

Generally it is a good idea to minimize yield in a taxable portfolio in order to reduce tax cost. Is that the actual question?

Topic Author
germark
Posts: 24
Joined: Thu Sep 06, 2018 2:18 pm

Re: Modeling "withdrawing just income generated"

Post by germark » Sun Jan 06, 2019 2:56 pm

AlohaJoe wrote:
Sat Jan 05, 2019 11:14 pm
germark wrote:
Fri Jan 04, 2019 6:37 pm
I'm interested in modeling questions like "What happens if I limit distributions to the income the portfolio generates, and invest in X% total stock market and Y% total bond market? What is the range of values I can expect to receive in terms of income each year, and final amount of capital at the end of Z years?"
Image

You can see that a bond ladder that only spends income will vary dramatically. Variability was as high as 36% over 30-year periods.

Image

The same is true when you add dividends to the mix. See how it can go from under $6,000 to over $10,000 back down to $4,000.
First, yields change over time, leading to inconsistent spending for yield-harvesting investors. Second, using a yield portfolio forces an inefficient tradeoff between retirement spending and estate goals. Third, drawing down an efficient portfolio with the same risk as a portfolio emphasizing yield has higher expected principal accumulation, and fourth, drawing down a portfolio is more tax-efficient than using yield to generate retirement income.
From "Is a Portfolio Built to Produce Yield a Sensible Retirement Income Portfolio?" by Sam Pittman
Hey AlohaJoe, I just wanted to thank you for this recommendation. I read the paper once last night and once again this morning. It's pretty much exactly what I was looking for :)

Figure 6 from the paper, which shows how real wealth of an initial portfolio has fared over 30 years when you don't reinvest the interest and dividends, is pretty much exactly the graph I was looking for.

Image

It looks like there were some pretty stressful times for people who did this approach starting in the 1963 or 1973, where the real wealth of their investments went down 50%! It's also amazing to me that real wealth in some cases went up so much. I guess that the variability really just surprises me.

Figure 5 is also, as you probably expected, both fascinating and surprising to me. I did not expect that there to be such wide variability in real spending! That's obviously an important number for someone considering this strategy, as it clearly matters whether the money you receive each month is steady or jumps around!

Do you have any other papers you recommend in this vein, or ways to follow this literature as it evolves?

AlohaJoe
Posts: 4048
Joined: Mon Nov 26, 2007 2:00 pm
Location: Saigon, Vietnam

Re: Modeling "withdrawing just income generated"

Post by AlohaJoe » Sun Jan 06, 2019 8:12 pm

germark wrote:
Sun Jan 06, 2019 2:56 pm
Figure 5 is also, as you probably expected, both fascinating and surprising to me. I did not expect that there to be such wide variability in real spending! That's obviously an important number for someone considering this strategy, as it clearly matters whether the money you receive each month is steady or jumps around!

Do you have any other papers you recommend in this vein, or ways to follow this literature as it evolves?
No, I don't know any other papers. It isn't really a subject that interests me much since it never made any sense to me. The kind of person who writes papers also doesn't think the strategy makes sense, which is why there are so few papers on it.

There's no a priori reason to think that dividends should be steady and looking at historical data clearly shows they aren't.

Image

"Everyone" knows that dividends yields have gone from well over 5% to well under 3% in the modern era. And there's no reason to think that 3% is some kind of "natural" number. During recessions, dividend yield regularly falls -- the last time it fell to under 1% for a few quarters. And if you look at, say, Japan, where dividend yield is usually 1% lower than the US (right now it is 1.7% but the average for the past few decades is 1%), we can see there's clearly no mathematical reason US yields can't permanently be lower than they are today.

Nowadays yields are so low that saying you're going to live on dividends is kinda-sorta equivalent to saying you're going to live on a 2.5% withdrawals, which is a rate so low, and so safe, that it doesn't matter how you pull it from your portfolio. And since you still need to be able to handle yields getting cut to 1% for at least a few quarters, your spending is so flexible that your margin or safety is huge, making it even less relevant how you pull withdrawals from your portfolio.

Topic Author
germark
Posts: 24
Joined: Thu Sep 06, 2018 2:18 pm

Re: Modeling "withdrawing just income generated"

Post by germark » Sun Jan 06, 2019 8:46 pm

AlohaJoe wrote:
Sun Jan 06, 2019 8:12 pm
germark wrote:
Sun Jan 06, 2019 2:56 pm
Figure 5 is also, as you probably expected, both fascinating and surprising to me. I did not expect that there to be such wide variability in real spending! That's obviously an important number for someone considering this strategy, as it clearly matters whether the money you receive each month is steady or jumps around!

Do you have any other papers you recommend in this vein, or ways to follow this literature as it evolves?
No, I don't know any other papers. It isn't really a subject that interests me much since it never made any sense to me. The kind of person who writes papers also doesn't think the strategy makes sense, which is why there are so few papers on it.
I meant the broader field of research-based portfolio withdrawal strategies.

