Missing the days of Dow 7000?

[audreyh1]

For the record, when doing spreadsheets to try to understand how asset allocation and rebalancing may provide benefit, it doesn't help to just look at the long-term returns. In a generally rising market, rebalancing will not outperform holding the riskier asset classes. Maximizing returns alone is NOT the goal of asset allocation and rebalancing.

What you want to study is the risk-adjusted return, which looks at the volatility (i.e. risk) as well as the performance. What asset allocation rebalancing is trying to do is to significantly reduce the volatility of a portfolio while capturing most of the performance.

It might just be easier to read the works done on Modern Portfolio Theory and The Efficient Frontier. These show the curves of performance versus risk (volatility in terms of standard deviations) for various asset class combinations rebalanced over various periods of time. You can then find the historical sweet spot on the curve that gives that optimizes performance versus volatility for a given set of asset classes. Modern portfolio theory - Wikipedia, the free encyclopedia

I got most of my understanding through Frank Armstrong's web posting. He now has a book that you can read part of it through google books The Informed Investor: a Hype-Free Guide to Constructing a Sound Financial Portfolio

Even Bogle's comments above indicated he ignored the volatility part of the equation.

Audrey

 

[NW-Bound]

Ah hah!

Audrey, I am glad you bring up MPT. Let's look at the link you provided, namely Modern portfolio theory - Wikipedia, the free encyclopedia.

Scroll down to where it says "Mathematically", then a box to enclose some equations.

This is the same kind of optimization problems we encounter in engineering. In fact, we have worked on problems with several hundreds or thousands of variables, and the subject dynamic system is described by hundreds of pages of equations.

But back to the set of equations on the Wiki page. Where does one get the variances and the correlation coefficients of the returns of different assets? From historical data? Establishing these statistical properties requires a long run. Do we have enough data? And most importantly, we are assuming these stochastic properties are constant. Is it a jump of faith? I claim to be ignorant in these matters and love to be educated by some economists among us.

Comparing this optimization problem to the ones that I deal with, engineers actually have it easier in several ways that I can think off-hand

1) We have manufacturing statistics from hundreds and even thousands of samples of a particular part, hence have very detailed statistics on each part.

2) The performance of each individual part is independent of another part or parameter. If that is not true, their correlation can be modeled and measured very accurately.

3) The effect of a part's performance onto the total system performance is well known, being the laws of physics applied to get those 100's of pages of equations. Using these equations, from the individual part characteristics, we could determine the outcome of the total system very accurately.

So, how could we go wrong?

1) I have seen a supplier changed its manufacturing process and either did not inform us, or we failed to understand its significance. The part new characteristics changed the total system behavior in a subtle and unpredictable way.

2) There were secondary-effect interactions between parts that we were not able to foresee.

3) The equations to propagate each component effect onto the total system were wrong, or most often just inaccurate or incomplete.

4) There are exogenous variables!


So, the equations in the Wiki link do not look that complex, but they bother me. How do they get those sigmas and rhos? And how do we know they stay constant with time?

Note how the assets that were supposed to be uncorrelated got all tangled up in this Great Recession. All except cash that is, causing people to cry out "Deflation".

The optimization result is only as accurate as the values one puts in for the parameters! Please, someone help me understand.

 

[haha]

The optimization result is only as accurate as the values one puts in for the parameters! Please, someone help me understand.

You already do.

Ha

 

[audreyh1]

So, the equations in the Wiki link do not look that complex, but they bother me. How do they get those sigmas and rhos? And how do we know they stay constant with time?

Note how the assets that were supposed to be uncorrelated got all tangled up in this Great Recession. All except cash that is, causing people to cry out "Deflation".

The optimization result is only as accurate as the values one puts in for the parameters! Please, someone help me understand.

That is true. All those sigmas and rhos come from historical data. You can waste a lot of time trying to absolutely optimize an allocation based on historical data and drill down to the nth degree, but that is not really the point. The general principle still applies - that diversification and rebalancing across poorly correlated asset classes provides benefit in terms of risk-adjusted return. It's easy to model the general asset classes and their historical correlation and behavior, and the principles of the theory are demonstrated. History is the best we have.

Perhaps part of what is bothering you with Wiki is that it presents a very high-level overview of the mathematical theory, and you need to study some of the published works to get any sense of how the rubber meets the road. That's why I went to an "implementation guide" rather than studying the math to design my portfolio.

