Amortization Based Withdrawal (ABW)

Non-investing personal finance issues including insurance, credit, real estate, taxes, employment and legal issues such as trusts and wills
User avatar
siamond
Posts: 5636
Joined: Mon May 28, 2012 5:50 am

Re: Amortization Based Withdrawal (ABW)

Post by siamond »

Ben Mathew wrote: Thu Nov 19, 2020 6:27 pmGreat. Works for me. I have changed it to "Rate of return to amortize extra income and expenses."
Thank you... Appreciated. :beer

Will review the updated wiki page later tonight.
User avatar
Kevin M
Posts: 11706
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: Amortization Based Withdrawal (ABW)

Post by Kevin M »

I've seen some debate about working in real or nominal terms. Has there been any discussion of preferring to work in real terms, but having to incorporate future cash flows that are determined in nominal terms, such as a non-inflation adjusted pension or annuity?

For example, I may prefer to work in real terms for my investment portfolio (for some reason referred to in this thread as a savings portfolio), so I would use a real expected return to determine real periodic payments using PMT. But if I have a non-inflation-adjusted annuity of some sort, I only know the nominal cash flows, so normally I would calculate a present value based on the nominal cash flows and a nominal discount rate.

Perhaps adjusting the discount rate to account for inflation might make sense, but this doesn't feel right to me, since the real PMT amount is not fixed, but is expected to decline (for positive discount rates).

I may not have looked closely enough, but handling this doesn't seem to be incorporated into the "advanced" calculator spreadsheet, nor does it seem to be addressed in the wiki article.

Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)
Wrench
Posts: 169
Joined: Sun Apr 28, 2019 10:21 am

Re: Amortization Based Withdrawal (ABW)

Post by Wrench »

Kevin M wrote: Thu Nov 19, 2020 6:51 pm I've seen some debate about working in real or nominal terms. Has there been any discussion of preferring to work in real terms, but having to incorporate future cash flows that are determined in nominal terms, such as a non-inflation adjusted pension or annuity?

For example, I may prefer to work in real terms for my investment portfolio (for some reason referred to in this thread as a savings portfolio), so I would use a real expected return to determine real periodic payments using PMT. But if I have a non-inflation-adjusted annuity of some sort, I only know the nominal cash flows, so normally I would calculate a present value based on the nominal cash flows and a nominal discount rate.

Perhaps adjusting the discount rate to account for inflation might make sense, but this doesn't feel right to me, since the real PMT amount is not fixed, but is expected to decline (for positive discount rates).

I may not have looked closely enough, but handling this doesn't seem to be incorporated into the "advanced" calculator spreadsheet, nor does it seem to be addressed in the wiki article.

Kevin
It occurs to me that "real versus nominal" is very dependent on your stage in life. If you are in accumulation mode, say at age 35-40, then real terms make the most sense because you will not be withdrawing money for 20-30 years and there is just no way to project inflation rates (or law changes, tax rates, retirement distribution rules, etc.) over that time frame. But if you are within a year or two of retirement, or in retirement, nominal makes more sense to me for two reasons: taxes and RMDs. Both are based on nominal dollars and are not inflation adjusted. Obviously they are inter-related. If I have to take a big RMD and jump to a higher tax bracket at some point in the future, I want to know about it and be able to plan for it, or maybe even prevent it. I don't see how to do that if all the projections are in real dollars. Or, am I missing something?

Wrench
User avatar
siamond
Posts: 5636
Joined: Mon May 28, 2012 5:50 am

Re: Amortization Based Withdrawal (ABW)

Post by siamond »

Kevin M wrote: Thu Nov 19, 2020 6:51 pm I've seen some debate about working in real or nominal terms. Has there been any discussion of preferring to work in real terms, but having to incorporate future cash flows that are determined in nominal terms, such as a non-inflation adjusted pension or annuity?

For example, I may prefer to work in real terms for my investment portfolio (for some reason referred to in this thread as a savings portfolio), so I would use a real expected return to determine real periodic payments using PMT. But if I have a non-inflation-adjusted annuity of some sort, I only know the nominal cash flows, so normally I would calculate a present value based on the nominal cash flows and a nominal discount rate.

Perhaps adjusting the discount rate to account for inflation might make sense, but this doesn't feel right to me, since the real PMT amount is not fixed, but is expected to decline (for positive discount rates).

I may not have looked closely enough, but handling this doesn't seem to be incorporated into the "advanced" calculator spreadsheet, nor does it seem to be addressed in the wiki article.

Kevin
This topic was discussed multiple times when this thread got started. We basically ended with a simple recommendation of expressing everything in real terms. So if you have a pension or annuity which is not inflation-adjusted, then the corresponding cash flows should be adjusted for [expected] inflation over the years before running the NPV/PMT amortization math. If you have a pension which is partly inflation-adjusted (most employees of the state in MA, including my wife, are in this situation), then view it as two pensions, one inflation-adjusted, the other not inflation-adjusted. The blog article I wrote and the corresponding calculator took all of this in account.

You are absolutely correct that this topic didn't make it yet in the wiki or in Ben's simple calculator.
User avatar
Ben Mathew
Posts: 1330
Joined: Tue Mar 13, 2018 11:41 am
Location: Seattle
Contact:

Re: Amortization Based Withdrawal (ABW)

Post by Ben Mathew »

Wrench wrote: Thu Nov 19, 2020 7:41 pm But if you are within a year or two of retirement, or in retirement, nominal makes more sense to me for two reasons: taxes and RMDs. Both are based on nominal dollars and are not inflation adjusted. Obviously they are inter-related. If I have to take a big RMD and jump to a higher tax bracket at some point in the future, I want to know about it and be able to plan for it, or maybe even prevent it. I don't see how to do that if all the projections are in real dollars. Or, am I missing something?
I find it easier to model future taxes in real dollars because tax brackets are inflation-adjusted. So, assuming no change in tax laws, you can just use current year's tax brackets if doing the analysis in real dollars. With nominal, you would have to increase future year tax brackets with the inflation rate.

