Real Rate of Return...
 
Notifications
Clear all

Real Rate of Returns in Monte Carlo simulations leads to far worse outcomes

18 Posts
5 Users
0 Likes
1,462 Views
(@howard230)
Eminent Member Customer
Joined: 3 years ago
Posts: 16
Topic starter  

I've struggled over the years with the ROR for my 8 asset classes in Pralana Gold. I understand the concept but given the supposed overvaluation of almost all asset classes right, projected Monte Carlo returns are all quite low for ROR. But (a big but), these tools with projected returns are normally not really looking past 10 years, let alone the 40 to 45 I am modeling. My problem is that I've used the current projected conservative ROR for all my asset classes which ends up giving me a far more pessimistic Monte Carlo outlook for my future versus Historical returns, Bear Market Analysis returns or the outlook in other calculators like Fidelity and Firecalc.

One tool that does at least project to 30 years is this one: https://www.blackrock.com/institutions/en-us/insights/charts/capital-market-assumptions

I'm curious if others use the same conservative RORs by asset class and do their Monte Carlo results look much worse than Historical and even "Bear Market Analysis" returns? My overall ROR for the 40 years using these estimates is 2.73%.


   
ReplyQuote
(@pizzaman)
Honorable Member Customer
Joined: 3 years ago
Posts: 430
 

I also model out 40 years and my Monte Carlo results are always worse then Historical (backtesting) returns. Measuring risk is at the heart of retirement planning and these two forms of computer simulations are the most common (with bootstrapping a third). However, I don't put much faith in Monte Carlo analyses for retirement planning. Backtesting is based on real market data, Mote Carlo randomly create data. Using Monte Carlo for retirement planning is only as good as the underlying model of the markets, which is problematic. There are no agreed standards for modeling markets in a realistic way. Backtesting takes into account asset allocation, return disposition, sequence of returns, volatility, inflation, momentum, etc. Looking at momentum, the stock market has a tendency to drop quickly and by large amounts, but then take a lot longer, sometimes years, to come back. This is one reason that annual rebalncing is not a good idea (but that's another discussion).

Pizza Man


   
ReplyQuote
(@hines202)
Reputable Member Customer
Joined: 3 years ago
Posts: 331
 

There are good arguments for not using historical backtesting as well (see other threads on monte carlo vs historical on this board). For one, as we've learned from past economic events and put in checks and balances to prevent those from happening again, they become moot. We've recovered very quickly from the last two downturns using this knowledge and techniques (but let's not get overly optimistic and assume every recovery will be quick).

That said, it's interesting and fun to see how your circumstance would have fared if for example, you retired in 1965, which is arguably the worst year to retire in history given all that transpired over the following years. Some historical tests go back to eras where economic markets and policy bear no resemblance to today.

I do have questions about the different monte carlo simulations - do they model typical economic trends (peak, expansion, trough, recession, etc) or just competely random ups and downs?


   
ReplyQuote
(@smatthews51)
Member Admin
Joined: 4 years ago
Posts: 689
 

@hines202 Bill, PRC's Monte Carlo simulations are based on completely random up's and down's (based on a normal distribution and the user-specified mean and standard deviation).


   
ReplyQuote
(@pizzaman)
Honorable Member Customer
Joined: 3 years ago
Posts: 430
 

The last two down turns are good examples of why Monte Carlo does not work well for retirement planning. Yes, check and balances were put in place during the last century, but that did not stop the 2008 global market melt down. Humans have an amazing ability to get around check and balances (and rules) - subprime mortgage anyone? Once in a century pandemic (made worst by human behavior) causing global market drop, albeit short term. The point is Monte Carlo does not handle these types of events very well. Historical back testing takes into account human lunacy, which has not changed since stock markets were first invented, so using as much historical data as possible will give you your worst case outcomes. I am not saying don't use Monte Carlo analyses, the more ways of looking at things the better (nothing is perfect) but I would not solely rely on it. Good discussions, keep it going!!!!!


   
ReplyQuote
(@howard230)
Eminent Member Customer
Joined: 3 years ago
Posts: 16
Topic starter  

I can tell you that my Monte Carlo results are far worse than all of the Bear Market simulations included in Pralana (great feature). I may just be far too pessimistic in my ROR for all my asset classes. I can't believe that the next 30 to 40 years are going to be worse than the 2.73% ROR I'm modeling, but am curious what others are using for their ROR for their asset classes. I am using many resources, including the ones Stuart recommends, that are attempting to predict future real rates of return and standard deviations, but they rarely go past 10 or 20 years. I will be overjoyed if the Historical returns are accurate for my scenarios.


   
ReplyQuote
 NC
(@nc-cpl)
Reputable Member Customer
Joined: 3 years ago
Posts: 246
 

Greg - what sources are you using? Stuart added a section to the forum and I've been posting sources I've found, and yet I'm still uncertain I'm using the "optimal" real rates of return, etc. Would be great if we could poll PRC users and see what emerges as the preferred sources.


   
ReplyQuote
(@howard230)
Eminent Member Customer
Joined: 3 years ago
Posts: 16
Topic starter  

I've used the Bogleheads that Stuart recommends, but they are quite old now:

https://www.bogleheads.org/wiki/Historical_and_expected_returns

Blackrock also seems good:

https://www.blackrock.com/institutions/en-axj/insights/capital-market-assumptions_AXJ

and this one from Vanguard:

https://advisors.vanguard.com/insights/article/marketperspectivesmay2021

Do not use GMO expectations unless you want really depressing returns.


