Diversification Can Improve Retirement Outcomes (SWR) [Very Long Post]

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
rkhusky
Posts: 6778
Joined: Thu Aug 18, 2011 8:09 pm

Re: Diversification Can Improve Retirement Outcomes (SWR) [Very Long Post]

Post by rkhusky » Sat Mar 30, 2019 4:10 pm

vineviz wrote:
Sat Mar 30, 2019 2:30 pm
rkhusky wrote:
Sat Mar 30, 2019 1:39 pm
You cannot differentiate these two cases of fluctuations with a period of 1 month versus a period of 1 year with the standard deviation and the mean.
I'm not sure I understand your question, then.

Annualized standard deviation is just monthly standard deviation times the square root of 12.

Annualized mean return is just (1 + monthly return) ^ 12.

If you know the monthly mean and variance, you know the annual mean and variance: they are literally just mathematical transformations of each other.
Mean and standard deviation throw out most of the information in a time series. Among that information is all ordering of the data. Two time series with identical mean and standard deviation can look quite different. The question is, I suppose, do 2 times series with the same mean and standard deviation have the same risk? In my example, does an investment that oscillates from peak to trough to peak consistently in roughly a month's time have the same risk as an investment that oscillates from peak to trough to peak consistently in roughly a year's time?

User avatar
One Ping
Posts: 701
Joined: Thu Sep 24, 2015 4:53 pm

Re: Diversification Can Improve Retirement Outcomes (SWR) [Very Long Post]

Post by One Ping » Sun Mar 31, 2019 10:29 am

vineviz wrote:
Fri Mar 29, 2019 12:26 pm
Curious about how much difference it would make if this hypothetical retiree had made some effort to better diversify this portfolio (call it Portfolio 1), I constructed two additional portfolios.

Portfolio 2 replaces some of the large cap stock with small allocations to small cap value and global ex US stocks: 48% large cap, 6% SCV, 6% international. This portfolio is designed (with the benefit of hindsight) to have the same overall volatility (aka risk) as Portfolio 1: stdev = 9.91%.

Portfolio 3 increases the allocation to bonds in order to have the same overall returns as Portfolio 1 (9.94%). It is 30% large cap, 7.5% SCV, 7.5% international, and 55% bonds.
Very interesting analysis, vineviz. :beer

While I agree with the general concept, I do have a question. How would a retiree apply the results from this one specific retrospective case study with one specific set of returns/correlations/allocations to their own retirement portfolio going forward? Return sequences and correlations vary over time and, it would seem to me, that the specific allocations to SCV and International to satisfy your constraints (same CAGR or same SD) would vary also.

Have you looked at other retirement dates that have come close to failing under the 4% 'rule' to test the robustness of your allocations? Would be very interesting ...

One Ping
"Re-verify our range to target ... one ping only."

Random Walker
Posts: 3961
Joined: Fri Feb 23, 2007 8:21 pm

Re: Diversification Can Improve Retirement Outcomes (SWR) [Very Long Post]

Post by Random Walker » Sun Mar 31, 2019 11:33 am

One Ping wrote:
Sun Mar 31, 2019 10:29 am
While I agree with the general concept, I do have a question. How would a retiree apply the results from this one specific retrospective case study with one specific set of returns/correlations/allocations to their own retirement portfolio going forward? Return sequences and correlations vary over time and, it would seem to me, that the specific allocations to SCV and International to satisfy your constraints (same CAGR or same SD) would vary also.
One Ping
The only data we have is backward looking. Data mining is a huge issue. But data can be strengthened. We can have out of sample data from different time periods, different geographic markets, even different asset classes. Beyond data we can have rational risk based and behavioral based explanations to expect premia to persist looking forward.
I think we need to take Vineviz’s data as simply a specific example of a more general concept. Combining less than perfectly correlated, uncorrelated, negatively correlated, sources of return improves portfolio efficiency. The more efficient portfolio will have a smoother ride, smaller maximal drawdowns, less volatility drag, narrower SD for same expected return, less subject to sequence of returns risk, less subject to Murphy’s law of retirement.
Yes, correlations change, markets evolve, correlations of risky assets tend to 1 in bad times. But with an unknown future, the investor can likely put the odds in his favor by maximally diversifying across sources of risk and return. Because the future is unknown, the benefits are only potential while the increased costs are certain. The first step to diversifying a US TSM portfolio is likely safe bonds. Second step International and EM equity. Third step tilts to size and value. Fourth step can be other factors/styles such as CS momentum, TS momentum, profitability. Fifth step alternatives such as reinsurance, alt lending, variance risk premium. Each step towards marginally more efficient portfolio comes at marginally increased cost. Each individual needs to decide where to draw the line.

Dave

User avatar
One Ping
Posts: 701
Joined: Thu Sep 24, 2015 4:53 pm

Re: Diversification Can Improve Retirement Outcomes (SWR) [Very Long Post]

Post by One Ping » Sun Mar 31, 2019 11:55 am

Random Walker wrote:
Sun Mar 31, 2019 11:33 am
One Ping wrote:
Sun Mar 31, 2019 10:29 am
While I agree with the general concept, I do have a question. How would a retiree apply the results from this one specific retrospective case study with one specific set of returns/correlations/allocations to their own retirement portfolio going forward? Return sequences and correlations vary over time and, it would seem to me, that the specific allocations to SCV and International to satisfy your constraints (same CAGR or same SD) would vary also.
One Ping
The only data we have is backward looking. Data mining is a huge issue. But data can be strengthened. We can have out of sample data from different time periods, different geographic markets, even different asset classes. Beyond data we can have rational risk based and behavioral based explanations to expect premia to persist looking forward.
I think we need to take Vineviz’s data as simply a specific example of a more general concept. Combining less than perfectly correlated, uncorrelated, negatively correlated, sources of return improves portfolio efficiency. The more efficient portfolio will have a smoother ride, smaller maximal drawdowns, less volatility drag, narrower SD for same expected return, less subject to sequence of returns risk, less subject to Murphy’s law of retirement.
Yes, correlations change, markets evolve, correlations of risky assets tend to 1 in bad times. But with an unknown future, the investor can likely put the odds in his favor by maximally diversifying across sources of risk and return. Because the future is unknown, the benefits are only potential while the increased costs are certain. The first step to diversifying a US TSM portfolio is likely safe bonds. Second step International and EM equity. Third step tilts to size and value. Fourth step can be other factors/styles such as CS momentum, TS momentum, profitability. Fifth step alternatives such as reinsurance, alt lending, variance risk premium. Each step towards marginally more efficient portfolio comes at marginally increased cost. Each individual needs to decide where to draw the line.

Dave
I agree with almost everything you say above. Having said that, testing the specific approach vineviz demonstrated above across different eras would provide an indication of the potential variability of the 'tilts'/allocations used to satisfy the constraints set forth in the OP. For an investor nearing retirement this would provide something actionable for them to consider.
"Re-verify our range to target ... one ping only."

dave_k
Posts: 331
Joined: Sat Dec 26, 2015 8:25 pm

Re: Diversification Can Improve Retirement Outcomes (SWR) [Very Long Post]

Post by dave_k » Sun Mar 31, 2019 11:59 am

Thank you for the post. I'm wondering about a couple things:
vineviz wrote:
Fri Mar 29, 2019 2:24 pm
I agree that diversification is still useful in accumulation. It's a central piece of my portfolio construction process, and unshakeable so.

That said, do I think it's useful to acknowledge that it LESS important for accumulators.

Just by way of illustration, here are the same portfolios and the same time periods I used above but with the scenario inverted from decumulation to accumulation. The investors started with $1 and contributed $335/month adjusted for inflation.

Unlike the withdrawal situation above, in which the three portfolios had significantly different outcomes, the situation here is much less dramatic.

Image

Here, unfortunate sequence of returns for Portfolio 3 were a slight drag on performance relative to the other two portfolios.
Is it possible that a large part of why it doesn't seem to matter as much for accumulation vs. decumulation during this time period is because of the poor performance in the early years, which is a worst case scenario for starting retirement, but makes much less difference to the relatively small amount invested at the beginning of an accumulation phase? Maybe there are other time periods where diversification makes as much of a difference for accumulators (or at least more of a difference than this period). My point being that diversification may make more of a difference to accumulators than this specific example illustrates, even if it's not quite as important as it is during decumulation.

vineviz wrote:
Fri Mar 29, 2019 5:43 pm
Tamarind wrote:
Fri Mar 29, 2019 4:54 pm
I concur. It appears Galeno actually took OP's post very much to heart by recommending a portfolio that includes small cap and international, albeit at market weight rather than underweight as in Portfolio 2.
I was unduly curt. The error that prompted my reply was in referring to the 50/50 VT/BNDW as "completely diversified". The global market cap portfolio has many useful traits, but "complete diversification" isn't one of them.