If you start arguing that we can't use the past to predict the future, or that market behavior changes during each decade so how cane we use the past, then for the same reason you have to throw out FIREcalc and have nothing on which to estimate portfolio survival. I think using the models makes way more sense than having nothing. And personally, the general principles make sense to me. I made my choices based on seeing graphs of various asset classes and how they behaved independently and as part of rebalanced allocation over time. Simple! The Frank Armstrong web tutorials provided a lot of these graphs and demonstrated how different mixes behaved over time. You could clearly see how small caps helped and to what degree, how international helped, and to what degree. You could make reasoned decisions about whether you even wanted to bother with them. Sure, asset class correlations will change over time, so you can't be guaranteed the exact same ratios will be optimal in the future, but it seems prudent to expect that they might behave in a generally similar fashion - or at least an approximation. It's better than nothing!

I provided the links for folks who question whether there is any "established science" or theory or studies behind the concepts of asset allocation and rebalancing. The answer is yes, there is an 50 year body of work so if you really want to know more, go study it. And studies and estimates of portfolio survival such as FIREcalc the Trinity Study and it's ilk assume that it is prudent to use investment strategies that have resulted from this body of work - that there some long-term benefit in asset allocation and rebalancing.

So don't get caught up on the details of optimization to squeeze out the last bit possible. The optimization is kind of like FIREcalc - 95% success rate is probably "good enough" given future uncertainties - no sense it trying to tweak things to ensure that 100% success rate. Just get things in the general ballpark, and that's probably the best you can do anyway.

And even if most asset classes (but not all) misbehaved similarly during the Great Recession, they did not all behave the same way - so there was some benefit with rebalancing.

It seems to me the questions have been - Is there any benefit to rebalancing, are there any studies that have shown historical benefit? And the answer is yes. Go study the stuff if you want to know.

Audrey

P.S. I was also an engineer - bachelors and masters degrees and 20 year career in the high-tech industry.

 

[NW-Bound]

That is true. All those sigmas and rhos come from historical data. You can waste a lot of time trying to absolutely optimize an allocation based on historical data and drill down to the nth degree, but that is not really the point.

Audrey, we actually agree more than you thought. If my post is a bit provocative, it is mainly to stir up interest of the "bystanders", not to challenge you at all. I will confess to being mischievous.

Let's go through through this. Heh heh heh...

All those sigmas and rhos come from historical data. You can waste a lot of time trying to absolutely optimize an allocation based on historical data and drill down to the nth degree, but that is not really the point.

YES! I wouldn't and you didn't.

If you start arguing that we can't use the past to predict the future, or that market behavior changes during each decade so how cane we use the past, then for the same reason you have to throw out FIREcalc and have nothing on which to estimate portfolio survival. I think using the models makes way more sense than having nothing.

NO! It was not my argument. As I said in an earlier post, I totally agree with the saying that although history does not repeat, it rhymes. Firecalc gives us a sense of the beat, so to speak. Way better than nothing. In fact, I love it.

I provided the links for folks who question whether there is any "established science" or theory or studies behind the concepts of asset allocation and rebalancing.

And the opportunity for me to poke fun at people who take it to the N-th degree.

So don't get caught up on the details of optimization to squeeze out the last bit possible. The optimization is kind of like FIREcalc - 95% success rate is probably "good enough" given future uncertainties - no sense it trying to tweak things to ensure that 100% success rate. Just get things in the general ballpark, and that's probably the best you can do anyway.

Music to my ears! I was making fun of the people who use these tools to tweak their portfolios to get a desired 3-significant digit answer, or worrying about their portfolio AA being out-of-whack by a few percents. I do use FireCalc as a rough guide as you do. But I have seen people worrying over little changes in the result. Come on people! Remember that history only rhymes and does not repeat. I would rather spend time watching out for those exogenous variables, any sighting of black swans.

By the way, Dory, the author of FireCalc said something about this. He said to not measure with a caliper what you are going to chop with an axe. He knew that if a computer spits out a number to 4 significant digits, people have a penchant to tweak it to get the last bit.

And even if most asset classes (but not all) misbehaved similarly during the Great Recession, they did not all behave the same way - so there was some benefit with rebalancing...

It seems to me the questions have been - Is there any benefit to rebalancing, are there any studies that have shown historical benefit? And the answer is yes.

NO! That was not my question. Without any math, people already know the intuitive answer: Buy low, sell high. The problem is only in the execution. Here is where we differ. One can buy little, or one can buy a lot. As long as one buys...