RMDs are a percentage of the account, so it is equally easy in nominal and real.
User avatar
Ben Mathew
Posts: 1330
Joined: Tue Mar 13, 2018 11:41 am
Location: Seattle
Contact:

Re: Amortization Based Withdrawal (ABW)

Post by Ben Mathew »

Kevin M wrote: Thu Nov 19, 2020 6:51 pm I may not have looked closely enough, but handling this doesn't seem to be incorporated into the "advanced" calculator spreadsheet, nor does it seem to be addressed in the wiki article.
I can add this to a version of the calculator that takes in nominal income and expenses, asks for the inflation rate, and converts it to real dollars.

Wiki section on nominal vs real needs some more fleshing out, explaining difference between nominal dollars and real dollars, how to convert nominal dollars to real dollars, and how to calculate real interest rate. Unless there's a convenient wiki for this that can be simply linked to.
User avatar
siamond
Posts: 5636
Joined: Mon May 28, 2012 5:50 am

Re: Amortization Based Withdrawal (ABW)

Post by siamond »

siamond wrote: Thu Nov 19, 2020 6:36 pmWill review the updated wiki page later tonight.
I did a first pass and made various minor edits, I believe primarily editorial.

Nice job... I think it hangs together pretty well until the (dreaded!) extra income and expenses topic we discussed so much, but there are still a few topics to rework, I think:
a) some terminology isn't consistent with typical spreadsheet functions (e.g. "Terminal Balance" instead of "Final Value")
b) the Final Value (aka Terminal Balance) discussion needs to elaborate on the likely need to adjust such value every year (e.g. due to realized inflation), which makes it tricky to use
c) we may need to discuss a sound approach to evaluate/re-evaluate the number of periods at some point. Life expectancy is NOT a good approach as a case in point. Hardcoding such number to 100 isn't the way to go either.
d) I'd move the discussion about expected returns before the discussion of extra income/expenses as it applies to the most basic case; this would seem a better built-up.
e) I continue to believe that the "Do It Yourself" nature of the intended audience isn't properly acknowledged. Personally, I would illustrate every section with corresponding (simple) spreadsheet formulas. This is NOT a pre-canned closed system, this really should be an open-source system (in philosophy!).

Now as to the dreaded section... I understand what Ben tried to do, but this just doesn't seem to work. First, the current text seems really hard to understand unless one knows exactly what is being described. We really need to go much more step in step, I suspect, starting by showing examples of future cash flows over time (including Kevin's point about non-inflation-adjusted cash flows). Next, the common core of principles we established quite a few posts ago (my points 1/2/3) isn't acknowledged. And consequently there is a myriad of scenarios between the two extremes (the bonds-centric approach and the portfolio-centric approach) which are not addressed. This being said, I realize I've been a real pain in the neck about this topic, so I guess maybe the onus is on me to propose some text. Won't be able to get to it until early next week though.
User avatar
Kevin M
Posts: 11706
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: Amortization Based Withdrawal (ABW)

Post by Kevin M »

siamond wrote: Thu Nov 19, 2020 8:25 pm Next, the common core of principles we established quite a few posts ago (my points 1/2/3) isn't acknowledged. And consequently there is a myriad of scenarios between the two extremes (the bonds-centric approach and the portfolio-centric approach) which are not addressed. This being said, I realize I've been a real pain in the neck about this topic, so I guess maybe the onus is on me to propose some text. Won't be able to get to it until early next week though.
I don't buy into your point 2 (or whatever the one is about using the investment portfolio expected rate of return as the discount rate for relatively safe future cash flows). From my brief read, it appears that you basically bulldozed over Ben's points, which I thought were quite sound (again, nothing to do with portfolio management, but about using appropriate discount rates). Here is my understanding of Ben's recommended approach, which seems thoroughly consistent with finance theory:
  1. Use an appropriate discount rate to calculate the present value of relatively safe cash flows. I would not call this bond-centric; it's just basic finance theory. I would use TIPS yields for inflation-indexed cash flows, and nominal Treasury yields for nominal cash flows. Yes, these are bonds, but calling it bond-centric seems somewhat pejorative to me.
  2. Include the PV of the safe cash flows as a component of the total portfolio (investment portfolio plus PV of future cash flows).
  3. Calculate the expected return of the total portfolio, using the safe discount rate as the expected return for the PV of the safe future cash flows.
Ben laid out quite clearly the possible issues with doing it the way that you are pushing, and I don't recall ever seeing a reasonable response to that.

I think what Ben has done in the wiki is a reasonable common ground, which lays out three approaches, one being the one you're pushing, and one being the one Ben has laid out.

I actually think the third one, keeping the cash flows separate, is the least contentious, and the one I probably would use. I would add some more verbiage to this section, pointing out that having additional safe cash flows affects the factors that go into selecting an AA, such as ability and need to take risk. I wouldn't have thought about it as similar to the sub-portfolios approach, but upon reflection, I see that that's another reasonable way to look at it; i.e., having additional safe cash flows reduces the need for an LMP, and enables the ability to have a larger risk portfolio (although it also reduces the need for a larger risk portfolio).

Fun stuff. I've been busy with personal stuff lately, so haven't been able to keep up with fun threads like this.

Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)
User avatar
Ben Mathew
Posts: 1330
Joined: Tue Mar 13, 2018 11:41 am
Location: Seattle
Contact:

Re: Amortization Based Withdrawal (ABW)

Post by Ben Mathew »

siamond wrote: Thu Nov 19, 2020 8:25 pm d) I'd move the discussion about expected returns before the discussion of extra income/expenses as it applies to the most basic case; this would seem a better built-up.
Good idea. I moved it. I think it flows better.
siamond wrote: Thu Nov 19, 2020 8:25 pm e) I continue to believe that the "Do It Yourself" nature of the intended audience isn't properly acknowledged. Personally, I would illustrate every section with corresponding (simple) spreadsheet formulas. This is NOT a pre-canned closed system, this really should be an open-source system (in philosophy!).
I think it's better not to make this article too Excel centric. Conceptual understanding--what's being done, why, and how it affects things--is more important. People with basic knowledge of Excel can do most or all of things being described. The Comparing Investments wiki has more on NPV formulas and so on. Currently, it has been linked to once in the text and once at the end under "See Also". I think it's better to link to that as needed than re-explain everything in this wiki. Also, the sample spreadsheets can serve as a template for people wanting further customization, and people can ask on the thread if they have questions.

