Page 1 of 1

How do you "estimate" future portfolio returns?

Posted: Thu Feb 14, 2013 2:41 pm
by countofmc
I've been trying out a number of retirement planning calculators, and one variable is that they ask you what the future returns on your portfolio would be. Not sure what a reasonable estimate would be for this. My AA is 40 stock/ 60 bonds, and I'm looking at a time horizon of about 30 years until retirement, and want to keep the AA fixed for that entire time (no reduction in stocks).

I usually punch in a 4% nominal return, or a 1% real return (with 3% estimated inflation). Is this reasonable?

Re: How do you "estimate" future portfolio returns?

Posted: Thu Feb 14, 2013 2:55 pm
by bottomfisher
I usually punch in a 4% nominal return, or a 1% real return (with 3% estimated inflation). Is this reasonable?
Is it possible? Yes. Is it reasonable? Let's all hope not. I toy with these compounding interest/return calculators from time to time. I always put a few different numbers in for real return to a get a better expectation of the range for good, average, and bad scenario over the given time frame

Re: How do you "estimate" future portfolio returns?

Posted: Thu Feb 14, 2013 2:58 pm
by NYBoglehead
The good news is if you use an extremely conservative estimate in your calculation and that conservative growth estimate says you have enough money, you should be in good shape. If you estimate 4% nominal return and that will provide you with enough money for retirement, anything above that will be a bonus.

Re: How do you "estimate" future portfolio returns?

Posted: Thu Feb 14, 2013 3:01 pm
by countofmc
NYBoglehead wrote:The good news is if you use an extremely conservative estimate in your calculation and that conservative growth estimate says you have enough money, you should be in good shape. If you estimate 4% nominal return and that will provide you with enough money for retirement, anything above that will be a bonus.
Wonderful point!

Re: How do you "estimate" future portfolio returns?

Posted: Thu Feb 14, 2013 3:04 pm
by Taylor Larimore
countofmc wrote:I've been trying out a number of retirement planning calculators, and one variable is that they ask you what the future returns on your portfolio would be. Not sure what a reasonable estimate would be for this. My AA is 40 stock/ 60 bonds, and I'm looking at a time horizon of about 30 years until retirement, and want to keep the AA fixed for that entire time (no reduction in stocks).

I usually punch in a 4% nominal return, or a 1% real return (with 3% estimated inflation). Is this reasonable?
This page in our wiki should be helpful:

Historical and Expected Returns

Best wishes.
Taylor

Re: How do you "estimate" future portfolio returns?

Posted: Thu Feb 14, 2013 3:26 pm
by Sidney
for a 40/60

4% real on stocks and -1% real on bonds gets you an average of 1% real. While that may be conservative, I don't think it is pessimistic. Mid-duration tips are running negative 1% or worse.

Re: How do you "estimate" future portfolio returns?

Posted: Thu Feb 14, 2013 3:33 pm
by countofmc
Sidney wrote:for a 40/60

4% real on stocks and -1% real on bonds gets you an average of 1% real. While that may be conservative, I don't think it is pessimistic. Mid-duration tips are running negative 1% or worse.
Maybe I need to take on more risk, although I can't see myself stomaching anything more than 50/50.

Re: How do you "estimate" future portfolio returns?

Posted: Thu Feb 14, 2013 3:34 pm
by NYBoglehead
Sidney wrote:for a 40/60

4% real on stocks and -1% real on bonds gets you an average of 1% real. While that may be conservative, I don't think it is pessimistic. Mid-duration tips are running negative 1% or worse.
I certainly hope that bonds won't lose 1% per year for 30 years!! While real returns right now might be negative, I think it's awfully pessimistic to think that will be the case for the next 30 years.

Re: How do you "estimate" future portfolio returns?

Posted: Thu Feb 14, 2013 3:55 pm
by SSSS
NYBoglehead wrote:I certainly hope that bonds won't lose 1% per year for 30 years!! While real returns right now might be negative, I think it's awfully pessimistic to think that will be the case for the next 30 years.
Pessimists tend to have a more realistic view of the world.

http://en.wikipedia.org/wiki/Depressive_realism

The best part of being a pessimist is that you'll be pleasantly surprised more often than negatively surprised.