The real question is how low is a good low. Hmmm... What was the thread topic again? Something about Dow 7000. Darn... I should have gone "all in" then. No one gave me a software model for "low detection".

P.S. I was also an engineer - bachelors and masters degrees and 20 year career in the high-tech industry.

I know you are a high-tech worker from your earlier posts. My friends who were the goldbugs I often told about were engineers too. You wouldn't believe what some people with masters and Ph.D.s do for their investment. Many people with common sense do much better than what they did. Oh well, it's their money. And they are still my friends.

By the way, people who are rich do not bother nor worry about these equations like we do. Buffet, Soros, and the like know how to "read" those exogenous variables that the academics do not even know where to stuff into their equations. I wish I had their knowledge and their skill. Bet you those aren't taught in school.

Sooo... Do you feel better now, Audrey? I had posted a congrat on your ER anniversary, but you probably saw it already.

 

[FUEGO]

The real question is how low is a good low. Hmmm... What was the thread topic again? Something about Dow 7000. Darn... I should have gone "all in" then. No one gave me a software model for "low detection".

Here's the model that I have given before: Buy a sufficiently diversified portfolio of relatively uncorrelated asset classes with very low turnover and very low expenses. Rebalance when reasonable to maintain your asset allocation. Invest regularly and don't try to time the market.

Guaranteed to get you investing during the absolute lows of the market. Also guaranteed to get you investing at the absolute highs of the market.

My point in the OP is not that I'm waiting for a crash or keeping money on the sidelines hoping for one. Rather I prefer to invest at lower prices we saw in the last year versus the higher prices of today. Really I just miss those days of buying stuff really cheap.

 

[NW-Bound]

...Really I just miss those days of buying stuff really cheap.

Yeah, easy for accumulators to say... People who ER'ed like me can only rebalance.

By going from less than 40% equity in late 08 to 60%, from the bottom in March 09 till now, I have "gained" enough to buy another house like the two I currently own. Wait, it would be even a better one, with the RE market as low as it is. Not that I would want another one. What's for?

Still, since you brought up the Dow 7000, I couldn't help thinking that if I had gone "all in", I could have bought a house like Westernskies just did and paid cash for it too with the gain. Oops, forgot that big would-be short-term cap gain... And weren't I thinking about downsizing for less maintenance?

See how we all vacillate between "greed and fear"... One can just turn the porfolio over to the computer like unclemick did, but what's the fun in that?

Should I be in "fear" mode today, seeing that the Dow is down more than 100 as we speak?

 

[ERD50]

I am enjoying this discussion - thanks to all who are adding to it. I'll have to take time later to fully absorb it all.

Audrey, thanks for mentioning the refs for re-balancing. It's been years since I read those and didn't recall that they went into that level of detail on re-balancing, so I will take a look.

There is one comment that I sorta-kinda don't agree with:

The optimization is kind of like FIREcalc - 95% success rate is probably "good enough" given future uncertainties - no sense it trying to tweak things to ensure that 100% success rate.

I agree that the absolute success % should not be taken too literally since, as has been said, the future will likely "rhyme" with the past, not repeat it. But I think that some people will infer that 95% and 100% are all "within a margin of error" and that we have little confidence that one is actually better than another. That is what I don't agree with.

As I see it, there is significance in the difference between a FIRECALC reported 95% success rate and a 100% rate. One represents a WR that actually did result in failure in 5% of the past scenarios, and one had zero failures in that time. For a 35 year portfolio, that gave me about an 11% delta in spending.

I think we can say with a very high degree of certainty that a WR that is 11% lower will also have a higher success rate in any future scenario. How can that not be? That is what FIRECALC is telling us, and I don't see how we can ignore that. 100% success is better than 95%. With future patterns, it might not by a 5% point boost, but it will be better.

This seems different than the analysis of optimizing an AA re-balancing point. In that case a point that was better in the past may be worse in the future.

So, I shoot for a 100% success rate. Not because I think that is going to "guarantee" success, but because I don't feel comfortable with a WR that I know would have failed in some past scenarios. I can't predict the future, but we can review the past. So while I don't know that the future will not be worse than the past, I at least have confidence that if it is as bad as the past, my portfolio will survive.

The succinct version: With FIRECALC, 100% does not necessarily mean 100% for future scenarios, but 100% is better than 95%.

And if we want to get technical in the measurement analogy, we could talk about absolute and relative accuracy, linearity, quantization (this applies to the FIRECALC success rates, since failures are discreet steps - it sometimes helps to look at the min/max/avg end values), and repeatability (and probably a few others I've forgotten about).

