A more realistic way to calculate for your retirement savings

[MooreBonds]

You've mentioned that you're willing to adjust spending downward if "times are bad".

Flexibility is definitely a big deal in terms of initial spending. But, you haven't specified the circumstances where you'll adjust downward. What if the market is flat instead of returning 3% real in the very first year, is that enough to make you adjust? If so, how much? What if it's down __%, what will you do then?

This is the biggest flaw in people who take the "I'll simply spend 5%/6% and can cut back if things start to look bleaker".

As Independent notes, what is the definition of "bleaker"? see-sawing market for 3 years with 0 nominal returns? Interest rates get cut in half (which hurts your fixed income allocation because you were 50% in bonds, and nearly all of that was in CDs that you can't sell, and can't reinvest in higher yielding instruments with a higher yield)? Interest rates double, and your 40% bond allocation gets hammered, causing 20% portfolio losses (on paper), while stocks barely rise?

If the market drops 10% in year #3, is that enough of a cause to slash spending 25%? When is it ok to resume? When it rises back up to its previous level? When it rises 20% from the low? From where do you draw your divine knowledge of how much to affect spending now to still turn out ok when things go sour?

The advantage with FireCalc is that based on historical actual returns, you know that things would have turned out ok with a given initial withdrawal rate.

With some homemade formula, you don't know what kind of impact a 5.5% initial WR will have if things are so-so for the first 5 years, then have a sizable drop in year 6-8. How much would you have had to cut back in spending in the past to weather the storm? How much do you have to cut back NOW to turn out ok?

Your homemade formula is not able to tell you this using historical events for comparison. I'd much rather at least have the past as a guide, rather than some formula that simply tells you what a given value will be in X years using simple compound interest (which is not helpful in a drawdown environment).

 

[daylatedollarshort]

Interest rates get cut in half (which hurts your fixed income allocation because you were 50% in bonds, and nearly all of that was in CDs that you can't sell, and can't reinvest in higher yielding instruments with a higher yield)? Interest rates double, and your 40% bond allocation gets hammered, causing 20% portfolio losses (on paper), while stocks barely rise?

I think for the bond portion that is why some people use ladders and get a rolling average of interest rates and hold to maturity.

TIPS and I bonds, even at a zero real return, over 40 years would provide a 2.5% annual withdrawal rate. Right now the thirty year TIPS are at .86% + inflation.

 

[Options]

...

I used firecalc, ********, financial engines, i-orp, my own spreadsheet, fidelity RIP and every other calculator I could get my hands on. Some gave me a smaller nest egg number than others... But they were only shooting for a 95% success rate.

For me, personally, I wanted a 100% with firecalc - plus left the option on the table to cut expenses and/or sell my primary home (not included in the nest egg) if things go south.

Others are fine with an 80% success rate in firecalc... Everyone has different levels of comfort and what will let them sleep at night.
...

Emphasis added.

+1
I don't like/trust my own spreadsheets, so also use (and like very much) ESPlanner, VG retirement estimator and Schwab calculator (more out of curiosity than anything else), and some others whose names I've forgotten. I too prefer 100% calculator success rate (with a cushion and Plan B, C, D) as my comfort levels require a more conservative SWAN PF. Compared to others here, I do not consider myself having an unreasonably large PF.

 

[itsme2]

The advantage with FireCalc is that based on historical actual returns, you know that things would have turned out ok with a given initial withdrawal rate.

Even with using FireCalc, if the market has been down X percent for Y number of consecutive years, one will still worry about running out of money sooner than he planned and need to decide how much he needs to cut his spending, no matter what "success rate" he got initially.

Therefore, this is not something a model can decide for you.

 

[brewer12345]

Is 78% success rate good enough? I have no idea. Because I don’t know how its calculation came about. That is why I asked “what the acceptable success rate should be” earlier in this thread.

Does FireCalc generate the number based on historical market index? If it does, does it include dividend? For example, your stock could go from $50 to $55 in one year (a nice 10% return), but it may also pay out $2 dividend. So your effective return is $7/$50 = 14% (instead of 10%, if calculated purely based on the stock price). The discrepancy will definitely get larger over 2-3 decades.

I would be more comfortable with my $461k result than most people here, since I derived the formular on my own and I tested with lots of different scenarios.

No matter what model we use, when “times are bad”, we will all have to face the reality and cut spending. Even for those who really have sufficient amount of savings, they are likely to spend less due to psychological reason. So, it just won’t be me using my model.

You brought up the criteria of downward adjustment in case of market downturn. I have not put much thought to it, as I have been mainly focusing on improving my model on the initial lump sum. But, in my recent reading, I did find this interesting article, on downward/upward adjustment rule.


Rethinking a 4% Withdrawal Rate

"The initial withdrawal rate is between 4.8% and 6%, based on the stock/bond mix. At the end of each year, the retiree takes the preceding year’s withdrawal amount [in dollars] and adjusts it for inflation. But there are guardrails against big market swings: If that amount divided by the current portfolio balance equals a withdrawal rate of 20% more or less than the initial rate, the retiree adjusts the amount they withdraw that year. No annual withdrawal is more than 10% more or less than the year before." (My interpretation is that, with +/- 20% market swing, you adjust by +/- 10%)

I have had occasion to fool with models (chiefly as a user) professionally and before you can really use a model with any confidence you really have to poke and prod it to make sure you understand how it works, what assumptions were made, and what shortcuts were taken. It is clear you have not done so with firecalc or the other models out there. I would suggest that whatever you use WRT your own spreadsheet, you spend the time with some of the other models out there.

Developing plan B, C, D... is a very important component of planning regardless of what the model(s) you use say. Worth thinking about.