Re: How do you "estimate" future portfolio returns?

Posted: Thu Feb 14, 2013 3:59 pm
by NYBoglehead
SSSS wrote:
NYBoglehead wrote:I certainly hope that bonds won't lose 1% per year for 30 years!! While real returns right now might be negative, I think it's awfully pessimistic to think that will be the case for the next 30 years.
Pessimists tend to have a more realistic view of the world.

http://en.wikipedia.org/wiki/Depressive_realism

The best part of being a pessimist is that you'll be pleasantly surprised more often than negatively surprised.
You're joking, right? What about all the pessimists here in the US, who kept their money on the sidelines last year while the rest of us enjoyed a 15% gain in stocks? Or missed out on the ~6% gains in the first 45 days of the year?

Re: How do you "estimate" future portfolio returns?

Posted: Thu Feb 14, 2013 4:07 pm
by ResNullius
Whenever I do sample runs with calculators, I always underestimate returns, overestimate inflation, and overestimate longevity. This produces the most conservative view possible, while still being somewhat reasonable.

Re: How do you "estimate" future portfolio returns?

Posted: Thu Feb 14, 2013 4:09 pm
by Sidney
Bonds are for safety. Also keep in mind that a fairly long sequence of negative real rates followed by increasing yields needs to do some real work to get the total return up by much in real terms. Once rates start to turn more positive, the NAV will take a hit for a while before the cumulative return moves positive.

A 60% load on bonds could put a drag on real returns for quite some time. It might be better to assume conservative returns on bonds and do more with equity -- either increase from 40% or tilt.

Re: How do you "estimate" future portfolio returns?

Posted: Thu Feb 14, 2013 4:19 pm
by red5
I personally use a weighted average of my asset location (bonds, us stocks, small value, etc) using Mr. Rick Ferri's 30 year forecast. I use his forecasts minus 1% for a low estimate and then I use his forecasts without any adjustment for a high estimate. This gives me a range that I may find my portfolio at sometime in the future.

I also take into account my AA for each year (in accordance with my IPS...basically I take into account an increase in bond allocation and a decrease in stock allocation). Thus I calculate an expected return for each year in accordance with Mr. Rick Ferri's forecast. Yes, this does entail finding expected returns for every single year until I turn 95 years old (I'm 30). Thank you Excel!

I chose to use Rick's forecasts because they seem to be lower than forecasts I found elsewhere (especially Vanguard) which (hopefully) gives me a conservative estimate.

Re: How do you "estimate" future portfolio returns?

Posted: Thu Feb 14, 2013 4:34 pm
by FillorKill
countofmc wrote:
Sidney wrote:for a 40/60

4% real on stocks and -1% real on bonds gets you an average of 1% real. While that may be conservative, I don't think it is pessimistic. Mid-duration tips are running negative 1% or worse.
Maybe I need to take on more risk, although I can't see myself stomaching anything more than 50/50.
If you're 40/60 and 50/50 would be your absolute limit.... Don't bother. The difference isn't that great - why push up against your limit? Saving more, making more, informed lifecycle tax management, TLH, etcetera can be far more valuable that going from 40/60 to 50/50 especially if it pushes your comfort level.

Re: How do you "estimate" future portfolio returns?

Posted: Fri Feb 15, 2013 12:22 am
by bobcat2
Question: How do you "estimate" future portfolio returns?

Answer: With a realistic confidence interval.

Before retirement I would ask myself how I would adjust my annual savings amounts, AA, and retirement date if the returns played out near the mean estimated returns, near the low bound estimated returns, and near the high bound estimated returns.