-ERD50

PS: By "better" I mean a higher success rate. There is the legitimate view that the extra years required to boost the portfolio may be "worse" for some than the small added risk. That's a different (and valid) discussion though.

 

[NW-Bound]

I think we can say with a very high degree of certainty that a WR that is 11% lower will also have a higher success rate in any future scenario. How can that not be? That is what FIRECALC is telling us, and I don't see how we can ignore that. 100% success is better than 95%. With future patterns, it might not by a 5% point boost, but it will be better.

This seems different than the analysis of optimizing an AA re-balancing point. In that case a point that was better in the past may be worse in the future.

So, I shoot for a 100% success rate.

Excellent point, ERD50, regarding enhancing FireCalc success rate vs. optimizing the AA of a portfolio.

Going to a lower WR means a higher chance of leaving a big estate behind. If one strives to avoid that and wants to enjoy as much material pleasure in life as one could, now that becomes a different kind of optimization problem. One chooses between "pleasure now and worry sh*tless later" and a more modest lifestyle. There is no computer model for that!

And then, we all are or had been comtemplating the trade-off between working a few more years vs. the risks of deteriorating health in the future. No computer model there either.

I am not that old but have grown less and less desirous of material things, never craving many of those things even in the past when I was younger. So, it works out OK with my goal of minimizing my WR to ensure 100% safety. Of course I have not told my kids about my assets and plans. They have to learn to earn theirs, regardless.

PS. I will not go as far as an engineer who left behind a $4M portfolio, sleeping on the floor of a homeless shelter in his last years, as reported in a recent thread. I have not asked, but suspect my wife wouldn't go along. That man had a WR of 0%!

 

[Running_Man]

All of my close friends in their 20's and 30's that view their portfolios as retirement funds "kept the faith" and continued DCA'ing in. Some older associates in their 40's-60's panicked and stopped contributing or sold most/all their equities. The latter group apparently were not comfortable with their asset allocation (but then again I haven't preached to them).

It definitely feels good to not see half your portfolio vaporize, and watch 3-5-10% drops in one day. Those dow 7000 days may not have left us for good. Time will tell. The economy certainly seems to be rebounding or at least getting worse at a much slower rate. But there remains a lot of hurdles to overcome on the road to recovery. That explains why the markets are still off significantly from highs of a couple years ago.

Do not confuse price with value. The market continues to be overvalued from my perspective from a US investor point of view. The consumption of income needed for the government to operate combined with the low cash payouts of dividends leads me to believe these stocks have a long way to go on the down side. Even though the government is offering $4500 for vehicles worth $1000 even that program has rapidly slowed as people are avoiding debt.

Until all the enthusiam has been drained away from young investors such as yourself the bubble will not truly have been popped. The market could even soar on optimism in the near term to above 10,000 and my view would not change as the news continues to be deflationary which is just horrible for stocks.

 

[audreyh1]

Sooo... Do you feel better now, Audrey? I had posted a congrat on your ER anniversary, but you probably saw it already.

Yes! I do feel much better, thanks. I thought I was having to communicate across a chasm. Turns out you are not standing that far away after all. Good, because I only have the energy for a few such discussions a year.

Yes, ERD50, I forgot the 95% success rate for FIREcalc meant that the portfolio did not survive in 5% of the cases. I thought I recalled some discussion that threw such percentages around and that there was a % above which the success rate was perhaps not that meaningful - was it 99%?, but I don't remember exactly what that discussion concluded. Maybe someone can post a link or two - it's been a while.

Audrey

 

[M Paquette]

Yes, ERD50, I forgot the 95% success rate for FIREcalc meant that the portfolio did not survive in 5% of the cases. I thought I recalled some discussion that threw such percentages around and that there was a % above which the success rate was perhaps not that meaningful - was it 99%?, but I don't remember exactly what that discussion concluded. Maybe someone can post a link or two - it's been a while.

Audrey

Ah, I bet that was William Bernstein's 80% number.

The Retirement Calculator from Hell, Part III

Let’s examine a small sampling of possible political, economic, and military failure modes:

The mildest scenario is that of catastrophic inflation, as experienced in Germany and Hungary in the 1920s or, more recently, in much of the developing world.

Political failures are slightly worse, since these threaten the basic human motivation to work and produce. The state, for whatever reason, can decide to confiscate your assets or, worse, society’s means of production. Anyone who judges this unlikely should turn on CNN during any G-8 or WTO conference.

