What's your FIREcalc success rate?

[SecondCor521]

Alan, it's tough out there. I was lucky to work for a good company for many years that was well known for not laying people off -- mostly because they never expanded too fast. But the founders died, the new guard came in, and the tech mania took over.

Les

It sounds like you and I worked for the same company. Did the founders have a garage and sell oscilloscopes to Mr. Disney? I was there 1994-2005; caught the tail end of the good times and watched what you described above.

2Cor521

 

[Lsbcal]

It sounds like you and I worked for the same company. Did the founders have a garage and sell oscilloscopes to Mr. Disney? I was there 1994-2005; caught the tail end of the good times and watched what you described above.

2Cor521

Yep, except after 2000 I worked for the spinoff which was a set of businesses Carly (or was it Platt?) thought she didn't need. Since then it's morphed into more separate entities. Hope it does well as I would like to keep my medical benefits in ER.

Les

 

[Alan]

On Monday we come into work to find that a higher bid has been tabled and we are in a bidding war.

Today we hear that the higher bidder has been accepted, but there won't be any communication meetings until late next week in case another higher bid comes in. I wish they would keep all this private and just tell us when the new owner is known for certain. The first bidder now receives a cancellation fee of $200M We need to sell $2B of products to make $200M profit - sheesh

 

[cute fuzzy bunny]

We need to sell $2B of products to make $200M profit

I think its safe to say that the people making more money off the higher bid wont be around to sell the $2B worth of stuff, or at least wont have to do any of the work.

 

[Meadbh]

You mean "lower" withdrawal rate? Something like: I got 100% success rate at 4%, 3.5%, 2%, and also 1%, right?

I think it would be a competition to see how MUCH you could safely withdraw. e.g. if your can safely withdraw 4% but not 5% and I can withdraw 6%, I must have more reserve.

 

[Sam]

Huh? Care to elaborate?

 

[Meadbh]

Sam, the original discussion several pages back was about shooting for a FireCalc success rate >100%. Obviously not mathematically possible, but CFB suggested

"It sort of does...if you're at 100% you can have it look for a higher withdrawal rate to suit the overage".

I think CFB is correct. The more you can withdraw while still keeping your success rate of 100%, the safer you are. Say you had 100% success at SWR of 2%. Most of us who want that 100% would not feel FI ready to RE. Let's say we save like mad and run it again in a year, and get 100% success at 3%. Close, but no cigar. Another year and bingo! 100% success at SWR of 4%. I would feel FI. But if I waited yet another year and it came out to 100% success at SWR 5%, I would feel positively rich and more likely to send in the resignation letter.

Other people might aim at a lower success rate. Both the success rate and the %SWR that you feel comfortable with reflect your risk tolerance (and confidence in FireCalc).

 

[Sam]

Oh that.

Anyway I was the one bringing up the >100% sucess rate. I was being sarcastic. I need to use emoticons more often.

As far as confidence in FIRECalc, I have no reason to doubt it at all. There is nothing arbitrary about its operation. It's just a model based on historical data.

 

[SecondCor521]

I think CFB is correct. The more you can withdraw while still keeping your success rate of 100%, the safer you are. Say you had 100% success at SWR of 2%. Most of us who want that 100% would not feel FI ready to RE. Let's say we save like mad and run it again in a year, and get 100% success at 3%. Close, but no cigar. Another year and bingo! 100% success at SWR of 4%. I would feel FI. But if I waited yet another year and it came out to 100% success at SWR 5%, I would feel positively rich and more likely to send in the resignation letter.

This makes no sense to me. You're actually safer the less you withdraw, not more. Unless you mean that you're safer because you have the option of withdrawing at a 4% rate but only need 2% of your portfolio to live on.

And waiting a year to get a success rate of 100% withdrawal at a 5% withdrawal rate in year 3 vs a 4% in year 2 will only happen if you significantly change your investment portfolio, investment expenses, or life expectancy inputs, not just because you saved more money.

2Cor521

 

[cute fuzzy bunny]

Hmm...I dont think there'd be a "S" WR at 5%, 4% is the max so far, without skulduggery. But your annual withdrawal amount could go up a whole hell of a lot more. So the thesis is correct, just the terminology needs a little help.

For example, if you have a million, and a 40k swr, but you cant live on that...its not workable. But if you squirrel away a million and a half, with a 60k swr, and you only need 50k...way good to go.

But if firecalc is telling you that you're at 100%, find that upper limit. Find your lowest workable budget. Theres your spread. I'm thinking that if your "happy budget" level is below 100% 4% SWR, and you can live at your baseline for 5 years without hating life...theres no financial reason to still be sitting at work.

 

[rmark]

Firecalc looks at U.S. history only, in essence looking at the worst of the best history, and therefore likely overstates the success rate. Stretching above 4% for typical retirement length periods requires heroic assumptions about complicated spending rules derived from small samples and short time periods. Either your portfolio works at 4% or you need to keep saving.

 

[Lusitan]

One of the things that'd be a nice addition to FIREcalc would be some options for modifying your asset allocation over time. During your working years, especially you're fairly young, most people saving for FIRE tend to invest primarily in equities; but when one reaches FIRE I would think that you'd want to modify your asset allocation to include maybe a bit more bonds.

Right now, I enter into FIREcalc the estimated asset allocation I'll use when I hit FIRE down the road, but the bond portion of that allocation is greater than I am currently holding during my accumulation phase.

I'm not sure to what degree that would actually matter, but just throwing it out there as something I've noticed.

 

[cute fuzzy bunny]

Firecalc looks at U.S. history only, in essence looking at the worst of the best history, and therefore likely overstates the success rate. Stretching above 4% for typical retirement length periods requires heroic assumptions about complicated spending rules derived from small samples and short time periods. Either your portfolio works at 4% or you need to keep saving.

I might agree in some part and disagree in others.

For starters, firecalcs data set starts right at the end of the civil war...a pretty dismal time economically. So we saw an awful lot of upside from that. Start the data set 20 years earlier or 20 years later and you get pretty different results.

Deflation, sometimes severe, was pretty common in the first 30-40 years of firecalc data. Pretty good chance that off the gold standard, with current monetary policy, that we'll be unlikely to see severe protracted deflation.

The US economy in the first 50+ years of firecalc data probably resembled an emerging market far more than a mature one. Returns for those periods may be pretty good compared to future returns.

On the plus side, firecalc (unlike monte carlo sims) give us correlated returns...for example, a monte carlo sim can produce a 25 year downturn but its unlikely that equities would go down for 25 years in the real world, with nobody deciding along the line that they might be a good buy.

Although you can sim your brains out, the real test is pretty simple: aside from Armageddon like scenarios, in which we'll all probably be equally screwed no matter what you invested in...survival through periods like the great depression and the stagflation/sideways markets of the 60's and 70's is probably a pretty good indication that you've got very little to worry about. That you can do anything about anyhow.

 

[Nords]

On the plus side, firecalc (unlike monte carlo sims) give us correlated returns...for example, a monte carlo sim can produce a 25 year downturn but its unlikely that equities would go down for 25 years in the real world, with nobody deciding along the line that they might be a good buy.

After reading statements like this about a thousand times, for once I'm struck with a new thought... how hard would it be to add a correlation component to the variations of Monte Carlo parameters?

Some persistence between one year's returns (up or down) and the next... no abrupt discontinuities... nothing exceeding the worst of the Great Depression or 1966-82.

Computational meteorologists are getting better & better at predicting hurricane tracks and forecasting the weather. Some of their techniques have to apply to financial modeling!

 

[GLM]

from Nords

"After reading statements like this about a thousand times, for once I'm struck with a new thought... how hard would it be to add a correlation component to the variations of Monte Carlo parameters?"

Financial engines sort of addresses this issue in its monte carlo simulations. They do corelate one years inflation to the next and real returns are impacted by inflation.

pulled from their site ....


To estimate the value of your investments, we need to account for the impact of inflation. If the growth rate of your investments is higher than the rate of inflation, then your investments will be worth more in the future than today.

Real rates of return
If your investments grow at 4% and inflation is also 4% that year, you're not making any real progress--you're just treading water financially.

The reason: if inflation of 4% requires you to spend 4% more next year to buy the same things you bought this year, the return means you are barely keeping up. In fact, if you need to pay income tax on the 4% earnings, your standard of living may actually go down.

To see how you're really doing financially, you need to know your real rate of return--after inflation. If your investments earn or grow by 10% and the inflation rate that year is 4%, then your real rate of return is the difference: 6%.

How we model inflation rates
Inflation rates tend to vary over time. When modeling portfolio values, it is very important to consider the possibility of different future inflation rates, as they will definitely affect your standard of living in retirement.

We assume that U.S. inflation rates will vary mostly (14 times out of 20) between 1% and 5% per year, but almost always (18 times out of 20) between 0% and 11%. In extreme cases (1 chance in 20), we assume that these rates can exceed 11% or fall below 0% (deflation).



Putting it all together
How do these three factors of future rate uncertainty, the current year's inflation rate affecting the following year's rate (persistence), and the tendency to return to an average rate (mean reversion) all fit together? Here's an example to explain how these three factors of inflation work with one another:

Suppose the inflation rate this year is 10% and the goal is to estimate next year's inflation rate.

The uncertainty factor alone means that a large range of inflation rates (say, from 0% to 20%) should be considered. But this range is too broad, considering the persistence factor.

The persistence factor says that a 10% inflation rate this year implies a similar rate next year. The combination of the uncertainty factor with the persistence factor narrows the likely range of inflation to 7% to 13%.

However, this 7% to 13% range of inflation doesn't consider the third factor (mean reversion) which says that inflation tends to return to its long-run average (now estimated at 3.5%). This factor lowers the estimate of next year's inflation to a range of 5% to 11%.