If the person is already retired, then most retirement income better becoming from annuity streams - SS, DB pension, and private life annuities. For the remainder of the retirement income coming from a portfolio of assets, I would expect the estimated returns to provide the remainder of my aspirational level of retirement income, but I would adjust the withdrawal rate (WR) every year conditional on the portfolio's performance. The risk in the portfolio is kept relatively low during retirement to avoid large unhappy surprises in WR changes. The longer one stays healthy the more additional annuity income is purchased over time during retirement.

In other words during retirement you are not that concerned with portfolio estimated returns, because you are not taking much risk in the portfolio.

BobK

Re: How do you "estimate" future portfolio returns?

Posted: Fri Feb 15, 2013 6:56 am
by Rodc
Depends on what question you are trying to answer.

What will you most likely have some years down the road?

What is the worst reasonable outcome? (whatever worst reasonable means?)

What is the worst possible?

What is a poor, but not super bad outcome (10th or 25th percentile outcome)?

4% nominal, 3% inflation for a 30 years is likely low for estimating a most likely, or a median outcome (two different things).

For estimating a poor but not truly epically bad outcome, maybe pretty good.

Too optimistic for worst thing that could ever happen.

But if you are 30 years out, today's economy is but a blink of an eye. Things will change in many ways, some totally surprising. And your future needs are totally unknown. Will you end up needing to retire in 20, forced out of work by a future depression or work at a very high pay for 40 years because you are having so much fun? Three divorces which wipe you out, or stay with one person and get a surprise inheritance? Great health or poor?

My point is only that investment uncertainty is dwarfed by all the other life factors at play, so don't spend too much time fine tuning at this point.

Try a few scenarios, what if things go pretty well, you retire in 30 years, get the long term average returns. What if you have to retire at 25 years after somewhat poor returns? Definitely make a plan that looks like it works for the first. Better yet. make a plan that works for the second if you reasonably can.Check, replan if needed, every 5 years or so.

Maybe give FireCalc a whirl as that does a simulation that looks at many scenarios for you. There are other Monte Carlo tools you can find for free as well.

Best of luck,
Rod

Re: How do you "estimate" future portfolio returns?

Posted: Fri Feb 15, 2013 8:42 am
by dbr
bobcat2 wrote:Question: How do you "estimate" future portfolio returns?

Answer: With a realistic confidence interval.

Before retirement I would ask myself how I would adjust my annual savings amounts, AA, and retirement date if the returns played out near the mean estimated returns, near the low bound estimated returns, and near the high bound estimated returns.
Right, you are describing a sensitivity analysis regarding the assumptions, which is an absolutely critical step in attempting a model like this.

I am not an MC modeler, but I agree that one is taken aback when an MC model starts out by asking for specification of exactly the thing that is stochastically variable, namely the returns. I understand the idea that the variability is generated by sampling from a distribution of returns, but isn't the point to drive the variability back to modeling a distribution of distributions of returns where expected return and variability of returns are the variables that are stochastic? (if that makes any sense) Also, of course, inflation should be one of the stochastic variables and not specified at a fixed value. In fact, also expected expenses should be modeled with a stochastic range.

Note, FireCalc finesses this problem by using actual historical returns in all their variable glory, the same for inflation. Jim Otar's model does the same and I think he argues this is a more appropriate picture than artificial assumptions applied in MC.

Re: How do you "estimate" future portfolio returns?

Posted: Fri Feb 15, 2013 9:39 am
by nisiprius
countofmc wrote:I've been trying out a number of retirement planning calculators, and one variable is that they ask you what the future returns on your portfolio would be. Not sure what a reasonable estimate would be for this. My AA is 40 stock/ 60 bonds, and I'm looking at a time horizon of about 30 years until retirement, and want to keep the AA fixed for that entire time (no reduction in stocks).

I usually punch in a 4% nominal return, or a 1% real return (with 3% estimated inflation). Is this reasonable?
Look for a better retirement calculator.

They all suck. but the kind you are using is just useless. To begin with, "we'll tell you the future if you'll tell us the future first" is a total cop-out. Second, there's no excuse for asking for two separate hard-to-predict numbers--returns AND inflation--when, in fact, everything can and should be done in real dollars and "real" (inflation-adjusted) returns. Real returns are bad enough to predict, but at least they are a bit more stable than nominal (raw-dollar-number) returns.

The insurmountable problem is that all financial numbers can be higher or lower than their historical averages by amounts that really matter, and they can stay higher or lower than their historical averages for decades.

You have (at least) two choices. Go with the "historical averages" or knock off two or three percent as a margin of safety, and in accord with what seems to be current universal pessimism going forward.

Doing some homebrew calculations using the SBBI data
--for 40% "large-company stocks"
--and "intermediate-term government bonds"
--rebalanced annually,

I find that the real return above inflation for 1926 through 2009 was 4.52%.

The SBBI yearbook says directly that the average annual return, nominal (before inflation) was 8.1% for 50/50 and 7.2% for 30/70, and inflation over the total period 1926 through 2009 was 3.7%, for real returns of 4.4% and 3.5%, respectively. I'm not going to bother to try to account for the different--might be because I used intermediate-term bonds--but, anyway, same ballpark.

But note that the same source says that the (nominal) returns for a 50/50 stocks/(long-term government) bonds portfolio, over 20-year holding periods, varied from 4.6% to 14.75%! I don't want to parse in detail other than to make the point that one 20-year period can be awfully different from another. Maybe there is stability in "the long run" but if so, the long run is very long. Do not be fooled into thinking that over a period of 20, 30, or 40 years, it will all average down to something pretty close to the historical average.

As for a better retirement calculator, I don't know what's available to you and I haven't used one myself in a while, but I think you want to be looking for "Monte Carlo simulators" that show you the potential range of outcomes that would have occurred, based on past experience. There are many philosophical problems with these, too, but they're less bad than "tell me what your investments will return and I'll tell you what your investments will return" calculators. Three examples of the Monte Carlo genre are Fidelity's Retirement Income Planner; Financial Engines; and FIRECalc. Often you can get access to such a calculator at no cost through your employer's retirement plan, if you have one. FIRECalc is just plain free, I think.

Re: How do you "estimate" future portfolio returns?

Posted: Fri Feb 15, 2013 9:46 am
by mptfan
NYBoglehead wrote:The good news is if you use an extremely conservative estimate in your calculation and that conservative growth estimate says you have enough money, you should be in good shape. If you estimate 4% nominal return and that will provide you with enough money for retirement, anything above that will be a bonus.
Of course the opposite is also true... if you use an extremely conservative estimate, you will end up saving much more than you need to save for retirement, and therefore deferring much more gratification when you are younger than is necessary to have a secure retirement. A bigger than necessary pile of cash that goes unused when you are older will not bring back your youth. You may find that when you are older, you may have health problems, or have less energy, or your significant other may have health problems, or be dead, and you may regret saving more than you needed to save when you were younger when you had the opportunity to more fully enjoy the fruit of your earnings.

In my opinion, it is wise to use an estimate that is neither extremely conservative or extremely optimistic.

Re: How do you "estimate" future portfolio returns?

Posted: Fri Feb 15, 2013 9:49 am
by NYBoglehead
mptfan wrote:
NYBoglehead wrote:The good news is if you use an extremely conservative estimate in your calculation and that conservative growth estimate says you have enough money, you should be in good shape. If you estimate 4% nominal return and that will provide you with enough money for retirement, anything above that will be a bonus.
Of course the opposite is also true... if you use an extremely conservative estimate, you will end up saving much more than you need to save for retirement, and therefore, depriving yourself unnecessarily while you are younger. A bigger than necessary pile of cash that goes unused when you are older will not bring back your youth. You may find that when you are older, you may have health problems, or have less energy, or your significant other may have health problems, or be dead, and you may regret saving more than you needed to save when you were younger.
I think that beats the alternative. If you are 65 and don't have enough money to live comfortably in retirement you're going to be eating a lot of ramen noodles and might be bagging groceries a few hours per week.