   
ReplyQuote
(@pizzaman)
Honorable Member Customer
Joined: 3 years ago
Posts: 430
 

Greg - for me I use a 70/30 stock/bond allocation, a real rate of return of 3.0% for stocks and 0.5% for bonds (I use 3.2% for inflation - 100 year average), healthcare inflation of 2%, and I assume that my social security (SS) will go down 15% starting in 2032 - that's because their is a quirk in how SS is determined, how much you will get from SS is partially based on how well the economy did in the year you turn 60. https://www.fool.com/retirement/2020/06/15/turning-60-this-year-a-weird-social-security-rule.aspx. Well I turned 60 in 2020, not go timing. Using all that, my historical (back-testing) simulations are always better than Monty Carlo. Guessing what the ROR will be over the next 40 years is almost pointless. Although the 4% rule has gotten kicked around some lately, I still like it for determining your worst case outcome. That's why I like back-testing historical results, includes some really bad times. If the world encounters something worst then the great depression of the 1930's, its not going to matter where you money is or what you thought your ROR would have been, we will all be in deep do-do.


   
ReplyQuote
(@hines202)
Reputable Member Customer
Joined: 3 years ago
Posts: 331
 

@giovanelli766 I was just reading a few articles that stated the big fear about those born in 1960 (as my wife was) getting screwed over was much ado about nothing. It turns out the calculation didn't burn them as it looked like it would in the past year or two, due to changing economic conditions.


   
ReplyQuote
(@pizzaman)
Honorable Member Customer
Joined: 3 years ago
Posts: 430
 

@hines202 You are correct, SSA released the National Average Wage Index for 2020 a couple days ago, and it came in at +2.83%, big exhale 🤑

https://www.ssa.gov/oact/cola/AWI.html


   
ReplyQuote
(@hines202)
Reputable Member Customer
Joined: 3 years ago
Posts: 331
 

@giovanelli766 Crisis averted, my wife is happy πŸ™‚


   
ReplyQuote
(@pizzaman)
Honorable Member Customer
Joined: 3 years ago
Posts: 430
 

@howard230 This is a portion of an article written by Michael Kitces in 2013 so it is a little dated (bonds have done even worse): https://www.kitces.com/blog/should-equity-exposure-decrease-in-retirement-or-is-a-rising-equity-glidepath-actually-better/

Of course, some will contend that today's environment will have diminished returns, even over the long run, relative to historical standards. Accordingly, the chart below shows the glidepath results with the return assumptions that Harold Evensky recommends for the popular MoneyGuidePro financial planning software package (arithmetic real returns of 5.5% for equities and 1.75% for bonds, which given their volatility result in geometric means of 3.4% and 1.5% respectively). Notably, these assumptions reflect both a lower overall return environment compared to historical averages, and also a reduced equity risk premium (i.e., the excess return of stocks over bonds is diminished).

I am not sure what the difference is between arithmetic and geometric in terms of real returns.


   
ReplyQuote
 NC
(@nc-cpl)
Reputable Member Customer
Joined: 3 years ago
Posts: 246
 

I ran across this and I'm sure there are other places where the difference is spelled out. At the end of the article it says:

The right set of scales for β€˜weighing’ returns

Self-directed investors are well-advised to ask themselves, or their advisers, how their investments have performed over the preceding 12 months and longer periods of 3, 5 and 10 years (or longer). Take an SMSF trustee without access to the necessary advice or tools. How do they do this? If they compile a spreadsheet with each return from the relevant periods and then simply average them, chances are they are over-estimating their returns.

A point-to-point measure of how an index has moved over a 12-month period is what it is. However, it is easy enough for a self-directed investor to average these 12-month measures incorrectly (ie using a simple arithmetic mean) over multiple periods and, even worse, using the result to estimate future wealth accumulation.

Investing is a journey across many financial periods and calls for a way to β€˜weigh’ those returns properly. The geometric mean is the appropriate set of scales for this job, at least until you look at the difference between time-weighted and money-weighted returns, but that is a topic for another day.

https://www.firstlinks.com.au/why-you-should-know-the-difference-between-arithmetic-and-geometric-investment-returns


   
ReplyQuote
 NC
(@nc-cpl)
Reputable Member Customer
Joined: 3 years ago
Posts: 246
 

an even better explanation (to me anyway - it has some good examples that spell it out): https://www.arborinvestmentplanner.com/geometric-average-vs-arithmetic-average-for-investment-returns/

Conclusion

These examples demonstrate that volatility lowers your investment returns. Arithmetic and geometric averages serve different purposes and only geometric averages will accurately reflect compounded investment returns. Arithmetic averages will always over state investment returns unless there is zero volatility.

The greater the volatility the greater the difference will be between arithmetic and geometric averages. When it comes to investment returns and retirement planning it is compounded (geometric) averages that matter.

The above example provides evidence that even small differences in investment returns can make huge differences in results over long periods of time. The consequence is investors need to put additional emphasis on the amount of volatility they are willing to accept.


   
ReplyQuote
Page 1 / 2
Share: