# The Safe Water Reservoir: (SWR): An analogy?

#### Zorba

##### Recycles dryer sheets
A mining town on the edge of the desert has built a dam to hold the
water it needs to survive. Each year the rains come. In some years
the rainfall is enough to replenish the reservoir, but in other
years evaporation depletes the water level. How much water can
the town draw from the reservoir each year to be reasonably certain
of not running it dry?

Analogies can be useful. They help us identify the essential elements
in any problem. Would the SWR experts here have a look at this toy
model and tell me what what is common and what is missing?

Zorba

Hmmm...
I smell a troll. zorba ~ hocoloco?
No. Not even close. I appreciate that everyone is on red alert these days
but you have just found the fastest way to chase away new contributors to
this board! I'm interested and I'm a newbie so please be gentle

I am trying to get some intuitive understanding of what the main factors
are in a SWR. Backtesting (with historical data) or simulation (via
Monte Carlo modeling) don't do it for me.

What if a bunch of people urinated in the reservoir?

Zorba:

I will give you the benefit of the doubt re: the troll issue.

FWIW, I think that the analogy you proposed is reasonable. The chief holes in it are:

- Every year the mining town's population grows by about 3%. If the reservoir isn't periodically expanded, eventually the annual rains won't come close to replenishing the water supply. I am referencing inflation risk here.

- The population of the town has the ability to attempt to influence rainfall amounts. If the town's attempts to boost rainfall are successful, ithe reservoir could receive much more than average annual rainfall. If they fail, the town could receive much less rainfall than required. I am referring to how much portfolio risk one can afford to take.

- The town can follow water use reduction strategies to use less water in times of extended drought (no watering the lawn, etc.). I am referring to cutting one's expenses/withdrawals in extended bad times.

Anyone else?

No. Not even close.  I appreciate that everyone is on red alert these days
but you have  just found the fastest way to chase away new contributors to
this board! I'm interested and I'm a newbie so please be gentle

I am trying to get some intuitive understanding of what the main factors
are in a SWR.  Backtesting (with historical data) or simulation (via
Monte Carlo modeling) don't do it for me.

Not sure what your objective is? - If the tools that you mention don't do it for you, and you are looking for a truly 'safe' WR - There isn't any. I don't care whether you take only 1% a year, I can concoct a scenerario where you'll be broke in 5 years.

Bottom line is that there is no way to predict the future. FireCalc is a good enough guideline. Anything else is just mental masturbation.

One point I can see where the analogy breaks with the portfolio and SWR system is that the amount of water in the reservoir doesn't affect how much falls from the sky. In the portfolio how much is there affects how much you gain. To keep to the analogy the portfolio would have to gain some specific set amount of dollars in a year rather than as a percentage of dollars in the portfolio. Just about all analogies break at some point - it's important to realize where.

Zorba:
Brewer beat me to the punch.
An interesting analogy.
So what's the answer? (We're busy retiring here, haven't got all day!)

Not sure what your objective is?

I can see the need for something like this. The financial concepts behind portfolio returns, withdrawals, inflation, etc. can be difficult for some to understand and even for those that understand it some can have difficulty "visualizing" it. When I was a physics undergrad I would always try to build a mental picture of how various systems worked - it gave me a better "intuitive" understanding beyond the equations.

Thanks Hyperborea. As a newbie I am trying to reduce this problem down
to its main elements so that I can intuitively understand what all of you
are talking about. In other words, what are the dominant factors that
influence the SWR?

To first order I thought we could discount inflation in my analogy
since both the annual returns (rainfall) and SWR (withdrawal) are
inflation adjusted. However, Inflation does introduce a correlation
that does not exist in my analogy. The other problem I see is that a
real SWR must depend somewhat on the age of the person. The limiting
case of large time is like an endowment.

The annual returns (aka rainfall) cannot be influenced. They are a
stochastic variable and cannot be predicted in advance. You might know
Rainfall like stock market returns have some mean reversion but that

What about the variance of the returns (uh, rainfall)? Whats better
for a SWR: a modest return with low variance, or a high return with
high variance?

What about the variance of the returns (uh, rainfall)? Whats better
for a SWR: a modest return with low variance, or a high return with high variance?
I would want the modest return with low variance. Predictability may not be accurate, but it would be more accurate with low variance. Now the question is, what would the modest return be? I think a 7% long-term total return with a 4% real return is the target.

Anything under that, you may drown. Anything over that, you will walk on water. I would be happy coasting for 40 years or more.

Analogies can be useful.
So can strawmen. But that doesn't make them relevant to solving the problem-- only to distracting everyone's attention from the real question by trying to build a good simulator.

