Blanchett Paper on 'Impact of Guaranteed Income on SWR'

[Huston55]

This subject is frequently discussed here, although usually in a tangential way regarding safe withdrawal rates & methods, what percentage of one's FIRE income is "guaranteed", whether to use 'fixed' or 'variable' withdrawals, and the like.

However, this paper analyzes it directly and produces some outputs & guidance that are useful for a range of scenarios that fit most situations. It also analyzes & compares 'past' results (back testing) with 'forward looking' results (actual forecasting), which I find especially useful. It's well worth a read and, for me, prompted some thought about our current FIRE income/SWR approach. The summary and link to the paper are below.

Well done Mr. Blanchett!

Conclusions
To better serve retirees and those saving for retirement, financial planners need to move beyond heuristic-based initial safe withdrawal rates. Results from this analysis suggest that optimal initial safe withdrawal rates varied significantly when guaranteed income was considered, from approximately 6 percent when 95 percent of wealth was in guaranteed income, versus approximately 2 percent when only 5 percent of wealth was in guaranteed income.

https://www.onefpa.org/journal/Page...drawals-on-Safe-Initial-Withdrawal-Rates.aspx

 

[joylesshusband]

Well done indeed.

Thanks for posting this!

 

[2017ish]

As with any of these analyses, there are nits to pick. The biggest one in this article, at least for me, was the limitation of spending cuts to 5% year over year. If one's discretionary spending exceeds nondiscretionary spending, that would be short sighted.

Nonetheless, a very good article that has been added to my "reread periodically" bookmark folder.

Thanks!

 

[DrRoy]

Looks like a good study of the subject.

 

[Gumby]

Table 5 is confusing. If you have high guaranteed income and your spending is entirely discretionary, your probability of success should be approaching 100%, not zero.

 

[marko]

...Results from this analysis suggest that optimal initial safe withdrawal rates varied significantly when guaranteed income was considered,...

Can one assume that this is what FireCalc (and others) takes into consideration when asking if/when SS will come into effect?

 

[audreyh1]

Well, I don't think I have the time to wade through this as I got bogged down easily, and Monte Carlo analysis is used as part of the modeling, and personally I'm not crazy about Monte Carlo for portfolio survival as it ignores history.

And I'm not sure the premise:

The potential impact of guaranteed income on safe withdrawal rates is important given its potential impact on consumption when the portfolio fails. For example, the safe initial withdrawal rate (or amount) for a household with $100,000 in annual guaranteed income and a $1 million portfolio should not be the same as a household with $10,000 in annual guaranteed income and a $1 million portfolio, because the impact of portfolio failure on consumption is significantly different for the two households; however, this fact is typically ignored when determining safe initial withdrawal rates based on the probability of success

as it always seemed clear to me that guaranteed income simply reduced the needed income from a portfolio. So in the above cases, the portfolio can provide the same annual income to both households, and household A simply has $90K more in total annual income than household B. So what differences would there be? Certainly household A can consume much more than household B. Might they choose higher risks with their investments? I guess it depends on how high their non-discretionary spending is. Ultimately you have to live within your means.

And where does 2% safe withdrawal rate if only 5% income guaranteed come from? That doesn't pass the smell test to me.

 

[bmcgonig]

Table 5 is confusing. If you have high guaranteed income and your spending is entirely discretionary, your probability of success should be approaching 100%, not zero.

I THINK what he's trying to say is that the probability of success that *you should aim at * when running MC ( or Firecalc) is close to 0. I.e., you have so much guaranteed income that you care much less about risk. (E.g., run firecalc with a success rate of 75% or down to 0 and not 95% as we regularly do here)

The drift that I get is that he's suggesting that the probability of success that we strive for when running Monte Carlos can be a lot less for someone that doesn't need the money i.e., that has a lot of guaranteed income, and that therefore they can use a higher SWR.

Seems intuitive but he's putting real numbers on it.

 

[Gumby]

I THINK what he's trying to say is that the probability of success that *you should aim at * when running MC ( or Firecalc) is close to 0. I.e., you have so much guaranteed income that you care much less about risk. (E.g., run firecalc with a success rate of 75% or down to 0 and not 95% as we regularly do here)

The drift that I get is that he's suggesting that the probability of success that we strive for when running Monte Carlos can be a lot less for someone that doesn't need the money i.e., that has a lot of guaranteed income, and that therefore they can use a higher SWR.

Seems intuitive but he's putting real numbers on it.

Your interpretation makes sense, but his writing is not entirely clear. To me, this seems like observing that the sun rises in the morning. The precise time it does so is of interest to only a few people.