I do not believe that saving for retirement means you are depriving yourself in your youth. Furthermore, if I find in retirement that I have too much money (the horror!) I will be more than happy to gift it to my children, give it away to charity, and buy absurdly expensive cars and go on frivolous vacations. Yes, the downside of saving "too much" is indeed scary.

As far as health concerns go, if I've got a lot of money I guess I won't have to burden my family with medical costs.

Re: How do you "estimate" future portfolio returns?

Posted: Fri Feb 15, 2013 9:52 am
by mptfan
NYBoglehead wrote: I do not believe that saving for retirement means you are depriving yourself in your youth.
You misunderstood my point. I did not say that saving for retirement means you are depriving yourself in your youth. I said that using an extremely conservative estimate of future returns may result in unnecessary deprivation in your youth, beyond what is necessary to have a secure retirement.

Re: How do you "estimate" future portfolio returns?

Posted: Fri Feb 15, 2013 10:21 am
by dbr
nisiprius wrote:
They all suck. but the kind you are using is just useless. To begin with, "we'll tell you the future if you'll tell us the future first" is a total cop-out. Second, there's no excuse for asking for two separate hard-to-predict numbers--returns AND inflation--when, in fact, everything can and should be done in real dollars and "real" (inflation-adjusted) returns. Real returns are bad enough to predict, but at least they are a bit more stable than nominal (raw-dollar-number) returns.

We are on the same page here, but don't forget if fixed annuities/pensions are in the picture you HAVE to model inflation somehow.


As for a better retirement calculator, I don't know what's available to you and I haven't used one myself in a while, but I think you want to be looking for "Monte Carlo simulators" that show you the potential range of outcomes that would have occurred, based on past experience. There are many philosophical problems with these, too, but they're less bad than "tell me what your investments will return and I'll tell you what your investments will return" calculators. Three examples of the Monte Carlo genre are Fidelity's Retirement Income Planner; Financial Engines; and FIRECalc. Often you can get access to such a calculator at no cost through your employer's retirement plan, if you have one. FIRECalc is just plain free, I think.

Don't forget FireCalc is not exactly a Monte-Carlo simulator in the sense that it rather simply tabulates the results of actual retirements as they would have played out historically rather than tabulating repeated samples from hypothetical distributions. I wonder of the MC programs that don't require the user to input the future to predict the future haven't simply made their own "arbitrary" prediction of the future. One can also refer to Rick Ferri's 30 year forecast and dump the onus on him. I like FireCalc because I would rather believe that the next thirty years will be like the last hundred years than believe that I can guess what parameters should go into the calculation. All previous caveats about this problem that have been mentioned in this and all other threads continue to be underlined for emphasis.

Re: How do you "estimate" future portfolio returns?

Posted: Fri Feb 15, 2013 10:32 am
by KyleAAA
I think a 1% real return on a 40/60 portfolio is EXTREMELY conservative. I think the 3-4% range is more realistic. I didn't look that up, but it feels about right. If a 40/60 portfolio only yields 1% real over the next 30 years, I think a LOT of Americans are in trouble.

Re: How do you "estimate" future portfolio returns?

Posted: Fri Feb 15, 2013 1:00 pm
by nisiprius
dbr wrote:I wonder of the MC programs that don't require the user to input the future to predict the future haven't simply made their own "arbitrary" prediction of the future.
Sure, but the big thing is that they at least incorporate a range of predictions and show a range of outcomes. Anything that shows you a range is better than anything that shows you a single number.

Re: How do you "estimate" future portfolio returns?

Posted: Fri Feb 15, 2013 4:51 pm
by Sidney
KyleAAA wrote:I think a 1% real return on a 40/60 portfolio is EXTREMELY conservative. I think the 3-4% range is more realistic. I didn't look that up, but it feels about right. If a 40/60 portfolio only yields 1% real over the next 30 years, I think a LOT of Americans are in trouble.
Assuming 3-4% real as an average, what do you see as the components -- what % real on bonds and what % real on equity to get to 3-4%?