AlohaJoe
Posts: 4048
Joined: Mon Nov 26, 2007 2:00 pm
Location: Saigon, Vietnam

Re: Modeling "withdrawing just income generated"

Post by AlohaJoe » Sun Jan 06, 2019 9:11 pm

germark wrote:
Sun Jan 06, 2019 8:46 pm
AlohaJoe wrote:
Sun Jan 06, 2019 8:12 pm
germark wrote:
Sun Jan 06, 2019 2:56 pm
Figure 5 is also, as you probably expected, both fascinating and surprising to me. I did not expect that there to be such wide variability in real spending! That's obviously an important number for someone considering this strategy, as it clearly matters whether the money you receive each month is steady or jumps around!

Do you have any other papers you recommend in this vein, or ways to follow this literature as it evolves?
No, I don't know any other papers. It isn't really a subject that interests me much since it never made any sense to me. The kind of person who writes papers also doesn't think the strategy makes sense, which is why there are so few papers on it.
I meant the broader field of research-based portfolio withdrawal strategies.
Oh, for that there's an overwhelming amount of research, sure. Four good authors to start with are Moshe Milevsky, David Blanchett, Javier Estrada, and Wade Pfau. Luke Delorme's paper "A Blueprint for Retirement Spending" is my favorite introduction/overview. Wade Pfau's "39 Modern Retirement Income Planning Techniques" is the best single-stop for a list of the most of the major ideas/theories for people who want to research more.

If you prefer books then Zwecher's "Retirement Portfolios" is the best of the "safety-first" school of thought. And McClung's "Living Off Your Money" is the best of the "probability-based" school of thought.
Last edited by AlohaJoe on Sun Jan 06, 2019 11:10 pm, edited 1 time in total.

User avatar
siamond
Posts: 4410
Joined: Mon May 28, 2012 5:50 am

Re: Modeling "withdrawing just income generated"

Post by siamond » Sun Jan 06, 2019 10:33 pm

dziuniek wrote:
Fri Jan 04, 2019 10:36 pm
Search for SIMBA BACKTESTING SPREADSHEET - there's a new, new version which has a lot of what you're looking for - historical data.

You can see the return of different asset classes using any asset mix you want.

Run it for US stocks, Intl stocks, as well as bonds.

This is probably the most helpful thing you'll be able to find.

You can mess with the data so you can account for dividends there maybe?
OP, I am the volunteer currently managing the Simba spreadsheet.

Dziuniek, thanks for the plug, but the Simba spreadsheet focuses on Total Returns, it doesn't separate dividends (and other forms of distributions) from the price appreciation, so I'm afraid it wouldn't help the OP. Sure, somebody could replace some of the raw total-return data series by price-only data series and then use the various tools to analyze, but this would be a significant amount of work, which might not be worth the effort (I agree with AlohaJoe's assessment a few posts ago).

This being said, I did include a couple of price-only data series (e.g. Dow Jones, S&P 500, Nikkei) in the Data_Misc tab, and maybe I could do a little more in this respect... Will ponder about it.
germark wrote:
Sat Jan 05, 2019 11:30 am
venkman wrote:
Fri Jan 04, 2019 10:48 pm
Go to the Backtest Portfolio section of Portfolio Visualizer.
https://www.portfoliovisualizer.com/backtest-portfolio
[...] I have to say, though, I wonder about the accuracy of this data. It looks like a free tool with no individual or corporation taking responsibility for its accuracy. Does anyone know of similar pay tools that stand by the accuracy of their data and analyses?
OP, I know the Portfolio Visualizer author and he's a very rigorous individual. He carefully documented his data sources on the FAQ section of his Web site and this is indeed solid data. His Web site is actually used by many professionals. Quite frankly, I think it is much more solid than most 'pay tools' who usually have something else to sell...

FactualFran
Posts: 766
Joined: Sat Feb 21, 2015 2:29 pm

Re: Modeling "withdrawing just income generated"

Post by FactualFran » Mon Jan 07, 2019 4:52 pm

dbr wrote:
Sat Jan 05, 2019 10:01 am
I wonder how that calculation handles capital gains distributions. It would be possible to do that for individual funds if the data has been tabulated as such. For a person interested in tax performance ST and LT would have to be separated. For the case of performance of asset classes I doubt there is meaningful data for capital gains distributions as such.
Portfolio Visualizer treats capital gain distributions as income.

As to the accuracy of results reported by Portfolio Visualizer, for a mutual fund for which I have made a comparison, it reports the same results that I calculate using the distribution and NAV data that I have.

Topic Author
germark
Posts: 24
Joined: Thu Sep 06, 2018 2:18 pm

Re: Modeling "withdrawing just income generated"

Post by germark » Thu Jan 10, 2019 6:05 pm

AlohaJoe wrote:
Sun Jan 06, 2019 9:11 pm
If you prefer books then Zwecher's "Retirement Portfolios" is the best of the "safety-first" school of thought. And McClung's "Living Off Your Money" is the best of the "probability-based" school of thought.
I actually read "Retirement Portfolios" many years ago. I hadn't heard of "Living Off Your Money" before, but bought upon your recommendation. It just arrived today :)

Post Reply