As I noted earlier, I don't agree that ABW is more for advanced or DIY users. I believe that the base wiki article should be accessible to all users, including those not using Excel.
siamond wrote: Thu Nov 19, 2020 8:25 pm Now as to the dreaded section... I understand what Ben tried to do, but this just doesn't seem to work. First, the current text seems really hard to understand unless one knows exactly what is being described. We really need to go much more step in step, I suspect, starting by showing examples of future cash flows over time (including Kevin's point about non-inflation-adjusted cash flows). Next, the common core of principles we established quite a few posts ago (my points 1/2/3) isn't acknowledged. And consequently there is a myriad of scenarios between the two extremes (the bonds-centric approach and the portfolio-centric approach) which are not addressed. This being said, I realize I've been a real pain in the neck about this topic, so I guess maybe the onus is on me to propose some text. Won't be able to get to it until early next week though.
My goal in this section was not to explain how our two strategies works. There is not nearly enough consensus to do so satisfactorily. The goal is simply to make the reader aware that there are different routes that people are taking and try to get them to click on those links and explore those routes more. That's it. Going into the somewhat superficial similarities of the NPV math and so on is not really useful IMO. The real story is that our two approaches take dramatically different approaches to risk management. After pages and pages of discussion, I still don't see how you calculate or account for risk in your approach. If we want to say more, it should be about that material difference rather than the superficial similarities in the math. Dwelling on the commonalities is burying the lede. I don't want to give people the impression that it's okay to replace safe cash flows with risky investments without worrying about the risk implications of doing that. It's easy to use one discount rate instead of the other, but hard to fully appreciate what that means. I tried to give examples and illustrations about the risk implications in this thread. But it's a tricky concept, and I don't want the wiki to be a source of false confidence that it's fine to do it one way or the other without getting people to understand it well. I think the separate articles being linked to is the right place to go into more detail. Both camps will have more room to explain how they account for and manage risk in their respective approaches. Nothing is served by more how-to and other details in the ABW page. It will just paper over the differences.
User avatar
WoodSpinner
Posts: 1771
Joined: Mon Feb 27, 2017 1:15 pm

Re: Amortization Based Withdrawal (ABW)

Post by WoodSpinner »

siamond wrote: Thu Nov 19, 2020 6:33 pm
GAAP wrote: Thu Nov 19, 2020 5:09 pm
siamond wrote: Wed Nov 18, 2020 8:32 pm Ben's way of setting bonds aside to create a pre-SS income flow is directly related to the way he manages his AA, we've been through it in past posts. This is the fundamental reason for which he's correct to use bonds expected return in the NPV/PMT math FOR HIS CASE, because he's going to sell bonds to generate short-term income while waiting for future income (e.g. SS). In my case (as for multiple posters on this thread), I keep a fixed AA for my (regular) portfolio, I do NOT adjust the portfolio's AA due to future income expectations, and the proper rate of return to use is the portfolio's expected rate of return, because those are the shares I'll sell in pre-SS years (hence losing the corresponding future return, i.e. Kitces opportunity cost). The key conclusion is ALWAYS the same, the rate of return for amortization is the [expected] rate of return of the shares to be used to enable the short-term extra spending. Differences come from ways to structure one's AA.
It occurs to me that a more general statement would be: match the discount rate to the method used to fund that short-term income. I think that actually covers both Siamond's and Ben's approach to funding the income, while also leaving open alternatives like shorter term funds, period certain annuities, etc. Then examples could be given for some of the alternatives.
Well, I wonder... If somebody has other PRESENT sources of income (e.g. annuity, ongoing pension or even simply part-time work), then those cash flows should be modeled as inputs to the NPV/PMT math as well, which should include future AND present cash flows, in truth. As to short-term funds, they are securities on their own right, so they match my definition.

Still, I see your point. The amortization process (and resulting extra spending) has to be funded by a given investment vehicle (while waiting for future cash flows to materialize). I was only thinking regular portfolio and corresponding holdings (stocks or bonds, CDs, money market or else), but there might be other examples. Then the 'rate' to be used for the amortization NPV/PMT process should be the expected rate of return for the investment vehicle which is going to be withdrawn from in the coming years. I just struggle to find examples going beyond regular portfolio holdings? And I fear we'll further confuse the matter if we try to be too abstract?
All,

It seems to me we should only Amortize Future Income that has not begun.

If you are currently receiving the income (e.g. currently receiving SS or a Pension) then it should simply be added (e.g. Current Income + Amortized Future Income + Amortized Portfolio Withdrawal) to come up with a projected spend amount.

For example:
- 60 year old, 30 year lifespan
- Currently receiving $12,000/ year pension
- Plans to collect SS at 70, $36,000/year
- Portfolio worth $1,000,000, AA 50/50, fixed
- Expected Real Return on Portfolio, 3%.

I would Amortize only the Future SS Income, not the pension.

Does this align with everyone’s thinking?

Ben, If you want ABW to be accessible to the casual DIY user then we need to step back and take a long look at the work LongVest has done to make VPW accessible and usable. I don’t think we are there yet and I am willing to help.

I suggest we start with a target audience of a somewhat more sophisticated DIYer and nail things down and then come back to the casual user.

When I was working, we used to develop persona(s) to use as we evaluate our designs. I suggest the following as a starting point:
  • Some familiarity with Excel, some familiarity with basic finance, currently retired and actively managing target withdrawals.
  • Some familiarity with Excel, some familiarity with basic finance, within 5 years of retirement, and planning target withdrawals.
Using these personas helps shift you to another mindset and let’s you see how others will perceive things, rather than how you perceive them. Have you done anything similar?

On a slightly different topic, I am still not seeing any discussions on one of the most important risks with ABW, the magnitude of income volatility. What should a retiree expect? What can they do to dampen the volatility? Any suggestions on how we should address this gap?

WoodSpinner
User avatar
Ben Mathew
Posts: 1330
Joined: Tue Mar 13, 2018 11:41 am
Location: Seattle
Contact:

Re: Amortization Based Withdrawal (ABW)

Post by Ben Mathew »

WoodSpinner wrote: Fri Nov 20, 2020 12:41 pm Ben, If you want ABW to be accessible to the casual DIY user then we need to step back and take a long look at the work LongVest has done to make VPW accessible and usable. I don’t think we are there yet and I am willing to help.
VPW is a good example to follow.

