Cash Allocation 5 Years Before Starting Social Security

[big-papa]

The ceiling and floor method gives leeway to draw more in good years and less in the bad years. Basing the withdrawals on market returns and years left(reduced longevity) makes more sense than just drawing more because you are younger. I understand wanting more in the early years, I just don't think that alone should get you to a dangerous withdrawal rate.

If the poster asking the question was using SWR for withdrawals and asking to increase the percentage for the first few years, then that wouldn't be very advisable since it increases the risk of portfolio depletion.

However, he is using VPW. VPW uses a % of remaining portfolio as the method of withdrawal (monotonically increasing over time until it hits 100%), so it already results in higher withdrawals during good years and reduced withdrawals during bad years. And because it's a % of remaining portfolio method, it's impossible to deplete the portfolio before the year designated to make the final withdrawal. On average, a small increase in withdrawal in the early years will result in a small reduction of withdrawals in later years, which is what the poster is looking for. As long as the increase in the early years is relatively small, there's little danger of the latter years' withdrawals falling below what's needed to pay the bills unless one was already in that situation to start with.

 

[Beer-man]

.

The alternative is to do what I posted in post #7 which is to calculate the value from which you are withdrawing as being the current value of your portfolio + the NPV of future SS payments. Call that value PortfolioPrime. Calculate the withdrawal $ amount from your portfolio using the value of PortfolioPrime with VPW. Each year recalculate the NPV of future SS payments and again add that to your actual portfolio to create a new PortfolioPrime value. Withdraw as prescribed by VPW. Rinse & Repeat.

.

Can you give me an example of how this would work with say a small pension and a modest SS amount with someone retiring around 55?
I’m not sure what NPV means either. Sorry I’m new to VPW but I’m thinking this will be the way to go since a large pension or huge portfolio will not be in my future. We do however forecast that SS will cover about 60-85% of our spending needs if taken@70 and that is taking into account a 25% haircut

 

[big-papa]

Can you give me an example of how this would work with say a small pension and a modest SS amount with someone retiring around 55?
I’m not sure what NPV means either. Sorry I’m new to VPW but I’m thinking this will be the way to go since a large pension or huge portfolio will not be in my future. We do however forecast that SS will cover about 60-85% of our spending needs if taken@70 and that is taking into account a 25% haircut

NPV = Net Present Value. It's a common financial calculation that calculates the equivalent dollar amount today of future streams of cash. In other words, what if I know I'm going to have some income stream for the next 10 years, what would the equivalent amount be if I had all of that money right now?

So to do that, you have to assume a certain discount rate. That would be different for something like SS which is COLA'd and, perhaps, some other stream like maybe a non-COLA'd pension.

Recently, over on Bogleheads, there was a good discussion of this idea complete with examples. As I mentioned above, this calculation is only used to calculate a total to input into VPW each year to calculate a withdrawal.

Here is the thread: Note that despite the title of the thread, they didn't spend much time talking about VPW - it was mainly limited to the concept of the time value of money as represented by NPV calculations. Pay attention to the posts by siamond - he does a good job explaining it and provides a link to some googlesheets with examples.

https://www.bogleheads.org/forum/viewtopic.php?f=2&t=263790

Cheers,
Big-Papa

 

[Beer-man]

NPV = Net Present Value. It's a common financial calculation that calculates the equivalent dollar amount today of future streams of cash. In other words, what if I know I'm going to have some income stream for the next 10 years, what would the equivalent amount be if I had all of that money right now?

So to do that, you have to assume a certain discount rate. That would be different for something like SS which is COLA'd and, perhaps, some other stream like maybe a non-COLA'd pension.

Recently, over on Bogleheads, there was a good discussion of this idea complete with examples. As I mentioned above, this calculation is only used to calculate a total to input into VPW each year to calculate a withdrawal.

Here is the thread: Note that despite the title of the thread, they didn't spend much time talking about VPW - it was mainly limited to the concept of the time value of money as represented by NPV calculations. Pay attention to the posts by siamond - he does a good job explaining it and provides a link to some googlesheets with examples.

https://www.bogleheads.org/forum/viewtopic.php?f=2&t=263790

Cheers,
Big-Papa

Perfect thank you. I like the spreadsheet he posted on that link too. Very helpful

 

[slowsaver]

I've been studying the VPW method over at Bogleheads and downloaded the spreadsheet. It seems to make sense for us.