Even just using those those assets (VT & BNDW), 50/50 is far from the most diversified combination. That'd be approximately 11% VT and 89% BNDW.
I'm curious why 11% VT and 89% BNDW is more diversified, is that because you are weighting for equal volatility?

I'm still trying to wrap my head around your reasoning for market cap weighting that includes domestic, international, and SCV not being good enough. It seems to me that the only difference between market cap weighting and having set allocations to each of these is the rebalancing between them with set allocations, but you also said there's no significant rebalancing benefit. It seems that at least some of the benefit has to be due to rebalancing, even if it's not real sensitive to the frequency. What am I missing? Is it just that the market cap weighting is too far from optimal?

User avatar
Topic Author
vineviz
Posts: 4431
Joined: Tue May 15, 2018 1:55 pm

Re: Diversification Can Improve Retirement Outcomes (SWR) [Very Long Post]

Post by vineviz » Sun Mar 31, 2019 12:20 pm

One Ping wrote:
Sun Mar 31, 2019 10:29 am
Return sequences and correlations vary over time and, it would seem to me, that the specific allocations to SCV and International to satisfy your constraints (same CAGR or same SD) would vary also.
The goal in creating a diversified portfolio is balancing the risks across the underlying assets, and it turns out that the most important characteristics (e.g. variance and correlation) are MUCH more stable across time than returns are. And the relative correlations and relative variance between assets is even more stable.

As a result, a retiree would not find that the weights of a diversified portfolio changed much. For instance, here's a chart showing what the most diversified mix of a large cap fund (VFINX) and small cap fund (NAESX) would have looked like from 1988 to 2019. This shows allocation based on rolling 36-month loopback periods for correlation and variance. The results were remarkably consistent across that 30 year period.

Image

The best diversification consistently came at about 40-45% small caps.

In practice you would surely NOT want to set a constraint on CAGR: I did that here to illustrate that the benefits were NOT coming from higher returns. If you did it in real life, you'd be susceptible to overweighting assets that had performed well recently and that's not a recipe for investment success.

Better, I think, use your risk tolerance to set a target range for volatility or just flat out set stock/bond ratio you stick to.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch

User avatar
Topic Author
vineviz
Posts: 4431
Joined: Tue May 15, 2018 1:55 pm

Re: Diversification Can Improve Retirement Outcomes (SWR) [Very Long Post]

Post by vineviz » Sun Mar 31, 2019 12:37 pm

dave_k wrote:
Sun Mar 31, 2019 11:59 am
Is it possible that a large part of why it doesn't seem to matter as much for accumulation vs. decumulation during this time period is because of the poor performance in the early years, which is a worst case scenario for starting retirement, but makes much less difference to the relatively small amount invested at the beginning of an accumulation phase? Maybe there are other time periods where diversification makes as much of a difference for accumulators (or at least more of a difference than this period).
I don't present any data here, and maybe its worthy of another thread, but the result generally holds true that sequence of returns risk is much less impactful in accumulation than in decumulation.

Diversification is still valuable, in my opinion, and not least because it provides something of a free lunch. It's just that the contributions themselves ameliorate most of the sequence of returns risk. This is especially when you couple contributions with a glide path that reduces equity exposure as you approach the end of accumulation.
dave_k wrote:
Sun Mar 31, 2019 11:59 am
% VT and 89% BNDW is more diversified, is that because you are weighting for equal volatility?
More or less. In a two asset portfolio you get similar weights from maximum diversification, risk parity, and inverse volatility.
dave_k wrote:
Sun Mar 31, 2019 11:59 am
I'm still trying to wrap my head around your reasoning for market cap weighting that includes domestic, international, and SCV not being good enough. It seems to me that the only difference between market cap weighting and having set allocations to each of these is the rebalancing between them with set allocations, but you also said there's no significant rebalancing benefit. It seems that at least some of the benefit has to be due to rebalancing, even if it's not real sensitive to the frequency. What am I missing? Is it just that the market cap weighting is too far from optimal?
Yes, this is pretty much what I meant. The frequency of rebalancing isn't the main thing driving the result, it's spreading the risks the portfolio is exposed too more evenly across independent sources.