 

[Gumby]

I have had occasion to fool with models (chiefly as a user) professionally and before you can really use a model with any confidence you really have to poke and prod it to make sure you understand how it works, what assumptions were made, and what shortcuts were taken. It is clear you have not done so with firecalc or the other models out there. I would suggest that whatever you use WRT your own spreadsheet, you spend the time with some of the other models out there.

Developing plan B, C, D... is a very important component of planning regardless of what the model(s) you use say. Worth thinking about.

Solid advice.

 

[gauss]

Neglecting any SS income is one of the most common conservative assumptions that I have seen. The fact that the information provided by SSA in the form of benefit statements and online modelers do not readily lend themselves to ER studies I believe compounds this effect.

I always ignored SS in my studies of ER as a young man, but in my mid-forties
I finally dug in to it to do a proper analysis which I will not repeat here.

As a very rough value of neglecting SS benefits consider the example where one has accrued a $2,500/month benefit if taken at age 70. To replace this income I pulled a quote today from immediateannuites.com. The rate shown was approximately $400,000. This quote, however, is for a non-inflation adjusted annuity. The $2500/month number was created in my case as working 22 years as a engineer.

Since DW and I are dual earners, that figure would be doubled to $800,000. Adding a true inflation adjustment and survivor benefits, the figure would likely be between 1M and 1.5M for a married couple similar to us.

Now some will say that there is risk in planning on SS. I agree. That is why I always discount this by about 1/3 to 1333/month or $20,000/year as opposed to setting it to zero.

It was in my mid-forties when I properly accounted for my expected value of SS payments, that I realized that I was way pass my goal for ER and when life events changed, I was ready to pull the plug.

Note that the 4% savings number is based on your expenses after subtracting any streams of income such as SS. If you ignore SS in this case you will get artificially high numbers IMHO.

Not really looking to start a debate on the future of SS here but just wanted to show the large value of SS and the potential error if ignoring it.
-gauss

 

[Big_Hitter]

No, not correct. Yes, you are missing a lot.

i.e. it ain't easy being an armchair actuary...

 

[Independent]

Is 78% success rate good enough? I have no idea.

1) Because I don’t know how its calculation came about. That is why I asked “what the acceptable success rate should be” earlier in this thread.

Does FireCalc generate the number based on historical market index? If it does, does it include dividend? For example, your stock could go from $50 to $55 in one year (a nice 10% return), but it may also pay out $2 dividend. So your effective return is $7/$50 = 14% (instead of 10%, if calculated purely based on the stock price). The discrepancy will definitely get larger over 2-3 decades.

2) I would be more comfortable with my $461k result than most people here, since I derived the formular on my own and I tested with lots of different scenarios.

No matter what model we use, when “times are bad”, we will all have to face the reality and cut spending. Even for those who really have sufficient amount of savings, they are likely to spend less due to psychological reason. So, it just won’t be me using my model.

3) You brought up the criteria of downward adjustment in case of market downturn. I have not put much thought to it, as I have been mainly focusing on improving my model on the initial lump sum. But, in my recent reading, I did find this interesting article, on downward/upward adjustment rule.


Rethinking a 4% Withdrawal Rate

"The initial withdrawal rate is between 4.8% and 6%, based on the stock/bond mix. At the end of each year, the retiree takes the preceding year’s withdrawal amount [in dollars] and adjusts it for inflation. But there are guardrails against big market swings: If that amount divided by the current portfolio balance equals a withdrawal rate of 20% more or less than the initial rate, the retiree adjusts the amount they withdraw that year. No annual withdrawal is more than 10% more or less than the year before." (My interpretation is that, with +/- 20% market swing, you adjust by +/- 10%)

1) Probably, if you're going to start a thread that says "I've built a better mousetrap", you ought to have at least a basic idea of what the current mousetraps do.

Yes, all these historical or stochastic calculators use or assume total return. Many have the option of specifying investment expenses as well.

2) I'm a big fan of DIY calculators because I want to see how things come together. I've built my own, like many people here. Realize that a simple present value calculation is just a start.

3) Yes, "flat with guardrails" is another class of adjustment strategies. All adjustment strategies involve trade-offs between "how much I might have to cut spending" vs. "how much I reduce the probability of outliving my money". Before I picked such a strategy, I'd want to so some sample calcs to see how severe those trade-offs are. I can't do that with a simple deterministic model.

 

[RetireAge50]

A stochastic model shows a range of results. It has hundreds or even thousands of different scenarios that could happen based on history (Firecalc) or chance (Monte Carlo). The model results show a distribution of outcomes with a few worst case scenarios, quite a few bad scenarios, a lot of middle of the road scenarios, some good scenarios, and a few great scenarios.

A deterministic model uses the average of all the stochastic results. It is the expected result (but in reality it will not happen as only one of the scenarios will actually happen).

Some observations

Some choose to go with a deterministic model without adjusting. This is the most dangerous as a bad sequence of returns or just a bad scenario in general will cause a drop in withdrawals or even loss of entire portfolio.

Others go with the Stochastic model but ignore all the scenarios except the worst of the worst. They reduce their withdrawals immediately just in case the worst happens (100% safe withdrawal rate).

Others are aware of the stochastic model results but base their current income on some kind of average of all the results (instead of only the worst case). They adjust these withdrawals going forward as needed.

It is interesting that some here criticize the OP for using a deterministic model while at the same time they are only using a single worst case scenario for their plan. Both of these approaches have advantages and disadvantages but I cannot say one is right or one is wrong.

I believe the OP stated he is adjusting his withdrawals going forward and I believe this is a good approach.