In other words, what are the dominant factors that
influence the SWR?
I think the most dominant factor is "spending no more money than the portfolio can replace." I'd say that other dominant factors include (1) anything that survives history with greater than 80% or (2) anything that survives Monte Carlo with greater than 80% or (3) predicting the future.

I think that the problem here is that you're over-analyzing the SWR. (Hence the ***** accusation.) Build a portfolio that's 25x your projected expenses, withdraw 4% of the initial amount and adjust that annually for inflation, and get on with your life.

Or try Bud Hebeler's feedback loop in "J.K. Lasser's Your Winning Retirement", and then you won't have to worry about history or methodology or predicting the future or whether a 4.00001% SWR is less survivable than 3.99999%.

Whats better for a SWR: a modest return with low variance, or a high return with high variance?
The one that survives more historical or Monte-Carlo situations, right? These are just a couple of factors in an extremely complex system whose behavior is not easily predicted (if it even can be). But returns can't be "too modest" and variance can't be "too high"-- it all has to be "about right." At which point the over-analysis tendencies kick in again...

I'm trying to be really open minded on new users. I did notice that most of the suspected 'hoco loco' variants stopped posting within a day or so of his banning.

There is a progression for most new users though...a "Hi I am..." post, a few questions, etc. I'm automatically suspicious of anyone who has a first post on the ever ridiculous 'safe withdrawal rates' or assailing firecalc. I'm even more suspicious of someone who doesnt ask "who is 'hoco-loco'".

That having been said, I have a solution to this question. The town stops drinking water and drinks beer instead. Its safer. Water is a terrible substance. Fish do unspeakable things in it.

What if it rains so much, that the reservoir overflows? And then you lose all that extra water cause it seeps into the ground?  Or what if it floods the town, and the inhabitants flee or drown?  Then who gets claim on the water?  What if global changes in climate cause all the water in the reservoir to freeze?  I don't get this analogy.

Can you kayak in it?

Well, I suppose I deserved an ass whopping for posting a *****-like
thread. Just for the record, I'm male, mid-40's, married, two kids,
living in the southwest. I have a net worth of \$1.5M+. Still
working... Christ, it does sounds alot like *****, doesn't it.

FIRECalc is a great tool. I've used it a lot but it has left me with
some questions that I am trying to answer. Over-analysis? To each
their own I guess. Some people like to just push the buttons on the
outside of the black box, while other people like to rip 'em apart to
see whats inside.

FIRECalc is a great tool. I've used it a lot but it has left me with
some questions that I am trying to answer.

Why don't you ask the specific questions? You'll probably get very specific answers.

Damn - And I was just about to build a tail water trout fishery just below the dam!

Q1: How important is the variance of the real return (inflation
adjusted) to the value of the SWR? (This is the rms variation in
rainfall in my lame analogy.)

Q2: If variance is important to SWR are there accepted strategies to
reduce the variance? I suspect you'll just tell me to go read
Bernstein but when people talk about asset allocation they rarely talk
about what their mix does in terms of return *and* risk.

Q3: How important is the age of the person to the SWR? Instead of a
fixed portfolio lifetime, FIRECalc could use mortality tables to give
some confidence interval. Doesn't someone in their mid-40's have a
different SWR than someone in their mid-60's?

Thanks.

Q3: How important is the age of the person to the SWR? Instead of a
fixed portfolio lifetime, FIRECalc could use mortality tables to give
some confidence interval. Doesn't someone in their mid-40's have a
different SWR than someone in their mid-60's?

Well, I'll try and deal with this one - Age is very important to SWR. When FireCalc asks you to input the Life of the Portfolio, this is essentially where you decide how long you are going to live. If you are 60 you might input 40 or 45 years. If you are only 40, you might input 60 or 65.

So, yes different aged folks have different SWR's.

Q1: How important is the variance of the real return (inflation
adjusted) to the value of the SWR? (This is the rms variation in
rainfall in my lame analogy.)

Q2: If variance is important to SWR are there accepted strategies to
reduce the variance? I suspect you'll just tell me to go read
Bernstein but when people talk about asset allocation they rarely talk
about what their mix does in terms of return *and* risk.

Q3: How important is the age of the person to the SWR? Instead of a
fixed portfolio lifetime, FIRECalc could use mortality tables to give
some confidence interval. Doesn't someone in their mid-40's have a
different SWR than someone in their mid-60's?

Thanks.

