Total Portfolio Allocation and Withdrawal (TPAW)

[Ben Mathew]

Hi Ben,

I have noticed that the stock allocation in the View Balance History changes with a couple percentage points (up or down) after the Monthly Rebalance (while none of the input parameter have changed). Is there an explanation for this?

Kind regards,
Michiel

I imagine the expected returns of stocks/bonds have changed

For the current portfolio balance estimation shown in “Current Portfolio Balance → View Balance History,” the “Monthly Rebalance” event updates the asset allocation used for the estimation to the asset allocation that is shown in “Tasks” at that time. The asset allocation shown in “Tasks” has most likely changed since the last “Monthly Rebalance.” This is due to changes in the expected returns of stocks and bonds, portfolio balance (because asset allocation is affected by the size of the portfolio relative to essential expenses and income during retirement), inflation rates that change the present value of nominal pensions, etc.

[Ben Mathew]

trying to look at suggested spend and actual expenses graphs and figure out the delta on my own is painstaking along the timeline.

I'm understanding that the "suggested spend" graph is the "Monthly Spending During Retirement" graph shown in the planner.

However I’m not sure what you mean by the “actual expenses graph.” Is it a graph in the planner, or a separate graph in a spreadsheet for example? If it is from the planner, can you give me the title of the graph as shown in the planner?

[Syner01]

trying to look at suggested spend and actual expenses graphs and figure out the delta on my own is painstaking along the timeline.

I'm understanding that the "suggested spend" graph is the "Monthly Spending During Retirement" graph shown in the planner.

However I’m not sure what you mean by the “actual expenses graph.” Is it a graph in the planner, or a separate graph in a spreadsheet for example? If it is from the planner, can you give me the title of the graph as shown in the planner?

expenses in your product show under 'extra spending' unless i am misunderstanding that section. i added my non discretionary expenses under 'spending - essential' and my discretionary expenses under 'spending - discretionary'. i want to see those together as 'total expenses' or at least 'essential' if nothing else against 'suggested spending' to see the delta. Am i the only one that wants to see and calculate the delta between what the product suggests spending to be (which is really a portfolio withdrawal strategy) vs my actual listed expenses?
maybe if you can't overlay on the graph you can create a 'report' section and show some basic bar/chart graphs with both together and delta or something like you would normally find in spreadsheets graph functions.

[Ben Mathew]

trying to look at suggested spend and actual expenses graphs and figure out the delta on my own is painstaking along the timeline.

I'm understanding that the "suggested spend" graph is the "Monthly Spending During Retirement" graph shown in the planner.

However I’m not sure what you mean by the “actual expenses graph.” Is it a graph in the planner, or a separate graph in a spreadsheet for example? If it is from the planner, can you give me the title of the graph as shown in the planner?

expenses in your product show under 'extra spending' unless i am misunderstanding that section. i added my non discretionary expenses under 'spending - essential' and my discretionary expenses under 'spending - discretionary'. i want to see those together as 'total expenses' or at least 'essential' if nothing else against 'suggested spending' to see the delta. Am i the only one that wants to see and calculate the delta between what the product suggests spending to be (which is really a portfolio withdrawal strategy) vs my actual listed expenses?
maybe if you can't overlay on the graph you can create a 'report' section and show some basic bar/chart graphs with both together and delta or something like you would normally find in spreadsheets graph functions.

The "Extra Spending" section is meant to be for expenses that you incur in certain years and not others—e.g. college tuition for 4 years which is an expense you won't have in other years. You don't need to enter your routine expenses here. The exception to this is if you want to create a spending floor at say $1,000 per month. Then you would enter $1,000 per month as an extra essential expense for all retirement years. (We'll eventually move spending floors to a different section. Then the "Extra Spending" section will need to be used only for extra spending needs, That will reduce some of the confusion around this.) But there is generally no need to enter $X per year as an Extra Discretionary Expense for all of retirement. So the sum of extra essential and extra discretionary expenses won't normally add up to an estimate of planned expenses.

So in your plan, keep the non-discretionary expenses you've entered under Extra Essential Expenses. But remove the discretionary expense estimates you have entered under Extra Discretionary Expenses unless it's truly extra for specific years. For example, if you need an extra $1,000 per month from ages 65-70 for say travel, enter $1,000 per month in Extra Discretionary Expenses for just those ages.

With that, you'll have the following graphs:

• The "Monthly Spending During Retirement" graph will show your total spending outcomes.
• The "Extra Spending > Essential" graph will show just your essential spending outcomes.
• The "General Spending" graph will show your discretionary spending outcomes.

If the discretionary spending outcomes shown in the "General Spending" graph is higher at certain ages and percentiles than you plan to spend, you can fix it by adjusting the inputs in the ways that I described earlier.

[Syner01]

trying to look at suggested spend and actual expenses graphs and figure out the delta on my own is painstaking along the timeline.

I'm understanding that the "suggested spend" graph is the "Monthly Spending During Retirement" graph shown in the planner.

However I’m not sure what you mean by the “actual expenses graph.” Is it a graph in the planner, or a separate graph in a spreadsheet for example? If it is from the planner, can you give me the title of the graph as shown in the planner?

expenses in your product show under 'extra spending' unless i am misunderstanding that section. i added my non discretionary expenses under 'spending - essential' and my discretionary expenses under 'spending - discretionary'. i want to see those together as 'total expenses' or at least 'essential' if nothing else against 'suggested spending' to see the delta. Am i the only one that wants to see and calculate the delta between what the product suggests spending to be (which is really a portfolio withdrawal strategy) vs my actual listed expenses?
maybe if you can't overlay on the graph you can create a 'report' section and show some basic bar/chart graphs with both together and delta or something like you would normally find in spreadsheets graph functions.

The "Extra Spending" section is meant to be for expenses that you incur in certain years and not others—e.g. college tuition for 4 years which is an expense you won't have in other years. You don't need to enter your routine expenses here. The exception to this is if you want to create a spending floor at say $1,000 per month. Then you would enter $1,000 per month as an extra essential expense for all retirement years. (We'll eventually move spending floors to a different section. Then the "Extra Spending" section will need to be used only for extra spending needs, That will reduce some of the confusion around this.) But there is generally no need to enter $X per year as an Extra Discretionary Expense for all of retirement. So the sum of extra essential and extra discretionary expenses won't normally add up to an estimate of planned expenses.

So in your plan, keep the non-discretionary expenses you've entered under Extra Essential Expenses. But remove the discretionary expense estimates you have entered under Extra Discretionary Expenses unless it's truly extra for specific years. For example, if you need an extra $1,000 per month from ages 65-70 for say travel, enter $1,000 per month in Extra Discretionary Expenses for just those ages.

With that, you'll have the following graphs:

• The "Monthly Spending During Retirement" graph will show your total spending outcomes.
• The "Extra Spending > Essential" graph will show just your essential spending outcomes.
• The "General Spending" graph will show your discretionary spending outcomes.

If the discretionary spending outcomes shown in the "General Spending" graph is higher at certain ages and percentiles than you plan to spend, you can fix it by adjusting the inputs in the ways that I described earlier.

Thanks Ben,
not sure i get what you are saying above. are you saying to add regular expenses like my avg heat, electric, mortgage monthly bill as a floor and then only include expenses with finite dates in the expenses section? how to keep track of my expenses then if i bunch them all as a 'floor'. also how to add those to the other expenses in the expenses sections and get a total to compare against suggested spend?

i have expenses in retirement. how to include them and compare them against suggested spending? they could be regular, one off, forever until i die, planned, unplanned, etc. no matter how you label them i have expenses. how can i add them as line items and add them up against my suggested spending like a budget? i am trying to model my retirement which is basically spending from retirement income streams offset by taxes and expenses. i want to be able to see what my total expenses are (no matter what type) and see the delta between what i need and what the model is suggesting i take out each month. bonus for detailed expenses reporting but for now just want to be able to add each in with start/end date/amount and build a graph and compare that one to the suggested spending. is there a basic way to do this? Sorry Ben trying to lump sum my expenses as a spending floor doesn't really help me budget well. it will get out of whack pretty quickly as these items change and i don't have a line item to update for each.

maybe this is the wrong tool for this kind of calculation though. sorry if i am going in a different direction than you intend. i was also hoping to see the full (total) expenses against the suggested spending on the graph so i could play around with the 'tilt' sliders and balance some sections out better. can i export your output to a spreadsheet maybe so i can add my own expenses and compare?

Thoughts?

[9-5 Suited]

For asset allocations focused on the Savings Portfolio, whether one agrees with the numbers or not, there are some readily accepted industry norms for aggressive-moderate-conservative allocations. Looking a asset allocation funds for example, you'll usually find 80-90% stock is aggressive, 60% is moderate, and 40% is conservative. The particulars don't matter here, what matters is that there are rough parameters the broad community understands for purposes of conversation.

If the TPAW approach moved toward the mainstream, it would end up generating its own conversational conventions for these broad terms. I imagine the entire curve would shift because the primary change is adding more income streams to the bond bucket. So would it be reasonable to think that 30% would become conservative, 50% moderate, and 70% aggressive? Trying to get a feel for the likely risk tolerance dispersion of allocations in a TPAW world.

My Total Portfolio allocation is 57%/43% on a roughly 70%/30% Savings Portfolio.

[Ben Mathew]

trying to look at suggested spend and actual expenses graphs and figure out the delta on my own is painstaking along the timeline.

I'm understanding that the "suggested spend" graph is the "Monthly Spending During Retirement" graph shown in the planner.

However I’m not sure what you mean by the “actual expenses graph.” Is it a graph in the planner, or a separate graph in a spreadsheet for example? If it is from the planner, can you give me the title of the graph as shown in the planner?

expenses in your product show under 'extra spending' unless i am misunderstanding that section. i added my non discretionary expenses under 'spending - essential' and my discretionary expenses under 'spending - discretionary'. i want to see those together as 'total expenses' or at least 'essential' if nothing else against 'suggested spending' to see the delta. Am i the only one that wants to see and calculate the delta between what the product suggests spending to be (which is really a portfolio withdrawal strategy) vs my actual listed expenses?
maybe if you can't overlay on the graph you can create a 'report' section and show some basic bar/chart graphs with both together and delta or something like you would normally find in spreadsheets graph functions.

The "Extra Spending" section is meant to be for expenses that you incur in certain years and not others—e.g. college tuition for 4 years which is an expense you won't have in other years. You don't need to enter your routine expenses here. The exception to this is if you want to create a spending floor at say $1,000 per month. Then you would enter $1,000 per month as an extra essential expense for all retirement years. (We'll eventually move spending floors to a different section. Then the "Extra Spending" section will need to be used only for extra spending needs, That will reduce some of the confusion around this.) But there is generally no need to enter $X per year as an Extra Discretionary Expense for all of retirement. So the sum of extra essential and extra discretionary expenses won't normally add up to an estimate of planned expenses.

So in your plan, keep the non-discretionary expenses you've entered under Extra Essential Expenses. But remove the discretionary expense estimates you have entered under Extra Discretionary Expenses unless it's truly extra for specific years. For example, if you need an extra $1,000 per month from ages 65-70 for say travel, enter $1,000 per month in Extra Discretionary Expenses for just those ages.

With that, you'll have the following graphs:

• The "Monthly Spending During Retirement" graph will show your total spending outcomes.
• The "Extra Spending > Essential" graph will show just your essential spending outcomes.
• The "General Spending" graph will show your discretionary spending outcomes.

If the discretionary spending outcomes shown in the "General Spending" graph is higher at certain ages and percentiles than you plan to spend, you can fix it by adjusting the inputs in the ways that I described earlier.

Thanks Ben,
not sure i get what you are saying above. are you saying to add regular expenses like my avg heat, electric, mortgage monthly bill as a floor and then only include expenses with finite dates in the expenses section? how to keep track of my expenses then if i bunch them all as a 'floor'. also how to add those to the other expenses in the expenses sections and get a total to compare against suggested spend?

i have expenses in retirement. how to include them and compare them against suggested spending? they could be regular, one off, forever until i die, planned, unplanned, etc. no matter how you label them i have expenses. how can i add them as line items and add them up against my suggested spending like a budget? i am trying to model my retirement which is basically spending from retirement income streams offset by taxes and expenses. i want to be able to see what my total expenses are (no matter what type) and see the delta between what i need and what the model is suggesting i take out each month. bonus for detailed expenses reporting but for now just want to be able to add each in with start/end date/amount and build a graph and compare that one to the suggested spending. is there a basic way to do this? Sorry Ben trying to lump sum my expenses as a spending floor doesn't really help me budget well. it will get out of whack pretty quickly as these items change and i don't have a line item to update for each.

maybe this is the wrong tool for this kind of calculation though. sorry if i am going in a different direction than you intend. i was also hoping to see the full (total) expenses against the suggested spending on the graph so i could play around with the 'tilt' sliders and balance some sections out better. can i export your output to a spreadsheet maybe so i can add my own expenses and compare?

Thoughts?

Currently there is no way to add up projected expenses like this and have it overlay the Monthly Spending During Retirement graph. You would have to do it manually in a spreadsheet. You can get the Monthly Spending During Retirement data into a spreadsheet by copying and pasting the relevant table from the PDF Report (located under "More" in the bottom right of the Results panel.) The tables are in the appendix. To keep the length of the tables reasonable, they show data for only one month per year, not all 12 months. So that will be a limitation.

It would be useful to be able to do this directly in the planner. We will create a separate budgeting section where you can enter projected expenses for a few different scenarios—a modest budget if money is tight, a more generous budget in case you have more money, etc. And then be able to see where the total expenses associated with these different budgets fall relative to the Monthly Spending During Retirement graph.

I have added this to the to-do list.

Thanks for explaining what you are looking to do. I think this will be a neat feature!

[Ben Mathew]

For asset allocations focused on the Savings Portfolio, whether one agrees with the numbers or not, there are some readily accepted industry norms for aggressive-moderate-conservative allocations. Looking a asset allocation funds for example, you'll usually find 80-90% stock is aggressive, 60% is moderate, and 40% is conservative. The particulars don't matter here, what matters is that there are rough parameters the broad community understands for purposes of conversation.

If the TPAW approach moved toward the mainstream, it would end up generating its own conversational conventions for these broad terms. I imagine the entire curve would shift because the primary change is adding more income streams to the bond bucket. So would it be reasonable to think that 30% would become conservative, 50% moderate, and 70% aggressive? Trying to get a feel for the likely risk tolerance dispersion of allocations in a TPAW world.

My Total Portfolio allocation is 57%/43% on a roughly 70%/30% Savings Portfolio.

Haghani and White surveyed "a group of 31 financially sophisticated friends, clients, and former colleagues from the finance industry" and found their average relative risk aversion (RRA) to be around 3, though they personally lean towards 2 (pages 175-180). I think we can safely say that an RRA of 6 would qualify as conservative and 1.5 as aggressive. On the risk tolerance input in the planner, these translate to:

Conservative: Risk tolerance = 7 (RRA = 6.48)
Moderate: Risk tolerance = 12 (RRA = 3.05)
Aggressive: Risk tolerance = 17 (RRA = 1.44)

To get some rough rules of thumb using these, let's use expected real returns of 5% for stocks and 2% for bonds.

Assume a 65 year old retiree, planning for 30 years (till age 94 yr 11 mo). Whether a particular asset allocation is conservative, moderate or aggressive will depend a lot on how much Social Security and pensions they have relative to their portfolio balance. So let's look at some combinations:

Current Portfolio Balance $1 million + Social Security and Pensions: $0

• Conservative: 14% stocks
• Moderate: 30% stocks
• Aggressive: 65% stocks

Current Portfolio Balance $1 million + Social Security and Pensions: $1,000/month ($12,000 per year)

• Conservative: 18% stocks
• Moderate: 39% stocks
• Aggressive: 82% stocks

Current Portfolio Balance $1 million + Social Security and Pensions: $2,000/month ($24,000 per year)

• Conservative: 22% stocks
• Moderate: 47% stocks
• Aggressive: 100% stocks

Current Portfolio Balance $1 million + Social Security and Pensions: $3,000/month ($36,000 per year)

• Conservative: 26% stocks
• Moderate: 55% stocks
• Aggressive: 100% stocks

Current Portfolio Balance $1 million + Social Security and Pensions: $4,000/month ($48,000 per year)

• Conservative: 30% stocks
• Moderate: 63% stocks
• Aggressive: 100% stocks

Current Portfolio Balance $1 million + Social Security and Pensions: $5,000/month ($60,000 per year)

• Conservative: 34% stocks
• Moderate: 72% stocks
• Aggressive: 100% stocks

Plan: https://tpawplanner.com/link?params=0Sq ... GXV1JxtJWb

This gives some sense of what conservative, moderate and aggressive asset allocations would look like with the lifecycle model.

Note that this table will scale without affecting the asset allocation. i.e. The asset allocation for $1 million + $2,000 per month will be the same as that of $500,000 + $1,000 per month.

These numbers assume riskless bonds. If bonds have interest rate risk and are not duration matched, that will increase the stock allocations listed above.

Based on the numbers above, a 60/40 portfolio would be moderate for a 65 year old retiree with a $1 million portfolio + $4,000 per month in Social Security and pensions (or a $500K portfolio + $2,000 per month, etc.) . The same 60/40 portfolio would be aggressive for a 65 year old retiree with no Social Security or pension.

[dogagility]

Whether a particular asset allocation is conservative, moderate or aggressive will depend a lot on how much Social Security and pensions they have relative to their portfolio balance.

Agree completely.

Your planner is elegantly simple in how it incorporates this concept into an easy to use retirement calculator.

[Ben Mathew]

Whether a particular asset allocation is conservative, moderate or aggressive will depend a lot on how much Social Security and pensions they have relative to their portfolio balance.

Agree completely.

Your planner is elegantly simple in how it incorporates this concept into an easy to use retirement calculator.

Thanks!

[Solicitorious]

Hi Ben

New poster here.

Firstly, a thousand thanks for all your hard work in creating and maintaining - and explaining - this wonderful tool. It's a game-changer, and streets ahead of anything I've ever seen, and I've seen a lot...

If I may be so presumptuous in my first post to suggest two slight enhancements, really just to improve the "at a glance" user experience.

a) perhaps a toggle on the legacy card, to switch to showing the 5/50/95 percentiles of total (lifetime) plan spend instead. [for those whose primary objective is extracting the most money while alive]

b) perhaps a toggle on the graph card, to switch to showing the 5/50/95 percentiles of yearly spend as a (scrollable) table, rather than a graph. [saves having to repeatedly hover over the graph bars to analyse what change occurs at year x, when plan inputs change]

Thanks again.

[Ben Mathew]

Hi Ben

New poster here.

Welcome to the forum!

Firstly, a thousand thanks for all your hard work in creating and maintaining - and explaining - this wonderful tool. It's a game-changer, and streets ahead of anything I've ever seen, and I've seen a lot...

Delighted to hear this. Thanks!

If I may be so presumptuous in my first post to suggest two slight enhancements, really just to improve the "at a glance" user experience.

Suggestions are always welcome. The planner has benefited a lot from the feedback we have received so far!

a) perhaps a toggle on the legacy card, to switch to showing the 5/50/95 percentiles of total (lifetime) plan spend instead. [for those whose primary objective is extracting the most money while alive]

Were you interested in the total or average spending across all retirement years?

Given that the graph shows spending by age, would you still find the total (or average) spending across ages to be useful? For example, if you increase spending tilt, it will push spending towards late retirement, which will cause the total/average spending to rise (because spending in the far future is cheaper—i.e. has lower present value—than the near future.) But that still wouldn't be a good enough reason on its own to increase the spending tilt. Total spending will always be maximized by delaying spending—but it's not the right metric to maximize because spreading spending across ages also matters. Given this, would the total/average spending statistic still be useful? I worry that publishing this statistic would risk being a distraction from the more nuanced decision of choosing the best spending profile across time.

b) perhaps a toggle on the graph card, to switch to showing the 5/50/95 percentiles of yearly spend as a (scrollable) table, rather than a graph. [saves having to repeatedly hover over the graph bars to analyse what change occurs at year x, when plan inputs change]

Thanks again.

I like the idea of a toggle that will display a table with the numbers behind the graph. I've added it to the to-do list.

Thanks for the suggestions!

[Circle the Wagons]

DON'T REBALANCE TOO FREQUENTLY

I would strongly caution against trading frequently to get to the target asset allocation. Besides taxes and bid/ask spreads, at short horizons there may also be things like price momentum that will need to be considered. We are also not adjusting for short term changes in risk based on the volatility of recent stock returns (I plan to add this down the road). So I would use the recommended asset allocation more as a target that points the way for new investments, with only an occasional rebalance when the allocation gets too far out of whack. In the absence of a clear understanding of costs and short horizon price artifacts, I would try to keep the target asset allocation bands broad enough that major rebalances don't have to happen more than a few times a year. At some point, I'll try to build a model of trading costs and rebalancing schedules into the planner to help with this decision, but that will be pretty far down the list.

Ben, noting a few things about Haghani and White's approach that I was hoping you might comment on:

- they use an aggressive rebalancing strategy - weekly with four sub-portfolios (partially but not fully due to an aggressive TLH approach)

- they use momentum as the risk input vs. historical volatility / SD; this seems to have the effect of much wilder frequent swings in target AA than what TPAW would recommend (swings which are being captured because of the frequent rebalancing)

- they don't seem to consider duration matching to be optimal / required for the fixed income portion of the portfolio; for example, the aforementioned wild swings include big swings in the mix of tbills vs. bond funds that wouldn't be driven by an investor's duration needs but rather by macros; the word "duration" is mentioned zero times in their book

- they are TIPS fans but not 100% of fixed income; depends on valuations and momentum

Thank you.

[RajaDada]

This may have been explained already. If so please point me to the post. I could not find an answer when I searched.

Why does my monthly inflation-adjusted spending amount (50th percentile) go up over time? I understand that I can adjust the spending tilt (in the risk section) but why does the default go up? Is the withdrawal rate set lower earlier on to manage risk? Or is it in anticipation of higher retirement costs later in life, perhaps medical costs?

My retirement budget calls for generous travel in the early years - about 30% of our entire budget. It will be wonderful if we can keep this up, but I suspect our travel will decline over time. My point being, I imagine spending less as time goes on, not more. But am I taking on unnecessary risk or not planning appropriately for the future if I adjust the tilt? The tilt adjustment is not intuitive to me. The picture below shows a -0.2% adjustment but I don't know what that means. Am I being reckless?

Thanks! (edited for clarity)

[Solicitorious]

a) perhaps a toggle on the legacy card, to switch to showing the 5/50/95 percentiles of total (lifetime) plan spend instead. [for those whose primary objective is extracting the most money while alive]

Were you interested in the total or average spending across all retirement years?

Given that the graph shows spending by age, would you still find the total (or average) spending across ages to be useful? For example, if you increase spending tilt, it will push spending towards late retirement, which will cause the total/average spending to rise (because spending in the far future is cheaper—i.e. has lower present value—than the near future.) But that still wouldn't be a good enough reason on its own to increase the spending tilt. Total spending will always be maximized by delaying spending—but it's not the right metric to maximize because spreading spending across ages also matters. Given this, would the total/average spending statistic still be useful? I worry that publishing this statistic would risk being a distraction from the more nuanced decision of choosing the best spending profile across time.

Hi Ben

I just think it would make TPAW consistent with other similar tools, e.g. FICalc/FireCalc, which do show the median (50th percentile) total plan (lifetime) spend.

Users of course could balance this hypothetical number versus the benefits of a (g<0). There is always risk that a user will not live to enjoy the hypothetical benefits of (g>=0).

I intend to use (g<0) in my plan, but would also like to evaluate total plan spend for other g, based on a hypothetical lifespan.

In a nutshell, I want to maximise my spend now, while ensuring, as far as possible, that I don't want to be bankrupt, should I live to a great age (unlikely, given my lifestyle and family history, medication, maximum family ages 85-93).
I'm in the UK, and even if I'm bankrupt @ 90+, the state should step-in to stop me from living in a cardboard box. I doubt I'll care @95 - should I get to that unlikely age - who is wiping my ass, or what my surroundings are...

Displaying total median spend would be consistent with similar tools. I can easily work out the annual average myself by dividing it by plan length, e.g. 30 years, although the average is of limited utility - as you say - the shape of the spend curve is the primary interest...

[Raspberry-503]

This may have been explained already. If so please point me to the post. I could not find an answer when I searched.

Why does my monthly inflation-adjusted spending rate (50th percentile) go up over time? I understand that I can adjust the tilt but why does the default go up? Is the withdrawal rate set lower earlier on to manage risk? Or is it in anticipation of higher retirement costs later in life, perhaps medical costs?

My retirement budget calls for generous travel in the early years - about 30% of our entire budget. It will be wonderful if we can keep this up until we die, but I have a sneaking suspicion that our travel will decline over time. My point being, I imagine spending less as time goes by, not more. But I am I taking on unnecessary risk or not planning appropriately for the future if I adjust the tilt? The tilt adjustment is not intuitive to me. The picture below shows a -0.2% adjustment but I don't know what that means. Am I being reckless?

Thanks!

I don't know the answer to your question, but what is did was add $1500 discretionary monthly expenses for the first 20 years, that created a "step" in spending rather than a smooth transition, which I think I prefer (my 75.yrsr old self may not thank me?)

I also played with the slider but I don't remember the default going up as you mention. In fact if I recall there are 2 sliders that work in concert.

[RajaDada]

I don't know the answer to your question, but what is did was add $1500 discretionary monthly expenses for the first 20 years, that created a "step" in spending rather than a smooth transition, which I think I prefer (my 75.yrsr old self may not thank me?)

I also played with the slider but I don't remember the default going up as you mention. In fact if I recall there are 2 sliders that work in concert.

Thanks. I tried adding $1000 per month for 20 years. After 20 years, the monthly amount is still larger than the original monthly amount, which is good.

[Ben Mathew]

DON'T REBALANCE TOO FREQUENTLY

I would strongly caution against trading frequently to get to the target asset allocation. Besides taxes and bid/ask spreads, at short horizons there may also be things like price momentum that will need to be considered. We are also not adjusting for short term changes in risk based on the volatility of recent stock returns (I plan to add this down the road). So I would use the recommended asset allocation more as a target that points the way for new investments, with only an occasional rebalance when the allocation gets too far out of whack. In the absence of a clear understanding of costs and short horizon price artifacts, I would try to keep the target asset allocation bands broad enough that major rebalances don't have to happen more than a few times a year. At some point, I'll try to build a model of trading costs and rebalancing schedules into the planner to help with this decision, but that will be pretty far down the list.

Ben, noting a few things about Haghani and White's approach that I was hoping you might comment on:

- they use an aggressive rebalancing strategy - weekly with four sub-portfolios (partially but not fully due to an aggressive TLH approach)

- they use momentum as the risk input vs. historical volatility / SD; this seems to have the effect of much wilder frequent swings in target AA than what TPAW would recommend (swings which are being captured because of the frequent rebalancing)

- they don't seem to consider duration matching to be optimal / required for the fixed income portion of the portfolio; for example, the aforementioned wild swings include big swings in the mix of tbills vs. bond funds that wouldn't be driven by an investor's duration needs but rather by macros; the word "duration" is mentioned zero times in their book

- they are TIPS fans but not 100% of fixed income; depends on valuations and momentum

Thank you.

That's interesting. Where did they talk about rebalancing weekly and their use of momentum as the risk input?

In The Missing Billionaires page 53, they said this about estimating the risk of the market portfolio:

Unfortunately, we cannot directly observe expectations of long-term market risk. Even if there were liquid, long-term markets in options that we could look to, it would be difficult to disentangle the true expectation from the insurance premium built into those prices. It's also not possible to derive expectations of risk from expected cash flows in the way we do when determining long-term expected returns for equities or bonds. What we can do is build up an estimate for market risk using a blend of long- and short-horizon realized volatility. This is the approach that options trading desks at most banks use to forecast market volatility.

[Ben Mathew]

This may have been explained already. If so please point me to the post. I could not find an answer when I searched.

Why does my monthly inflation-adjusted spending amount (50th percentile) go up over time? I understand that I can adjust the spending tilt (in the risk section) but why does the default go up? Is the withdrawal rate set lower earlier on to manage risk? Or is it in anticipation of higher retirement costs later in life, perhaps medical costs?

The bolded part above is right. Because future spending is risky, if you valued spending in all ages equally (time preference rate = 0%), you'll want to underspend a little in early retirement in case markets do poorly. This is precautionary savings. Spending grows over time, not because you prefer to spend more later, but simply because you spend cautiously early on.

To offset this, you can increase the spending tilt or the time preference rate (or add extra spending in early retirement). It makes sense to do that if you value spending in early retirement more than spending in late retirement. If you value all ages the same, then leave the spending tilt and time preference rate at the default 0% and allow expected spending to rise over time.

My retirement budget calls for generous travel in the early years - about 30% of our entire budget. It will be wonderful if we can keep this up, but I suspect our travel will decline over time. My point being, I imagine spending less as time goes on, not more. But am I taking on unnecessary risk or not planning appropriately for the future if I adjust the tilt? The tilt adjustment is not intuitive to me. The picture below shows a -0.2% adjustment but I don't know what that means. Am I being reckless?

Thanks! (edited for clarity)

No, a negative spending tilt does not mean that you're being reckless. It's perfectly legitimate (and common) to want to spend more in early retirement when you are in better health. You just need to weigh the value to you of doing that against the risk of ending up with too little in late retirement. Make sure you can tolerate the 5th percentile spending outcome at all ages. There is no right or wrong in how you structure your spending—only tradeoffs!

[Ben Mathew]

a) perhaps a toggle on the legacy card, to switch to showing the 5/50/95 percentiles of total (lifetime) plan spend instead. [for those whose primary objective is extracting the most money while alive]

Were you interested in the total or average spending across all retirement years?

Given that the graph shows spending by age, would you still find the total (or average) spending across ages to be useful? For example, if you increase spending tilt, it will push spending towards late retirement, which will cause the total/average spending to rise (because spending in the far future is cheaper—i.e. has lower present value—than the near future.) But that still wouldn't be a good enough reason on its own to increase the spending tilt. Total spending will always be maximized by delaying spending—but it's not the right metric to maximize because spreading spending across ages also matters. Given this, would the total/average spending statistic still be useful? I worry that publishing this statistic would risk being a distraction from the more nuanced decision of choosing the best spending profile across time.

Hi Ben

I just think it would make TPAW consistent with other similar tools, e.g. FICalc/FireCalc, which do show the median (50th percentile) total plan (lifetime) spend.

Users of course could balance this hypothetical number versus the benefits of a (g<0). There is always risk that a user will not live to enjoy the hypothetical benefits of (g>=0).

I intend to use (g<0) in my plan, but would also like to evaluate total plan spend for other g, based on a hypothetical lifespan.

Thanks for explaining how you'd like to use it. I've added this to the list.

[Circle the Wagons]

DON'T REBALANCE TOO FREQUENTLY

I would strongly caution against trading frequently to get to the target asset allocation. Besides taxes and bid/ask spreads, at short horizons there may also be things like price momentum that will need to be considered. We are also not adjusting for short term changes in risk based on the volatility of recent stock returns (I plan to add this down the road). So I would use the recommended asset allocation more as a target that points the way for new investments, with only an occasional rebalance when the allocation gets too far out of whack. In the absence of a clear understanding of costs and short horizon price artifacts, I would try to keep the target asset allocation bands broad enough that major rebalances don't have to happen more than a few times a year. At some point, I'll try to build a model of trading costs and rebalancing schedules into the planner to help with this decision, but that will be pretty far down the list.

Ben, noting a few things about Haghani and White's approach that I was hoping you might comment on:

- they use an aggressive rebalancing strategy - weekly with four sub-portfolios (partially but not fully due to an aggressive TLH approach)

- they use momentum as the risk input vs. historical volatility / SD; this seems to have the effect of much wilder frequent swings in target AA than what TPAW would recommend (swings which are being captured because of the frequent rebalancing)

- they don't seem to consider duration matching to be optimal / required for the fixed income portion of the portfolio; for example, the aforementioned wild swings include big swings in the mix of tbills vs. bond funds that wouldn't be driven by an investor's duration needs but rather by macros; the word "duration" is mentioned zero times in their book

- they are TIPS fans but not 100% of fixed income; depends on valuations and momentum

Thank you.

That's interesting. Where did they talk about rebalancing weekly and their use of momentum as the risk input?

In The Missing Billionaires page 53, they said this about estimating the risk of the market portfolio:

Unfortunately, we cannot directly observe expectations of long-term market risk. Even if there were liquid, long-term markets in options that we could look to, it would be difficult to disentangle the true expectation from the insurance premium built into those prices. It's also not possible to derive expectations of risk from expected cash flows in the way we do when determining long-term expected returns for equities or bonds. What we can do is build up an estimate for market risk using a blend of long- and short-horizon realized volatility. This is the approach that options trading desks at most banks use to forecast market volatility.

See the "Momentum" sub-section that starts on pg. 59.

In their cited 2016 paper, they discuss monthly rebalancing. But at Elmwealth.com, it is the more aggressive weekly approach.

[Ben Mathew]

DON'T REBALANCE TOO FREQUENTLY

I would strongly caution against trading frequently to get to the target asset allocation. Besides taxes and bid/ask spreads, at short horizons there may also be things like price momentum that will need to be considered. We are also not adjusting for short term changes in risk based on the volatility of recent stock returns (I plan to add this down the road). So I would use the recommended asset allocation more as a target that points the way for new investments, with only an occasional rebalance when the allocation gets too far out of whack. In the absence of a clear understanding of costs and short horizon price artifacts, I would try to keep the target asset allocation bands broad enough that major rebalances don't have to happen more than a few times a year. At some point, I'll try to build a model of trading costs and rebalancing schedules into the planner to help with this decision, but that will be pretty far down the list.

Ben, noting a few things about Haghani and White's approach that I was hoping you might comment on:

- they use an aggressive rebalancing strategy - weekly with four sub-portfolios (partially but not fully due to an aggressive TLH approach)

- they use momentum as the risk input vs. historical volatility / SD; this seems to have the effect of much wilder frequent swings in target AA than what TPAW would recommend (swings which are being captured because of the frequent rebalancing)

- they don't seem to consider duration matching to be optimal / required for the fixed income portion of the portfolio; for example, the aforementioned wild swings include big swings in the mix of tbills vs. bond funds that wouldn't be driven by an investor's duration needs but rather by macros; the word "duration" is mentioned zero times in their book

- they are TIPS fans but not 100% of fixed income; depends on valuations and momentum

Thank you.

That's interesting. Where did they talk about rebalancing weekly and their use of momentum as the risk input?

In The Missing Billionaires page 53, they said this about estimating the risk of the market portfolio:

Unfortunately, we cannot directly observe expectations of long-term market risk. Even if there were liquid, long-term markets in options that we could look to, it would be difficult to disentangle the true expectation from the insurance premium built into those prices. It's also not possible to derive expectations of risk from expected cash flows in the way we do when determining long-term expected returns for equities or bonds. What we can do is build up an estimate for market risk using a blend of long- and short-horizon realized volatility. This is the approach that options trading desks at most banks use to forecast market volatility.

See the "Momentum" sub-section that starts on pg. 59.

In their cited 2016 paper, they discuss monthly rebalancing. But at Elmwealth.com, it is the more aggressive weekly approach.

Got it. When I read the book, I had missed that they are using momentum to estimate risk. I had assumed that momentum was only used in their estimate of expected return. But you're right—they talk about using momentum to estimate risk as well, with positive momentum reducing risk and negative momentum increasing risk ("grind higher and plummet lower"). My guess is that they're using momentum to enrich the basic volatility estimate derived from historical realized volatility—not abandoning historical realized volatility altogether. While they may condition on momentum, the risk estimate still must ultimately be derived in some way from the historical volatility.

This would explain why they can rebalance at a high frequency—they are accounting for momentum in their model.

Momentum would push asset allocation in the opposite direction that the earnings yield moves it. When the price goes up, the earnings yield goes down (reducing stock allocation) but momentum goes up (increasing stock allocation). So there is some offsetting at work in short horizons.

Trying to capture momentum effects by trading/rebalancing at short horizons would require a good deal of technical sophistication, frequent monitoring, and staying on top of trading costs. Elm Wealth might be able to do it, but it's unlikely that DIY investors will have the time, expertise or resources to replicate what they're doing. I'd stick with my earlier recommendation to refrain from rebalancing too frequently and simply trying to avoid getting on the wrong side of momentum.

Regarding bond duration: I'm not sure how Elm Wealth is set up, but if they have just a bond fund and a stock fund and mix the two in different proportions for different clients based on their Merton share, it might not be feasible to duration match bonds to individual clients' spending horizons. But duration matching will have a significant impact on spending risk for floors and for risk portfolios with a high allocation to bonds. So I think DIY investors should take bond duration into account if they can.

[AlohaJoe]

[- they don't seem to consider duration matching to be optimal / required for the fixed income portion of the portfolio;

Duration matching definitely isn't optimal. See Milevsky & Huang's 2010 paper "Should Retirees Hedge Inflation or Just Worry About it?"

Super short version: your spending doesn't track CPI-U so a CPI-U hedged instrument is clearly not optimal. It's just convenient and probably good enough in practice given all the other not quite perfect assumptions we make in retirement planning.

[Circle the Wagons]

DON'T REBALANCE TOO FREQUENTLY

I would strongly caution against trading frequently to get to the target asset allocation. Besides taxes and bid/ask spreads, at short horizons there may also be things like price momentum that will need to be considered. We are also not adjusting for short term changes in risk based on the volatility of recent stock returns (I plan to add this down the road). So I would use the recommended asset allocation more as a target that points the way for new investments, with only an occasional rebalance when the allocation gets too far out of whack. In the absence of a clear understanding of costs and short horizon price artifacts, I would try to keep the target asset allocation bands broad enough that major rebalances don't have to happen more than a few times a year. At some point, I'll try to build a model of trading costs and rebalancing schedules into the planner to help with this decision, but that will be pretty far down the list.

Ben, noting a few things about Haghani and White's approach that I was hoping you might comment on:

- they use an aggressive rebalancing strategy - weekly with four sub-portfolios (partially but not fully due to an aggressive TLH approach)

- they use momentum as the risk input vs. historical volatility / SD; this seems to have the effect of much wilder frequent swings in target AA than what TPAW would recommend (swings which are being captured because of the frequent rebalancing)

- they don't seem to consider duration matching to be optimal / required for the fixed income portion of the portfolio; for example, the aforementioned wild swings include big swings in the mix of tbills vs. bond funds that wouldn't be driven by an investor's duration needs but rather by macros; the word "duration" is mentioned zero times in their book

- they are TIPS fans but not 100% of fixed income; depends on valuations and momentum

Thank you.

That's interesting. Where did they talk about rebalancing weekly and their use of momentum as the risk input?

In The Missing Billionaires page 53, they said this about estimating the risk of the market portfolio:

Unfortunately, we cannot directly observe expectations of long-term market risk. Even if there were liquid, long-term markets in options that we could look to, it would be difficult to disentangle the true expectation from the insurance premium built into those prices. It's also not possible to derive expectations of risk from expected cash flows in the way we do when determining long-term expected returns for equities or bonds. What we can do is build up an estimate for market risk using a blend of long- and short-horizon realized volatility. This is the approach that options trading desks at most banks use to forecast market volatility.

See the "Momentum" sub-section that starts on pg. 59.

In their cited 2016 paper, they discuss monthly rebalancing. But at Elmwealth.com, it is the more aggressive weekly approach.

Got it. When I read the book, I had missed that they are using momentum to estimate risk. I had assumed that momentum was only used in their estimate of expected return. But you're right—they talk about using momentum to estimate risk as well, with positive momentum reducing risk and negative momentum increasing risk ("grind higher and plummet lower"). My guess is that they're using momentum to enrich the basic volatility estimate derived from historical realized volatility—not abandoning historical realized volatility altogether. While they may condition on momentum, the risk estimate still must ultimately be derived in some way from the historical volatility.

This would explain why they can rebalance at a high frequency—they are accounting for momentum in their model.

Momentum would push asset allocation in the opposite direction that the earnings yield moves it. When the price goes up, the earnings yield goes down (reducing stock allocation) but momentum goes up (increasing stock allocation). So there is some offsetting at work in short horizons.

Trying to capture momentum effects by trading/rebalancing at short horizons would require a good deal of technical sophistication, frequent monitoring, and staying on top of trading costs. Elm Wealth might be able to do it, but it's unlikely that DIY investors will have the time, expertise or resources to replicate what they're doing. I'd stick with my earlier recommendation to refrain from rebalancing too frequently and simply trying to avoid getting on the wrong side of momentum.

Regarding bond duration: I'm not sure how Elm Wealth is set up, but if they have just a bond fund and a stock fund and mix the two in different proportions for different clients based on their Merton share, it might not be feasible to duration match bonds to individual clients' spending horizons. But duration matching will have a significant impact on spending risk for floors and for risk portfolios with a high allocation to bonds. So I think DIY investors should take bond duration into account if they can.

Thank you for your thoughts and confirmation on these points.

On their site, you can see the historical impact of their approach under different assumptions. It looks to the untrained eye like chasing the market. E.g., stock allocations stay high for a long while despite high CAPE / low ERP, until suddenly a momentum trigger hits and it plummets (or vice versa). Whether "riding the wave" offsets the risk of getting caught out over-exposed when valuations are high ... I can't say.

The other motivation for the more frequent trading (though not necessarily rebalancing of AA) is TLH, which I know is an entirely separate topic and will be specific to individual tax situations.

[Circle the Wagons]

[- they don't seem to consider duration matching to be optimal / required for the fixed income portion of the portfolio;

Duration matching definitely isn't optimal. See Milevsky & Huang's 2010 paper "Should Retirees Hedge Inflation or Just Worry About it?"

Super short version: your spending doesn't track CPI-U so a CPI-U hedged instrument is clearly not optimal. It's just convenient and probably good enough in practice given all the other not quite perfect assumptions we make in retirement planning.

Your comment may be more relevant to the fourth point:

- they are TIPS fans but not 100% of fixed income; depends on valuations and momentum

I can only access the abstract, but it seems to advocate a mix of nominals and TIPS. Treasuries despite no apparent inflation risk premium? Credit bonds despite the correlation with stocks? Can't be sure how they're coming to something superior to all TIPS.

It's not enough to say CPI-U will not perfectly track an individual investor's experienced inflation, as TIPS are really the only game in town for fixed income inflation hedging.

On duration matching -- it seems smart to do whether your fixed income is real, nominal or a mix.

[GAAP]

Hedging basically means to invest in something with the potential to reduce exposure to a particular risk.

TIPS reduce exposure to inflation risk to the extent that the individual's inflation experience matches the specific inflation measured by CPI-U. Even if the individual's expenses are in the same proportion as those in the CPI-U measure, local factors will still be in play and reduce the degree of correlation between expenses and index. There's a reason why the BLS publishes regional statistics -- they're different from the overall total. The correlation of CPI-U to expenses will never be perfect. CPI-E is likely a better match to retirement expenses - but it won't be perfect either.

TIPS, Social Security, and inflation-linked pensions also lag actual inflation. The inflation adjustment that occurs this year is for inflation that occurred last year. Unfortunately, expenses do not lag inflation. That difference compounds over time.

Duration matching hedges interest rate risk -- but not perfectly, unless cash flows exactly balance for the entire duration.

Diversification hedges against poor performance from a single asset or asset class -- but if everything drops, it doesn't protect against more rapid portfolio depletion.

So, a degree of equity exposure combined with a degree of TIPS exposure is a better method for protecting the portfolio (and related spending) against more risks than just investing in either TIPS or equities. Duration-matching the fixed-income portion of the portfolio will help to the degree that future interest rate risk is realized.

I have a more fundamental problem with the assumption that inflation hedging is sufficient for retirement -- even if it were truly available. Such hedging does not protect against standard of living growth in the general population, and therefore effectively ensures that the retiree will not maintain the same standard of living as the general population. Computers, mobile phones, and internet access are all examples of things that have moved from luxury to essential in the past 30 years. A degree of portfolio growth -- or at least supportable spending growth -- is necessary to maintain standard of living.

Asset allocation does not disappear as a concern once retired. Allocations that are considered by industry norms to be conservative may in fact actually be quite risky when considering all of the risks faced. Volatility is only one of those risks -- and behavioral factors may lead an investor to make decisions that are actually not well correlated to the overall degree of risk.

[Ben Mathew]

I have a more fundamental problem with the assumption that inflation hedging is sufficient for retirement -- even if it were truly available. Such hedging does not protect against standard of living growth in the general population, and therefore effectively ensures that the retiree will not maintain the same standard of living as the general population. Computers, mobile phones, and internet access are all examples of things that have moved from luxury to essential in the past 30 years.

Agree. I made a related point here:

I would also caution against converting a large portion of the portfolio to a TIPS ladder. While having all of one's planned consumption funded by a TIPS ladder will guarantee that level of real spending, it's also important to remember that if stocks and bonds do well overall, others will be spending more—potentially much more. I think most of us are social creatures and our desired level of spending is influenced by what others are doing. So keeping up with the market is nice. And conversely, dropping with the market is not as bad as it otherwise would be because it will be easier to cut spending when others are cutting as well—because of the same social aspect. This would also be an argument for holding the global market cap weighted stock-bond fund as the risk portfolio (which Bill Sharpe has advocated.)

[AlohaJoe]

TIPS, Social Security, and inflation-linked pensions also lag actual inflation. The inflation adjustment that occurs this year is for inflation that occurred last year. Unfortunately, expenses do not lag inflation. That difference compounds over time.

There was a good article in the Journal of Financial Planning about this way back in 2013: "Inflation, Hyperinflation, Adjustment Lags: Why TIPS Don’t Guarantee Real Rates of Return"[1]. The effects aren't especially large for the range of inflation we're likely to expect in developed countries. But for people who demand 0% risk of any decrease in standard of living (i.e. people who slavishly follow SWR) then even slight drops in standard of living breach that SWR contract.

[1]: https://www.financialplanningassociatio ... tes-return

I have a more fundamental problem with the assumption that inflation hedging is sufficient for retirement -- even if it were truly available. Such hedging does not protect against standard of living growth in the general population, and therefore effectively ensures that the retiree will not maintain the same standard of living as the general population. Computers, mobile phones, and internet access are all examples of things that have moved from luxury to essential in the past 30 years.

Complete agreement here. The entire point of economic progress is that life gets better than the rate of inflation. If all you do is keep up with inflation then you are gradually getting poorer. Over a short retirement or periods of economic stagnation the effects might not be particularly large. But I think many of us know people who rely entirely on Social Security (which uses a differently, slightly lower inflation adjustment than TIPS does) and after 30+ years of that are not exactly what we'd consider middle class anymore.

Smart phones didn't even really exist until 2008 when the iPhone 3G was released. There are plenty of actual retirees on this board whose "retirement budget" didn't include anything about smartphones because they retired before that. Yet now they're probably spending $200+ a month when you amortize the cost of the phone hardware, the data & voice monthly spending, subscription to picture backup services, etc, etc.

And that's where I think framing it as "social" or "keeping up with the Joneses" is wrong because this isn't (purely) positional conspicuous consumption from marketing brainwashing. Economic progress is largely driven by getting people to spend money by addressing real needs. All the Bogleheads who claim they would never want or need to increase spending in retirement to "keep up with the Joneses" ... while reading Bogleheads on their smart phone and drinking much better coffee than any American drank at home 30 years ago and using an air fryer to make actually crispy french fries at home....

A long time ago I wrote a pair of articles exploring this in more detail. In the first[2] I compared data from the BLS on how American spending changed over the first half of the 20th century.

In 1901, the average annual expenditures for an American family were $769. [...] Now let’s fast forward to 1950. If we look at an inflation calculator then it will tell us that $769 in 1901 is $2,061 at the start of 1950. Except in 1950 the average expenditure was actually $3,808.

If you only kept up with inflation then you went from middle-class to poor.

[2]: https://justusjp.medium.com/inflation-m ... 78bd0b7283

In a later post[3] I use the "US Consumer Bundle" (all the money spending by consumers in the US) as an alternative to CPI-U in typical SWR calculations, just to demonstrate the size of the effect we're talking about. The SWR drops to 2.4% and the worst years aren't from the the Great Depression or 1960s/70s Stagflation -- they're from times like 1940s when inflation was relatively low (2.2%) but standards of living were increasing dramatically (+9% a year). If you only kept up with inflation then you'd be living a Great Depression lifestyle while everyone around is getting a TV, a car, and buying a home (home ownership was only about 20% before this period).

[3]: https://justusjp.medium.com/safe-withdr ... 1c19f9a8fd

To be clear: I think for a typical 65ish retiree these effects are pretty minimal. You will almost certainly die in ~20 years anyway. Your spending will slow down as you get sicker & more frail, so you can (for example) redirect your travel budget to buying smart phones instead. Your equities will almost certainly not experience the worst returns in history. You will get old and discover that reducing your spending by 2% a year or whatever isn't actually the end of the world -- sure, the grandkids will complain you don't have Netflix and Hulu but you just watch Andy Griffith reruns on your VHS anyway.....

But for early retirees the effects are more pronounced and need to be taken into consideration.

[itsmeagain]

Thanks @GAAP and @AlohaJoe for these clear and compelling explanations of why 'keeping up with inflation' may not be good enough. As a fairly conservative investor, I need to keep the issue you raise in mind.

[ScubaHogg]

Thanks @GAAP and @AlohaJoe for these clear and compelling explanations of why 'keeping up with inflation' may not be good enough. As a fairly conservative investor, I need to keep the issue you raise in mind.

Though if you go down that path remember while some expenses creep up (smartphones) the real price of other products go down (laptops are significantly cheaper than they used to be).

Personally with an inflation indexed LMP of tips/social security, etc. and some equities I wouldn’t stress it too much

[TipsQuestions]

I have a more fundamental problem with the assumption that inflation hedging is sufficient for retirement -- even if it were truly available. Such hedging does not protect against standard of living growth in the general population, and therefore effectively ensures that the retiree will not maintain the same standard of living as the general population. Computers, mobile phones, and internet access are all examples of things that have moved from luxury to essential in the past 30 years.

I don't agree here. The CPI takes into account these changes. Anything that moves into the mainstream from the realm of the rich (for example, expensive health care) will get a growing share of the CPI. America's standard of living has increased dramatically since 1940 - and it's all been captured in the CPI.

[ScubaHogg]

I have a more fundamental problem with the assumption that inflation hedging is sufficient for retirement -- even if it were truly available. Such hedging does not protect against standard of living growth in the general population, and therefore effectively ensures that the retiree will not maintain the same standard of living as the general population. Computers, mobile phones, and internet access are all examples of things that have moved from luxury to essential in the past 30 years.

I don't agree here. The CPI takes into account these changes. Anything that moves into the mainstream from the realm of the rich (for example, expensive health care) will get a growing share of the CPI. America's standard of living has increased dramatically since 1940 - and it's all been captured in the CPI.

Well, no. GAAP is correct. CPI measures inflation. But we have so many nice things compared to our grandparents because the economy has grown faster than inflation.

So CPI “will” capture how your current standard of living increases in price

It won’t, and isn’t trying to, capture how standard of livings improve

This is all off topic of TPAW though

[TipsQuestions]

So CPI “will” capture how your current standard of living increases in price
It won’t, and isn’t trying to, capture how standard of livings improve

No, the basket of goods and services that compose the CPI are updated on a regular basis, incorporating anything new that improves your standard of living, as well as quality improvements to "legacy" goods. If consumers began purchasing nothing but widgets, the CPI would (eventually) be 100% widgets. It does not capture changes in static standards of living - I could get the same standard of living as a person in 1970 for around half the 2023 CPI, I'd guess.

[Ben Mathew]

So CPI “will” capture how your current standard of living increases in price
It won’t, and isn’t trying to, capture how standard of livings improve

No, the basket of goods and services that compose the CPI are updated on a regular basis, incorporating anything new that improves your standard of living, as well as quality improvements to "legacy" goods. If consumers began purchasing nothing but widgets, the CPI would (eventually) be 100% widgets. It does not capture changes in static standards of living - I could get the same standard of living as a person in 1970 for around half the 2023 CPI, I'd guess.

The idea behind the CPI is to measure how the price of a fixed basket of goods and services (with a composition representative of consumer spending patterns) changes over time. There are a lot of complications in measuring this due to changes in relative prices, quality increases, and the introduction of new goods and services. Updating the basket changes the composition of the basket to reflect new spending patterns, but the new basket is not meant to be a better basket. When the quality of a good increases, they try to adjust for that by reducing the price increase—i.e. they are trying to keep the basket the same and not allowing it to drift up due to quality improvements. People have argued that they haven't fully controlled for quality increases and so have overstated inflation. But the goal at least is to measure a basket of constant quality.

To illustrate that the CPI is designed to capture cost of living changes and not average spending levels, take the simple example where neither prices nor spending composition has changed, but people have gotten wealthier and consume twice as much of all goods and services as they did before. CPI would remain the same because the reference basket and prices have not changed. But spending doubled. So the CPI is capturing the cost of living, not average spending. Keeping up with the CPI would not be enough to keep up with average spending.

[AlohaJoe]

America's standard of living has increased dramatically since 1940 - and it's all been captured in the CPI.

In 2000 the BLS said the average annual expenditure of all consumer units was $38,045 [1]. In 2022 it was $72,967 [2]. The BLS inflation calculator[3] says that $38,045 in 2000 is the same as $63,366 in 2022. But consumer spending was actually 15% higher than simply adjusting for inflation.

CPI does not capture all increases in consumer spending.

[1]: https://www.bls.gov/cex/tables/calendar ... 0-2005.pdf
[2]: https://www.bls.gov/cex/tables/calendar ... 1-2022.pdf
[3]: https://www.bls.gov/data/inflation_calculator.htm

[TipsQuestions]

In 2000 the BLS said the average annual expenditure of all consumer units was $38,045 [1]. In 2022 it was $72,967 [2]. The BLS inflation calculator[3] says that $38,045 in 2000 is the same as $63,366 in 2022. But consumer spending was actually 15% higher than simply adjusting for inflation.

CPI does not capture all increases in consumer spending.

At a glance I'd say you can't compare "consumer unit" income and CPI. If households split in half tomorrow, "consumer unit" income would halve, but CPI would be unchanged.

[GAAP]

I have a more fundamental problem with the assumption that inflation hedging is sufficient for retirement -- even if it were truly available. Such hedging does not protect against standard of living growth in the general population, and therefore effectively ensures that the retiree will not maintain the same standard of living as the general population. Computers, mobile phones, and internet access are all examples of things that have moved from luxury to essential in the past 30 years.

I don't agree here. The CPI takes into account these changes. Anything that moves into the mainstream from the realm of the rich (for example, expensive health care) will get a growing share of the CPI. America's standard of living has increased dramatically since 1940 - and it's all been captured in the CPI.

Standard of living is a measure of available economic output per person. It is generally computed for a country by dividing the total economic output of that country (GDP) by the number of people in the population -- yielding per-capita GDP. That number must be adjusted for inflation to make any reasonable comparisons between economies, or over time periods. Real per-capita GDP is thus the useful measure of standard of living for a country.

Growth in standard of living means that the real per-capita GDP ratio had to increase over time. Since that number is already adjusted for inflation, the growth has to be in excess of inflation. IOW, each person is getting more economic benefit after inflation than they did previously.

[GAAP]

I have a more fundamental problem with the assumption that inflation hedging is sufficient for retirement -- even if it were truly available. Such hedging does not protect against standard of living growth in the general population, and therefore effectively ensures that the retiree will not maintain the same standard of living as the general population. Computers, mobile phones, and internet access are all examples of things that have moved from luxury to essential in the past 30 years.

I don't agree here. The CPI takes into account these changes. Anything that moves into the mainstream from the realm of the rich (for example, expensive health care) will get a growing share of the CPI. America's standard of living has increased dramatically since 1940 - and it's all been captured in the CPI.

Well, no. GAAP is correct. CPI measures inflation. But we have so many nice things compared to our grandparents because the economy has grown faster than inflation.

So CPI “will” capture how your current standard of living increases in price

It won’t, and isn’t trying to, capture how standard of livings improve

This is all off topic of TPAW though

I don't actually use TPAW, but this seems to me like one thing that TPAW users should consider when determining if they want a spending tilt, and how great that tilt should be. Truly matching long-term (since 1929) per-capita GDP growth would require a tilt (to the future) of about 2%/year (assuming long term trends hold).

Per-capita GDP growth has never been consistent with annual changes ranging from 13% drops to 17% increases. Ten-year periods ranged from slight drops (about 0.5%) up to 8%. 20 year periods ranged from less than 1% up to 5%. The 40-year range has been about 1.5-3.5%. All of those shorter periods averaged about 2%.

Ben can probably comment about how to use TPAW to achieve the "retirement smile" combined with a long-term 2% standard of living tilt.

[ahc19081]

Ben can probably comment about how to use TPAW to achieve the "retirement smile" combined with a long-term 2% standard of living tilt.

We're gonna need a bigger g.

I dunno. Over a 35-year retirement 2%/annum would be almost precisely double the spending in real terms. I take your point about rising standards of living, but I have complete spending data for my household going back 20+ years. When I compare the basket of goods & services we purchased in the early 2000s (ex taxes and charitable giving) with my best estimate of our expenses going forward, there's only a 0.7%/annum real increase. And that's over a time period where for various reasons our lifestyle got quite a bit more luxurious (e.g., we drove a Passat then, we drive a Tesla now). Yes, we are spending much more in some categories than we did then, but in others we're spending considerably less. I'm not sure I see us expanding our real expenses at 2% year after year in our older years no matter what happens to GDP.

[GAAP]

Ben can probably comment about how to use TPAW to achieve the "retirement smile" combined with a long-term 2% standard of living tilt.

We're gonna need a bigger g.

I dunno. Over a 35-year retirement 2%/annum would be almost precisely double the spending in real terms. I take your point about rising standards of living, but I have complete spending data for my household going back 20+ years. When I compare the basket of goods & services we purchased in the early 2000s (ex taxes and charitable giving) with my best estimate of our expenses going forward, there's only a 0.7%/annum real increase. And that's over a time period where for various reasons our lifestyle got quite a bit more luxurious (e.g., we drove a Passat then, we drive a Tesla now). Yes, we are spending much more in some categories than we did then, but in others we're spending considerably less. I'm not sure I see us expanding our real expenses at 2% year after year in our older years no matter what happens to GDP.

A St. Louis Fed article on the topic: https://fredblog.stlouisfed.org/2024/04 ... enditures/

Mucking around with the data in that article yields a quarterly increase of 0.38%/quarter from the start of 2000 -- about 1.5% annually. It looks like you've managed to avoid the hedonistic treadmill to a degree.

I prefer to plan/attempt to achieve real portfolio growth sufficient to handle a 1% increase. That's much easier to say than to actually achieve when withdrawing from the portfolio. However, any year that I don't actually need the 1% is a year that increases the sustainability of the portfolio -- and thus the sustainability of the spending.

