Won the game? Where to set AA?

[Animorph]

Quote:

Originally Posted by daylatedollarshort If WR is the withdrawal rate, does that withdrawal rate include drawing down the portfolio principal? Sorry, if I am being a bit dense, I usually use my own spreadsheets so I don't know a lot about Firecalc.

Keep in mind that you only "draw down principal" in bad years, and restore it in good years, including inflation. In the median cases, given a reasonable equity allocation, the value of your principal goes up in real terms by the end of the retirement period. Or if your portfolio throws off enough dividends to cover your withdrawals, you can leave principal untouched.

[audreyh1]

Quote:

Originally Posted by Lsbcal Good observations Audrey. I was partially retired in 2003 and fully about 2006. At the start of 2008 our portfolio had grown nicely as shown here: You can see that 2008 was a nasty surprise as we were only at 55% equities. We had some expenses too which we don't now have: 1) DS in college 2) high medical costs 3) no Social Security 4) much higher taxes So this just illustrates that the picture is very specific to one's life circumstances.

Yes, I can see how you would want to avoid a repeat of the 2008 experience where you were 15% ahead and then suddenly 15% behind. I think our experience was something like 32% ahead of starting value in real terms in 2007, then 12% below our starting value in 2008. Same equity exposure as you.

I sometimes forget that we didn't start fully drawing down on our retirement fund until this past January, so even though we had a couple of nasty bears we were rebalancing and allowing the retirement fund to recover, so that by the end of 2010 we had more than recovered in real terms even after taxes paid.

Quote:

Originally Posted by Lsbcal To me this feels counter intuitive. Does it survive a 1930's scenario? Even if it does, will anyone really put it into action should such a terrible set of years occur?

That results of their study are counterintuitive. But they explain it well. Given your penchant for analysis, I really think you should study their article and recommendations.

Actually - if you treated it like they recommend, cut back to 40% equities, and then let equity percent gradually drift higher, you would be doing close to what they recommend for new retirees.

I don't think I'm comfortable dropping below 50% equities in the long run for various reasons, a major one in the short run being the assumption that fixed income will not show any positive real return. Our retirement fund is now about 60% above starting value in real terms (100% nominal) so that may be why I'm fairly comfortable with the volatility.

[daylatedollarshort]

Quote:

Originally Posted by Animorph Keep in mind that you only "draw down principal" in bad years, and restore it in good years, including inflation. In the median cases, given a reasonable equity allocation, the value of your principal goes up in real terms by the end of the retirement period. Or if your portfolio throws off enough dividends to cover your withdrawals, you can leave principal untouched.

I was curious if the posters here are adjusting their withdrawal rates to draw down their principals to zero at the end of 40 years or whatever time frame they are putting in for retirement or leaving their inflation adjusted principals in tact.

[audreyh1]

Quote:

Originally Posted by daylatedollarshort I was curious if the posters here are adjusting their withdrawal rates to draw down their principals to zero at the end of 40 years or whatever time frame they are putting in for retirement or leaving their inflation adjusted principals in tact.

The FIREcalc model of SWR in general assumes drawing down the portfolio to 0 although you can give it a terminal minimum if you wish.

It can also model the % of remaining portfolio where it defines "failure" as going below initial portfolio value inflation adjusted over the time frame. That comparison is apples and oranges to the original SWR method which defines failure as going to zero within the time frame specified.

In general, these withdrawal models do assume you "tap into principal" at times when needed. Otherwise recommended withdrawal rates would be much, much lower. So low someone might not be able to retire because they can't amass a large enough portfolio.

[Lsbcal]

Quote:

Originally Posted by audreyh1 ...(snip)... That results of their study are counterintuitive. But they explain it well. Given your penchant for analysis, I really think you should study their article and recommendations. Actually - if you treated it like they recommend, cut back to 40% equities, and then let equity percent gradually drift higher, you would be doing close to what they recommend for new retirees. I don't think I'm comfortable dropping below 50% equities in the long run for various reasons, a major one in the short run being the assumption that fixed income will not show any positive real return. ...

I'll check out the Pfau study. I agree that going below 50% equities right now is probably not a great move. When real rates get towards historical rates, then I'll revisit this.

[Lsbcal]

Quote:

Originally Posted by NW-Bound The runs I made was without any additional income sources. The fixed income part was "Long Interest Rate" and the equity portion was Total Market. I just now reran the two 40% equities cases with the WR shown in red in the table I posted, but with 5-year Treasury instead. The WR numbers turned out better, at 2.00% and 3.08%. I am currently drawing 3.5%, and expect our SS to replace at least 1% WR, but have not bothered to figure it out exactly. Of course that would also depend on when we start SS. And our portfolio has foreign stocks and is not as simple as the test cases. These do provide an understanding of how equities are needed in a portfolio, however.

Likewise our portfolio is much more complicated then the FIRECalc options allow. Also I do some market timing in a very controlled way. We have international equities too. Also a fair component of corporate bonds and EM bonds, etc. in our intermediate bond funds. But I would move the nominal bond funds to Treasuries based on a model I use for equity selloffs. If TIPS ever get to historical levels, I'll probably replace the nominal bond funds.

Still the FIRECalc results are good to exam on a relative basis, changing equity mixtures and affect on spending levels available in down market environments.

[daylatedollarshort]

Quote:

Originally Posted by audreyh1 The FIREcalc model of SWR in general assumes drawing down the portfolio to 0 although you can give it a terminal minimum if you wish. It can also model the % of remaining portfolio where it defines "failure" as going below initial portfolio value inflation adjusted over the time frame. That comparison is apples and oranges to the original SWR method which defines failure as going to zero within the time frame specified. In general, these withdrawal models do assume you "tap into principal" at times when needed. Otherwise recommended withdrawal rates would be much, much lower. So low someone might not be able to retire because they can't amass a large enough portfolio.

I am looking at the TIPS historical rates of return and many issues bought at auction would have provided a ~2% relatively risk free real return over inflation while keeping an inflation adjusted principal intact for comparison sake:

Daily Treasury Real Yield Curve Rates

I guess I am a total wuss on the risk front because I look at those returns and think, hey good enough without the gut wrenching ups and downs and we don't ever spend down the portfolio principal.

Added -

If we spent down principal then we could get a safe withdrawal over 2% just with TIPS. I don't get all the articles on 2% being the new safe withdrawal rate because you can almost get that now with longer term TIPS on yield alone and keep the inflation adjusted portfolio intact.

[Katsmeow]

Quote:

Originally Posted by daylatedollarshort I was curious if the posters here are adjusting their withdrawal rates to draw down their principals to zero at the end of 40 years or whatever time frame they are putting in for retirement or leaving their inflation adjusted principals in tact.

I don't have drawing principal down to zero as a goal so I would never purposefully increase withdrawal rates (i.e. increasing spending) to achieve it. However, I am willing to have principal be zero at the end of the plan.

Firecalc in its basic form finds a portfolio successful if at the end of the plan you have more than zero. Being close to zero counts as a success. As Audrey mentions you can run various options that define success differently.

That said, in most instances with successful portfolios the odds are that you end up with more money at the end than you had at the beginning.

I just did a Firecalc run the other day. One that wasn't particularly conservative in terms of spending. The lowest successful run ended up with about $29k at the end of the plan, the next lowest had an ending about of about $159k. There were 113 historical periods in the run. In 21 of them the ending value was less than the plan starting with. In all the others, the ending amount was more than the plan started with. The median run was about double the starting amount.

All of that said, I don't believe in using Firecalc to determine one's actual withdrawal rate. It is simply data and a guideline. It is clear that the most risky time period is the first few years of retirement. Like most everyone, if things went south the first few year I would certainly be cutting many discretionary expenses (or deferring them).

Quote:

Originally Posted by daylatedollarshort I guess I am a total wuss on the risk front because I look at those returns and think, hey good enough without the gut wrenching ups and downs and we don't ever spend down the portfolio principal.

I think that is fine for people who can do that and who have a large enough portfolio or enough inflation protected income that they don't have to worry about losing purchasing power through inflation. But, that isn't us. If we tried to come up with a plan that would withstand 30 years of inflation without using principal (in some years) we wouldn't be able to do it without living a lifestyle that would not appeal to us. So we are OK with using principal in some years.

[NW-Bound]

Ah, talking about more complex portfolios than the basic two-component default portfolios, one can play with the more sophisticated options in FIRECalc, and get some better results.

Caveat: I do not know how FIRECalc exactly defines US small cap, small cap value, etc..., and also where the author obtained historical performance values. Also, we do not know how well future performance can be predicted from the past.

If I had access to the FIRECalc computation engine, I could wrap an optimizer around it, and write an optimization program based on the common Simplex algorithm. But as I can only play with FIRECalc by varying manual inputs, I can obtain a better portfolio than the simple 2-component portfolio, but cannot claim it to be the globally optimal (meaning the solution I present may be simply a local maxima).

Before we claim something is better, we must define "goodness". Following in the OP's spirit, I will say that portfolio A is better than portfolio B if A allows a higher WR rate during a 30-year retirement period, while also not drawing the portfolio below a certain threshold, such as 40% or 50% of initial portfolio value,

With that definition, I have found that the following portfolio using FIRECalc model is superior to the simplistic 2-component portfolios.

NW-Bound Optimal Portfolio
US Microcap: 6%
US Small Cap: 6%
US Small Cap Value: 6%
US Large Cap Value: 6%
S&P 500: 16%
US Long-term Treasury: 0%
US Long-term Corporate Bond: 0%
US 1-month Treasury: 60%

Note that the above portfolio is 40% in US equities with an emphasis on small stocks, and the rest of 60% in short-term fixed income almost like cash!

If you take the 40% equity simplistic portfolio, using a $1M initial value and a WR of $30,791 (see my post #20 above), you will get this following FIRECalc result.

The lowest and highest portfolio balance throughout your retirement was $400,930 to $4,944,019, with an average of $1,454,631. (Note: values are in terms of the dollars as of the beginning of the retirement period for each cycle.)

That $400,930 is the minimum value that we desire.

Now, keeping the same WR of $30,791, and use the more complex portfolio as I describe above, you will get

The lowest and highest portfolio balance throughout your retirement was $619,352 to $3,054,433, with an average of $1,448,184.

Note that the 2nd portfolio brings up the minimum, while reducing the maximum such that the average value is the same. In other words, the more complex portfolio has less volatility, and provides the same average return.

All that for 60% in almost cash! Can you believe that?

PS. Alternatively, if you can settle for $400K minimum portfolio value instead of the $619K above, you can draw $33,081 instead of $30,791. Resultant FIRECalc summary follows.

The lowest and highest portfolio balance throughout your retirement was $464,486 to $2,868,181, with an average of $1,296,244.

Note that the worst case end balance is $464K, but the lowest point in the 30-year period is $400K, meaning that it bounces back.

[samclem]

Quote:

Originally Posted by NW-Bound If the idea is to maximize the lowest value your portfolio might reach during a 30-year retirement while still drawing a decent WR, then a 40% equity is the optimal point, as the OP indicated. . . . So, I made a bunch of runs with minimum portfolio value and % equities as the two parameters, in order to see what WR would go with these values. Below is the result. Having zero equities is not as safe as one would think. Red WRs are the highest for the respective minimum portfolio values of 50% and 40% of initial portfolios. Note that the WR does not change much between 40% equity and 50% equity. I would go with 50% as the upside is better, while the downside is about the same as 40% equity. % Equities Minimum Portfolio Value Annual Withdrawal 0% 60% 0.23% 0% 55% 0.92% 0% 50% 1.06% 0% 40% 1.31% ---- ---- ---- 30% 60% Not possible 30% 55% 1.33% 30% 50% 2.14% 30% 40% 2.74% ---- ---- ---- 40% 60% Not possible 40% 55% 1.22% 40% 50% 2.15% 40% 40% 2.97% ---- ---- ---- 50% 60% Not possible 50% 55% 1.06% 50% 50% 2.13% 50% 40% 2.89% ---- ---- ---- 55% 60% Not possible 55% 55% 0.98% 55% 50% 2.12% 55% 40% 2.82% --- --- --- 60% 60% Not possible 60% 55% 0.89% 60% 50% 2.01% 60% 40% 2.76% PS. I added the 55% and 60% equity cases to address Audrey's post below.

Just a note: Nice charts, and you've stated your assumptions/conditions well, but there's one highly conservative assumption that might be worth emphasizing: The table represents the "zero percent failure rate" cases. For example, for 50% equities and a min acceptable portfolio value of 50% of starting value and a 2.13% WR, no portfolio in the scores of FIRECalc 30-year runs ever dropped below this value.
This, plus the "X % of starting amount adjusted for inflation" rather than the "X% of year end value" withdrawal method, plus the "rebalance back to target AA rather than draw down bonds/fixed in down years" would, I think produce a more pessimistic WR than many people might be able to achieve. I think some of these variables might also be optimized with a higher % of equities than the assumed case.
Anyway, great discussion topic.

[NW-Bound]

Yes, we all know that this represents a very conservative WR for scroogy or chicken-hearted retirees who do not want to see their portfolio ever dropped to 50% or 40% of the initial value.

It also means that they can be sure to leave a sizeable inheritance to their heir, if they are willing to cut back from the common 4%WR to 3%.

And all the above applies to the historical worst case. In many other periods, they would leave behind assets greater than what they started their retirement with, often greater than 2X or 3X, and at inflation-adjusted values no less, meaning in nominal terms it would be even greater.

Quote:

Originally Posted by samclem Just a note: Nice charts, and you've stated your assumptions/conditions well, but there's one highly conservative assumption that might be worth emphasizing: The table represents the "zero percent failure rate" cases. For example, for 50% equities and a min acceptable portfolio value of 50% of starting value and a 2.13% WR, no portfolio in the scores of FIRECalc 30-year runs ever dropped below this value. This, plus the "X % of starting amount adjusted for inflation" rather than the "X% of year end value" withdrawal method, plus the "rebalance back to target AA rather than draw down bonds/fixed in down years" would, I think produce a more pessimistic WR than many people might be able to achieve. I think some of these variables might also be optimized with a higher % of equities than the assumed case. Anyway, great discussion topic.