I don't want to overstate my position: market cap weighting may often be "good enough". Even though it is never the MOST diversified portfolio, a global market cap weight portfolio is still WELL diversified and has other advantages (generally low expense ratios, low trading costs and tracking error, etc.).

My examples were, admittedly, somewhat lopsided in that the base case was sort of "worst case": an investor whose assets were 100% US and whose stocks were 100% large cap. A global market cap weighted portfolio would have been an improvement over Portfolio A, though obviously it would still have had room for improvement. I purposely didn't use maximally diversified portfolios for B and C so as to not cloud the message that even small improvements in diversification can be important.

That said, it has become increasingly easy and cheap to build a portfolio that is MORE diversified than the global market cap portfolio without it becoming unduly complicated to manage.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch

dave_k
Posts: 331
Joined: Sat Dec 26, 2015 8:25 pm

Re: Diversification Can Improve Retirement Outcomes (SWR) [Very Long Post]

Post by dave_k » Sun Mar 31, 2019 12:50 pm

Thanks for clarifying. Would you mind posting your allocation as an example (assuming you are following this)? Maybe you have elsewhere (haven't checked - you have a lot of posts :happy), but an example would be helpful here.

User avatar
Topic Author
vineviz
Posts: 4431
Joined: Tue May 15, 2018 1:55 pm

Re: Diversification Can Improve Retirement Outcomes (SWR) [Very Long Post]

Post by vineviz » Sun Mar 31, 2019 1:00 pm

dave_k wrote:
Sun Mar 31, 2019 12:50 pm
Thanks for clarifying. Would you mind posting your allocation as an example (assuming you are following this)? Maybe you have elsewhere (haven't checked - you have a lot of posts :happy), but an example would be helpful here.
The allocations in the OP, you mean?

Portfolio 1 was 60% large cap U.S. stocks (basically S&P 500) and 40% intermediate US treasuries.

Portfolio 2 was 48% large cap, 6% SCV, 6% international, and 40% intermediate US treasuries.

Portfolio 3 was 30% large cap, 7.5% SCV, 7.5% international, and 55% intermediate US treasuries.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch

dave_k
Posts: 331
Joined: Sat Dec 26, 2015 8:25 pm

Re: Diversification Can Improve Retirement Outcomes (SWR) [Very Long Post]

Post by dave_k » Sun Mar 31, 2019 1:26 pm

vineviz wrote:
Sun Mar 31, 2019 1:00 pm
dave_k wrote:
Sun Mar 31, 2019 12:50 pm
Thanks for clarifying. Would you mind posting your allocation as an example (assuming you are following this)? Maybe you have elsewhere (haven't checked - you have a lot of posts :happy), but an example would be helpful here.
The allocations in the OP, you mean?

Portfolio 1 was 60% large cap U.S. stocks (basically S&P 500) and 40% intermediate US treasuries.

Portfolio 2 was 48% large cap, 6% SCV, 6% international, and 40% intermediate US treasuries.

Portfolio 3 was 30% large cap, 7.5% SCV, 7.5% international, and 55% intermediate US treasuries.
I understood those to be examples to illustrate the point given the data for that time period, where the constant 4% real withdrawal rate lasted the minimum 30 years for a 60/40 portfolio. What is your portfolio now, if you don't mind sharing? Or what would you recommend to someone following this advice now, who otherwise would be comfortable with a 70/30 portfolio of market cap weighted total world stocks to bonds?

HEDGEFUNDIE
Posts: 2791
Joined: Sun Oct 22, 2017 2:06 pm

Re: Diversification Can Improve Retirement Outcomes (SWR) [Very Long Post]

Post by HEDGEFUNDIE » Sun Mar 31, 2019 1:32 pm

dave_k wrote:
Sun Mar 31, 2019 1:26 pm
vineviz wrote:
Sun Mar 31, 2019 1:00 pm
dave_k wrote:
Sun Mar 31, 2019 12:50 pm
Thanks for clarifying. Would you mind posting your allocation as an example (assuming you are following this)? Maybe you have elsewhere (haven't checked - you have a lot of posts :happy), but an example would be helpful here.
The allocations in the OP, you mean?

Portfolio 1 was 60% large cap U.S. stocks (basically S&P 500) and 40% intermediate US treasuries.

Portfolio 2 was 48% large cap, 6% SCV, 6% international, and 40% intermediate US treasuries.

Portfolio 3 was 30% large cap, 7.5% SCV, 7.5% international, and 55% intermediate US treasuries.
I understood those to be examples to illustrate the point given the data for that time period, where the constant 4% real withdrawal rate lasted the minimum 30 years for a 60/40 portfolio. What is your portfolio now, if you don't mind sharing? Or what would you recommend to someone following this advice now, who otherwise would be comfortable with a 70/30 portfolio of market cap weighted total world stocks to bonds?
viewtopic.php?f=10&t=276307

dave_k
Posts: 331
Joined: Sat Dec 26, 2015 8:25 pm

Re: Diversification Can Improve Retirement Outcomes (SWR) [Very Long Post]

Post by dave_k » Sun Mar 31, 2019 1:38 pm

Perfect - I see it includes a lot of follow-up posts with explanations. Thanks!

User avatar
siamond
Posts: 4806
Joined: Mon May 28, 2012 5:50 am

Re: Diversification Can Improve Retirement Outcomes (SWR) [Very Long Post]

Post by siamond » Fri Apr 05, 2019 3:40 pm

rkhusky wrote:
Fri Mar 29, 2019 4:50 pm
I've always had an issue with volatility being equated with risk. If one is using past performance anyways, it seems like a better and more functional measure for risk would be related to SWR. Standard deviation is obviously much easier to calculate and work with, but it's correlation with true risk is nebulous.
[...]
It is also interesting that, in terms of past performance, a 100% stock portfolio had a higher SWR than did a 100% bond portfolio. So, were bonds really safer than stocks?
Totally agreed, volatility (std-deviation) has very little to do with a measure that is concretely palatable to retirees. I made the case that SWR is both a measure of risk and reward in this blog article and its follow-ups, and a much more palatable measure than volatility vs. CAGR.

PS. there is actually a remarkably simple way to compute SWR (in hindsight) with a harmonic mean function. Please check this other blog article of mine (the SWR section).
azanon wrote:
Fri Mar 29, 2019 3:15 pm
So much effort and time is put into avoiding the potential perils of using the 4% rule for retirement income (aka the "constant dollar" method), and often it is never realized that the greatest issue with that is actually the use of that withdrawal method itself. Said another way, trying to find the optimum portfolio for that retirement income method is the equivalent of trying to put lipstick on a pig.
I used to agree with this view and was almost ready to declare the SWR metric (and all related analysis) as meaningless garbage. I changed my tune though. There is actually a fascinating linkage between SWR and variable withdrawal methods. Please check the risk/reward blog article I just described above (up to and including the third part).

User avatar
siamond
Posts: 4806
Joined: Mon May 28, 2012 5:50 am

Re: Diversification Can Improve Retirement Outcomes (SWR) [Very Long Post]

Post by siamond » Fri Apr 05, 2019 4:30 pm

Continuing with SWR as a risk & reward perspective, here is a Simba backtesting chart that I believe may capture the OP's idea in a more systematic manner. Click on it to see a larger version.

Horizontal axis is a given starting year of a 30 years retirement cycle. Vertical axis is the corresponding maximum withdrawal rate that would have brought the portfolio to exactly zero after 30 years (hence SWR-like). Yup, portfolio 2 definitely performed better during the dreaded mid-60s (a quarter of a point of improvement is nothing to sneeze at!) and seemed to hold water pretty nicely for the other starting years.

I wouldn't say the same of Portfolio 3, but then its definition in the OP is rather artificial and period-specific. Back to the OP's core point, yup, the addition of SCV and Int'l as 'diversifiers' in Portfolio 2 would have resulted in a truly palatable improvement in the life of corresponding retirees. My own (early retiree) portfolio was constructed with this kind of consideration in mind...