Local military action. Probably the lowest-probability item on this list, but something to think about on other continents.

The Big One: Some deranged prime minister or colonel in central Russia, Pyongyang, or South Asia could let loose the four horsemen upon the planet.

So, think about what a 97% 40-year success rate means: the absence of all of the above for approximately the next 1,200 years. (A 97% success rate means a 3% failure rate; those 40 years divided by 0.03 is 1,200 years.) Ignore for a minute the uncertainties of the less-developed world and think only about the winners: Germany—in this century alone, three episodes of military and/or economic disaster, the first two associated with mass starvation. Japan—wartime devastation even worse than Germany’s. England—near brushes with disaster in 1812-1814 and in both world wars. And even the United States—repeated banking failures, civil war, and the near-bankruptcy of the Treasury in the 19th century. The near collapse of the capitalist economy in the 1930s. And oh yes, I almost forgot—the entire globe barely missed mass incineration in October 1962.

History’s best-case scenario was the Roman Empire, which survived more or less intact for about seven centuries (if you ignore the odd sackings of the capital after 200 A.D.).

A wildly optimistic historian might give us another few centuries of economic, political, and military continuity. Back-of-the-envelope, that’s about an 80% survival rate over the next 40 years. Thus, any estimate of long-term financial success greater than about 80% is meaningless.

 

[NW-Bound]

Excellent, excellent...

Thank you M Paquette, for reminding us of this. I have seen this writing from Bernstein a while back, then forgot all about it, probably the same as many members of this forum. Nowadays, we have access to so much information and data, and having conditioning ourselves to be fast readers lest we miss something important, we become infoholics who often fail to retain the crucial things we ran across.

Those exogenous variables can throw a monkey wrench into our plan and ruin it. "Eat, drink, and be merry" indeed, as we can never be sure. The human mind hates uncertainties, but sadly there is never any guarantee. Least from a computer program.

 

[audreyh1]

Ah! And it was William Bernstein of the The Online Asset Allocator no less! That name was vaguely stuck in my memory as being associated with the x% survival rate, but I couldn't remember the context exactly or the precise numbers. Thank you so much M Paquette for the link! What a genius you are to figure out what I was mumbling about!!!

Bernstein's site is an excellent reference/overview page for those seeking to learn more about AA. It may be about the best "launching point". The Frank Armstrong link is no longer valid, but you can easily find his book and other materials on the web by searching.

Audrey

 

[ERD50]

Ah, I bet that was William Bernstein's 80% number.

The Retirement Calculator from Hell, Part III

Thus, any estimate of long-term financial success greater than about 80% is meaningless.

Thanks for digging up the quote. If Bernstein wants to look at it that way fine, but I don't agree with that view at all. Maybe my kids will thank me some day.

Sure, there is a chance something bad will happen over 40 years. I don't see that as any reason to add to that risk by choosing a WR that historically would not survive in 2 out of 10 cases.

Consider someone who RE'd @ age 50 in 1969 in the US. What world events in that time rendered having or not having a portfolio left "meaningless" for that 90 year old person today?

Even if bad times come, I would think one could weather them better with some extra cushion in the portfolio.

I do agree that people have to make the personal decision regarding the "risks" involved with working long enough to achieve the portfolio size required for 100% versus 80% ( ~ 25% larger according to FIRECALC). But I'm not going to try to rationalize it as anything other than what it is, trading one risk for another.

-ERD50

 

[NW-Bound]

I forgot to add that I will continue to keep my WR low. In fact I will strive to keep to 4% or less of the current value of portfolio. One can never run out of money this way. It means I will splurge in good years, and go back to my LBYM way in bad years. If we have more people like myself, the recession would get worse, but I cannot help being selfish. Would you do the reverse?

Not that I discounted Bernstein's possible black swan events. But in case it doesn't happen, I hate having to kick myself for having to panhandle while the world parties on.

In the case a black swan happens, we would be going down the tube together. In that condition, I doubt that any of us would say "I wish I had bought that Porsche instead of sticking with my 3.5%WR".

Still, Bernstein had a good point about not sweating these things too much.

 

[ERD50]

Further to my previous post:

And even the United States—repeated banking failures, civil war, and the near-bankruptcy of the Treasury in the 19th century. The near collapse of the capitalist economy in the 1930s. And oh yes, I almost forgot—the entire globe barely missed mass incineration in October 1962.