VPW is an implementation of ABW, just like siamond's and mine are different implementations. I am suggesting that siamond and I create pages for our implementations, just like Longinvest did for VPW. Go into it in as much detail as is required there, complete with formulas and worksheets and everything else. The ABW page can stay focused on the general concepts and not on specific implementations.
User avatar
Ben Mathew
Posts: 1330
Joined: Tue Mar 13, 2018 11:41 am
Location: Seattle
Contact:

Re: Amortization Based Withdrawal (ABW)

Post by Ben Mathew »

WoodSpinner wrote: Fri Nov 20, 2020 12:41 pm On a slightly different topic, I am still not seeing any discussions on one of the most important risks with ABW, the magnitude of income volatility. What should a retiree expect? What can they do to dampen the volatility? Any suggestions on how we should address this gap?
This is a very important question. It would depend on the specific implementation--the combination of asset allocation + withdrawal strategy. I plan to address this in my implementation with the use of confidence intervals that I discussed in an earlier post.
User avatar
siamond
Posts: 5636
Joined: Mon May 28, 2012 5:50 am

Re: Amortization Based Withdrawal (ABW)

Post by siamond »

WoodSpinner wrote: Fri Nov 20, 2020 12:41 pmIt seems to me we should only Amortize Future Income that has not begun.

If you are currently receiving the income (e.g. currently receiving SS or a Pension) then it should simply be added (e.g. Current Income + Amortized Future Income + Amortized Portfolio Withdrawal) to come up with a projected spend amount. [...]
What you suggested is mathematically equivalent to what I suggested. The present value of a present value is equal to itself! :wink:

Now some pensions aren't [fully] inflation-adjusted, so it is more convenient to include them in the amortization math, even if they already started. And personally, I like to include all sources of income in the amortization math, because the outcome is a spending budget (as opposed to just a withdrawal amount) and this makes it easier to track as part of my regular expense tracking.

No big deal, both ways work.
User avatar
siamond
Posts: 5636
Joined: Mon May 28, 2012 5:50 am

Re: Amortization Based Withdrawal (ABW)

Post by siamond »

Kevin M wrote: Thu Nov 19, 2020 10:59 pm
siamond wrote: Thu Nov 19, 2020 8:25 pm Next, the common core of principles we established quite a few posts ago (my points 1/2/3) isn't acknowledged. And consequently there is a myriad of scenarios between the two extremes (the bonds-centric approach and the portfolio-centric approach) which are not addressed. This being said, I realize I've been a real pain in the neck about this topic, so I guess maybe the onus is on me to propose some text. Won't be able to get to it until early next week though.
I don't buy into your point 2 (or whatever the one is about using the investment portfolio expected rate of return as the discount rate for relatively safe future cash flows). From my brief read, it appears that you basically bulldozed over Ben's points, which I thought were quite sound (again, nothing to do with portfolio management, but about using appropriate discount rates). Here is my understanding of Ben's recommended approach, which seems thoroughly consistent with finance theory:
  1. Use an appropriate discount rate to calculate the present value of relatively safe cash flows. I would not call this bond-centric; it's just basic finance theory. I would use TIPS yields for inflation-indexed cash flows, and nominal Treasury yields for nominal cash flows. Yes, these are bonds, but calling it bond-centric seems somewhat pejorative to me.
  2. Include the PV of the safe cash flows as a component of the total portfolio (investment portfolio plus PV of future cash flows).
  3. Calculate the expected return of the total portfolio, using the safe discount rate as the expected return for the PV of the safe future cash flows.
Ben laid out quite clearly the possible issues with doing it the way that you are pushing, and I don't recall ever seeing a reasonable response to that.

I think what Ben has done in the wiki is a reasonable common ground, which lays out three approaches, one being the one you're pushing, and one being the one Ben has laid out.

I actually think the third one, keeping the cash flows separate, is the least contentious, and the one I probably would use. I would add some more verbiage to this section, pointing out that having additional safe cash flows affects the factors that go into selecting an AA, such as ability and need to take risk. I wouldn't have thought about it as similar to the sub-portfolios approach, but upon reflection, I see that that's another reasonable way to look at it; i.e., having additional safe cash flows reduces the need for an LMP, and enables the ability to have a larger risk portfolio (although it also reduces the need for a larger risk portfolio).

Fun stuff. I've been busy with personal stuff lately, so haven't been able to keep up with fun threads like this.

Kevin
Kevin, I would really urge you to do a bit more than a 'brief read' and truly analyze the precise use case we're discussing here. Please set aside the preconceived ideas you have about so-called financial theory, and just think about the underlying math of amortizing future cash flows (the NPV/PMT 2-step process). It is really quite simple.

If you follow the pointers I provided, you would notice Kitces' article on this precise topic, as well as the blog article I wrote myself a while ago. Both articles go through the amortization math in a lot of details (I called it 'annuitize' in the blog, but Ben's terminology is better). Which leads straight to point #2, the proper amortization rate is the rate of the assets you're going to sell in the short-term. You can use the type of assets you want, depending on the risk/reward profile you prefer, but the rule stands. If you find a flaw in the math, you tell me, but I doubt it...

No, I didn't go in a lengthy debate to compare pros and cons of Ben's approach with mine, because I know very well both are valid. And many other strategies in-between are valid. They just have pros and cons. I am not 'bulldozing' anybody, I am trying to get to a flexible withdrawal method which can address a wide variety of situations. If I am vocal, this is because I am trying to avoid ABW being cornered in a single narrow approach, however valid such approach might seem to some.

If you follow the pointer to my blog (or read the first few pages of this thread), you might also perceive a bit better than I am NOT speaking lightly about this topic. I studied it at length and I am well aware of the risk/reward balance, and I practice it myself every day, having a somewhat complex future cash flow situation between my wife and I. What is very clear to me is that when I early retired in my early 50s, I wasn't going to set aside 15+ years-equivalent of bonds while planning to postpone SS. Maybe YOU would have carved out TIPS, I might have pointed out how expensive this would be, but I would still respect your choice. So please give me a little bit of respect back and avoid the slights? We all have different financial circumstances, different financial goals, different behavioral profiles, and our personal answers differ.
User avatar
Ben Mathew
Posts: 1330
Joined: Tue Mar 13, 2018 11:41 am
Location: Seattle
Contact:

Re: Amortization Based Withdrawal (ABW)