Image

User avatar
galeno
Posts: 1551
Joined: Fri Dec 21, 2007 12:06 pm

Re: Diversification Can Improve Retirement Outcomes (SWR) [Very Long Post]

Post by galeno » Fri Apr 05, 2019 6:15 pm

It's the old question of which is better? A simple port or a complex port? I'm a simple port advocate.

The OP's idea of diversification is to use more asset classes in a non-cap weighed fashion. Bernstein advocates this approach in his older books. So do many money managers and now money managing BOTS.

I'd rather SIMPLIFY a portfolio vs complicate it. Especially if we're talking about chasing ephemeral factors. Back testing can be a dangerous tool.

What's wrong with the simple two fund globalist portfolio of X% world stocks + Y% world bonds? Or the nationalist two fund porfolio of X% US stocks + Y% TBM?
AA = 40/55/5. Expected CAGR = 3.8%. GSD (5y) = 6.2%. USD inflation (10 y) = 1.8%. AWR = 4.0%. TER = 0.4%. Port Yield = 2.82%. Term = 33 yr. FI Duration = 6.0 yr. Portfolio survival probability = 95%.

User avatar
Topic Author
vineviz
Posts: 4431
Joined: Tue May 15, 2018 1:55 pm

Re: Diversification Can Improve Retirement Outcomes (SWR) [Very Long Post]

Post by vineviz » Mon Jul 08, 2019 5:03 pm

siamond wrote:
Fri Apr 05, 2019 4:30 pm
Continuing with SWR as a risk & reward perspective, here is a Simba backtesting chart that I believe may capture the OP's idea in a more systematic manner. Click on it to see a larger version.
I've been meaning to thank siamond for pointing out the usefulness of the Simba spreadsheet in exploring this question.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch

larryswedroe
Posts: 15945
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Re: Diversification Can Improve Retirement Outcomes (SWR) [Very Long Post]

Post by larryswedroe » Mon Jul 08, 2019 6:56 pm

Galeno, there is nothing wrong with that two fund portfolio, (in fact I recommend it to investors who cannot deal with tracking variance) but it does concentrate risks in market beta and term and nothing else. And as I have been noting, there are very long periods when TSM underperforms totally riskless tbills. Three of them in fact in US of at least 12 years, and at least 13 for the S&P. You can historically build more of diversified risk parity portfolio which would have been more efficient, with lot less tail risk which means lot less sequence risk. And like I have said, if you believe markets efficient than why concentrate risks in one unique factor rather than diversifying across multiple ones. Note also you can do it with just two funds for equities, just like the two Vanguard TSM funds. So it is not more complex in that sense, and it has meant historically more efficiency. I don't know what the future holds, which is why I diversify, it's an admission of not knowing. It's NOT making a bet on one factor which TSM is doing, yet most frame the problem in the wrong way leading them to conclude that they don't want to bet on other factors. That to me is a framing problem mistake.
If interested in at least learning about the history and evidence and strategy before drawing conclusions, if have not read Reducing the Risk of Black Swans you might want to pick it up.
Best wishes
Larry

User avatar
siamond
Posts: 4806
Joined: Mon May 28, 2012 5:50 am

Re: Diversification Can Improve Retirement Outcomes (SWR) [Very Long Post]

Post by siamond » Mon Jul 08, 2019 7:57 pm

vineviz wrote:
Mon Jul 08, 2019 5:03 pm
I've been meaning to thank siamond for pointing out the usefulness of the Simba spreadsheet in exploring this question.
Thank you! Yes, I really like this kind of chart. I think this provides much richer information than the usual "Is SWR 3% or 4%" discussion.

Actually, my thinking about diversification has evolved a bit over time. I always looked at 'factor diversification' statements with a very skeptical eye. Factors are very artificial constructs, mathematical ways to slice and dice past history of the market with one arbitrary formula or another, and there is an ever evolving zoo of those. It is really hard to believe that there is something truly fundamental at play here. This seems way too academic in other words.

On the other hand, I do tilt towards Mid Value, Small Value, Int'l Small and Emerging... The historical performance is hard to deny (and as you indicated, directly applicable to metrics relevant to both accumulators AND retirees), and some underlying explanations do seem to make common sense. PLUS this gives me a comfortable feeling about hedging my bets. But I would never qualify that as 'factor diversification'.