Which of these scenarios would have not gone better with more money? I recall my father saying that "cash was King" in the Depression. Hmmm, I'd rather come out of a "near miss" with my portfolio intact - I'd have enough money to buy a round of drinks. I just don't get his reasoning.

Worse, and very surprising for Bernstein, I think his numbers are off.

First, we only live in one industrialized country at a time. You can't add events from different places on the globe together to make a point. Calculated as a failure rate ( Bad Event / One person's 40 year exposure to Bad Events ) , he is increasing the numerator (multiple country bad events), but not increasing the denominator ( The Germany unique bad events apply to the person in Germany, not to the person in the US and Germany). He sort of implies that when he says: And even the United States...

That is like calculating *my* odds of getting hit by lightening based on the number of lightening strikes in the entire world over 40 years, rather than the number of lightening strikes that hit the space I occupy over 40 years.

Second - even accepting his 80% factor for external events

80% success from external events combined with an 80% success portfolio = .8*.8 = 64% success.

80% success from external events combined with an 97% success portfolio = .8*.97 = 77% success.

80% success from external events combined with a 100% success portfolio = .8*1 = 80% success.

I'm not so sure we should be referring to a difference in odds going from 64% to 80% as "meaningless".

I think Bernstein is great, but I'm not buying into this one. Maybe this is what is bugging me about this - we run FIRECALC and we see a 5% failure rate due to the market events (that would affect all people with that portfolio, AA, WR, etc) over the past periods since 1871. I don't see external events (that would affect all people with that portfolio, AA, WR, etc) occur in 5% of these cases that make having money unimportant. Couple that with a surviving spouse, I think there are good reasons to shoot for above 80% success rates.

You could almost turn his viewpoint around to say that there is not much sense in planning for anything past your 40th birthday. World events might wipe you out anyway, so why bother?

edit/add after seeng NWBs post - yes, you captured my feelings with the party/panhandle analogy.

-ERD50

 

[youbet]

it sometimes helps to look at the min/max/avg end values), and repeatability (and probably a few others I've forgotten about).

-ERD50

.

Absolutely! I find output data that simply states the number of failures per trial to have little meaning to me. Reading posts here, it's obvious others get confused too. For example, I read posts frequently where the poster assumes a 100% success rate means his/her portfolio performs wonderfully during the withdrawal phase and there will never be concern over breathtaking dips, slides or prolonged periods of inflation causing degradation in buying power.

I always take the output of Firecalc by year and organize it into a simple histogram. Then I can see (and show DW) that historically our plan would have resulted in zero failures but , depending on the period, might have had constant growth and a huge ending value or might have exposed us to scary dips and left us with near zero assets during our final years. Or anything inbetween.

An earlier version of Firecalc provided this histogram. I think eliminating it from the current version was a mistake.

 

[M Paquette]

The way I interpret the probability of success numbers and Bernstein's suggestion that 80% success is 'good enough' is pretty much summed up by:

Don't sweat the small stuff.

Unlike a computer program, I can respond to possible problems by making adjustments. That is, if I have a reasonable asset allocation, and a reasonable withdrawal rate and plan for my available assets, I'm not going to worry about whether one online calculator says I have only a 95% chance of making it, or another calculator says I have a 100% success rate.

If I find myself on a path that looks prone to hitting a failure, I can adjust. I can lower my withdrawal rate a bit, for example, or I can alter when I want to take Social Security, or when I want to start IRA withdrawals. I can even (shudder!) by a Single Payment Immediate Annuity.

 

[M Paquette]

Just for fun, I punched in my current numbers into FireCalc, and tried various variations on withdrawal plans, from the "spend more now, because old pharts spend less" option to Bob Clyatt's "95% of the previous year" rule for handling downturns, as well as the flat rate. I also put in possible asset allocation changes I would permit myself to make, limited to a 15% max change.

At the 80% confidence level, lowest to highest annual safe withdrawal rates had a 1.75:1 ratio. My current withdrawal rate was lower than the lowest safe rate. When I went for a 95% confidence level things tightened up a bit, but again my current withdrawal rate was low.

I obviously have too much money saved for retirement. What to do, what to do?

The wide range of safe withdrawal rates just shows that I have the room to adjust things significantly within my overall plan, my only real point in this pointless weekend.

 

[NW-Bound]

Unlike a computer program, I can respond to possible problems by making adjustments...