[ScubaHogg]

So CPI “will” capture how your current standard of living increases in price
It won’t, and isn’t trying to, capture how standard of livings improve

No, the basket of goods and services that compose the CPI are updated on a regular basis, incorporating anything new that improves your standard of living, as well as quality improvements to "legacy" goods. If consumers began purchasing nothing but widgets, the CPI would (eventually) be 100% widgets. It does not capture changes in static standards of living - I could get the same standard of living as a person in 1970 for around half the 2023 CPI, I'd guess.

CPI measures inflation

Inflation != real economic growth

I don’t know how else to say it.

If it was you wouldn’t even need CPI. Nominal GDP/per capita, much easier to measure, would be all you needed

[ScubaHogg]

I don't actually use TPAW, but this seems to me like one thing that TPAW users should consider when determining if they want a spending tilt, and how great that tilt should be. Truly matching long-term (since 1929) per-capita GDP growth would require a tilt (to the future) of about 2%/year (assuming long term trends hold).

Per-capita GDP growth has never been consistent with annual changes ranging from 13% drops to 17% increases. Ten-year periods ranged from slight drops (about 0.5%) up to 8%. 20 year periods ranged from less than 1% up to 5%. The 40-year range has been about 1.5-3.5%. All of those shorter periods averaged about 2%.

Ben can probably comment about how to use TPAW to achieve the "retirement smile" combined with a long-term 2% standard of living tilt.

I just don’t see it being a big deal. The 90 year olds in my life aren’t stressing new cars, iPhones, vuori pants, or many of those items that make up increased SOL. Indeed their travel budgets have shrunk to near zero, houses are smaller, little to no driving, etc. All that frees up cash to offset your concerns

LTC is a different story

[TipsQuestions]

The idea behind the CPI is to measure how the price of a fixed basket of goods and services (with a composition representative of consumer spending patterns) changes over time.

The BLS defines the CPI in its white papers as: "a measure of the average change over time in the prices paid by urban consumers for a market basket of consumer goods and services." Market, not fixed. Again, if consumers begin buying nothing but widgets, the CPI basket will eventually consist of nothing but widgets.

To illustrate that the CPI is designed to capture cost of living changes and not average spending levels, take the simple example where neither prices nor spending composition has changed, but people have gotten wealthier and consume twice as much of all goods and services as they did before. CPI would remain the same because the reference basket and prices have not changed. But spending doubled. So the CPI is capturing the cost of living, not average spending. Keeping up with the CPI would not be enough to keep up with average spending.

That's true, if some alien force doubled everyone's income and consumption, and producers could supply goods and services at the same marginal cost, the CPI wouldn't change. But I was responding to this:

I have a more fundamental problem with the assumption that inflation hedging is sufficient for retirement -- even if it were truly available. Such hedging does not protect against standard of living growth in the general population, and therefore effectively ensures that the retiree will not maintain the same standard of living as the general population. Computers, mobile phones, and internet access are all examples of things that have moved from luxury to essential in the past 30 years.

My TIPS portfolio will give me access to standard of living improvements from the items mentioned above (especially with that 2% real return). I don't expect any steady decline in my standard of living vs the general population, however you might define that.

[Ben Mathew]

The idea behind the CPI is to measure how the price of a fixed basket of goods and services (with a composition representative of consumer spending patterns) changes over time.

The BLS defines the CPI in its white papers as: "a measure of the average change over time in the prices paid by urban consumers for a market basket of consumer goods and services." Market, not fixed.

That is referring to the relative weights of goods and services in the basket: how much hamburgers vs hot dogs are in there. The weights are chosen to reflect what people are consuming. So if spending patterns change, the weights will change. If everyone consumed twice as many hamburgers and twice as many hot dogs as before, the basket would not change. If they consumed 2x as many hamburgers and 1.5x as many hot dogs as before, hamburgers will become weighted more than before and hot dogs less. But the basket is not supposed to get better. The goal is to set it up so that people would be indifferent between the new and old baskets. In practice, that's hard to do, so baskets end up getting better. But that's a shortcoming of the implementation that the Bureau of Labor Statistics (BLS) tries to correct—it's not the intended goal of CPI.

From Essentials of Economics by Greg Mankiw, 5th edition:

The Bureau of Labor Statistics does its best to account for quality change. When the quality of a good in the basket changes—for example, when a car model has more horsepower or gets better gas mileage from one year to the next—the Bureau adjusts the price of the good to account for the quality change. It is, in essence, trying to compute the price of a basket of goods of constant quality. Despite these efforts, changes in quality remain a problem because quality is so hard to measure.

There is still much debate among economists about how severe these measurement problems are and what should be done about them. Several studies written during the 1990s concluded that the consumer price index overstated inflation by about 1 percentage point per year. In response to this criticism, the Bureau of Labor Statistics adopted several technical changes to improve the CPI, and many economists believe the bias is now only about half as large as it once was. [Pages 351-352]

Again, if consumers begin buying nothing but widgets, the CPI basket will eventually consist of nothing but widgets.

Yes, but this refers to the composition of the basket (100% widgets), not how many widgets there are in the basket. The number of widgets in the basket would be adjusted so that you are indifferent between the new 100% widget basket and the original basket. So if the new widgets are better than whatever was in the basket before, then there would be fewer widgets in the basket to adjust for that.

To illustrate that the CPI is designed to capture cost of living changes and not average spending levels, take the simple example where neither prices nor spending composition has changed, but people have gotten wealthier and consume twice as much of all goods and services as they did before. CPI would remain the same because the reference basket and prices have not changed. But spending doubled. So the CPI is capturing the cost of living, not average spending. Keeping up with the CPI would not be enough to keep up with average spending.

That's true, if some alien force doubled everyone's income and consumption, and producers could supply goods and services at the same marginal cost, the CPI wouldn't change.

This doesn't require a hypothetical alien force. This is just economic growth making societies richer over time, allowing people to consume more goods ands services than before.

My TIPS portfolio will give me access to standard of living improvements from the items mentioned above (especially with that 2% real return). I don't expect any steady decline in my standard of living vs the general population, however you might define that.

The 2% real return of TIPS won't translate to increasing payouts over time unless you structure the payouts to increase over time. The typical flat bond ladder would have fewer TIPS funding later years to offset the 2% yield and get a constant payout. To make payouts a constant $X per year, the payout t years from now would get $X/(1+.02)^t worth of TIPS today. So the 2% return reduces the cost of the TIPS needed for the target payout, but doesn't increase the target payout.

[GAAP]

My TIPS portfolio will give me access to standard of living improvements from the items mentioned above (especially with that 2% real return). I don't expect any steady decline in my standard of living vs the general population, however you might define that.

The 2% real return of TIPS won't translate to increasing payouts over time unless you structure the payouts to increase over time. The typical flat bond ladder would have fewer TIPS funding later years to offset the 2% yield and get a constant payout. To make payouts a constant $X per year, the payout t years from now would get $X/(1+.02)^t worth of TIPS today. So the 2% return reduces the cost of the TIPS needed for the target payout, but doesn't increase the target payout.

I think @TipsQuestions may be talking about TIPS in a more general portfolio like the risk portfolio. In that case, the ability of the portfolio to support standard of living increases is a function of the overall net real performance after withdrawals compared to the rate of improvement in the general economy.

TIPS instead of nominal bonds in a classic fixed AA offer the potential for returns with a lower downside deviation. TIPS purchased about a year ago also offered real returns greater than the rates achieved historically by nominal bonds. Whether or not that level of performance occurs again remains to be seen.

In any case, we're getting way off-topic here. The typical TPAW user should probably focus more on what they hope to achieve in order to align their plans with those objectives. Having done that, the user should periodically revisit the plan, revising as necessary to adjust for actual performance. Growth or no-growth in standard of living is only one of the potential factors to consider.

[TipsQuestions]

But the basket is not supposed to get better. The goal is to set it up so that people would be indifferent between the new and old baskets. In practice, that's hard to do, so baskets end up getting better. But that's a shortcoming of the implementation that the Bureau of Labor Statistics (BLS) tries to correct—it's not the intended goal of CPI.

Ok, thanks for you responses. I couldn't find anything on the BLS website about attempts to make consumers indifferent to baskets over time. But that's not possible, it doesn't matter, I'll take your word for it, etc. My point of contention was holding inflation linked investments "effectively ensures that the retiree will not maintain the same standard of living as the general population". As you point out, standards of living and portfolio withdrawals are separate concepts from CPI changes. I can adjust my withdrawals from my investments, TIPS or otherwise, to enjoy standard of living improvements as needed.

Also, thinking about this more, "maintaining the same standard of living as the general population" in retirement isn't a rational goal. Naturally, retirees are going to miss out on real wage growth workers may enjoy, and a person retiring at 50 is going to have a lower standard of living in retirement than one who retires at 70, all else equal. But so what?