A1. It's pretty important. If you take a look around for 10 minutes on http://www.retireearlyhomepage.com/ you can find discussions of how the portfolio structure affects the SWR, examples of different portfolio structures and their associated SWR's, etc. For the record, I believe the canonical 4% assumes a roughly 75% S&P 500 / 25% commercial paper mix.

A2. Actually what you want to look at is the variance or portfolio makeup plotted against the SWR and pick the portfolio makeup / SWR local maximum. Again, there are graphs of this stuff over at the website mentioned in A1.

A3. Actually it depends on the person's life expectancy, which does in turn depend on one's age (and health, and family history...). The crude answer is to pick a payout period that roughly matches your life expectancy. Another approach is outlined at http://www.retireearlyhomepage.com/swrlife.html.

malakito

I used to live next to a reservoir. No worries about water...I have a pump and a hose. I did in fact kayak in it extensively.

What used to make me chuckle was that when we had heavy early rains, they'd start 'preemptively' letting water out just in case we had more heavy rains they didnt want to have to let it out later. Or some such other dumb reason. Inevitably, after one of these substantial water dumpings, we'd get no water for the rest of the rainy season. Reservoir looked llike a big bomb went off and left a crater, in which a small amount of muddy rain water had collected.

I would refer to that state as the "Safe Water Reservoir After Use Of Hoco Magic Tool", which would make a dumb and fairly useless acronym of SWRAUOHMT. But feel free to use it anyhow.

No. Not even close.  I appreciate that everyone is on red alert these days
but you have  just found the fastest way to chase away new contributors to
this board! I'm interested and I'm a newbie so please be gentle

I am trying to get some intuitive understanding of what the main factors are in a SWR.  Backtesting (with historical data) or simulation (via Monte Carlo modeling) don't do it for me.

To understand SWR you need at least a passing aquaintance with arithmetic. If you can't fathom a simple historical backtest of the data, the best thing to do is just sit back and dream about retirement -- that may be as close as you'll get to it.

intercst

We have an excellent supply of clean, cool water in Canada.

Maybe if you'd stop screwing us around on beef and softwood lumber, we'd sell you some.

.......and there is an awful lot of oil in the tarsands.

Thanks Hyperborea. As a newbie I am trying to reduce this problem down
to its main elements so that I can intuitively understand what all of you
are talking about. In other words, what are the dominant factors that
influence the SWR?

Hi Zorba,

It is a combination of factors.

It is the likely return from each asset class and how the mix of different assets work together. If they all move down together and you have to sell 4% when they have all lost 80%, then your capital is in trouble. If all the assets sit 80% underwater from where you retired at, you'll eat through the remainder of your capital in no time. To use actual numbers, if you needed \$4 to live each year, you started with \$100 capital and it fell to \$20, you would still need to sell \$4 each year (plus inflation) to live and you would eat thru your remaining capital in less than five years.

There are some mitigating factors.

Your portfolio has a cash dividend yield. Depending on your mix of assets, this reduces the impact on capital in bad years because you have to sell less. So if your portfolio had \$2 a year in cash dividends, you lived off \$4, you need only sell \$2. In the above example, this would double how long your money could last. So considerably better.

What some investors have done in the past is add between 40-60% on standard treasury bonds to provide a cash income stream of 4%-7%. The real returns pre-tax on the bonds varied enormously but were around 2%-3% in the US. Bonds didn't make a great investment for returns but they provided a yield which reduced the damage when stocks fell. You could partially live off the dividends and sell some bonds instead of selling your stocks. Adding 20% in bonds rather than holding 100% in stocks quite often offered better survival rates and higher withdrawal rates even though you're adding a lower returning asset class (bonds). Kinda of counterintuitive. The reason was partly the income which avoid selling stocks when they were underwater and partly because sometimes bonds rose when stocks fell.

Today, investors thankfully have more options than merely stocks and bonds. Global REITs now provide the ability to link capital returns to inflation and provide higher yields than bonds too. These are becoming increasingly attractive as a real alternative.

Timber provides cash from the timber harvests each year 4%-5% and inflation increases in the value of standing timber (plum creek & rayonier are US REITs).

TIPS & I-Bonds provide direct linkage to inflation which beats the old form of treasury bonds that lost out with inflation. They don't provide much in the way of yields unless you plan to spend the capital down over 40 years in which case the returns are around 3.4% for TIPS today. Over 40 years you sell 2.5% of capital and use part of your interest. This may be suitable to someone risk averse for a small part of their portfolio. A 60-year old retiree who has a max life expectancy of 100 might find this appealing. Your other assets are not structured to spend the capital down leaving you with plenty if you do get past 100.