Of course, the key is to have the basic necessities being only a small percentage, say 60% or 75%, of the 4% WR. It means that you have leeway to cut out expenses in bad years. Since our lifestyle is frugal despite having two houses, my fixed costs are fairly low, and I must remind myself to keep it that way.

In a way, it is good that my expenses in the past varied wildly from year to year. One year, we spent more than 2.5 times the level of the previous year, and did not even notice until I reviewed it much later. It was great living in a bull-market year, and when you still had work income to boot! The high expenses included "one-time" events like home remodeling, but also more potentially recurrent items such as travels. We just need to go back to our frugal way until the storm is past. Somebody else will have to provide the stimulus the world economy needs.

PS. If there are further signs that we are out of the woods, there is still time for us to book a trip later this year. Maybe go to the town of Cognac to see my favorite spirit being made, or any wine town to see the vendange. You can't take it with you, heh heh heh...

 

[ERD50]

The way I interpret the probability of success numbers and Bernstein's suggestion that 80% success is 'good enough' is pretty much summed up by:

Don't sweat the small stuff.

... I'm not going to worry about whether one online calculator says I have only a 95% chance of making it, or another calculator says I have a 100% success rate.

If I find myself on a path that looks prone to hitting a failure, I can adjust. I can lower my withdrawal rate a bit, ...

Yes, but I wonder how this would work out in practice. Let's stick to Bernstein's 80% number, FIRECALC says:

80% success for 40 years = 4.15% WR; a $41,500 annual WD on a $1M portfolio
100% success for 40 years = 3.34% WR; a $33,400 annual WD on a $1M portfolio

So hard times come along, and Mr 80% has to cut back. Since he was taking out a higher average than Mr 100%, it would seem that he would now need to cut back below what Mr 100% was spending in order to get back on track. That is a pretty steep cut - I'm sure it's more complex than a simple average, but if that is ballpark, then Mr 80% might need to cut back to ~ $25,300 in some years. While Mr 100% can keep rolling along at $33,400.

I don't see either as right/wrong. Some might see it as a good trade-off to live a bit higher (or retire a bit earlier) if they have some confidence that the cut back is only a 20% risk, and are willing to make those kind of cuts. Others might not feel comfortable with that idea at all.

As to how this works in practice - I really wonder about that. People keep saying they would cut back, but what exactly triggers that? How much, when, how often? If one cannot answer those questions before the stuff hits the fan, I think they will have trouble figuring it out in the midst of a mess. For example - do you cut back after a single bad year? Two, three, five? A certain % drop in NW? I wonder how many people who have said they will do this actually have a plan, and have an understanding of how it would work. I'm not saying that as a criticism or a challenge, I'd actually like to see it, maybe I would do some of it if I felt comfortable enough.

I've done some crude modeling of this in a spreadsheet, and cutting back had to be pretty extreme to right the ship. I just ran the 95% spending rule in FIRECALC - that took the spending $41,500 spending down to ~ $18,000 ( a 57% cut!) for some long stretches (kinda tough to follow the lines, but the 10-20 year period was bunched up with lines in that area).

Are people really going to cut back that much for that long? And was it worth spending extra in the early years, versus a continuous $33,400? Maybe for some - but I wonder how many have thought this through.

edit/add: I keep cross posting with NWB - Of course, the key is to have the basic necessities being only a small percentage, say 60% or 75%, of the 4% WR. Or ~ 43% in the case above, for a 5-10 year stretch...

-ERD50

 

[M Paquette]

I think ERD50 and NW-bound both have it right.

The key to making this whole early retirement thing work financially is it have a plan. A good, detailed plan of testable options and regular checkup points, with an honest-to-gosh budget for the spending side.

I've been doing this for myself so long that I forget to state what is obvious to me. (Senior moments?) Mea culpa.

Anywho, I do a quarterly checkup to see if I'm on a reasonably survivable path. This resulted in a small adjustment downward in withdrawals last March, which put me on what is currently a 3.09% withdrawal rate for the next 4 quarters after March. The changes in spending came off the luxury tranch of the budget. (I have tranches for bare essentials, basics, comfort & nice to have, and luxury items.) For example, we got an outside view cabin instead of a balcony cabin for a cruise, and skipped a few other very nonessential items.

The gotcha with this, as ERD50 correctly points out, is that a repeat of the Great Depression, or even the 1960's will result in the luxury tranch, and some of the comfort & nice to have tranch going empty. We'll be comfortable, but we won't be going on any cruises. That's a decision we have decided we can live with. (I'll be out back in the tomato patch...)

