I've been too scared to do it personally, but the Permanent Portfolio (1/4 Gold, 1/4 Long term treasuries, 1/4 Stocks, 1/4 cash or short term treasuries) has done surprisingly well compared to a 60/40, with much less volatility.
Affordability of Insuring Against Long-Tail Risk: Personally I think that trying to drive risk to Zero isn't affordable, at least for me.
But there is probably an ideal rate of "insurance" between 0 and 100%, e.g. 5%-10% Gold on the efficient frontier, maybe commodities.
@jkandell - Agreed.
And that's why I chose to buy a 5-15 year TIPS ladder, not a 30-50 year tips ladder.
To me, the incremental risk reduction isn't worth the incremental cost.
As far as retirement planning/finances, I am not sure what you guys mean by Insurance? A few posts up, the type of major challenges enumerated weren't Black swan events, they were more like end of life as we know it events, for which there is no form of "Insurance" coverage, unless Stuart adds a tab for the cost of building and fully stocking a doomsday shelter in Arizona 😱. The best Insurance for Black Swan events like the global pandemic and great recession is a healthy equity stake in your asset allocation (AA). Your stock funds build up a financial buffer to handle these events. What was the best thing to have been done during the pandemic, assuming you had a roughly 60/40 AA for the past decade or so with about two years of emergency money set aside? Nothing! Except maybe do some Roth conversions during the stock fire sale. During the great recession? Same thing, nothing, except add to your equity AA during the huge fire sale on stocks and/or replenish your bonds/funds.
We are starting to talk about things that PRC is not designed to handle and maybe out of the scope of this thread.
Would anyone here be interested in sharing and discussing their Pralana retirement plans? Could be publicly here in the forum, or via a community Zoom call, or privately via emails/calls.
Personally I'm greedy to receive feedback from other experienced users regarding what I may have missed in my own planning, but I'm also hoping to hear such discussions about the specific retirement plans of other Pralana users in order to learn best-practices, and even share my own feedback on others' plans where I think I might be able to add value.
Thank you @PizzaMan, @JKandell, and @Ricke for your thoughtful feedback on my retirement plan.
For me, this exercise was a success largely because of the valuable critiques that you provided.
Since I want you to know that your time and energy was well spent and is appreciated, I'd like to let you know what I changed in my own retirement plan based on your feedback on my plan and others shared in this thread.
How this thread helped improve my retirement plan:
- Roth Conversions
- I found an additional ~$85k (~13%) in spendable cash by hand-tuning the Roth Conversion plan suggest by Pralana-optimizer
- LTC
- Increased LTC from $300k (couple, net incremental LTC) to $500k (couple, average full spend)
- Pulled in LTC expenses from age 97-99 (end of plan) to age 85-89 to make Monte-Carlo analyses @ age 90 (my life expectancy) more representative of my most likely lived experience
- HSA
- Adjusted HSA plan to ensure HSA spends-down before death; Withdrawing from HSA before ROTH for all qualified medical expenses.
- Capital Market Assumptions
- Updated my CMA from a 1-period model to a 3-period model
- Areas identified for further study:
- Asset Location: Evaluating locating TIPS Bond Ladder in Tax Deferred v. Taxable
- LTC: Consider assuming 25/10%-ile LTC costs
Thank you all so much!
-Kevin
Capital Market Assumptions--Updated my CMA from a 1-period model to a 3-period model
After reading this thread I did the same, switching to (1) one set of returns/sd for the next 10y and then (2) another set of returns/sd for the remainder of my 41 year span after that (with some gliding in between). If I indeed think returns will be much lower for the next decade or so because of high valuation and return-to-mean, that doesn't imply those rates will carry over for the entire forty years of my plan!
The Income Lab folks did an interesting simulation of this. They refer to using different capital market expectations in different periods as "regime based monte carlo". To test its accuracy, they simulate using monte carlos with two CMA periods: one ten years from the year in question, the other for years 20-30. To determine CMA for the shorter period of "the next ten years", they used a cape-filtered analysis of the historical returns previous that point. For the "long" 30 year period they used the 30 year averages previous that point. They compared this "regime based" monte carlo to normal Monte carlo (with a single set of market expectations) and to historical simulations. By comparing predicted success versus what actually happened, they found that a historical method and regime monte carlo both jived with reality the best overall. Both were highly inaccurate, but 25% more accurate than traditional monte carlo. What they call the “historical method”, which was the most accurate of all, is similar to Pralana's, tabulating each previous trajectory up to that point. “Traditional Monte Carlo“ (based on a single CMA for the entire plan, using the previous 30 year averages) were less predictive. The authors postulate that the historical analysis was the most accurate because it incorporated momentum and mean reversion, whereas the other methods didn't.
Given these results, I wish the authors had also compared regime-based monte carlo against a regime-based historical analysis like Karsten Jeske does at earelyretirementnow. Jeske sifts for only the historical trajectories matching the current level of the CAPE and SP500 at the start of the period (e.g. currently cape > 20 and Sp500 at a high), but is otherwise a straight historical analysis like they simulate in the article (or like Pralana’s). If taking into account current conditions works for monte carlo, why wouldn't it also work for historical analysis? Since the raw historical method was the most accurate of all (and more accurate than regime-based monte carlo), you'd think they would test the same idea for itf.
One important takeaway from their analysis: don't confuse a monte carlo "probability of success" with what you'll see in the real world. That doesn't make them useless by any means, as they show the span of what you might get. But don't take their numbers ("_ probability of success") too literally.