Post by Ben Mathew »

siamond wrote: Fri Nov 20, 2020 7:38 pm [...] point #2, the proper amortization rate is the rate of the assets you're going to sell in the short-term. You can use the type of assets you want, depending on the risk/reward profile you prefer [...].
I suggest that language along these lines not be included in the ABW wiki. This would suggest to the reader that replacing a safe income stream with bonds vs portfolio AA vs anything in between is all part of a continuum of reasonable choices depending on their risk/reward preferences, just like choosing a safe or risky AA would be. But the risk implications of this choice is qualitatively different and more complex. It's not just about more risk vs less. It's also about the timing of the risk. Explaining that properly in a wiki article will be hard. I gave examples earlier showing some of the unexpected ways the risk can play out. But here's a simple example: a safe payout expected 10 years from today, amortized using the expected return of a risky portfolio will generate risk-free consumption for the next 10 years and risky consumption thereafter. A first time reader will not pick up on that. There is still confusion about this after much discussion. This is just not easy to understand, and we can't treat this in sufficient depth in the space of a general wiki article on ABW. Moreover, it's not necessary for people following the other two strategies because those don't have the same unexpected timing issues. The timing issues are relevant only to users adopting your strategy because of the rate you're using to amortize. So why introduce this issue in the general article? Why not save the discussion for the wiki article where you go in-depth into the specifics of your implementation?
Last edited by Ben Mathew on Sat Nov 21, 2020 12:42 am, edited 1 time in total.
User avatar
Ben Mathew
Posts: 1330
Joined: Tue Mar 13, 2018 11:41 am
Location: Seattle
Contact:

Re: Amortization Based Withdrawal (ABW)

Post by Ben Mathew »

siamond wrote: Fri Nov 20, 2020 7:38 pm If I am vocal, this is because I am trying to avoid ABW being cornered in a single narrow approach, however valid such approach might seem to some.
I think we are all in agreement that the ABW wiki should be left flexible/general and not adhere too closely to one method. Many of us don't like the raw historical return estimate, yet it's represented in the wiki. I didn't include your approach to income/expense in the first draft only because I didn't realize you were doing it that way. Once it was clear that you and several others were discounting using the portfolio rate of return, and we couldn't change each others' minds about the right way to do it, I promptly updated the article to reflect that. I think you can rest assured that your voice came through loud and clear and the article was changed to include your approach. The discussion we're having now is not whether to include your approach or my approach, but how much detail to include. That's a tough one because, despite our best efforts, there isn't consensus on the details that truly matter. I think the best way forward would be a minimalist approach that doesn't get us bogged down in never-ending circular debates. I'm sure you and I would both prefer developing our own implementations (like Longinvest did) rather than endless discussions about what goes on in the ABW page.
KarenC
Posts: 157
Joined: Mon Apr 27, 2015 7:25 am

Re: Amortization Based Withdrawal (ABW)

Post by KarenC »

Ben Mathew wrote: Fri Nov 20, 2020 11:39 pm
siamond wrote: Fri Nov 20, 2020 7:38 pm [...] point #2, the proper amortization rate is the rate of the assets you're going to sell in the short-term. You can use the type of assets you want, depending on the risk/reward profile you prefer [...].
I suggest that language along these lines not be included in the ABW wiki. This would suggest to the reader that replacing a safe income stream with bonds vs portfolio AA vs anything in between is all part of a continuum of reasonable choices depending on their risk/reward preferences, just like choosing a safe or risky AA would be. […]
From what I've read so far, with regards to replacing a safe income stream, using withdrawals from bonds or from the portfolio AA are both reasonable choices.
"The first principle is that you must not fool yourself, and you are the easiest person to fool." — Richard P. Feynman
User avatar
Ben Mathew
Posts: 1330
Joined: Tue Mar 13, 2018 11:41 am
Location: Seattle
Contact:

Re: Amortization Based Withdrawal (ABW)

Post by Ben Mathew »

KarenC wrote: Sat Nov 21, 2020 6:55 am
Ben Mathew wrote: Fri Nov 20, 2020 11:39 pm
siamond wrote: Fri Nov 20, 2020 7:38 pm [...] point #2, the proper amortization rate is the rate of the assets you're going to sell in the short-term. You can use the type of assets you want, depending on the risk/reward profile you prefer [...].
I suggest that language along these lines not be included in the ABW wiki. This would suggest to the reader that replacing a safe income stream with bonds vs portfolio AA vs anything in between is all part of a continuum of reasonable choices depending on their risk/reward preferences, just like choosing a safe or risky AA would be. […]
From what I've read so far, with regards to replacing a safe income stream, using withdrawals from bonds or from the portfolio AA are both reasonable choices.
We will have to agree to disagree. I believe there are problems with siamond's approach, and I tried to show why. I realize I haven't convinced everyone, including you. So siamond's approach is included in the wiki, like the other two approaches, without any suggestion as to whether it is reasonable or not. Keep in mind that there is no suggestion in the writeup that my approach is reasonable either. Both approaches are presented equally, without commentary, with links provided for further exploration. I even put siamond's approach before my own.

What I am opposed to is any sort of language that would appear to suggest that these are all reasonable approaches. You believe it is. I believe it is not. So how about we keep the wiki language neutral about it?
User avatar
Kevin M
Posts: 11706
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: Using the Amortization Based Withdrawal Approach to Determine Withdrawals: Year 2000 Retiree Example

Post by Kevin M »

Ben Mathew wrote: Tue Nov 10, 2020 1:59 pm
siamond wrote: Sun Nov 08, 2020 5:57 pm Ben, could you please read Kitces? Otherwise, we'll swap posts until we're blue on the face and this won't help.
I read Kitces's article: Choosing an appropriate discount rate for retirement planning strategies. I have read several of his articles, and find many of them informative. But I disagree with the analysis he did here and the conclusions he drew.

The problem is that he is not controlling for risk in his comparisons. He says:
In the context of financial planning strategies, the proper discount rate to use is [...] what return could be generated over time by the money, if it were in fact available today to be invested.
Instead it should be:
In the context of financial planning strategies, the proper discount rate to use is [...] what return could be generated over time by the money in an investment of comparable risk, if it were in fact available today to be invested.
He does acknowledge the need to control for risk in one spot. In the paragraph after example 1, he says (my bolding):
Jeremy’s decision to use a 10% discount rate was based on having a (comparable risk) alternative investment that already had an expected return of 10%, which effectively became the “hurdle rate” that the new investment had to clear
The "comparable risk" caveat is crucial. But he seems to forget about it in the rest of the article and draws conclusions that are clearly false if risk is not in fact comparable. Take Example 2:
Example 2. Charlie is a 65-year-old male, and is eligible for a $40,000/year lifetime pension. The company has also offered Charlie a $500,000 lump sum option, in exchange for his pension, and he’s trying to decide if this is a “good” deal or not. According to the Society of Actuaries’ Longevity Illustrator, as a non-smoking male with reasonable health, Charlie’s life expectancy is approximately age 87, which means he can expect to receive $40,000/year for an average of 22 years.

Charlie’s alternative, if he takes the lump sum, would be investing it in a diversified portfolio, in which he estimates that he could earn a 7% average annual growth rate. Which means that potential portfolio return is the opportunity cost he’s giving up if he sticks with the pension and doesn’t take the lump sum. Thus, 7% would be Charlie’s discount rate.

If Charlie then calculates the present value of his $40,000 year for 22 years at a 7% discount rate, he’ll find that the net present value is “only” about $451,000. In other words, it would only take $451,000 to provide $40,000/year for 22 years at a 7% rate of return. Since the available lump sum provides more than that – a full $500,000 – this implies that it’s a slightly better deal to take the lump sum, since it would literally give Charlie “more than enough” to replace his 22 years of pension payments at a 7% rate of return (with almost $50,000 to spare!).
Kitces is discounting the safe pension payout of $40,000 per year at the 7% interest rate that his risky portfolio provides. So the pension is valued at $451,000 (I get $442,450, but that's not material). if the lump sum offered is more than that, then he says take the lump sum. But this only works if the portfolio is as safe as the pension. It is clearly wrong if the portfolio is riskier than the pension. Suppose the lump sum is a dollar extra: $451,001. Should Charlie cash in the safe pension payouts and place all of it in the risky portfolio to harvest a very tiny increase in expected return? That would be an enormous increase in risk for a trivial increase in expected return. So clearly not. To avoid such misleading conclusions, we have to use risk-adjusted returns for the discount rate to value any asset. It's a core principle in finance, and you can't violate it without running into a lot of trouble.

To see why bond return is the more appropriate discount rate for the safe pension payouts, look at it this way: To hold risk constant between the two alternatives (lump sum vs keep the pension), Charlie would have to invest the lump sum in safe bonds. If Charlie invests it instead at the risky AA of the whole portfolio, then he is taking on more risk. To keep risk the same, he would have to invest the lump sum in bonds. So his comparison is not "pension vs lump sum invested in current portfolio AA" but rather "pension vs lump sum invested in bonds". So using the discount rate offered by bonds (say 2% nominal), he calculates the PV of the pension to be $706,322. If the lump sum offer is less than that, then keep the pension because bonds pay less. If the lump sum offer is more than that, then take the lump sum and invest it in bonds because bonds pay more. Charlie's lump sum offer of $500,000 was less than this threshold, so he should decline.
Read the Kitces article, and agree 100% with Ben's critique. Kitces notes that the alternative investment should be of comparable risk, yet ignores that in most of the article.

It makes no financial sense to use a risky portfolio rate of return as the discount rate for a safe, inflation-adjusted income stream like Social Security benefits. Financial advisor Alan Roth agrees:
Another thing I love is that, rather than just add the total benefit payments, SSAnalyze takes into account the cost of money, calculating the net present value of the payments and allowing users to pick the discount rate.

The default discount rate indicated is 2%, with which I agree, as this is a U.S. government-backed inflation-protected income stream. Don’t game it by using a high stock-like rate that would result in taking the benefit early.
Source: https://www.financial-planning.com/news ... calculator

However, I think Roth made a mistake in mentioning the 2% yield, as that was about the nominal 30-year yield in August 2016; the real 30-year yield was about 0.6%.

Note also that Mike Piper's OpenSocialSecurity currently uses a default discount rate of -0.46%, which was the 20-year TIPS yield on 11/19/2020. Of course the tool allows one to enter one's own discount rate if one clicks the additional input link near the top. Here is what the tool tip says:
A "discount rate" is necessary to reflect the fact that the sooner a dollar is received, the sooner it can be invested. The default discount rate is the yield on 20-year TIPS, (which is currently -0.46%). See this article for a discussion of why the yield on 20-year TIPS is used as the default.
Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)
marcopolo
Posts: 3615
Joined: Sat Dec 03, 2016 10:22 am

Re: Amortization Based Withdrawal (ABW)

Post by marcopolo »

Ben Mathew wrote: Sat Nov 21, 2020 1:54 pm
KarenC wrote: Sat Nov 21, 2020 6:55 am
Ben Mathew wrote: Fri Nov 20, 2020 11:39 pm
siamond wrote: Fri Nov 20, 2020 7:38 pm [...] point #2, the proper amortization rate is the rate of the assets you're going to sell in the short-term. You can use the type of assets you want, depending on the risk/reward profile you prefer [...].
I suggest that language along these lines not be included in the ABW wiki. This would suggest to the reader that replacing a safe income stream with bonds vs portfolio AA vs anything in between is all part of a continuum of reasonable choices depending on their risk/reward preferences, just like choosing a safe or risky AA would be. […]
From what I've read so far, with regards to replacing a safe income stream, using withdrawals from bonds or from the portfolio AA are both reasonable choices.
We will have to agree to disagree. I believe there are problems with siamond's approach, and I tried to show why. I realize I haven't convinced everyone, including you. So siamond's approach is included in the wiki, like the other two approaches, without any suggestion as to whether it is reasonable or not. Keep in mind that there is no suggestion in the writeup that my approach is reasonable either. Both approaches are presented equally, without commentary, with links provided for further exploration. I even put siamond's approach before my own.

What I am opposed to is any sort of language that would appear to suggest that these are all reasonable approaches. You believe it is. I believe it is not. So how about we keep the wiki language neutral about it?
Thanks for all the work you are doing on this.

I am curious, do you also believe that the only reasonable approach in real life would also be to use assets with the same risk as Soc Sec to bridge the gap until that safe cash flow is turn on?

For example, someone retiring at age 50, and delaying SS until age 70, is your position that the only reasonable thing to do would be to have 20 years worth (NPV) of SS in TIPS at the start of retirement?

Also, is SS really as risk free as TIPS? Under current law, SS will be reduced by ~30%, so do you include that in discount rate?
Once in a while you get shown the light, in the strangest of places if you look at it right.
User avatar
Ben Mathew
Posts: 1330
Joined: Tue Mar 13, 2018 11:41 am
Location: Seattle
Contact:

Re: Amortization Based Withdrawal (ABW)

Post by Ben Mathew »

marcopolo wrote: Sat Nov 21, 2020 8:39 pm I am curious, do you also believe that the only reasonable approach in real life would also be to use assets with the same risk as Soc Sec to bridge the gap until that safe cash flow is turn on?

For example, someone retiring at age 50, and delaying SS until age 70, is your position that the only reasonable thing to do would be to have 20 years worth (NPV) of SS in TIPS at the start of retirement?
No. What you are describing is a safe subportfolio strategy, described in the wiki article under the first approach: "Keep it separate", second paragraph (my bolding):
A retiree formally pursuing the sub-portfolio approach described earlier may indeed find it convenient to use conservative estimates of social security and pensions to fund the safe sub-portfolios. The income streams are already duration matched. Gaps remaining can be filled with duration matched bonds.
The third approach ("safe cash flow valued using using bond rates") does not require filling in the gaps. It only requires valuing the safe cash flows and counting it as a bond in the portfolio. So the PV of the SS is already counted as a bond. Whether or not a retiree wants to top that up with more bonds depends on their risk preference. A risk averse retiree will add a lot of actual bonds to their "SS bonds". A risk tolerant retiree may not add any. So this is not about more or less bonds. That depends on risk aversion. Rather, this is about accounting for all assets properly so that risk can be held consistent across time (by maintaining a consistent asset allocation on the total portfolio).
marcopolo wrote: Sat Nov 21, 2020 8:39 pm Also, is SS really as risk free as TIPS? Under current law, SS will be reduced by ~30%, so do you include that in discount rate?
I handle this by using a conservative estimate of SS. So based on what you describe, 70% of SS could be the safe cash flow discounted by the bond rate. The wiki writeup does address the need to make a conservative safe estimate before discounting with the bond rate: "Create a conservatively low estimate of safe extra income."

To incorporate risky cash flows, I think the easiest and most intuitive approach is to re-run the analysis using low, medium, and high scenarios and evaluate if the AA and withdrawal strategy looks satisfactory across all three scenarios.
marcopolo
Posts: 3615
Joined: Sat Dec 03, 2016 10:22 am

Re: Amortization Based Withdrawal (ABW)

Post by marcopolo »

Ben Mathew wrote: Sat Nov 21, 2020 10:26 pm
marcopolo wrote: Sat Nov 21, 2020 8:39 pm I am curious, do you also believe that the only reasonable approach in real life would also be to use assets with the same risk as Soc Sec to bridge the gap until that safe cash flow is turn on?

For example, someone retiring at age 50, and delaying SS until age 70, is your position that the only reasonable thing to do would be to have 20 years worth (NPV) of SS in TIPS at the start of retirement?
No. What you are describing is a safe subportfolio strategy, described in the wiki article under the first approach: "Keep it separate", second paragraph (my bolding):
A retiree formally pursuing the sub-portfolio approach described earlier may indeed find it convenient to use conservative estimates of social security and pensions to fund the safe sub-portfolios. The income streams are already duration matched. Gaps remaining can be filled with duration matched bonds.
The third approach ("safe cash flow valued using using bond rates") does not require filling in the gaps. It only requires valuing the safe cash flows and counting it as a bond in the portfolio. So the PV of the SS is already counted as a bond. Whether or not a retiree wants to top that up with more bonds depends on their risk preference. A risk averse retiree will add a lot of actual bonds to their "SS bonds". A risk tolerant retiree may not want to add any. So this is not about more or less bonds. That depends on risk aversion. Rather, this is about accounting for all assets properly so that risk can be held consistent across time (by maintaining a consistent asset allocation on the total portfolio).
marcopolo wrote: Sat Nov 21, 2020 8:39 pm Also, is SS really as risk free as TIPS? Under current law, SS will be reduced by ~30%, so do you include that in discount rate?
I handle this by using a conservative estimate of SS. So based on what you describe, 70% of SS could be the safe cash flow discounted by the bond rate. The wiki writeup does address the need to make a conservative safe estimate before discounting with the bond rate: "Create a conservatively low estimate of safe extra income."

To incorporate risky cash flows, I think the easiest and most intuitive approach is to re-run the analysis using low, medium, and high scenarios and evaluate if the AA and withdrawal strategy looks satisfactory across all three scenarios.
Thanks for the detailed explanation.
Once in a while you get shown the light, in the strangest of places if you look at it right.
KarenC
Posts: 157
Joined: Mon Apr 27, 2015 7:25 am

Re: Amortization Based Withdrawal (ABW)

Post by KarenC »

Ben Mathew wrote: Sat Nov 21, 2020 1:54 pm
KarenC wrote: Sat Nov 21, 2020 6:55 am
Ben Mathew wrote: Fri Nov 20, 2020 11:39 pm
siamond wrote: Fri Nov 20, 2020 7:38 pm [...] point #2, the proper amortization rate is the rate of the assets you're going to sell in the short-term. You can use the type of assets you want, depending on the risk/reward profile you prefer [...].
I suggest that language along these lines not be included in the ABW wiki. This would suggest to the reader that replacing a safe income stream with bonds vs portfolio AA vs anything in between is all part of a continuum of reasonable choices depending on their risk/reward preferences, just like choosing a safe or risky AA would be. […]
From what I've read so far, with regards to replacing a safe income stream, using withdrawals from bonds or from the portfolio AA are both reasonable choices.
We will have to agree to disagree. I believe there are problems with siamond's approach, and I tried to show why. I realize I haven't convinced everyone, including you. So siamond's approach is included in the wiki, like the other two approaches, without any suggestion as to whether it is reasonable or not. Keep in mind that there is no suggestion in the writeup that my approach is reasonable either. Both approaches are presented equally, without commentary, with links provided for further exploration. I even put siamond's approach before my own.