Thanks for mentioning this spreadsheet. I had not heard of it before now, and I'm really enjoying it.

 

[big-papa]

Thanks for mentioning this spreadsheet. I had not heard of it before now, and I'm really enjoying it.

It's a method that makes a lot of sense. The author, when developing VPW, created math that is equivalent to what the PMT function in excel already does, but has chosen not to use the function. No matter.

Though VPW is a specific implementation, at its highest level, it was a rediscovery. If you google, you'll find the same general idea elsewhere. One term used is the "actuarial method" with a specific version called "ARVA" or "annually recalculated virtual annuity" and another one is called "planner's method". I have links below for those interested.

The primary difference in all of these approaches is the choice of expected returns in the PMT calculation that generates the % to withdraw from your portfolio, though the VPW spreadsheet adds some pretty cool features as a practical implementation.

VPW: This uses extremely long term real returns for worldwide stocks and bonds from Credit Suisse's website. During any retirement duration, returns may be higher or lower than this. The returns used are weighted between stocks and bonds according to your AA

ARVA: ARVA is pretty conservative and uses current TIPs real rates of return as the expected returns.

"planners's method" as described by Bud Hebeler in the link below back in 2015 used a fixed 5% interest rate and 3.5% inflation rate.

Regardless of the choice of expected returns, by definition, they are always "wrong". And the degree of wrong-ness of the expected returns for any sequence of returns that occur over a retirement will determine what the trajectory of the withdrawals over the years will look like. If the actual sequence of returns that occur are better than the expected returns, then the trajectory usually is such that withdrawals are low in the early years and can become extremely high in the later years. Vice versa if the returns are less than the expected returns. The reality, however, is that over a 30 or 40 year retirement, there can be periods of both over and underperforming relative to the expected returns used, so the trajectories can take on all sorts of different shapes.

One idea that came up on BH was an alternative to using a fixed expected return in the calculation. And that is to use the latest expected real returns available each year.

For stocks, that calculation can be something as simple as 1/CAPE (with CAPE oftentimes referred to as PE10). Or you can look on any number of websites and get 5 or 10 year expected returns. Just remember, you need expected "real" returns, not nominal. I know of one BH'er that averages 1/CAPE with an real stock estimate from elsewhere, for example.

Bonds, at least nominal bonds like total bond market or intermediate treasuries, pay nominal rates. And the nominal returns over the duration of a bond fund are highly correlated with the current yield. There are a number of methods used to estimate future real returns of bonds including just taking the nominal SEC yield as reported by Morningstar and geometrically subtracting the expected inflation rate. The expected inflation rate might be estimated by using last year's inflation rate, averaging several years inflation rates, or using the current breakeven rate between treasuries and TIPs or general knowledge from the web-gods that the Fed wants to target between 2.25 and 2.5% long term inflation. Some of the site mentioned above for stocks also have inflation estimates.

Once you have those two numbers, then calculate the expected returns by weighting the two numbers according to your AA. Simplest thing is to do this once per year when calculating this year's withdrawal. Historically, such a method helps to smooth withdrawals over the years and helps to prevent withdrawals from becoming extremely high or extremely low. Not perfect by any means, but it does seem to help quite a bit. And most importantly, extreme accuracy in the expected return calculation is not required in order to see some improvement in the trajectories. To use this method, though, you'll either need to modify the VPW spreadsheet or roll your own.

Side note: If your AA contains international stocks, EM, Small Caps, etc. there are some websites that give expected future returns for them as well. There are some users that just ignore that and use the calculation for US stocks only and consider any upside that occurs by holding these other assets as a pure bonus that may, from time to time, give them a larger withdrawal.


VPW:
https://www.bogleheads.org/wiki/Variable_percentage_withdrawal

ARVA:
https://larrysiegeldotorg.files.wor...nly-spending-rule-article-youll-ever-need.pdf

Planner's Method:
How Much Can I Afford to Spend in Retirement?: Bud Hebeler's Autopilot Withdrawal Rule

Cheers,
Big-Papa

 

[SumDay]

Thanks to the Big-Papa for that VPW spreadsheet. I've fiddled with it and just downloaded the advanced version. Good stuff!

While doing some research, I stumbled across this Kitces article which may (or may not) be helpful to some: https://www.kitces.com/blog/valuing...s-as-an-asset-on-the-household-balance-sheet/

 

[Cut-Throat]

It's a method that makes a lot of sense. The author, when developing VPW, created math that is equivalent to what the PMT function in excel already does, but has chosen not to use the function. No matter.

Though VPW is a specific implementation, at its highest level, it was a rediscovery. If you google, you'll find the same general idea elsewhere. One term used is the "actuarial method" with a specific version called "ARVA" or "annually recalculated virtual annuity" and another one is called "planner's method". I have links below for those interested.

The primary difference in all of these approaches is the choice of expected returns in the PMT calculation that generates the % to withdraw from your portfolio, though the VPW spreadsheet adds some pretty cool features as a practical implementation.

VPW: This uses extremely long term real returns for worldwide stocks and bonds from Credit Suisse's website. During any retirement duration, returns may be higher or lower than this. The returns used are weighted between stocks and bonds according to your AA

ARVA: ARVA is pretty conservative and uses current TIPs real rates of return as the expected returns.

"planners's method" as described by Bud Hebeler in the link below back in 2015 used a fixed 5% interest rate and 3.5% inflation rate.

Regardless of the choice of expected returns, by definition, they are always "wrong". And the degree of wrong-ness of the expected returns for any sequence of returns that occur over a retirement will determine what the trajectory of the withdrawals over the years will look like. If the actual sequence of returns that occur are better than the expected returns, then the trajectory usually is such that withdrawals are low in the early years and can become extremely high in the later years. Vice versa if the returns are less than the expected returns. The reality, however, is that over a 30 or 40 year retirement, there can be periods of both over and underperforming relative to the expected returns used, so the trajectories can take on all sorts of different shapes............................................................................

And of course here is the Author of VPW reasoning, which makes more sense to me.

https://www.bogleheads.org/forum/viewtopic.php?f=2&t=263790#p4210735 :

"VPW's internal rate isn't based on a future return prediction. It's an asset-allocation-specific fixed rate that intends to be lower than high returns, and higher than low returns. It doesn't require precision but it has to be fixed; it's part of what gives VPW robustness and simplicity. The fixed rate will allow for higher spending in good years, and lower spending in bad years.Many users of the general PMT method take a different approach. They try to estimate future returns and use their prediction as internal rate to calculate a withdrawal. Each year, they update their prediction to calculate the next withdrawal. A series of perfect prediction would obviously lead to perfect constant-dollar withdrawals. In other words, their objective is to recreate SWR, but with more safety (no premature depletion).There are various reasons (other than simplicity) why I didn't adopt this prediction approach for VPW. Here are some important ones:- Regardless of all the ink about the accuracy (or not) of various medium or long-term future return metrics, I do not believe that any single metric can predict the future returns of a single stock investment. I see no way, as an example, for a ratio of current price related to past few years of earnings to have told me anything trustworthy about the future returns of Enron (which went bankrupt) or Apple (which became the largest worldwide capitalization). We simply don't know the future.- I don't believe that anyone who really knew how to predict future returns with sufficient accuracy to make a difference would tell me about it; he would be insane to do so instead of keeping the secret to himself and becoming a billionaire. By telling so many others about it that a random person like me would end up hearing about it, he would kill the advantage of the prediction. Think about it, if everybody knew years in advance the exact date at which Enron was going bankrupt and that Apple would soar, nobody would have put any money into Enron, or they would have sold their shares, dropping its price to the point where the remaining upcoming dividends, until bankruptcy, would return as much as Apple ahead. Of course, at the same time, people would fight to buy Apple shares, increasing their price and reducing their future returns appropriately so that they're more or less equal to Enron's. Markets are adaptive, no static.-Lastly, even if we were able to perfectly predict future returns, using this perfect prediction would lead to bad portfolio management with PMT-based withdrawals: "sell more shares when low", and "sell less shares when high". In other words, it would result into a reverse dollar cost averaging, a damaging thing for a portfolio.
A good time to withdraw more money from a portfolio is when shares are expensive. A good time to withdraw less is when their share values are down. VPW does just that, naturally, without needing to predict future returns as it's a purely reactive system."

 

[big-papa]

And of course here is the Author of VPW reasoning, which makes more sense to me.

https://www.bogleheads.org/forum/viewtopic.php?f=2&t=263790#p4210735 :

"VPW's internal rate isn't based on a future return prediction. It's an asset-allocation-specific fixed rate that intends to be lower than high returns, and higher than low returns. It doesn't require precision but it has to be fixed; it's part of what gives VPW robustness and simplicity. The fixed rate will allow for higher spending in good years, and lower spending in bad years.Many users of the general PMT method take a different approach. They try to estimate future returns and use their prediction as internal rate to calculate a withdrawal. Each year, they update their prediction to calculate the next withdrawal. A series of perfect prediction would obviously lead to perfect constant-dollar withdrawals. In other words, their objective is to recreate SWR, but with more safety (no premature depletion).There are various reasons (other than simplicity) why I didn't adopt this prediction approach for VPW. Here are some important ones:- Regardless of all the ink about the accuracy (or not) of various medium or long-term future return metrics, I do not believe that any single metric can predict the future returns of a single stock investment. I see no way, as an example, for a ratio of current price related to past few years of earnings to have told me anything trustworthy about the future returns of Enron (which went bankrupt) or Apple (which became the largest worldwide capitalization). We simply don't know the future.- I don't believe that anyone who really knew how to predict future returns with sufficient accuracy to make a difference would tell me about it; he would be insane to do so instead of keeping the secret to himself and becoming a billionaire. By telling so many others about it that a random person like me would end up hearing about it, he would kill the advantage of the prediction. Think about it, if everybody knew years in advance the exact date at which Enron was going bankrupt and that Apple would soar, nobody would have put any money into Enron, or they would have sold their shares, dropping its price to the point where the remaining upcoming dividends, until bankruptcy, would return as much as Apple ahead. Of course, at the same time, people would fight to buy Apple shares, increasing their price and reducing their future returns appropriately so that they're more or less equal to Enron's. Markets are adaptive, no static.-Lastly, even if we were able to perfectly predict future returns, using this perfect prediction would lead to bad portfolio management with PMT-based withdrawals: "sell more shares when low", and "sell less shares when high". In other words, it would result into a reverse dollar cost averaging, a damaging thing for a portfolio.
A good time to withdraw more money from a portfolio is when shares are expensive. A good time to withdraw less is when their share values are down. VPW does just that, naturally, without needing to predict future returns as it's a purely reactive system."

First let me say that there is absolutely nothing wrong with VPW. One could do a whole lot worse. But we also all know that investing and retiring involves some amount of math, along with some amount of philosophy/faith.

Some comments:
- I can't speak for anybody else, but I know I'm not trying to recreate SWR with a PMT function - only trying to smooth things a little bit and trying to prevent the "dive to the ground" and "shoot to the moon" sequences I've seen with classic VPW. If using imperfect valuation metrics occasionally results in some withdrawals along the way that are a little higher than expected or lower than expected, that's OK by me.
- As noted in my original post, extreme accuracy of the future predictions isn't even required to smooth things out some and mitigate what the sequences described above do.
- Not sure about the Apple/Enron points: you either accept the boglehead philosophy of spreading your risk among many stocks via low cost index funds or you don't. And again, accuracy of prediction isn't required here.
- Finally it boils down to one thing: either valuations matter or they don't. If one either believes that valuations don't matter, or one doesn't care if it matters or not, then the simplicity of VPW is a beautiful thing. If one believes that valuations do matter, while still believing that using valuations to change an asset allocation (market timing) is a bad thing, then why not take advantage of it to smooth things out a little bit?

And to be clear, valuations haven't always mattered: The early 1970's were a good example where the churn and volatility was more of a result of the oil embarago/stagflation than it was about valuations. So definitely not perfect. And while every possible scenario that could happen in the future has not happened in the past, I haven't seen any of the damage he mentions to any simulated portfolio I've looked at when looking at historical sequences.

Cheers,
Big-Papa

 

[Cut-Throat]

- I can't speak for anybody else, but I know I'm not trying to recreate SWR with a PMT function - only trying to smooth things a little bit and trying to prevent the "dive to the ground" and "shoot to the moon" sequences I've seen with classic VPW.

My portfolio's Asset Allocation does not have the "dive to the ground" and "shoot to the moon" sequences... Backtested to the worst periods of Market History.


But then again, I am retired and employing VPW, and not taking unnecessary risks with my Investing at this point...

 

[big-papa]

My portfolio's Asset Allocation does not have the "dive to the ground" and "shoot to the moon" sequences... Backtested to the worst periods of Market History.


But then again, I am retired and employing VPW, and not taking unnecessary risks with my Investing at this point...

Yup - you mentioned earlier a high allocation to bonds, so I would expect that a goodly portion of the variations would be due to the difference in inflation vs. bond yields over time and less so on stock return sequences.