Programmers are like deer in the forest, you don't want to scare them away, you want to lure them in and make them think they are going to be able to do the request easily. 😀
Here, that "easy" feature would be basic bootstrap MC. Instead of a pure historical analysis, the user would tell it how many historical years to pick at a time and a random number generator would pick the index year, fill in the requested number of years and repeat until that iteration of the plan was full and then do that all over again for each iteration. You wait until you get the basic version before you start lobbying for more advanced versions.
Programmers are like deer in the forest, you don't want to scare them away, you want to lure them in and make them think they are going to be able to do the request easily. 😀 Here, that "easy" feature would be basic bootstrap MC. Instead of a pure historical analysis, the user would tell it how many historical years to pick at a time and a random number generator would pick the index year, fill in the requested number of years and repeat until that iteration of the plan was full and then do that all over again for each iteration. You wait until you get the basic version before you start lobbying for more advanced versions.
My post wasn't aimed at Stuart and Charlie to add anything to Pralana. It was simply validating what Kevin had decided to do with regard to regime monte carlo. In my feature request to Pralana some time ago, I stuck with historical bootstrap like you describe since it would be quite easy to implement and intuitive for users to understand.
Retirement Income
- Deferred compensation: 2026 = $735k, 2027 = $205k
- Social Security_SELF = ~$40k/yr starting age 70
- Social Security_Partner = ~$30k/yr starting at HER age 70
I know i already commented on your plan, and this is late to the party: but have you run the numbers at opensocialsecurity.com to see the optimal claiming strategy for both of you as a pair? That site is terrific because it allows you to calculate total longevity-weighted npv. And you can fiddle with everything: discount rate (eg 20y tips rate or your portfolio rate?), various longevity tables, cuts in ss, etc. Most couples find the maximum value results from lower paying partner claiming at 62 while the higher at 70. If you work in claiming options to your roth conversions, you might end up with a different final plan, too. Then again, without getting “into your business”, you may not desire to merge finances if you’re not married.
Retirement Income
- Social Security_SELF = ~$40k/yr starting age 70
- Social Security_Partner = ~$30k/yr starting at HER age 70
I know i already commented on your plan, and this is late to the party: but have you run the numbers at opensocialsecurity.com to see the optimal claiming strategy for both of you as a pair? That site is terrific because it allows you to calculate total longevity-weighted npv. And you can fiddle with everything: discount rate (eg 20y tips rate or your portfolio rate?), various longevity tables, cuts in ss, etc. Most couples find the maximum value results from lower paying partner claiming at 62 while the higher at 70. If you work in claiming options to your roth conversions, you might end up with a different final plan, too. Then again, without getting “into your business”, you may not desire to merge finances if you’re not married.
Hello @jkandell,
Thank you for the suggestion! I always appreciate ideas for improvement.
I had not tried opensocialsecurity.com before, although I'd certainly heard of it. Upon your referral, I checked it out and it seems like a useful specialty resource.
Since we are not married, it's not clear exactly how best to use opensocialsecurity.com for our case. Without marriage, there is no social security survivor benefits and that means that the typically optimal "couples strategy" for the low earning partner to claim at 62 and for the high earning partner to claim at 70 doesn't pay off as well.
In our current plan we've been thinking of Social Security like an inflation-adjusted annuity, in that it's the best longevity insurance we can buy, and we're both planning to delay claiming until age 70 in order to maximize our guaranteed income in the worst-case scenarios where our retirement plan fails.
I sometimes run scenarios exploring the financial consequences of marrying and filing our taxes jointly; as you might imagine the outcomes look materially better in the Married scenarios due to the higher tax brackets. Going forward I will incorporate this 62/70 claiming strategy into those scenarios as well.
Thanks so much,
Kevin
I had not tried opensocialsecurity.com before, although I'd certainly heard of it. Upon your referral, I checked it out and it seems like a useful specialty resource.
Since we are not married, it's not clear exactly how best to use opensocialsecurity.com for our case. Without marriage, there is no social security survivor benefits and that means that the typically optimal "couples strategy" for the low earning partner to claim at 62 and for the high earning partner to claim at 70 doesn't pay off as well.
It's still valuable as a way to optimize individual claiming ages. (And to see the opportunity cost of staying single.)
In our current plan we've been thinking of Social Security like an inflation-adjusted annuity, in that it's the best longevity insurance we can buy, and we're both planning to delay claiming until age 70 in order to maximize our guaranteed income in the worst-case scenarios where our retirement plan fails.
Since it generates the longevity-weighted NPV (i.e. taking into account both the opportunity cost of not claiming, and also that each year's PV only "counts" according to its probability of you still being alive), you can think of opensocialsecurity as pricing and optimizing that very longevity insurance. It's going to depend heavily on the discount rate you use and the actuarial longevity table you choose, but within those parameters the age with the maximum payout is the optimal longevity insurance. You can also explore e.g. "break even" discount rates where two claiming strategies are equal.
I sometimes run scenarios exploring the financial consequences of marrying and filing our taxes jointly; as you might imagine the outcomes look materially better in the Married scenarios due to the higher tax brackets. Going forward I will incorporate this 62/70 claiming strategy into those scenarios as well.
Presenting a pralana report would be a romantic marriage proposal IMO-- Heck, most men just do a ring. Down on one knee: "Honey, look at this tabular report column 3." But then you'll need to invite Pralana to the wedding.