All of these estimated calculations are based on the history of inflation rates over the past 50 years, as well as changes from year to year and from decade to decade.

 

[Lsbcal]

I have free access to Financial Engines via Vanguard. However, from what I know of this program it is designed for the accumulation years to tell you whether you will meet your income objective in retirement. I believe it is not designed for retirees -- too bad and frustrating. You can tell it that you are retiring this year and it will tell you if all is well. Also it appears to be stagnant, not being enhanced just supported.

Les

 

[cute fuzzy bunny]

Yeah thats all well and good. Problem is we dont KNOW what most of the correlative factors are and a bunch of them are driven by emotional issues, short to intermediate group psychology, world events, economic situations, war in the middle east, beavers storming downtown portland, godzilla, red death, green death, incorrect line voltage, etc.

I think it'd be easier to pick winning stocks 95% of the time, figure out the weather, or know what the hell my wife really wants to have for dinner when she gets home.

 

[GLM]

I have free access to Financial Engines via Vanguard. However, from what I know of this program it is designed for the accumulation years to tell you whether you will meet your income objective in retirement. I believe it is not designed for retirees -- too bad and frustrating. You can tell it that you are retiring this year and it will tell you if all is well. Also it appears to be stagnant, not being enhanced just supported.

Les

Yes, I stand corrected. Having a way to go before FIRE I am alas still focused on accumulation. Perhaps by the time we get there programs will be available that apply similar logic to scenarios!

 

[dory36]

I thought it was fixed by throwing out the incomplete runs, which pretty much eliminates the false optimism of partial runs by replacing it with the false optimism of avoiding all the bad stuff which happened in the last century. It's not as if we could glue together a bunch of 25-year runs, double their failure rate, and proclaim victory.

Yep -- the earlier version included those partial periods, but after chatting with many folks here, I dropped that in the new version to reduce the false optimism of partial runs.

The tricky part of the partial runs is that, practically, FIRECalc "failures" seem to occur mainly when someone is retiring into a prolonged downturn. Late-term downturns don't have anywhere near the impact. So without real data, it is hard to factor in the later years.

One possible way to ballpark the impact... "stress test" your model by increasing the withdrawals and shortening the term to maintain the same success rates but in terms that include the later years.

For example, with your own allocations and other input, find the withdrawals that get to 100% and 80% success rates for a 15 year retirement, and again for a 60 year retirement (or whatever your long term plan is).

In a quick test with the default input, I found that the 60 year withdrawals were roughly 60% of the 15 year withdrawals, for similar success rates.

So I can include the more recent years by selecting a 15 year term instead of the longer term, and scale the withdrawals accordingly.

dory36
back from the three hour cruise

 

[cute fuzzy bunny]

Good to see that the professor figured out how to use coconut shells and banana peels to patch that hole in the side of the minnow...

 

[Rambler]

Despite a success rate of 100% indicated by FIRE, I have not ERed. I am waiting for 4 years until my youngest daughter enters college.

Same here, except 2 years for me (maybe a few extra months). I have run a variety of scenarios, almost all have given me 100%, except for the scenarios that give me a lot more play money than I need...the other thing is that I'm also working on my psychological readiness to eject...I'm not there yet.

 

[Cadence]

Maybe folks here are concentrating too hard on the scenario success rate. For all Firecalc's strong points, it should be understood that it has chosen the year 1871 to start the first scenario. Had it chosen 1900, also completely arbitrary, the percentages would be a little different. There is no real world distinction between 99% and 100%.

 

[dory36]

Maybe folks here are concentrating too hard on the scenario success rate. For all Firecalc's strong points, it should be understood that it has chosen the year 1871 to start the first scenario. Had it chosen 1900, also completely arbitrary, the percentages would be a little different. There is no real world distinction between 99% and 100%.

Using later starting years is something you can do in the advanced panel. That's mainly to let people skip past the Great Depression, as many asked.

In typical runs (30 year terms or longer), a 1% difference is roughly a single extra failing sequence, since the success/total equation has a denominator of around 100. If you start in, say, 1927 or so, then a single failing sequence will change your success rate by roughly 2%, since the denominator is around 50.

So yes, starting later will perhaps change the numerator (fewer potential failing sequences) and always change the denominator (fewer total sequences.)

 

[deserat]

I've played with the numbers slowly ratcheting down and up the different options in FIRECALC (w SS, w/o SS, changing dates of retirement, changing yearly withdrawals) and frankly, the point at which it goes under 100% is way above my lifestyle costs figure - big factor was the different pensions available to me that are staggered plus the fact that I've saved a pretty good chunk -looks like the magic date is 2009 if I want it to be - it may be for me but not for my husband i.e. I'll retire and he can still work....I'm tired of working. It's amazing how lowering your current lifestyle costs (LBYM) and keeping it that way for a good time gives you so many options early on - and I didn't really start saving until I was 29.