What I am opposed to is any sort of language that would appear to suggest that these are all reasonable approaches. You believe it is. I believe it is not. So how about we keep the wiki language neutral about it?
My sense is that siamond's suggested wording for the amortization rate to use is completely neutral.
"The first principle is that you must not fool yourself, and you are the easiest person to fool." — Richard P. Feynman
User avatar
siamond
Posts: 5636
Joined: Mon May 28, 2012 5:50 am

Re: Amortization Based Withdrawal (ABW)

Post by siamond »

Ben Mathew wrote: Sat Nov 21, 2020 12:32 am
siamond wrote: Fri Nov 20, 2020 7:38 pm If I am vocal, this is because I am trying to avoid ABW being cornered in a single narrow approach, however valid such approach might seem to some.
I think we are all in agreement that the ABW wiki should be left flexible/general and not adhere too closely to one method. Many of us don't like the raw historical return estimate, yet it's represented in the wiki. I didn't include your approach to income/expense in the first draft only because I didn't realize you were doing it that way. Once it was clear that you and several others were discounting using the portfolio rate of return, and we couldn't change each others' minds about the right way to do it, I promptly updated the article to reflect that. I think you can rest assured that your voice came through loud and clear and the article was changed to include your approach. The discussion we're having now is not whether to include your approach or my approach, but how much detail to include. That's a tough one because, despite our best efforts, there isn't consensus on the details that truly matter. I think the best way forward would be a minimalist approach that doesn't get us bogged down in never-ending circular debates. I'm sure you and I would both prefer developing our own implementations (like Longinvest did) rather than endless discussions about what goes on in the ABW page.
I've tried hard to focus on the commonalities between our respective approaches (and other possible approaches in-between) and I thought we had made good progress, identifying that differences are about personal choices to structure assets (e.g. driven by one's perceptions about risks and rewards) instead of the withdrawal procedure itself. Unfortunately, since then, we not only went in never-ending circles, but we even went backwards in the past few days, where even basic math is being disputed and personal bias about what 'risk' means muddies what should indeed be a neutral discussion... And I am growing tired of such unproductive and borderline acrimonious discussions.

I have to reluctantly concede that this minimalistic wiki approach you've been suggesting, combined with complementary links, might indeed be the only way to make progress. This seems an unfortunate outcome to me, but yes, quite possibly the only way to proceed.

I don't plan to develop another wiki page, I just don't think this is a terribly efficient format to explain things clearly (in addition of triggering endless and unproductive discussions), so I will tweak my blog (and associated spreadsheet) to improve consistency with the ABW wiki page, notably adding emphasis on amortization terminology, and I'll beef it up over time (first thing to do is to expand on dynamic expected returns and show backtested trajectories in a Part 3 - a reprisal of an old thread of mine from years ago!).

Overall, although we obviously disagree on the pros & cons of respective approaches, as well as the ideal format of such wiki page (I still have a lot of qualms about not acknowledging the DYI spreadsheet-oriented nature of the intended audience), I do appreciate the patience you displayed and the openness to consider other ideas than yours. Let's try to keep going in a non-judgmental manner, shall we?
User avatar
Ben Mathew
Posts: 1330
Joined: Tue Mar 13, 2018 11:41 am
Location: Seattle
Contact:

Re: Amortization Based Withdrawal (ABW)

Post by Ben Mathew »

siamond wrote: Sun Nov 22, 2020 10:05 am I have to reluctantly concede that this minimalistic wiki approach you've been suggesting, combined with complementary links, might indeed be the only way to make progress. This seems an unfortunate outcome to me, but yes, quite possibly the only way to proceed.
Great. I know that it's not the ideal representation of either of our viewpoints. But I believe that it is a good compromise.
siamond wrote: Sun Nov 22, 2020 10:05 am Overall, although we obviously disagree on the pros & cons of respective approaches, as well as the ideal format of such wiki page (I still have a lot of qualms about not acknowledging the DYI spreadsheet-oriented nature of the intended audience), I do appreciate the patience you displayed and the openness to consider other ideas than yours.
Thank you for your genuine attempt to find common ground. I'm sorry we couldn't make more progress on that front. I think we are coming at this from very different places, and trying to understand what the other person is saying through very different lenses/models. That might have been the source of some frustration on both sides. But the fact that I was able to express my concerns about a popular method used by you and several others on this thread, without being attacked and silenced, is a credit to you and the Bogleheads community in general.
siamond wrote: Sun Nov 22, 2020 10:05 am Let's try to keep going in a non-judgmental manner, shall we?
Wholeheartedly agree!
User avatar
1210sda
Posts: 1713
Joined: Wed Feb 28, 2007 8:31 am

Re: Amortization Based Withdrawal (ABW)

Post by 1210sda »

Is there a way to directly access the ABW calculator on the Wiki without having to go back to Willthrill's original post and clicking on the wiki page link??
User avatar
Kevin M
Posts: 11706
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: Amortization Based Withdrawal (ABW)

Post by Kevin M »

1210sda wrote: Wed Nov 25, 2020 6:11 pm Is there a way to directly access the ABW calculator on the Wiki without having to go back to Willthrill's original post and clicking on the wiki page link??
Basic: https://drive.google.com/file/d/1XOnBaU ... sp=sharing
Advanced: https://drive.google.com/file/d/1v8KZfX ... sp=sharing

Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)
User avatar
1210sda
Posts: 1713
Joined: Wed Feb 28, 2007 8:31 am

Re: Amortization Based Withdrawal (ABW)

Post by 1210sda »

Thank you Kevin M.
klaus14
Posts: 522
Joined: Sun Nov 25, 2018 7:43 pm

Re: Amortization Based Withdrawal (ABW)

Post by klaus14 »

This is a good method.
Personally, I would come up with two return numbers: one is the expected case and the other is a pessimistic case. And I would make sure i am fine in the pessimistic scenario.

To come up with both numbers i would use Vanguard return numbers from here. I would put them into PV calculator for my portfolio.

Then i would get 50percentile real return (~ 2.62%) from PV and put it into ABW calculator.
I would use 25p real return (~1.33%) as pessimistic return expectation and put it into ABW calculator as well.
If 25p number produces an income that I am comfortable with then i would retire. After retirement i would use 50p real return to determine each year's withdrawal

For a 35 year retirement, that leaves 25% at the end, for a 1M portfolio numbers are:
Expected: $38,536
Pessimistic: $ 29,542

I am leaving 25% so that i could put it into an annuity. Together with Social Security that solves the longevity risk.
Post Reply