 

[audreyh1]

Someone doesn't need the money so the FA should point out that they can take more risk with their investments and/or draw more heavily if they'd prefer to spend more sooner?

He did say something about targeting more of a 75% success rate than a 95% success rate might be appropriate, I assume for people who can afford it, but since he is using Monte Carlo analysis which tends to lower success rates I don't know how that corresponds to FIRECALC succes rates.

 

[Hermit]

Well, I don't think I have the time to wade through this as I got bogged down easily, and Monte Carlo analysis is used as part of the modeling, and personally I'm not crazy about Monte Carlo for portfolio survival as it ignores history...

Someone doesn't need the money so the FA should point out that they can take more risk with their investments and/or draw more heavily if they'd prefer to spend more sooner?

He did say something about targeting more of a 75% success rate than a 95% success rate might be appropriate, I assume for people who can afford it, but since he is using Monte Carlo analysis which tends to lower success rates I don't know how that corresponds to FIRECALC succes rates.

Monte Carlo simulations should not ignore history. They should select values from a range that represents history. The problem, IMHO, is that often the calculators are not clear on how the numbers are selected. With FIRECALC you know exactly how the numbers are selected, but the chance of history repeating to that extent is very slim also. I have used many calculators and I have not seen a major difference in results that would change my retirement perspective. Then again, I will be one of those who has adequate guaranteed income to support my lifestyle when I turn 70.

 

[Independent]

I THINK what he's trying to say is that the probability of success that *you should aim at * when running MC ( or Firecalc) is close to 0. I.e., you have so much guaranteed income that you care much less about risk. (E.g., run firecalc with a success rate of 75% or down to 0 and not 95% as we regularly do here)

The drift that I get is that he's suggesting that the probability of success that we strive for when running Monte Carlos can be a lot less for someone that doesn't need the money i.e., that has a lot of guaranteed income, and that therefore they can use a higher SWR.

Seems intuitive but he's putting real numbers on it.

Yes, that's what I get.

Years ago, I happened to be involved in a project at w*rk on this topic, at the same time I was trying to plan my own retirement. It took some puzzling to come to that basic conclusion.

In the words we used - people have "basic" income needs, they need to be nearly 100% certain of covering that amount. In addition, most have "target" or "aspirational" income wants. Unless you're very risk averse, you can be more aggressive with this additional amount.

 

[Huston55]

Updated Commentary

Here's some updated commentary on the original article from Wade Pfau. I see no new revelations but, it's interesting to see another's comments.

https://retirementresearcher.com/important-factor-determining-retirement-withdrawal-rate/

 

[Huston55]

Another (somewhat) Related Article

Many (Most?) of our E-R.org community members seem to already do this analysis themselves, in one way or another. But, I think this article, and example, dovetail well with the OP and are interesting enough to post & discuss.

https://retirementresearcher.com/wh...4VipndHehqhcVGnjX8LzaR18dXfy8g&_hsmi=54264849

 

[Gumby]

There are a substantial number of variables that can make it difficult to calculate the present value of either assets or liabilities. How long will the post retirement income last, when will the major health expenses occur, how long will George and Martha live, what discount rate should you use for future income, what inflation rate for costs? I'm sure you can think of others.

Many people will like this approach, but I find it easier to think in terms of cash flows needed and able to be generated, which is what FIREcalc does for me.

 

[Huston55]

There are a substantial number of variables that can make it difficult to calculate the present value of either assets or liabilities. How long will the post retirement income last, when will the major health expenses occur, how long will George and Martha live, what discount rate should you use for future income, what inflation rate for costs? I'm sure you can think of others.



Many people will like this approach, but I find it easier to think in terms of cash flows needed and able to be generated, which is what FIREcalc does for me.

True. But, all those same variables must also be estimated for any MC/backtesting analysis (FIREcalc included). So, I don't think one approach is necessarily more accurate than another.

While I'm like you (I typically like to approach it from the cash flow side), the primary message I got from this article/approach is the "guaranteed" (or "reliable" if you prefer) income message; similar to the message in the OP.

 

[Blue Collar Guy]

I didnt let some details get in my way of rerunning Firecalc, hahah. It now says at 75.6 % success i can SPEND 170k for 51 years. I deducted 30 % from my portfolio for taxes and this is what i got. As usual Im spending a bit below my means. This is how most of us in here got to this point. That fire engine red caddy is getting closer and closer to my driveway.

 

[foxfirev5]

My simple minded interest and divy plan allows me to forego these studies and calculations. Now let's go for a boat ride ( as soon as DW fixes her hair , makeup etc.)