Annually, I'll rebalance the portfolio and tap off enough to bring short term cash to a year's expenses. As part of this I'll do a sanity check on asset allocation and expected dividend flow, as well as the budget going forward.

Realistically, I know that we won't be traveling as much or as far in our 70s and 80s as we do in our 50s. I also know that at least part of our Social Security income is likely to be available when we are in our late 60s. Countering that will be things like possible long term medical expenses, downsizing the home, a possibility of needing to move to assisted living, and so on, all of which affect the future budget.

Risk is everywhere. There is no such thing as certainty. I accept that, and will deal with it. I've planned as best I can to cover a range of scenarios, as as for the rest, well, the thing about black swans is that they're black, and hard to spot, especially at night.

Rule 19: Satisfaction is not guaranteed.

 

[Running_Man]

Further to my previous post:


Which of these scenarios would have not gone better with more money? I recall my father saying that "cash was King" in the Depression. Hmmm, I'd rather come out of a "near miss" with my portfolio intact - I'd have enough money to buy a round of drinks. I just don't get his reasoning.

Worse, and very surprising for Bernstein, I think his numbers are off.

First, we only live in one industrialized country at a time. You can't add events from different places on the globe together to make a point. Calculated as a failure rate ( Bad Event / One person's 40 year exposure to Bad Events ) , he is increasing the numerator (multiple country bad events), but not increasing the denominator ( The Germany unique bad events apply to the person in Germany, not to the person in the US and Germany). He sort of implies that when he says: And even the United States...

That is like calculating *my* odds of getting hit by lightening based on the number of lightening strikes in the entire world over 40 years, rather than the number of lightening strikes that hit the space I occupy over 40 years.

Second - even accepting his 80% factor for external events

80% success from external events combined with an 80% success portfolio = .8*.8 = 64% success.

80% success from external events combined with an 97% success portfolio = .8*.97 = 77% success.

80% success from external events combined with a 100% success portfolio = .8*1 = 80% success.

I'm not so sure we should be referring to a difference in odds going from 64% to 80% as "meaningless".

I think Bernstein is great, but I'm not buying into this one. Maybe this is what is bugging me about this - we run FIRECALC and we see a 5% failure rate due to the market events (that would affect all people with that portfolio, AA, WR, etc) over the past periods since 1871. I don't see external events (that would affect all people with that portfolio, AA, WR, etc) occur in 5% of these cases that make having money unimportant. Couple that with a surviving spouse, I think there are good reasons to shoot for above 80% success rates.

You could almost turn his viewpoint around to say that there is not much sense in planning for anything past your 40th birthday. World events might wipe you out anyway, so why bother?

edit/add after seeng NWBs post - yes, you captured my feelings with the party/panhandle analogy.

-ERD50

Wow! Absolutely spot on, I totally agree with you on this and could not have thought it better, forget posting it! If you have a world event that happens 20% of the time in someone's life in the world somewhere, but only effects 20% of the world population the effect is only 4% not 20%!!!!

 

[bosco]

Further to my previous post:


Second - even accepting his 80% factor for external events

80% success from external events combined with an 80% success portfolio = .8*.8 = 64% success.

80% success from external events combined with an 97% success portfolio = .8*.97 = 77% success.

80% success from external events combined with a 100% success portfolio = .8*1 = 80% success.

I'm not so sure we should be referring to a difference in odds going from 64% to 80% as "meaningless".

-ERD50

I agree with your analysis of the "geographical flaw" in Bernstein's arguments, but the above analysis is not correct.

The multiplication of probabilities as you perform them above is only valid if the individual events are statistically independent.

In this case, many of the black swan events that Bernstein is contemplating would likely precipitate the portfolio failure--therefore the events are related, not independent. Unless you honestly believe that you can design a portfolio that would be independent from total banking system collapse, war, etc. Even if you could (without ruining its performance in more "normal" times), your money might not help you much. You might be better off to invest in reloaders, generators, bullets, etc.

I think the "true" percentage lies somewhere between your calculation and Bernstein's, but probably closer to his. The implicit assumption in your calculation is that the financial and the political are not related to one another. We all know that is not true.

Your argument gains some validity due to the fact that the Firecalc data series doesn't include most, if any, of these events in its history. But that's just really another way of saying that it is a very limited data set, and wrapping one's financial expectations around this limited series is analagous to a market timer that designs a system that optimizes and "curve-fits" past data.

So I think that the spirit of Bernstein's argument stands.