[daylatedollarshort]

For an all TIPS portfolio yielding 1.5% real, what is the safe withdrawal rate over 40 years, spending down the portfolio to zero at the end of the 40 years?

[audreyh1]

Remember that 40% or 50% of initial value is in real terms, so the retiree might not notice that. What would be noticeable would be it dropping below those values or the initial value in nominal terms.

[NW-Bound]

If you can get 1.5% real return net of all investment costs, then you will draw your stash down to 0 after 40 years with an initial WR of 3.29%, which can be adjusted with inflation.

Quote:

Originally Posted by daylatedollarshort For an all TIPS portfolio yielding 1.5% real, what is the safe withdrawal rate over 40 years, spending down the portfolio to zero at the end of the 40 years?

[Lsbcal]

Quote:

Originally Posted by NW-Bound ...(snip)... With that definition, I have found that the following portfolio using FIRECalc model is superior to the simplistic 2-component portfolios. NW-Bound Optimal Portfolio US Microcap: 6% US Small Cap: 6% US Small Cap Value: 6% US Large Cap Value: 6% S&P 500: 16% US Long-term Treasury: 0% US Long-term Corporate Bond: 0% US 1-month Treasury: 60% Note that the above portfolio is 40% in US equities with an emphasis on small stocks, and the rest of 60% in short-term fixed income almost like cash! ...

Interesting portfolio NW-Bound. So I assume you are 60% in the 1-month Treasuries paying 0.06% now.

I do value tilt our portfolio but am willing to shift to growth based on an algorithm. We see that Wellesley and Wellington have done quite well with equities that tend towards the value side and with intermediate bonds. They both don't seem to need small caps.

Mid-cap value funds have done very well over the last 20 years -- not generally acknowledged. Micro caps ... too hot for my tastes.

[daylatedollarshort]

Quote:

Originally Posted by NW-Bound If you can get 1.5% real return net of all investment costs, then you will draw your stash down to 0 after 40 years with an initial WR of 3.29%, which can be adjusted with inflation.

Okay, thanks. How does that compare to the safe withdrawal rate for a 50/50 stock / bond portfolio spent down to zero?

[NW-Bound]

No, not 60% but only 26% now, in cash or cash equivalent paying around 1.5% as we speak.

And please note my caveat as posted earlier.

Quote:

Originally Posted by NW-Bound Caveat: I do not know how FIRECalc exactly defines US small cap, small cap value, etc..., and also where the author obtained historical performance values. Also, we do not know how well future performance can be predicted from the past.

[NW-Bound]

With a simplistic portfolio of 50% 5-yr Treasury and 50% Total Market, you can draw $34,219/yr from a $1M portfolio. In the worst case, you would end up with $0 after 40 years. In the best case, you will leave behind almost $7M.

FIRECalc Summary:

The lowest and highest portfolio balance throughout your retirement was $319 to $6,951,646, with an average of $1,727,040. (Note: values are in terms of the dollars as of the beginning of the retirement period for each cycle.)

Quote:

Originally Posted by daylatedollarshort Okay, thanks. How does that compare to the safe withdrawal rate for a 50/50 stock / bond portfolio spent down to zero?

[daylatedollarshort]

Quote:

Originally Posted by NW-Bound With a simplistic portfolio of 50% 5-yr Treasury and 50% Total Market, you can draw $34,219/yr from a $1M portfolio. In the worst case, you would end up with $0 after 40 years. In the best case, you will leave behind almost $7M. FIRECalc Summary: The lowest and highest portfolio balance throughout your retirement was $319 to $6,951,646, with an average of $1,727,040. (Note: values are in terms of the dollars as of the beginning of the retirement period for each cycle.)

Okay, thanks again. I guess then that Firecalc's lowest SWR is much more than a lot of the revised SWR rates being thrown around these days in various papers.

I have read about 1.8% being the new SWR, but that is close to the real return on longer term TIPS alone even these days, keeping the inflation adjusted principal intact.

[NW-Bound]

Anyway, speaking of "optimizing" a portfolio, one must first ask what it is that he wants to optimize. In other words, one must decide what is the "goodness" quality that he is after.

The first thing I learned more than 35 years ago in Optimal Control, a graduate level engineering class, was that there were infinitely many optimal control solutions. Why? Because there are infinitely many optimal criteria one can define.

Back on investment strategies, if one wants the lowest volatility, the solution is going to be vastly different than one to provide the maximum expected return, the latter can leave one broke in the worst case. And of course there are many shades of gray in between.

I think that the OP criteria of "worst case minimal portfolio value" is quite practical, as most people want to give up some of the upside to protect the downside. In engineering, it's called the "minimax" problem, where one wants to maximize the minimum, in contrast with maximizing the average, or god forbid, maximizing the maximum (going for broke!).