IMHO, diversification should really be much simpler than that. It's really about having more individual securities at play, more eggs having a significant impact on one's basket. Cap-weighted indices have a lot of great properties, but they totally skew the index towards a small number of individual mega-caps which happen to be in great shape at that point in time (yesterday's winners?), while basically ignoring the rest of the haystack. Sure, I could reformulate the previous statement in terms of 'small' and 'value' considerations (and emerging is kind of the small value of the full set of countries while US is the epitome of 'large growth'), but those are only two arbitrary metrics. Using corresponding index funds is definitely an effective way to make more eggs more significant though, and hopefully give a better chance to future winners.

So... I now view my tilts primarily as a practical way to hedge my bets, reaching some kind of coarse happy medium between equal-weighting (which is definitely too extreme) and cap-weighting (the other extreme). I wish there would be a form of low-cost index that better captures my line of thinking about diversification (logarithmic cap-weighting? :D), but this will do.

visualguy
Posts: 1408
Joined: Thu Jan 30, 2014 1:32 am

Re: Diversification Can Improve Retirement Outcomes (SWR) [Very Long Post]

Post by visualguy » Mon Jul 08, 2019 8:35 pm

The problem with this study is that it starts with the same initial withdrawal balance ($100K) for all portfolios, thus ignoring what happened during the accumulation period where the different portfolios would have resulted in different final accumulation balances/initial withdrawal balances.

User avatar
Topic Author
vineviz
Posts: 4431
Joined: Tue May 15, 2018 1:55 pm

Re: Diversification Can Improve Retirement Outcomes (SWR) [Very Long Post]

Post by vineviz » Mon Jul 08, 2019 8:41 pm

visualguy wrote:
Mon Jul 08, 2019 8:35 pm
The problem with this study is that it starts with the same initial withdrawal balance ($100K) for all portfolios, thus ignoring what happened during the accumulation period where the different portfolios would have resulted in different final accumulation balances/initial withdrawal balances.
That's not a "problem", it's a key feature of any proper systematic inquiry: hold the extraneous variables constant, so far as constraints allow.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch

visualguy
Posts: 1408
Joined: Thu Jan 30, 2014 1:32 am

Re: Diversification Can Improve Retirement Outcomes (SWR) [Very Long Post]

Post by visualguy » Mon Jul 08, 2019 8:51 pm

vineviz wrote:
Mon Jul 08, 2019 8:41 pm
visualguy wrote:
Mon Jul 08, 2019 8:35 pm
The problem with this study is that it starts with the same initial withdrawal balance ($100K) for all portfolios, thus ignoring what happened during the accumulation period where the different portfolios would have resulted in different final accumulation balances/initial withdrawal balances.
That's not a "problem", it's a key feature of any proper systematic inquiry: hold the extraneous variables constant, so far as constraints allow.
No, ignoring the fact that the balances at the end of accumulation/beginning of withdrawal would be different for these portfolios leads to a misleading picture.

User avatar
Topic Author
vineviz
Posts: 4431
Joined: Tue May 15, 2018 1:55 pm

Re: Diversification Can Improve Retirement Outcomes (SWR) [Very Long Post]

Post by vineviz » Mon Jul 08, 2019 9:46 pm

visualguy wrote:
Mon Jul 08, 2019 8:51 pm
No, ignoring the fact that the balances at the end of accumulation/beginning of withdrawal would be different for these portfolios leads to a misleading picture.
The topic of this post has nothing to say about the accumulation phase: the example portfolios are examples of decumulation portfolios, to illustrate the impact of diversification on portfolios under conditions of withdrawal.

The nature of the portfolios during accumulation is not addressed here, one way or the other, and are therefore irrelevant: i.e. an extraneous variable.

If you disagree, feel free to PM me and we can discuss it offline so that this thread isn't further derailed.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch

986racer
Posts: 122
Joined: Thu Aug 11, 2016 10:09 am

Re: Diversification Can Improve Retirement Outcomes (SWR) [Very Long Post]

Post by 986racer » Thu Jul 11, 2019 7:11 pm

I’d highly suggest picking up “Living Off Your Money” by Michael McClung. He covers all of this and more.

Post Reply