Besides these, the ability to balance a stock allocation with small cap, value stocks, int'l and emerging market stocks provides additional diversification within the global equities asset class itself. As the last five years showed, US small cap did well, US large cap did not. Emerging did well, Asia large cap did not. And so on.

Taken together, these assets can be used to reduce the overall portfolio volatility, increase the cash dividends (depending on how much you put into high cash yielding investments) and greatly smooth out results. This in turn reduces the loss on returns caused by needing to live every year regardless of whether the stock market did well that year. Hopefully you can tell from this run-thru that investing in higher returning asset classes is only half the battle when living off investments. The other half is getting the mix of assets right, balancing the goal of higher long-run returns (assets like ScV) with other assets that pay higher dividends and give more assurance of having enough to live off. Going for broke can actually make you broke, whereas some caution serves you better. The goal is different from the accumulation phase where you want do build assets not provide income, year-to-year consistent returns or any of that. Different lifecycle, different needs.

Let me know if that is helpful for you or if you have any questions. Happy to help.

Petey

TIPS & I-Bonds provide direct linkage to inflation

Small correction: TIPS and I-Bonds provide direct linkage to the CPI. CPI <> inflation, depending on where you live and what your lifestyle is. Many experts claim actual inflation is 1-1.5% more than CPI for the average person. CPI comes nowhere near covering inflation for me in Northern CA with my lifestyle. In the last 2 years, beef has gone up more than 35%, gas has gone from ~1.65 to 2.45, milk and other dairy products are up 25%, and housing has doubled in the last 6 years.

If the 'experts' are right and CPI understates inflation by 1%, your life 25 years from now on an all-tips diet wouldnt be so tiperific.

Q1: How important is the variance of the real return (inflation
adjusted) to the value of the SWR? (This is the rms variation in
rainfall in my lame analogy.)

Q2: If variance is important to SWR are there accepted strategies to
reduce the variance? I suspect you'll just tell me to go read
Bernstein but when people talk about asset allocation they rarely talk
about what their mix does in terms of return *and* risk.

Q3: How important is the age of the person to the SWR? Instead of a
fixed portfolio lifetime, FIRECalc could use mortality tables to give
some confidence interval. Doesn't someone in their mid-40's have a
different SWR than someone in their mid-60's?

Thanks.

Hopefully I've covered question 2 from my POV.

With variance of returns, it depends on how it affects your withdrawals. A lot of times variances just mean you sell the thing that hasn't gone down. Sometimes you use cash or bonds instead of selling stocks if they all went down. You have a spread of assets so you have more options. The wider the asset allocation, the more choices available each year.

In terms of different SWR for a 40 year old vs a 60 year old, yes quite likely. It depends on how long an asset will last out. You could plan to take 5% from an asset and it will work in some scenarios and not others.

The S&P 500 broke between 1966-95 when you withdrew more than 4%, mixed with 80% or more of bonds. 1966-82 was a bear market and you got killed if you retired right at the start of it in 1966. If however one had owning large cap value, small cap value, oil & gas, int'l value, emerging markets, real estate securities - a broader mix of assets - the volatility would have been less and the returns higher not only because those assets perform better over long periods but because they do not correlate together (move to the same degree or even in the same direction sometimes).

At a certain point, a specific asset allocation will last long enough for most retirees with the right blend of assets. That certainly won't be 80% US total market, 20% US bonds. Investors are gradually seeing that a higher than 10% allocation to real estate and timber works better than bonds alone. Ibbotson has done studies that show that adding 20% for REITS with a lower allocation to stocks & bonds added to returns and reduced portfolio volatility, for instance.

Petey

Small correction: TIPS and I-Bonds provide direct linkage to the CPI.  CPI <> inflation, depending on where you live and what your lifestyle is.  Many experts claim actual inflation is 1-1.5% more than CPI for the average person.  CPI comes nowhere near covering inflation for me in Northern CA with my lifestyle.  In the last 2 years, beef has gone up more than 35%, gas has gone from ~1.65 to 2.45, milk and other dairy products are up 25%, and housing has doubled in the last 6 years.

If the 'experts' are right and CPI understates inflation by 1%, your life 25 years from now on an all-tips diet wouldnt be so tiperific.

Very true, TH.

I was talking more in the general idea sense of it. To convey the idea of what the products do because I was not sure of the original poster's level of investment knowledge. Also there are usually some lurkers who benefit from posts, build up knowledge reading over a period of time and only then feel able to join the discussion. Lot of it will go over people's heads - sure did me when I first joined some other boards.

I agree the US inflation numbers do not seem correct. They seem about right in the UK. Perhaps lagging 0.5% if anything but not too much out.

Petey