Finding Middle Ground on the LMP

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
User avatar
Topic Author
Horton
Posts: 682
Joined: Mon Jan 21, 2008 3:53 pm

Finding Middle Ground on the LMP

Post by Horton » Thu Jun 25, 2020 3:33 pm

I recently re-read "Viability of the Spend Safely in Retirement Strategy", a paper by Wade Pfau, Joe Tomlinson, and Steven Vernon that follows up on their original paper on the Spend Safely in Retirement Strategy (SSiRS). I like the follow-up paper more than the original because it focuses specifically on the SSiRS strategy and contains a variety of practical information.

The SSiRs includes two key steps:
1. Optimize expected Social Security benefits through a careful delay strategy; in this case, many middle income retirees may have all the guaranteed lifetime income they need.
2. Generate retirement income from savings using the IRS required minimum distribution (RMD) rules, coupled with a low-cost index fund, target date fund, or balanced fund.
For individuals retiring before the optimal age to begin Social Security, the SSiRS suggest the following approach:
In this case, the retiree would set aside a “retirement transition fund” that equals the total amount of the Social Security bridge payments that are expected to be withdrawn until actual Social Security benefits start.
A common topic of debate on the forum is the concept of the Liability Matching Portfolio (LMP) popularized by William Bernstein. Most seem to either love or hate the concept with little middle ground. The opponents often argue, among other things, that it is impossible for an individual to forecast his/her liabilities from retirement date through death. This is a true and valid argument, but proponents of the LMP would argue that current expenses are the best estimate of future expenses (after including some level of inflationary increases) and that the plan can always be modified if needed or desired.

In an effort to set the above disagreement aside, I propose a middle ground inspired by the SSiRS. One approach to implement the LMP concept in conjunction with the SSiRS is to
  • (a) delay Social Security,
  • (b) build an LMP that contains an amount equal to Social Security at the optimal age multiplied by the number of years from retirement age to Social Security commencement, and
  • (c) use the remainder of the portfolio to support a desired variable withdrawal strategy (e.g., VPW or RMD) using a desired asset allocation


For example:
  • An individual wishes to retire at age 60
  • His Social Security at age 70 will be $75,000
  • His LMP would be $750,000 ($75,000 x 10 years)
  • The LMP would be invested in bonds and the remainder of the portfolio can be invested however desired
In this case, the LMP is easy to quantify and is directly tied to an observable amount (i.e., Social Security benefits). In addition, most retirees can reasonably expect to live until the commencement of Social Security, so longevity is not a factor to consider. The paper notes that Social Security benefits optimized for middle-income retirees "might comprise two-thirds to over 80% of the total retirement income." As a result, a retiree may have little need for a "safety first" withdrawal strategy above and beyond the LMP + SS, and can instead rely on a variable withdrawal method to support discretionary expenses in retirement.

Note: I personally suspect that much of the debate about the LMP rests on when people desire to retire. Those who plan to retire prior to say age 55 will likely find the cost of an LMP prohibitively expensive. So, in the interest of discussing the merits of this approach, let's assume that the above concepts are focused entirely on individuals planning to retire on or after age 55.

User avatar
Mountain Doc
Posts: 172
Joined: Tue Aug 08, 2017 3:15 pm
Location: Life Elevated

Re: Finding Middle Ground on the LMP

Post by Mountain Doc » Thu Jun 25, 2020 3:52 pm

This is basically what I plan to do. I expect SS to meet all of my basic spending needs, and I'll likely use a TIPS ladder to bridge the gap between retirement and drawing SS at 70. The rest of the portfolio will likely go into a balanced fund (something like a LifeStrategy fund) and will fund discretionary spending and charitable giving.

User avatar
Svensk Anga
Posts: 702
Joined: Sun Dec 23, 2012 5:16 pm

Re: Finding Middle Ground on the LMP

Post by Svensk Anga » Thu Jun 25, 2020 5:45 pm

I planned our retirement using the Bernstein LMP method as laid out in "Ages of the Investor". It worked out that we needed the LMP only to bridge from retirement at 57 to SS at 70, so we wound up with something like the Vernon/Pfau/Tomlinson strategy. If we both take SS at 70 (probably not optimum for the lower earner though), we need very little more to cover our definition of basic needs. I think this system can work very well, especially for retirees with either modest basic expenses or substantial SS benefits. My benefit is near max plus I get about 25% more in delayed retirement credits if I claim at 70. DW will get about half that. Four years in, I am very pleased and comfortable with how it is going. It helps to stay the course for market events like late 2018 or this spring, when one has a ladder of CD's/TIPS maturing at intervals to cover planned spending. I also don't worry much about bond yields because I plan to be spending chunks of principal as rungs mature. There is no reinvestment risk.

One thing I think a lot of folks miss on the LMP idea is that Bernstein advocates using it to fund only "basic" expenses. Discretionary expenses are to be funded from a risk portfolio which can be invested as aggressively as the retiree desires. In one's planning, one is free to place the split between basic and discretionary wherever one likes and so make the plan as conservative or as aggressive as desired. The basic level spending has smaller estimation error potential than the whole retirement budget, so should make sizing the LMP easier than if it was for full funding.

User avatar
WoodSpinner
Posts: 1480
Joined: Mon Feb 27, 2017 1:15 pm

Re: Finding Middle Ground on the LMP

Post by WoodSpinner » Thu Jun 25, 2020 8:00 pm

OP,

I am also using a similar approach with a bit of a twist....

I have a cash flow spreadsheet projecting out my expenses (core+discretionary) and income through age 90. Based on this, it seems like once SS begins at 70, I will have minimal need to tap my portfolio. I have built an LMP portfolio (Short (3 years cash-flow)/Intermediate Treasuries (remaining years Cash-flow) ) to cover the years from 58 till start of SS at 70.

Spreadsheet is updated yearly and I rebalance as needed (as expenses change) at least yearly.

I do hold some bonds in the Growth part of the Portfolio as well — this allows me to hit an overall target AA of 55/45. I rebalance this section of the portfolio yearly as well, plus 5/20 bands for shorter term volatility.

The LMP funds are reserved for expenses and are never used for rebalancing.

This slide illustrates my approach (note, I currently have 10 years till start of SS)
Image

WoodSpinner
Last edited by WoodSpinner on Fri Jun 26, 2020 11:38 am, edited 2 times in total.

User avatar
FIREchief
Posts: 4560
Joined: Fri Aug 19, 2016 6:40 pm

Re: Finding Middle Ground on the LMP

Post by FIREchief » Thu Jun 25, 2020 9:19 pm

I like this thread. Many good thoughts. I think the biggest challenge might be LTC costs, which could be a lump sum LMP reserve of x years times SNF costs in the area where a person(s) live.
I am not a lawyer, accountant or financial advisor. Any advice or suggestions that I may provide shall be considered for entertainment purposes only.

bigskyguy
Posts: 141
Joined: Sat Jan 24, 2015 4:59 pm

Re: Finding Middle Ground on the LMP

Post by bigskyguy » Thu Jun 25, 2020 9:20 pm

I find your approach completely defensible, and logical. We have taken a more comprehensive approach, which includes LMP funding in conjunction with delayed SS so that what we have determined to be our core expenses for our duration. Our plan includes a TIPS ladder to age 80 for my dearest (my age 90), to be followed by a deferred annuity. We have chosen to insure a relatively high level of consumption, both because we can, and because we choose to. We realize that we are clearly in the minority, with 75% of our retirement savings committed to this approach. But it works for us.
The Safety First approach is not for everyone, but for us, it is rather compelling, as well as reassuring. What we lose in potential upside we gain in personal reassurance that we are very unlikely to become a financial burden on any of our extended family.

User avatar
Svensk Anga
Posts: 702
Joined: Sun Dec 23, 2012 5:16 pm

Re: Finding Middle Ground on the LMP

Post by Svensk Anga » Thu Jun 25, 2020 9:37 pm

FIREchief wrote:
Thu Jun 25, 2020 9:19 pm
I like this thread. Many good thoughts. I think the biggest challenge might be LTC costs, which could be a lump sum LMP reserve of x years times SNF costs in the area where a person(s) live.
For a single surviving spouse in a LTC facility, I think this still works well. They would have the most SS income possible on either spouse's work record. Home equity could cover the remaining costs in a lot of cases. Ideally, there is some risk portfolio left. Most other expenses are done with by then: no travel, no house, done driving, etc. The tougher cases are: both spouses needing care; and one in care (perhaps with dementia) and one outside and healthy and incurring all their old normal costs.

I do earmark the fixed income portion of our risk portfolio to supplement pension and SS for LTC, so we will probably have a floor for rebalancing there.

User avatar
Ben Mathew
Posts: 1045
Joined: Tue Mar 13, 2018 11:41 am
Location: Seattle
Contact:

Re: Finding Middle Ground on the LMP

Post by Ben Mathew » Fri Jun 26, 2020 8:51 am

Horton wrote:
Thu Jun 25, 2020 3:33 pm
  • (a) delay Social Security,
  • (b) build an LMP that contains an amount equal to Social Security at the optimal age multiplied by the number of years from retirement age to Social Security commencement, and
  • (c) use the remainder of the portfolio to support a desired variable withdrawal strategy (e.g., VPW or RMD) using a desired asset allocation
I think this is basically a good plan. Some things to consider:

1. Does an LMP make sense for your preferences?

LMP is not the only way to buy safety. It is safety of a particular kind. You could get safety with:

(A) A large LMP + an aggressively invested risk portfolio or
(B) A small LMP + a conservatively invested risk portfolio

Which kind of safety makes sense for you? If you can draw a bright line under a certain base level spending below which you are not willing to go--say $40,000 per year--even if you are offered very attractive terms, then that's a reason to lock away $40,000 per year as LMP. Put differently, even if your risk portfolio goes down to zero, and all you are left with is the LMP, you would choose not to put even a small fraction of the LMP at risk, no matter how attractive the terms may be (i.e. even if the stock market offers very high expected return and very low risk.)

2. Why should the LMP match Social Security?

The social security payout is an arbitrary amount determined by a government program. It would be a coincidence if it matched up with your desired LMP level based on your preferences. I would suggest coming up with your own number independent of social security. If it happens to be the same, that's fine. If it's not, then go with your own preferences. Maybe you want to protect more than social security pays out. Maybe less. No reason to tie the size of the LMP to social security payouts.

3. VPW will need to be tweaked for most people (but it's easy to do)

VPW is close to being an excellent strategy for withdrawing from your risk portfolio. It's much better than fixed withdrawal strategies like the 4% rule. But it has a significant drawback. It uses annuitization to calculate how much money you can withdraw per year given expected portfolio returns. Makes sense at one level. The problem is that you are solving for a flat consumption profile, but face more risk in later years, so utility is actually falling over time. You know to a high degree of certainty what you will consume next year. But there is a lot of uncertainty over how much you will consume 20 years from now. For example, you might know that next year, your consumption will equal $70,000 +/- $5,000. But 20 years from now, your consumption will equal $70,000 +/- $30,000. Expected consumption is constant at $70,000 per year. But the risk is increasing. So utility is decreasing. That's unlikely to be the optimal strategy for most people. When faced with uncertainty, utility maximization suggests that we will save extra for the future ("precautionary saving") so consumption doesn't fall too low if things go bad. A person wishing to do "precautionary savings" will want to solve for increasing consumption over time, not constant consumption. The good news is that it's easy to fix this in the formulas. It just needs a small tweak to solve for a growing consumption profile. Withdrawal rates, of course, will be lower in the early years to generate growing expected consumption.
Last edited by Ben Mathew on Fri Jun 26, 2020 9:38 am, edited 1 time in total.

aristotelian
Posts: 7590
Joined: Wed Jan 11, 2017 8:05 pm

Re: Finding Middle Ground on the LMP

Post by aristotelian » Fri Jun 26, 2020 9:10 am

Ben Mathew wrote:
Fri Jun 26, 2020 8:51 am

2. Why should the LMP match Social Security?

The social security payout is an arbitrary amount determined by a government program. It would be a coincidence if it matched up with your desired LMP level based on your preferences. I would suggest coming up with your own number independent of social security. If it happens to be the same, that's fine. If it's not, then go with your own preferences. Maybe you want to protect more than social security pays out. Maybe less. No reason to tie the size of the LMP to social security payouts.
My thinking on this is that Liability Matching with Social Security provides a way to quantify the value of SS and how much it reduces the work your investment portfolio will have to do after claiming SS. Social Security is guaranteed income that you know you will have covered in the future but not at the start of retirement. When determining your SWR, your portfolio has to cover your non-SS spending needs over your full retirement, plus the portion equivalent to SS during your bridging period when you are retired but haven't claimed SS. Since the latter portion is going to have a relatively short timeframe, it needs to be covered by safe assets. Setting aside enough safe funds in the LMP to cover the SS portion in the bridging period, you can then determine your SWR for the investment portfolio to cover the non-SS portion over the duration of the retirement.

Depending on how close you are to SS, a LMP for the bridge years is going to be a lot less than 25X expenses, so in effect this enables you to hit your FI number much earlier than simply not counting SS.

Here is some back of napkin math:

Say you are seeking $60K inflation adjusted spending. Following the 4% rule, you would need $1.5M. However, say you are in your late 40's/early 50's, expecting say $20K social security and reasonably confident that it will be there. You can create a liability-matching portfolio to provide that chunk of income just for the bridge years before SS kicks in. In this case, say your are 45, planning to claim at 65, that would require $400K. Set that aside in bonds or TIPS and plan to spend it down, $20K per year. Your remaining portfolio needs to cover $40K, which you can do with $1M. That means your total FI number is now $1.4M ($1M + $400K), which reduces your FI number by $100K.

The difference is even greater the closer your are to being able to claim SS. Say you are 55, the liability matching side of your portfolio only needs to be $200K, making your FI number only $1.2M.

User avatar
willthrill81
Posts: 19097
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: Finding Middle Ground on the LMP

Post by willthrill81 » Fri Jun 26, 2020 10:23 am

As long as the gap between retirement and beginning SS benefits isn't too long, this is a perfectly plausible approach that many are already using.

However, it becomes increasingly 'expensive' as the gap lengthens. For the same reason, an LMP approach does not really work out well for early retirees.

For instance, if 50% of a retiree's expenses are essential, then the typical 25x used for a planned 30 year retirement period would mean that 12.5x would be used to buy fixed income, potentially including something like a SPIA. If the remaining 12.5x was invested aggressively, this means that the retiree's total portfolio would have somewhere between a 30/70 to 50/50 AA, which is conservative but plausible. But as the planned retirement period lengthens, this would tilt the portfolio even more towards fixed income. It could easily result in a historically unnecessarily low withdrawal rate as well, potentially 2% or lower.

IMHO, almost any approach that results in a total AA with less than about 25% in stocks or in a starting total portfolio withdrawal rate lower than about 3% is unnecessarily, possibly recklessly, conservative. An exception to this assessment is retirees who have accumulated far more than they need and so have an extremely low withdrawal rate in order to leave behind the bulk of their assets to others.
“It's a dangerous business, Frodo, going out your door. You step onto the road, and if you don't keep your feet, there's no knowing where you might be swept off to.” J.R.R. Tolkien,The Lord of the Rings

User avatar
Topic Author
Horton
Posts: 682
Joined: Mon Jan 21, 2008 3:53 pm

Re: Finding Middle Ground on the LMP

Post by Horton » Fri Jun 26, 2020 10:55 am

willthrill81 wrote:
Fri Jun 26, 2020 10:23 am
As long as the gap between retirement and beginning SS benefits isn't too long, this is a perfectly plausible approach that many are already using.

However, it becomes increasingly 'expensive' as the gap lengthens. For the same reason, an LMP approach does not really work out well for early retirees.
I agree and noted this in the OP. If you retire in your 40’s or early 50’s, then flexibility is key. Flexibility in withdrawals and utilization of your human capital. I’m trying to make this beyond the scope of this thread.

aristotelian
Posts: 7590
Joined: Wed Jan 11, 2017 8:05 pm

Re: Finding Middle Ground on the LMP

Post by aristotelian » Fri Jun 26, 2020 10:55 am

willthrill81 wrote:
Fri Jun 26, 2020 10:23 am
As long as the gap between retirement and beginning SS benefits isn't too long, this is a perfectly plausible approach that many are already using.

However, it becomes increasingly 'expensive' as the gap lengthens. For the same reason, an LMP approach does not really work out well for early retirees.

For instance, if 50% of a retiree's expenses are essential, then the typical 25x used for a planned 30 year retirement period would mean that 12.5x would be used to buy fixed income, potentially including something like a SPIA. If the remaining 12.5x was invested aggressively, this means that the retiree's total portfolio would have somewhere between a 30/70 to 50/50 AA, which is conservative but plausible. But as the planned retirement period lengthens, this would tilt the portfolio even more towards fixed income. It could easily result in a historically unnecessarily low withdrawal rate as well, potentially 2% or lower.

IMHO, almost any approach that results in a total AA with less than about 25% in stocks or in a starting total portfolio withdrawal rate lower than about 3% is unnecessarily, possibly recklessly, conservative. An exception to this assessment is retirees who have accumulated far more than they need and so have an extremely low withdrawal rate in order to leave behind the bulk of their assets to others.
If you look at my math above, it works out quite well for "normal-early" retirees. The LMP plan suggested above is not covering *all* of bridge period expenses with fixed income, only the portion that will be covered by SS upon claiming. Say you are retiring at 57 with expected SS $25k at 67, you would have $250K in fixed income in the LMP. At 47 it would be $500K. I don't know where you are pulling the 12.5X number--it does not have to be that high.

I do agree that longer early retirement (and larger expected SS) will result in more conservative overall allocation and that is something to be aware of. Using LMP also reduces your FI number by a lot more the closer you get to SS.

User avatar
Topic Author
Horton
Posts: 682
Joined: Mon Jan 21, 2008 3:53 pm

Re: Finding Middle Ground on the LMP

Post by Horton » Fri Jun 26, 2020 10:59 am

Ben Mathew wrote:
Fri Jun 26, 2020 8:51 am
2. Why should the LMP match Social Security?

The social security payout is an arbitrary amount determined by a government program. It would be a coincidence if it matched up with your desired LMP level based on your preferences. I would suggest coming up with your own number independent of social security. If it happens to be the same, that's fine. If it's not, then go with your own preferences. Maybe you want to protect more than social security pays out. Maybe less. No reason to tie the size of the LMP to social security payouts.
Social Security is a payment that most retirees can expect to receive. For many, it may cover 60-80% of expenses. As such, it serves as a simple reference point for creating a mini-LMP. We can obviously make this more complicated, but I’m recommending a very simple strategy that virtually anyone could implement.

User avatar
Steve Reading
Posts: 1859
Joined: Fri Nov 16, 2018 10:20 pm

Re: Finding Middle Ground on the LMP

Post by Steve Reading » Fri Jun 26, 2020 11:03 am

Ben Mathew wrote:
Fri Jun 26, 2020 8:51 am
3. VPW will need to be tweaked for most people (but it's easy to do)

VPW is close to being an excellent strategy for withdrawing from your risk portfolio. It's much better than fixed withdrawal strategies like the 4% rule. But it has a significant drawback. It uses annuitization to calculate how much money you can withdraw per year given expected portfolio returns. Makes sense at one level. The problem is that you are solving for a flat consumption profile, but face more risk in later years, so utility is actually falling over time. You know to a high degree of certainty what you will consume next year. But there is a lot of uncertainty over how much you will consume 20 years from now. For example, you might know that next year, your consumption will equal $70,000 +/- $5,000. But 20 years from now, your consumption will equal $70,000 +/- $30,000. Expected consumption is constant at $70,000 per year. But the risk is increasing. So utility is decreasing. That's unlikely to be the optimal strategy for most people. When faced with uncertainty, utility maximization suggests that we will save extra for the future ("precautionary saving") so consumption doesn't fall too low if things go bad. A person wishing to do "precautionary savings" will want to solve for increasing consumption over time, not constant consumption. The good news is that it's easy to fix this in the formulas. It just needs a small tweak to solve for a growing consumption profile. Withdrawal rates, of course, will be lower in the early years to generate growing expected consumption.
Ben, this is not the same as the retirement withdrawal strategy you previously presented here right (I got it bookmarked):
viewtopic.php?t=289064

For those willing to do a little bit of math, would the link above be the technically superior choice in your opinion, or a VPW with rising consumption?

User avatar
Topic Author
Horton
Posts: 682
Joined: Mon Jan 21, 2008 3:53 pm

Re: Finding Middle Ground on the LMP

Post by Horton » Fri Jun 26, 2020 11:11 am

Ben Mathew wrote:
Fri Jun 26, 2020 8:51 am
3. VPW will need to be tweaked for most people (but it's easy to do)

VPW is close to being an excellent strategy for withdrawing from your risk portfolio. It's much better than fixed withdrawal strategies like the 4% rule. But it has a significant drawback. It uses annuitization to calculate how much money you can withdraw per year given expected portfolio returns. Makes sense at one level. The problem is that you are solving for a flat consumption profile, but face more risk in later years, so utility is actually falling over time. You know to a high degree of certainty what you will consume next year. But there is a lot of uncertainty over how much you will consume 20 years from now. For example, you might know that next year, your consumption will equal $70,000 +/- $5,000. But 20 years from now, your consumption will equal $70,000 +/- $30,000. Expected consumption is constant at $70,000 per year. But the risk is increasing. So utility is decreasing. That's unlikely to be the optimal strategy for most people. When faced with uncertainty, utility maximization suggests that we will save extra for the future ("precautionary saving") so consumption doesn't fall too low if things go bad. A person wishing to do "precautionary savings" will want to solve for increasing consumption over time, not constant consumption. The good news is that it's easy to fix this in the formulas. It just needs a small tweak to solve for a growing consumption profile. Withdrawal rates, of course, will be lower in the early years to generate growing expected consumption.
In the OP, I suggested use of a “desired variable withdrawal strategy (e.g., VPW or RMD).” VPW and RMD methods provide pretty good bookends depending on the retiree’s preferences.

User avatar
Ben Mathew
Posts: 1045
Joined: Tue Mar 13, 2018 11:41 am
Location: Seattle
Contact:

Re: Finding Middle Ground on the LMP

Post by Ben Mathew » Fri Jun 26, 2020 11:26 am

aristotelian wrote:
Fri Jun 26, 2020 9:10 am
Ben Mathew wrote:
Fri Jun 26, 2020 8:51 am

2. Why should the LMP match Social Security?

The social security payout is an arbitrary amount determined by a government program. It would be a coincidence if it matched up with your desired LMP level based on your preferences. I would suggest coming up with your own number independent of social security. If it happens to be the same, that's fine. If it's not, then go with your own preferences. Maybe you want to protect more than social security pays out. Maybe less. No reason to tie the size of the LMP to social security payouts.
My thinking on this is that Liability Matching with Social Security provides a way to quantify the value of SS and how much it reduces the work your investment portfolio will have to do after claiming SS. Social Security is guaranteed income that you know you will have covered in the future but not at the start of retirement. When determining your SWR, your portfolio has to cover your non-SS spending needs over your full retirement, plus the portion equivalent to SS during your bridging period when you are retired but haven't claimed SS. Since the latter portion is going to have a relatively short timeframe, it needs to be covered by safe assets. Setting aside enough safe funds in the LMP to cover the SS portion in the bridging period, you can then determine your SWR for the investment portfolio to cover the non-SS portion over the duration of the retirement.

Depending on how close you are to SS, a LMP for the bridge years is going to be a lot less than 25X expenses, so in effect this enables you to hit your FI number much earlier than simply not counting SS.

Here is some back of napkin math:

Say you are seeking $60K inflation adjusted spending. Following the 4% rule, you would need $1.5M. However, say you are in your late 40's/early 50's, expecting say $20K social security and reasonably confident that it will be there. You can create a liability-matching portfolio to provide that chunk of income just for the bridge years before SS kicks in. In this case, say your are 45, planning to claim at 65, that would require $400K. Set that aside in bonds or TIPS and plan to spend it down, $20K per year. Your remaining portfolio needs to cover $40K, which you can do with $1M. That means your total FI number is now $1.4M ($1M + $400K), which reduces your FI number by $100K.

The difference is even greater the closer your are to being able to claim SS. Say you are 55, the liability matching side of your portfolio only needs to be $200K, making your FI number only $1.2M.
If you count social security as bonds and adjust for that in the risk portfolio (and that seems to me to be the right way to go about it), then there shouldn't be anything special about choosing an LMP level equal to social security. The spreadsheet below shows options for a person retiring at 50 with a $1 million portfolio, with social security expected to be $20,000 per year starting age 65.

The first column shows the case where there is no LMP. All assets are in the risk portfolio (RP). This includes the $1,000,000 in the retirement accounts, as well as the PV of social security discounted at real interest rate of 0%, which is $720,000. The total portfolio therefore is $1,720,000. All of it is in the RP invested at 40/60, which implies $688,000 in stocks and $1,032,000 in bonds. ($720,000 is already bonds in the form of social security. So that means $312,000 in bonds from the $1 million portfolio). If the retiree's plan is to consume 4% of the RP at age 50, then expected consumption at age 50 is 4%*$1,720,000 = $68,800.

The next column shows how things change if LMP is $10,000 per year. The expected consumption at age 50 declines to $58,400, a decline of $10,400. If you increase LMP to $20,000 to match social security, again expected consumption reduces by $10,400 to $48,000. And if you increase LMP to $30,000 which is higher than social security, expected consumption reduces again by $10,400 to $37,600.

Image

User avatar
Ben Mathew
Posts: 1045
Joined: Tue Mar 13, 2018 11:41 am
Location: Seattle
Contact:

Re: Finding Middle Ground on the LMP

Post by Ben Mathew » Fri Jun 26, 2020 11:31 am

Horton wrote:
Fri Jun 26, 2020 10:59 am
Ben Mathew wrote:
Fri Jun 26, 2020 8:51 am
2. Why should the LMP match Social Security?

The social security payout is an arbitrary amount determined by a government program. It would be a coincidence if it matched up with your desired LMP level based on your preferences. I would suggest coming up with your own number independent of social security. If it happens to be the same, that's fine. If it's not, then go with your own preferences. Maybe you want to protect more than social security pays out. Maybe less. No reason to tie the size of the LMP to social security payouts.
Social Security is a payment that most retirees can expect to receive. For many, it may cover 60-80% of expenses. As such, it serves as a simple reference point for creating a mini-LMP. We can obviously make this more complicated, but I’m recommending a very simple strategy that virtually anyone could implement.
That is reasonable. It can be a good starting point for people to think about how much, if any, they should lock away in LMP.

aristotelian
Posts: 7590
Joined: Wed Jan 11, 2017 8:05 pm

Re: Finding Middle Ground on the LMP

Post by aristotelian » Fri Jun 26, 2020 11:48 am

Ben Mathew wrote:
Fri Jun 26, 2020 11:26 am
aristotelian wrote:
Fri Jun 26, 2020 9:10 am
Ben Mathew wrote:
Fri Jun 26, 2020 8:51 am

2. Why should the LMP match Social Security?

The social security payout is an arbitrary amount determined by a government program. It would be a coincidence if it matched up with your desired LMP level based on your preferences. I would suggest coming up with your own number independent of social security. If it happens to be the same, that's fine. If it's not, then go with your own preferences. Maybe you want to protect more than social security pays out. Maybe less. No reason to tie the size of the LMP to social security payouts.
My thinking on this is that Liability Matching with Social Security provides a way to quantify the value of SS and how much it reduces the work your investment portfolio will have to do after claiming SS. Social Security is guaranteed income that you know you will have covered in the future but not at the start of retirement. When determining your SWR, your portfolio has to cover your non-SS spending needs over your full retirement, plus the portion equivalent to SS during your bridging period when you are retired but haven't claimed SS. Since the latter portion is going to have a relatively short timeframe, it needs to be covered by safe assets. Setting aside enough safe funds in the LMP to cover the SS portion in the bridging period, you can then determine your SWR for the investment portfolio to cover the non-SS portion over the duration of the retirement.

Depending on how close you are to SS, a LMP for the bridge years is going to be a lot less than 25X expenses, so in effect this enables you to hit your FI number much earlier than simply not counting SS.

Here is some back of napkin math:

Say you are seeking $60K inflation adjusted spending. Following the 4% rule, you would need $1.5M. However, say you are in your late 40's/early 50's, expecting say $20K social security and reasonably confident that it will be there. You can create a liability-matching portfolio to provide that chunk of income just for the bridge years before SS kicks in. In this case, say your are 45, planning to claim at 65, that would require $400K. Set that aside in bonds or TIPS and plan to spend it down, $20K per year. Your remaining portfolio needs to cover $40K, which you can do with $1M. That means your total FI number is now $1.4M ($1M + $400K), which reduces your FI number by $100K.

The difference is even greater the closer your are to being able to claim SS. Say you are 55, the liability matching side of your portfolio only needs to be $200K, making your FI number only $1.2M.
If you count social security as bonds and adjust for that in the risk portfolio (and that seems to me to be the right way to go about it), then there shouldn't be anything special about choosing an LMP level equal to social security. The spreadsheet below shows options for a person retiring at 50 with a $1 million portfolio, with social security expected to be $20,000 per year starting age 65.

The first column shows the case where there is no LMP. All assets are in the risk portfolio (RP). This includes the $1,000,000 in the retirement accounts, as well as the PV of social security discounted at real interest rate of 0%, which is $720,000. The total portfolio therefore is $1,720,000. All of it is in the RP invested at 40/60, which implies $688,000 in stocks and $1,032,000 in bonds. ($720,000 is already bonds in the form of social security. So that means $312,000 in bonds from the $1 million portfolio). If the retiree's plan is to consume 4% of the RP at age 50, then expected consumption at age 50 is 4%*$1,720,000 = $68,800.

The next column shows how things change if LMP is $10,000 per year. The expected consumption at age 50 declines to $58,400, a decline of $10,400. If you increase LMP to $20,000 to match social security, again expected consumption reduces by $10,400 to $48,000. And if you increase LMP to $30,000 which is higher than social security, expected consumption reduces again by $10,400 to $37,600.

Image
Lol, sure, you can make your LMP smaller or bigger than SS.

User avatar
Ben Mathew
Posts: 1045
Joined: Tue Mar 13, 2018 11:41 am
Location: Seattle
Contact:

Re: Finding Middle Ground on the LMP

Post by Ben Mathew » Fri Jun 26, 2020 11:49 am

Steve Reading wrote:
Fri Jun 26, 2020 11:03 am
Ben Mathew wrote:
Fri Jun 26, 2020 8:51 am
3. VPW will need to be tweaked for most people (but it's easy to do)

VPW is close to being an excellent strategy for withdrawing from your risk portfolio. It's much better than fixed withdrawal strategies like the 4% rule. But it has a significant drawback. It uses annuitization to calculate how much money you can withdraw per year given expected portfolio returns. Makes sense at one level. The problem is that you are solving for a flat consumption profile, but face more risk in later years, so utility is actually falling over time. You know to a high degree of certainty what you will consume next year. But there is a lot of uncertainty over how much you will consume 20 years from now. For example, you might know that next year, your consumption will equal $70,000 +/- $5,000. But 20 years from now, your consumption will equal $70,000 +/- $30,000. Expected consumption is constant at $70,000 per year. But the risk is increasing. So utility is decreasing. That's unlikely to be the optimal strategy for most people. When faced with uncertainty, utility maximization suggests that we will save extra for the future ("precautionary saving") so consumption doesn't fall too low if things go bad. A person wishing to do "precautionary savings" will want to solve for increasing consumption over time, not constant consumption. The good news is that it's easy to fix this in the formulas. It just needs a small tweak to solve for a growing consumption profile. Withdrawal rates, of course, will be lower in the early years to generate growing expected consumption.
Ben, this is not the same as the retirement withdrawal strategy you previously presented here right (I got it bookmarked):
viewtopic.php?t=289064

For those willing to do a little bit of math, would the link above be the technically superior choice in your opinion, or a VPW with rising consumption?
I have not updated that thread, but the short answer is that I set up the problem correctly and got the solution wrong. Once I corrected the mistakes, the strategy I arrived at is the same as the standard lifecycle portfolio model with CRRA utility will imply: hold a fixed allocation and use variable percentage withdrawal. So I didn't get a new strategy out of it. But I understand better how the standard solution does a good job of minimizing sequence of return risk over past returns.

Longinvest's VPW which solves for constant consumption falls within the category of solutions that will minimize sequence of return risk over past returns. But a more general solution will allow for rising or falling consumption based on time preferences as well as precautionary savings. Longinvest's VPW is a special case where time discounting exactly cancels out precautionary savings, leaving expected consumption flat.
Last edited by Ben Mathew on Fri Jun 26, 2020 12:07 pm, edited 2 times in total.

User avatar
Ben Mathew
Posts: 1045
Joined: Tue Mar 13, 2018 11:41 am
Location: Seattle
Contact:

Re: Finding Middle Ground on the LMP

Post by Ben Mathew » Fri Jun 26, 2020 11:52 am

aristotelian wrote:
Fri Jun 26, 2020 11:48 am
Lol, sure, you can make your LMP smaller or bigger than SS.
I thought you were saying there is a special benefit to equating social security to LMP (versus above or below), but maybe I misunderstood you.

User avatar
Topic Author
Horton
Posts: 682
Joined: Mon Jan 21, 2008 3:53 pm

Re: Finding Middle Ground on the LMP

Post by Horton » Fri Jun 26, 2020 12:05 pm

Ben Mathew wrote:
Fri Jun 26, 2020 11:49 am
Livesoft's VPW which solves for constant consumption falls within the category of solutions that will minimize sequence of return risk over past returns. But a more general solution will allow for rising or falling consumption based on time preferences as well as precautionary savings. Livesoft's VPW is a special case where time discounting exactly cancels out precautionary savings motive and leaves expected consumption flat.
Just to give credit where credit is due, I think longinvest created VPW (with feedback from a variety of BHs):

viewtopic.php?t=120430

User avatar
Ben Mathew
Posts: 1045
Joined: Tue Mar 13, 2018 11:41 am
Location: Seattle
Contact:

Re: Finding Middle Ground on the LMP

Post by Ben Mathew » Fri Jun 26, 2020 12:10 pm

Horton wrote:
Fri Jun 26, 2020 12:05 pm
Ben Mathew wrote:
Fri Jun 26, 2020 11:49 am
Livesoft's VPW which solves for constant consumption falls within the category of solutions that will minimize sequence of return risk over past returns. But a more general solution will allow for rising or falling consumption based on time preferences as well as precautionary savings. Livesoft's VPW is a special case where time discounting exactly cancels out precautionary savings motive and leaves expected consumption flat.
Just to give credit where credit is due, I think longinvest created VPW:

viewtopic.php?t=120430
Oops. I meant Longinvest. Fixed!

Livesoft posts so much that he is imprinted in my brain.

User avatar
willthrill81
Posts: 19097
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: Finding Middle Ground on the LMP

Post by willthrill81 » Fri Jun 26, 2020 2:54 pm

aristotelian wrote:
Fri Jun 26, 2020 10:55 am
willthrill81 wrote:
Fri Jun 26, 2020 10:23 am
As long as the gap between retirement and beginning SS benefits isn't too long, this is a perfectly plausible approach that many are already using.

However, it becomes increasingly 'expensive' as the gap lengthens. For the same reason, an LMP approach does not really work out well for early retirees.

For instance, if 50% of a retiree's expenses are essential, then the typical 25x used for a planned 30 year retirement period would mean that 12.5x would be used to buy fixed income, potentially including something like a SPIA. If the remaining 12.5x was invested aggressively, this means that the retiree's total portfolio would have somewhere between a 30/70 to 50/50 AA, which is conservative but plausible. But as the planned retirement period lengthens, this would tilt the portfolio even more towards fixed income. It could easily result in a historically unnecessarily low withdrawal rate as well, potentially 2% or lower.

IMHO, almost any approach that results in a total AA with less than about 25% in stocks or in a starting total portfolio withdrawal rate lower than about 3% is unnecessarily, possibly recklessly, conservative. An exception to this assessment is retirees who have accumulated far more than they need and so have an extremely low withdrawal rate in order to leave behind the bulk of their assets to others.
If you look at my math above, it works out quite well for "normal-early" retirees. The LMP plan suggested above is not covering *all* of bridge period expenses with fixed income, only the portion that will be covered by SS upon claiming. Say you are retiring at 57 with expected SS $25k at 67, you would have $250K in fixed income in the LMP. At 47 it would be $500K. I don't know where you are pulling the 12.5X number--it does not have to be that high.

I do agree that longer early retirement (and larger expected SS) will result in more conservative overall allocation and that is something to be aware of. Using LMP also reduces your FI number by a lot more the closer you get to SS.
My example was for a traditional LMP approach, not what is suggested by the OP.
“It's a dangerous business, Frodo, going out your door. You step onto the road, and if you don't keep your feet, there's no knowing where you might be swept off to.” J.R.R. Tolkien,The Lord of the Rings

aristotelian
Posts: 7590
Joined: Wed Jan 11, 2017 8:05 pm

Re: Finding Middle Ground on the LMP

Post by aristotelian » Fri Jun 26, 2020 3:19 pm

Ben Mathew wrote:
Fri Jun 26, 2020 11:52 am
aristotelian wrote:
Fri Jun 26, 2020 11:48 am
Lol, sure, you can make your LMP smaller or bigger than SS.
I thought you were saying there is a special benefit to equating social security to LMP (versus above or below), but maybe I misunderstood you.
No, I agree it is an arbitrary choice but I can see why OP would use it as a starting point. Since SS does represent a new source of guaranteed income during retirement, LMP can make a big difference in that case versus a typical SWR method that ignores SS.

User avatar
Ben Mathew
Posts: 1045
Joined: Tue Mar 13, 2018 11:41 am
Location: Seattle
Contact:

Re: Finding Middle Ground on the LMP

Post by Ben Mathew » Fri Jun 26, 2020 3:48 pm

aristotelian wrote:
Fri Jun 26, 2020 3:19 pm
Ben Mathew wrote:
Fri Jun 26, 2020 11:52 am
aristotelian wrote:
Fri Jun 26, 2020 11:48 am
Lol, sure, you can make your LMP smaller or bigger than SS.
I thought you were saying there is a special benefit to equating social security to LMP (versus above or below), but maybe I misunderstood you.
No, I agree it is an arbitrary choice but I can see why OP would use it as a starting point. Since SS does represent a new source of guaranteed income during retirement, LMP can make a big difference in that case versus a typical SWR method that ignores SS.
I agree. An SWR strategy that does not include SS in the portfolio will have problems. Setting LMP equal to SS, while not first best, can solve some of those problems. But I would still encourage people to go further and pick an LMP based on their preferences, and then base the withdrawal strategy on the full portfolio, including SS as a bond.

Unfortunately including SS as a bond in the portfolio does not seem to be widely accepted, based on this recent thread: Factoring Social Security into Bond Allocation.

User avatar
willthrill81
Posts: 19097
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: Finding Middle Ground on the LMP

Post by willthrill81 » Fri Jun 26, 2020 3:51 pm

aristotelian wrote:
Fri Jun 26, 2020 3:19 pm
Ben Mathew wrote:
Fri Jun 26, 2020 11:52 am
aristotelian wrote:
Fri Jun 26, 2020 11:48 am
Lol, sure, you can make your LMP smaller or bigger than SS.
I thought you were saying there is a special benefit to equating social security to LMP (versus above or below), but maybe I misunderstood you.
No, I agree it is an arbitrary choice but I can see why OP would use it as a starting point. Since SS does represent a new source of guaranteed income during retirement, LMP can make a big difference in that case versus a typical SWR method that ignores SS.
Ben Mathew wrote:
Fri Jun 26, 2020 3:48 pm
I agree. An SWR strategy that does not include SS in the portfolio will have problems. Setting LMP equal to SS, while not first best, can solve some of those problems. But I would still encourage people to go further and pick an LMP based on their preferences, and then base the withdrawal strategy on the full portfolio, including SS as a bond.

Unfortunately including SS as a bond in the portfolio does not seem to be widely accepted, based on this recent thread: Factoring Social Security into Bond Allocation.
A very plausible alternative to all of those approaches is to use a spreadsheet and the time value of money formula, including expected returns for one's portfolio and multiple income streams beginning (or ending) at various points in the future to determine how much one can withdraw at the current point in time and adjust subsequent withdrawals as expectations change. I personally find this to be the most logical and flexible approach to making withdrawals.
“It's a dangerous business, Frodo, going out your door. You step onto the road, and if you don't keep your feet, there's no knowing where you might be swept off to.” J.R.R. Tolkien,The Lord of the Rings

User avatar
Ben Mathew
Posts: 1045
Joined: Tue Mar 13, 2018 11:41 am
Location: Seattle
Contact:

Re: Finding Middle Ground on the LMP

Post by Ben Mathew » Fri Jun 26, 2020 4:38 pm

willthrill81 wrote:
Fri Jun 26, 2020 3:51 pm
A very plausible alternative to all of those approaches is to use a spreadsheet and the time value of money formula, including expected returns for one's portfolio and multiple income streams beginning (or ending) at various points in the future to determine how much one can withdraw at the current point in time and adjust subsequent withdrawals as expectations change. I personally find this to be the most logical and flexible approach to making withdrawals.
I am assuming this is amortizing the full portfolio (including all future income streams like pension and social security) over years left. The primary difference between this and Longinvest's VPW is would be (1) include PV of all income streams in the portfolio (2) use your own forward looking estimate for portfolio return and (3) use your own estimate of years left instead of (100-age). Is that right?

If so, I think it is at base a good strategy for the risk portfolio. The main problem is that it is solving for flat consumption, which increases risk in later years (same expected consumption, higher std deviation of consumption, therefore lower utility). I think the amortization formula should be used to solve more generally for consumption growing at rate g to accomodate time preferences and precautionary saving. g could be negative (expected consumption falling), but I think that many people would choose positive g. The flat consumption path is obtained by setting g=0, which might be optimal for some, but unlikely to be optimal for all.

The other question is whether a person should have an LMP in addition to the risk portfolio. That depends very much on individual preferences. For some people, the answer will be no. A person with CRRA preferences, for example, would say no. They would hold only a RP. Individuals have to decide if locking up a certain amount of base consumption in an LMP makes sense for them. A helpful question would be: if you had only the LMP left, would you put some of it at risk? If so, then you should reduce your LMP and pick more safety if desired through a more conservative allocation in the RP.

User avatar
willthrill81
Posts: 19097
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: Finding Middle Ground on the LMP

Post by willthrill81 » Fri Jun 26, 2020 4:54 pm

Ben Mathew wrote:
Fri Jun 26, 2020 4:38 pm
willthrill81 wrote:
Fri Jun 26, 2020 3:51 pm
A very plausible alternative to all of those approaches is to use a spreadsheet and the time value of money formula, including expected returns for one's portfolio and multiple income streams beginning (or ending) at various points in the future to determine how much one can withdraw at the current point in time and adjust subsequent withdrawals as expectations change. I personally find this to be the most logical and flexible approach to making withdrawals.
I am assuming this is amortizing the full portfolio (including all future income streams like pension and social security) over years left. The primary difference between this and Longinvest's VPW is would be (1) include PV of all income streams in the portfolio (2) use your own forward looking estimate for portfolio return and (3) use your own estimate of years left instead of (100-age). Is that right?
You can fully amortize the portfolio, but you certainly don't have to. You can select the terminal value to be anything you want, including a greater balance than you started with. That's the beauty of the approach; it's extremely flexible.
Ben Mathew wrote:
Fri Jun 26, 2020 4:38 pm
If so, I think it is at base a good strategy for the risk portfolio. The main problem is that it is solving for flat consumption, which increases risk in later years (same expected consumption, higher std deviation of consumption, therefore lower utility). I think the amortization formula should be used to solve more generally for consumption growing at rate g to accomodate time preferences and precautionary saving. g could be negative (expected consumption falling), but I think that many people would choose positive g. The flat consumption path is obtained by setting g=0, which might be optimal for some, but unlikely to be optimal for all.
Again, using the TVM formula at the core makes this approach exceedingly flexible. You can adjust your expectations for returns to whatever you want, including negative real returns.

In our instance, I'm planning on retiring at around age 52. I fully expect that we'll spend more in our 50s and 60s than in our 70s and 80s. Consequently, slightly front-loading our withdrawals by increasing the returns in the formula by around 1% beyond what we actually expect will accomplish this and result in naturally smaller withdrawals than otherwise later on. Of course, portfolio returns may mean that we see increases or larger than expected decreases in subsequent withdrawals. That's just the nature of uncertainty. But we're planning on our starting withdrawals being around double what our essential spending will be (e.g. property taxes, insurance, food, home maintenance, etc.), so we're planning to allow for substantial flexibility if needed. And once I reach 70, our expected SS benefits should cover all of our essential spending, leaving whatever we have remaining for discretionary spending, subsequent large medical expenses not covered by Medicare, and subsequent LTC expenses.
Ben Mathew wrote:
Fri Jun 26, 2020 4:38 pm
The other question is whether a person should have a LMP in addition to the risk portfolio. That depends very much on individual preferences. For some people, the answer will be no. A person with CRRA preferences, for example, would say no. They would hold only a RP. Individuals have to decide if locking up a certain amount of base consumption in an LMP makes sense for them.
I agree. Personal preferences, non-portfolio income, spending flexibility, etc. all impact the decision.

Hybrid approaches that don't involve a true LMP are also possible. One of which is McClung's Prime Harvesting approach, the core tenet of which is that stocks can be used to rebalance into bonds, but bonds are never sold to rebalance into stocks. This means that the entirety of one's bonds (including all fixed income assets) represent a quasi-LMP in that there is no chance of 'rebalancing into oblivion'. This is an AA decision though, not a withdrawal decision. One could incorporate such an approach with any of the withdrawal methods one desired.
“It's a dangerous business, Frodo, going out your door. You step onto the road, and if you don't keep your feet, there's no knowing where you might be swept off to.” J.R.R. Tolkien,The Lord of the Rings

User avatar
Topic Author
Horton
Posts: 682
Joined: Mon Jan 21, 2008 3:53 pm

Re: Finding Middle Ground on the LMP

Post by Horton » Fri Jun 26, 2020 4:58 pm

Ben Mathew wrote:
Fri Jun 26, 2020 4:38 pm
willthrill81 wrote:
Fri Jun 26, 2020 3:51 pm
A very plausible alternative to all of those approaches is to use a spreadsheet and the time value of money formula, including expected returns for one's portfolio and multiple income streams beginning (or ending) at various points in the future to determine how much one can withdraw at the current point in time and adjust subsequent withdrawals as expectations change. I personally find this to be the most logical and flexible approach to making withdrawals.
I am assuming this is amortizing the full portfolio (including all future income streams like pension and social security) over years left. The primary difference between this and Longinvest's VPW is would be (1) include PV of all income streams in the portfolio (2) use your own forward looking estimate for portfolio return and (3) use your own estimate of years left instead of (100-age). Is that right?

If so, I think it is at base a good strategy for the risk portfolio. The main problem is that it is solving for flat consumption, which increases risk in later years (same expected consumption, higher std deviation of consumption, therefore lower utility). I think the amortization formula should be used to solve more generally for consumption growing at rate g to accomodate time preferences and precautionary saving. g could be negative (expected consumption falling), but I think that many people would choose positive g. The flat consumption path is obtained by setting g=0, which might be optimal for some, but unlikely to be optimal for all.

The other question is whether a person should have a LMP in addition to the risk portfolio. That depends very much on individual preferences. For some people, the answer will be no. A person with CRRA preferences, for example, would say no. They would hold only a RP. Individuals have to decide if locking up a certain amount of base consumption in an LMP makes sense for them.
I agree that willthrill’s approach is reasonable for more sophisticated retirees who want use their own discount rate and longevity assumptions.

Ben Mathew - the “time preferences” and CRRA concepts are way too theoretical for me. I may be the dunce in the room, but I suspect that most people won’t be able to integrate this into their planning. That said, I do think it’s understandable to incorporate some cost of living adjustments, be it inflation or some other measure. In the OP, by taking SS multiplied by the number of years until SS begins, I have basically assumed a real discount rate of zero and, therefore, incorporated some level of inflation.

User avatar
Steve Reading
Posts: 1859
Joined: Fri Nov 16, 2018 10:20 pm

Re: Finding Middle Ground on the LMP

Post by Steve Reading » Fri Jun 26, 2020 5:19 pm

Ben Mathew wrote:
Fri Jun 26, 2020 11:49 am
Steve Reading wrote:
Fri Jun 26, 2020 11:03 am
Ben Mathew wrote:
Fri Jun 26, 2020 8:51 am
3. VPW will need to be tweaked for most people (but it's easy to do)

VPW is close to being an excellent strategy for withdrawing from your risk portfolio. It's much better than fixed withdrawal strategies like the 4% rule. But it has a significant drawback. It uses annuitization to calculate how much money you can withdraw per year given expected portfolio returns. Makes sense at one level. The problem is that you are solving for a flat consumption profile, but face more risk in later years, so utility is actually falling over time. You know to a high degree of certainty what you will consume next year. But there is a lot of uncertainty over how much you will consume 20 years from now. For example, you might know that next year, your consumption will equal $70,000 +/- $5,000. But 20 years from now, your consumption will equal $70,000 +/- $30,000. Expected consumption is constant at $70,000 per year. But the risk is increasing. So utility is decreasing. That's unlikely to be the optimal strategy for most people. When faced with uncertainty, utility maximization suggests that we will save extra for the future ("precautionary saving") so consumption doesn't fall too low if things go bad. A person wishing to do "precautionary savings" will want to solve for increasing consumption over time, not constant consumption. The good news is that it's easy to fix this in the formulas. It just needs a small tweak to solve for a growing consumption profile. Withdrawal rates, of course, will be lower in the early years to generate growing expected consumption.
Ben, this is not the same as the retirement withdrawal strategy you previously presented here right (I got it bookmarked):
viewtopic.php?t=289064

For those willing to do a little bit of math, would the link above be the technically superior choice in your opinion, or a VPW with rising consumption?
I have not updated that thread, but the short answer is that I set up the problem correctly and got the solution wrong. Once I corrected the mistakes, the strategy I arrived at is the same as the standard lifecycle portfolio model with CRRA utility will imply: hold a fixed allocation and use variable percentage withdrawal. So I didn't get a new strategy out of it. But I understand better how the standard solution does a good job of minimizing sequence of return risk over past returns.

Longinvest's VPW which solves for constant consumption falls within the category of solutions that will minimize sequence of return risk over past returns. But a more general solution will allow for rising or falling consumption based on time preferences as well as precautionary savings. Longinvest's VPW is a special case where time discounting exactly cancels out precautionary savings, leaving expected consumption flat.
Very helpful, thanks Ben.

User avatar
Ben Mathew
Posts: 1045
Joined: Tue Mar 13, 2018 11:41 am
Location: Seattle
Contact:

Re: Finding Middle Ground on the LMP

Post by Ben Mathew » Fri Jun 26, 2020 6:15 pm

willthrill81 wrote:
Fri Jun 26, 2020 4:54 pm
Ben Mathew wrote:
Fri Jun 26, 2020 4:38 pm
willthrill81 wrote:
Fri Jun 26, 2020 3:51 pm
A very plausible alternative to all of those approaches is to use a spreadsheet and the time value of money formula, including expected returns for one's portfolio and multiple income streams beginning (or ending) at various points in the future to determine how much one can withdraw at the current point in time and adjust subsequent withdrawals as expectations change. I personally find this to be the most logical and flexible approach to making withdrawals.
I am assuming this is amortizing the full portfolio (including all future income streams like pension and social security) over years left. The primary difference between this and Longinvest's VPW is would be (1) include PV of all income streams in the portfolio (2) use your own forward looking estimate for portfolio return and (3) use your own estimate of years left instead of (100-age). Is that right?
You can fully amortize the portfolio, but you certainly don't have to. You can select the terminal value to be anything you want, including a greater balance than you started with. That's the beauty of the approach; it's extremely flexible.
That's good. I like the flexibility of inputting a terminal value besides zero. Could be used for LTC expenses, longevity insurance, bequest motive, or just a general cushion.
willthrill81 wrote:
Fri Jun 26, 2020 4:54 pm
Ben Mathew wrote:
Fri Jun 26, 2020 4:38 pm
If so, I think it is at base a good strategy for the risk portfolio. The main problem is that it is solving for flat consumption, which increases risk in later years (same expected consumption, higher std deviation of consumption, therefore lower utility). I think the amortization formula should be used to solve more generally for consumption growing at rate g to accomodate time preferences and precautionary saving. g could be negative (expected consumption falling), but I think that many people would choose positive g. The flat consumption path is obtained by setting g=0, which might be optimal for some, but unlikely to be optimal for all.
Again, using the TVM formula at the core makes this approach exceedingly flexible. You can adjust your expectations for returns to whatever you want, including negative real returns.

In our instance, I'm planning on retiring at around age 52. I fully expect that we'll spend more in our 50s and 60s than in our 70s and 80s. Consequently, slightly front-loading our withdrawals by increasing the returns in the formula by around 1% beyond what we actually expect will accomplish this and result in naturally smaller withdrawals than otherwise later on.
Entering a lower expected return than the true value is a way to hack an increasing consumption profile (and vice versa higher for decreasing). But it is hard to visualize how this will play out for planning purposes. I think explicitly allowing for consumption growth would be neater and easier to think through than doing it indirectly by underestimating portfolio return. I don't know if there are ready-made functions in Excel, but if not, the explicit formula for amortization with growth would not be hard to manually enter.
Last edited by Ben Mathew on Fri Jun 26, 2020 7:03 pm, edited 1 time in total.

User avatar
Ben Mathew
Posts: 1045
Joined: Tue Mar 13, 2018 11:41 am
Location: Seattle
Contact:

Re: Finding Middle Ground on the LMP

Post by Ben Mathew » Fri Jun 26, 2020 6:48 pm

Horton wrote:
Fri Jun 26, 2020 4:58 pm
Ben Mathew - the “time preferences” and CRRA concepts are way too theoretical for me. I may be the dunce in the room, but I suspect that most people won’t be able to integrate this into their planning.
These concepts are probably not as hard as they sound. Time preference is just your willingness to give up consumption today to have more tomorrow. The more you are willing to do that, the more you will want to save, and the more your consumption will increase over time. CRRA (constant relative risk aversion) utility is a bit more involved, but the upshot is that it's the most reasonable of the simple utility functions. The standard models of saving and investing tend to use this specification. Not only is it reasonable, it also gives neat closed form solutions. The main result of interest to us is that a person with CRRA utility will hold a fixed fraction of their wealth in stocks, rebalancing back to the same AA every year. The AA chosen depends on risk aversion, not wealth or time horizon.

That said, ideally, people should be able to arrive at a good strategy without having to understand any of these concepts. IIRC Merton said in some talk that you don't need to know the details of how a car works to be able to drive it. Similarly, financial choices should be packaged and presented in terms where people can pick and choose based on their preferences and needs, without having to understand all the mechanics of what went into the choices being presented.

So instead of asking "What's your time preference?" or "What's your risk aversion" or "Do you want an LMP?", we could translate all of this into a range of possible consumption sequences (with error bands), and ask, do you like this one or this one or this one? e.g. Offer them increasing, decreasing, or flat consumption sequences. By picking the increasing sequence, they are implicitly saying that they value the future highly--that's their time preference. Similarly, by picking a consumption sequence with wide error bands, they are implicitly announcing that their risk aversion is low. And by choosing a consumption sequence where consumption will never fall below $20,000 per year, they are implicitly announcing that they don't have CRRA utility and would like an LMP holding $20,000 per year of guaranteed consumption. They don't need to know what time preference or risk aversion or LMP or CRRA utility is. They just have to pick one consumption sequence over another.

ionball
Posts: 220
Joined: Wed Jan 31, 2018 12:17 pm

Re: Finding Middle Ground on the LMP

Post by ionball » Fri Jun 26, 2020 7:57 pm

I hope the OP plan described is sound because I have implemented a similar version. Rather than LMP based on SS values, I used desired expenditures. For me, SS after age 70 will easily cover 80% or more of expenses. At age 70, the plan is to terminate LMP tracking and switch to VPW as a guideline to not exceed. At or near age 80 will consider an immediate annuity for longevity protection. The annuity would probably be a small portion of the remaining portfolio value and might be used as a cognitive decline protection too.

Regarding LTC, I have tagged a reserve amount that is within the risk portfolio. I monitor that reserve amount to be sure it falls within the bond portion of the asset allocation of the risk portfolio. Currently tracking this plan is more complex than I prefer, but it gives me assurance of safety. At age 70 the tracking will be much simpler and the asset allocation can shift more heavily to stocks.

This plan better work or I'll be kicking myself!

User avatar
Topic Author
Horton
Posts: 682
Joined: Mon Jan 21, 2008 3:53 pm

Re: Finding Middle Ground on the LMP

Post by Horton » Fri Jun 26, 2020 8:21 pm

ionball wrote:
Fri Jun 26, 2020 7:57 pm
This plan better work or I'll be kicking myself!
Sounds like a good plan to me! :beer

If anything, you might be able to increase your spending.

User avatar
Ben Mathew
Posts: 1045
Joined: Tue Mar 13, 2018 11:41 am
Location: Seattle
Contact:

Re: Finding Middle Ground on the LMP

Post by Ben Mathew » Fri Jun 26, 2020 8:32 pm

ionball wrote:
Fri Jun 26, 2020 7:57 pm
I hope the OP plan described is sound because I have implemented a similar version. Rather than LMP based on SS values, I used desired expenditures. For me, SS after age 70 will easily cover 80% or more of expenses. At age 70, the plan is to terminate LMP tracking and switch to VPW as a guideline to not exceed. At or near age 80 will consider an immediate annuity for longevity protection. The annuity would probably be a small portion of the remaining portfolio value and might be used as a cognitive decline protection too.

Regarding LTC, I have tagged a reserve amount that is within the risk portfolio. I monitor that reserve amount to be sure it falls within the bond portion of the asset allocation of the risk portfolio. Currently tracking this plan is more complex than I prefer, but it gives me assurance of safety. At age 70 the tracking will be much simpler and the asset allocation can shift more heavily to stocks.

This plan better work or I'll be kicking myself!
Sounds like you have covered a lot of bases.

Can you explain what you mean by "tracking" in the bolded instances above?

Dandy
Posts: 6283
Joined: Sun Apr 25, 2010 7:42 pm

Re: Finding Middle Ground on the LMP

Post by Dandy » Sat Jun 27, 2020 7:59 am

Bernstein's approach is for those who have enough. He goes beyond LMP for the bridge to SS at age 70 and suggests safe, short term bonds/TIPS latter -not just bonds. I think after the bridge to collect SS at age 70 is the key. Bernstein would suggest a longer "safe" portfolio to fund basic retirement expenses for another decade or more. I suggest that healthy retirees plan to fund retirement to age 90 at a minimum. That gives you 20 years beyond age 70 of retirement funding needs.

I think you could get that extra funding by an estimate of your net RMDs. Any additional funding to me is a using a flexible withdrawal approach to the "risk" portfolio (including RMDs) i.e. Taking more from equities when they do well and less when they don't. Also, that the assets beyond the bridge to SS at age 70 are invested in both equities and fixed income.

ionball
Posts: 220
Joined: Wed Jan 31, 2018 12:17 pm

Re: Finding Middle Ground on the LMP

Post by ionball » Sat Jun 27, 2020 10:35 am

Ben Mathew wrote:
Fri Jun 26, 2020 8:32 pm
Can you explain what you mean by "tracking" in the bolded instances above?
I made a poor word choice and should have said "monitoring". My monitoring spreadsheet aggregates several accounts, checks AA, and stress tests for a 50% decline in stocks.

ionball
Posts: 220
Joined: Wed Jan 31, 2018 12:17 pm

Re: Finding Middle Ground on the LMP

Post by ionball » Sat Jun 27, 2020 10:52 am

Horton wrote:
Fri Jun 26, 2020 8:21 pm
If anything, you might be able to increase your spending.
Tried to spend on travel this year including an expensive trip to Switzerland booked for September. COVID-19 impact on travel was not in my financial plan, but I'm thankful for what I have. :beer

User avatar
Ben Mathew
Posts: 1045
Joined: Tue Mar 13, 2018 11:41 am
Location: Seattle
Contact:

Re: Finding Middle Ground on the LMP

Post by Ben Mathew » Sat Jun 27, 2020 10:54 am

ionball wrote:
Sat Jun 27, 2020 10:35 am
Ben Mathew wrote:
Fri Jun 26, 2020 8:32 pm
Can you explain what you mean by "tracking" in the bolded instances above?
I made a poor word choice and should have said "monitoring". My monitoring spreadsheet aggregates several accounts, checks AA, and stress tests for a 50% decline in stocks.
Got it. Thanks for clarifying.

grok87
Posts: 9100
Joined: Tue Feb 27, 2007 9:00 pm

Re: Finding Middle Ground on the LMP

Post by grok87 » Sun Jun 28, 2020 9:01 am

Good thread.

One thing i have not seen discussed though is that, under current law, Social Security benefits will be cut by an amount currently estimated at 24% starting in 2034.

I'm wondering how folks are factoring that in. i think for my personal situation it means i should probably start collecting social security as early as i can rather than waiting to age 70.- i.e. i want to get as many years as possible of the "non-haircut" benefit.

cheers,
grok
RIP Mr. Bogle.

User avatar
Topic Author
Horton
Posts: 682
Joined: Mon Jan 21, 2008 3:53 pm

Re: Finding Middle Ground on the LMP

Post by Horton » Sun Jun 28, 2020 9:30 am

grok87 wrote:
Sun Jun 28, 2020 9:01 am
Good thread.

One thing i have not seen discussed though is that, under current law, Social Security benefits will be cut by an amount currently estimated at 24% starting in 2034.

I'm wondering how folks are factoring that in. i think for my personal situation it means i should probably start collecting social security as early as i can rather than waiting to age 70.- i.e. i want to get as many years as possible of the "non-haircut" benefit.

cheers,
grok
If I were within 10 or 15 years of beginning Social Security, I wouldn’t worry too much about it. If SS benefits are changed there are a wide range of options available, and generations may be impacted differently. On the off chance that existing SS benefits were reduced, the other thing to consider is that 75% of a higher number is better than 75% of a lower number.

If I were more than 10 or 15 years away from SS, then I would build a plan that’s flexible enough to handle reductions in the benefit.

In the end, this all goes back to my last point in the OP - the strategy should work well for those retiring between 55-70. If you are retiring earlier, then all bets are off - flexibility is the key.

dbr
Posts: 32820
Joined: Sun Mar 04, 2007 9:50 am

Re: Finding Middle Ground on the LMP

Post by dbr » Sun Jun 28, 2020 9:40 am

Considering that the purest, maybe even the only real, LMP is an inflation indexed annuity, which is what SS is, it would seem that advocating a fixed income bridge to SS and then SS is pretty much as 100% pure an LMP as you are going to find.

I always thought the LMP idea was for "absolutely" needed spending and that there is also a "risk" portfolio for discretionary spending.

I suspect in a lot of cases there is no decision power here but just words about what people want to call things.

User avatar
willthrill81
Posts: 19097
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: Finding Middle Ground on the LMP

Post by willthrill81 » Sun Jun 28, 2020 9:47 am

dbr wrote:
Sun Jun 28, 2020 9:40 am
I suspect in a lot of cases there is no decision power here but just words about what people want to call things.
Throughout the forum (and much of human life), what to label something is considered to be a big deal. IMHO, it seldom is.

"What's in a name? That which we call a rose / By Any Other Name would smell as sweet."
-Shakespeare
“It's a dangerous business, Frodo, going out your door. You step onto the road, and if you don't keep your feet, there's no knowing where you might be swept off to.” J.R.R. Tolkien,The Lord of the Rings

grok87
Posts: 9100
Joined: Tue Feb 27, 2007 9:00 pm

Re: Finding Middle Ground on the LMP

Post by grok87 » Sun Jun 28, 2020 9:57 am

Horton wrote:
Sun Jun 28, 2020 9:30 am
grok87 wrote:
Sun Jun 28, 2020 9:01 am
Good thread.

One thing i have not seen discussed though is that, under current law, Social Security benefits will be cut by an amount currently estimated at 24% starting in 2034.

I'm wondering how folks are factoring that in. i think for my personal situation it means i should probably start collecting social security as early as i can rather than waiting to age 70.- i.e. i want to get as many years as possible of the "non-haircut" benefit.

cheers,
grok
If I were within 10 or 15 years of beginning Social Security, I wouldn’t worry too much about it. If SS benefits are changed there are a wide range of options available, and generations may be impacted differently. On the off chance that existing SS benefits were reduced, the other thing to consider is that 75% of a higher number is better than 75% of a lower number.

If I were more than 10 or 15 years away from SS, then I would build a plan that’s flexible enough to handle reductions in the benefit.

In the end, this all goes back to my last point in the OP - the strategy should work well for those retiring between 55-70. If you are retiring earlier, then all bets are off - flexibility is the key.
thanks for your thoughts.
i'm going to try to quantify this somehow using the baseline approach that current law will not change. this may be a slightly tricky exercise but i think it is worth doing...
RIP Mr. Bogle.

ionball
Posts: 220
Joined: Wed Jan 31, 2018 12:17 pm

Re: Finding Middle Ground on the LMP

Post by ionball » Sun Jun 28, 2020 10:41 am

grok87 wrote:
Sun Jun 28, 2020 9:01 am
Good thread.

One thing i have not seen discussed though is that, under current law, Social Security benefits will be cut by an amount currently estimated at 24% starting in 2034.

I'm wondering how folks are factoring that in. i think for my personal situation it means i should probably start collecting social security as early as i can rather than waiting to age 70.- i.e. i want to get as many years as possible of the "non-haircut" benefit.

cheers,
grok
Good point! The projected SS shortfall is potentially an important consideration. It seems like a good idea to stress test your retirement plan for that possibility before ending the employment income phase. My approach was/is simplistic: run the retirement plan numbers in a 2nd scenario with a 25% reduction of currently expected SS benefit. For me, a cut would cause me to stomp and grumble, but not as painful as it would have been to extend employment.

Perhaps the opensocialsecurity.com calculator would help one to decide when to claim SS benefits. I have not crossed that bridge yet.

User avatar
Ben Mathew
Posts: 1045
Joined: Tue Mar 13, 2018 11:41 am
Location: Seattle
Contact:

Re: Finding Middle Ground on the LMP

Post by Ben Mathew » Sun Jun 28, 2020 11:50 am

dbr wrote:
Sun Jun 28, 2020 9:40 am
Considering that the purest, maybe even the only real, LMP is an inflation indexed annuity, which is what SS is, it would seem that advocating a fixed income bridge to SS and then SS is pretty much as 100% pure an LMP as you are going to find.
I'm disappointed that inflation indexed annuities are not available except for social security. A compromise for those seeking more than what social security provides would be to buy a TIPS that pays out at age 90, and use the proceeds to purchase a nominal SPIA at age 90. You avoid inflation risk till age 90, and longevity risk after. It's a middle ground between inflation risk and longevity risk.
dbr wrote:
Sun Jun 28, 2020 9:40 am
I always thought the LMP idea was for "absolutely" needed spending and that there is also a "risk" portfolio for discretionary spending.

I suspect in a lot of cases there is no decision power here but just words about what people want to call things.
Thinking about it from a utility function perspective can help sharpen concepts. Suppose you are down to a portfolio, which if 100% invested in bonds, is just enough to fund $20,000 per year in guaranteed spending for the rest of your life. Would you still invest a portion of your portfolio in stocks? Or would you in no way consider the possibility of dropping below $20,000 per year of spending. If you would decline to invest in stocks, and choose to live on $20,000 per year thereafter, then that means you should fence off $20,000 in an LMP. Your consumption will never fall below $20,000 per year and the cost of achieving that is worth it to you.

grok87
Posts: 9100
Joined: Tue Feb 27, 2007 9:00 pm

Re: Finding Middle Ground on the LMP

Post by grok87 » Sun Jun 28, 2020 11:51 am

ionball wrote:
Sun Jun 28, 2020 10:41 am
grok87 wrote:
Sun Jun 28, 2020 9:01 am
Good thread.

One thing i have not seen discussed though is that, under current law, Social Security benefits will be cut by an amount currently estimated at 24% starting in 2034.

I'm wondering how folks are factoring that in. i think for my personal situation it means i should probably start collecting social security as early as i can rather than waiting to age 70.- i.e. i want to get as many years as possible of the "non-haircut" benefit.

cheers,
grok
Good point! The projected SS shortfall is potentially an important consideration. It seems like a good idea to stress test your retirement plan for that possibility before ending the employment income phase. My approach was/is simplistic: run the retirement plan numbers in a 2nd scenario with a 25% reduction of currently expected SS benefit. For me, a cut would cause me to stomp and grumble, but not as painful as it would have been to extend employment.

Perhaps the opensocialsecurity.com calculator would help one to decide when to claim SS benefits. I have not crossed that bridge yet.
thanks.
yes- i was thinking of using the calculator. Currently i am assuming my benefits will be cut, on average, by 12%. But i'm thinking of revising my assumptions and just using the full 24% cut for planning purposes.

I hope this doesn't start some sort of off-topic debate. The way i think about it the 24% is what is projected to happen under current law. Hopefully that is a non-controversial statement.
cheers,
grok
RIP Mr. Bogle.

User avatar
vineviz
Posts: 6666
Joined: Tue May 15, 2018 1:55 pm

Re: Finding Middle Ground on the LMP

Post by vineviz » Sun Jun 28, 2020 12:01 pm

Ben Mathew wrote:
Sun Jun 28, 2020 11:50 am

I'm disappointed that inflation indexed annuities are not available except for social security. A compromise for those seeking more than what social security provides would be to buy a TIPS that pays out at age 90, and use the proceeds to purchase a nominal SPIA at age 90. You avoid inflation risk till age 90, and longevity risk after. It's a middle ground between inflation risk and longevity risk.
What you describe is very close to the lowest risk solution, but not quite right. The goal would be to buy a TIPS with a duration which matches the current duration of the annuity you intend to purchase later.

In other words if you are currently 70 and want to fund a SPIA to be purchased at age 90, you’d need a TIPS that currently has a duration of 24 years and not a maturity of 20 years.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch

grok87
Posts: 9100
Joined: Tue Feb 27, 2007 9:00 pm

Re: Finding Middle Ground on the LMP

Post by grok87 » Sun Jun 28, 2020 12:04 pm

vineviz wrote:
Sun Jun 28, 2020 12:01 pm
Ben Mathew wrote:
Sun Jun 28, 2020 11:50 am

I'm disappointed that inflation indexed annuities are not available except for social security. A compromise for those seeking more than what social security provides would be to buy a TIPS that pays out at age 90, and use the proceeds to purchase a nominal SPIA at age 90. You avoid inflation risk till age 90, and longevity risk after. It's a middle ground between inflation risk and longevity risk.
What you describe is very close to the lowest risk solution, but not quite right. The goal would be to buy a TIPS with a duration which matches the current duration of the annuity you intend to purchase later.

In other words if you are currently 70 and want to fund a SPIA to be purchased at age 90, you’d need a TIPS that currently has a duration of 24 years and not a maturity of 20 years.
sounds about right
RIP Mr. Bogle.

User avatar
Ben Mathew
Posts: 1045
Joined: Tue Mar 13, 2018 11:41 am
Location: Seattle
Contact:

Re: Finding Middle Ground on the LMP

Post by Ben Mathew » Sun Jun 28, 2020 1:06 pm

vineviz wrote:
Sun Jun 28, 2020 12:01 pm
Ben Mathew wrote:
Sun Jun 28, 2020 11:50 am

I'm disappointed that inflation indexed annuities are not available except for social security. A compromise for those seeking more than what social security provides would be to buy a TIPS that pays out at age 90, and use the proceeds to purchase a nominal SPIA at age 90. You avoid inflation risk till age 90, and longevity risk after. It's a middle ground between inflation risk and longevity risk.
What you describe is very close to the lowest risk solution, but not quite right. The goal would be to buy a TIPS with a duration which matches the current duration of the annuity you intend to purchase later.

In other words if you are currently 70 and want to fund a SPIA to be purchased at age 90, you’d need a TIPS that currently has a duration of 24 years and not a maturity of 20 years.
Yes. I think of bonds in terms of their payout date, but maybe did not make that clear. You'd ideally structure this so the payout comes at age 90. With zero coupon bonds, it would be straightforward, duration=maturity=years left till age 90. With a coupon bond like TIPS, you'd want longer maturity to compensate for the earlier coupons. Put zero coupon TIPS as yet another financial product that I think should exist, but does not (as far as I know).

Post Reply