When to take Social Security

[ERD50]

Quote:

Originally Posted by Spanky ERD50, The graph is from this article written a professor in accounting/finance: http://www.nysscpa.org/cpajournal/20...ntials/p42.htm He suggests that age 62 is the best time to take SS benefits.

The trouble is - the definition of 'best' is a purely fiscal one. He who dies with the most toys 'wins'.

C-T is working under a different definition of 'best'. He is not concerned with leaving money to heirs, but does not want to run out if he happens to survive to a ripe, old age. If you think about SS as an annuity, delaying SS effectively annuitizes more of his income. And this appears to help meet his stated goals.

As you effectively sell off a portion of your portfolio to fund the increased SS payment, you are decreasing the Nest Egg - not as much needed with the larger SS payment and not as much left for heirs. But, a stream of income from Uncle Sam for as long as C-T is breathing.

I have little doubt that the professor in accounting/finance did his calculations correctly. But he was not addressing C-T's situation, so it may not be fully relevant.

-ERD50

[jdw_fire]

First some housekeeping for my analysis of Bikerdude's numbers.  Some of my numbers need to be changed: the amount divided over 8 years should have been $323k which drops the yearly income between age 62 & 70 by $500 to $53,455.

And now to your comment

Quote:

Originally Posted by FIRE'd@51 The problem that I have with this analysis is that the multiplier you are using should not be 25.  I know you are using it because it is the reciprocal of the 4% SWR.  But the 4% SWR doesn't apply to the average 70 year-old who, according to the table Spanky posted (which is from the SS web site) has a life expectancy of 12.8 years.  I think you should be using a much lower multiplier. I went to the Vanguard annuity calculator and priced a lifetime inflation-adjusted (using CPI-U) annuity for a 70 year-old man.  The SWR is 6.85%, which implies a multiplier of 14.6.  Redoing your analysis with a 14.6 multiplier leaves you with 14.6 x $13,080 = $190,968 to spend down over the 8 years from 62-70, or $23,870 per year.  So the residual portfolio becomes 650K - 191K = 459K.  Applying the 4% SWR to the 459K yields $18,360.  So you have a total amount to spend from age 62-70 of $18,360 + $23,870 = $42,230 about the same as the original $43K you would get from taking 17K SS at 62 and adding it to 26K (4% of 650K). So it seems to me that all this additional spending power from age 62-70 is the result of choosing a multiplier which is way too high.

My example is from the perspective of a 62 yo person deciding whether to take SS at 62 or wait and take it at 70, i.e. what can he do at age 62 to increase his yearly income for the next 8 years.  I don't think using the 4% rule at age 62 is way out of line (heck even the bunny agreed to it).  Remember that the choice here is whether to have a plan that starts at age 62 and produces a real income of $43k/yr ($17k/yr from SS plus $26k/yr from the $650k portfolio using the 4% rule)  OR  have a plan that produces a real income of $53,455/yr from age 62 to age 70 ($40,375/yr from $323k/8 plus $13,080/yr from the $327k portfolio using the 4% rule) and a real income of $43k from age 70 on ($29,920/yr from SS plus $13,080/yr from the $327k portfolio using the 4% rule).  Which ever plan you choose, the WDs from the portfolio using the 4% rule start at age 62.

[ERD50]

Quote:

Originally Posted by FIRE'd@51 The problem that I have with this analysis is that the multiplier you are using should not be 25. I went to the Vanguard annuity calculator and priced a lifetime inflation-adjusted (using CPI-U) annuity for a 70 year-old man. The SWR is 6.85%, which implies a multiplier of 14.6.

Fire'd@51, I think we went around on this a bit on the previous thread, but let me look at it a bit differently this time.

What if we try to value the SS difference as you did? Not by how much it reduces a portfolio that we are only willing to draw 4% from, but by looking straight at the 'cost' of the additional payments received for SS. Like this:

Assume SS would be $10,000 for a 62yo. $17,600 if delayed to 70yo. The 70yo gets an added $7,600. But he had to 'pay' for it by not getting the $10,000 for 8 years. So that is $80,000 plus add in a compounded real rate of return of 4% on that increasing balance should get you about $95,800. So the added $7,600 SS annuity cost him $95,800, or a SWR of about 7.93%, or a multiplier of 12.61.

Is that correct?

The reason I went through that exercise is that it appears the Feds are giving a better 'deal' than Vanguard is (easy when it is not 'your' money!). If that were not the case, I might suggest that C-T should not delay SS, but use the money instead to buy an annuity from Vanguard. It looks like SS is the better deal, but C-T may want to supplement that at age 70, as the govt only offers X amount.

-ERD50

[youbet]

An article over at the Retire Early Home Page discusses the pros and cons of cola'd annunities and concludes that an older retiree could increase his/her SWR by 1% or more, at the expense of giving up some/all of their portfolio to the insurance company. The author does not recommend cola'd annuities for younger (example = 45 yo) retirees.

To the extent that delaying SS is in effect purchasing a cola'd annuity, the article has some relavance.

http://www.retireearlyhomepage.com/inflannu.html

[jdw_fire]

Quote:

Originally Posted by youbet An article over at the Retire Early Home Page discusses the pros and cons of cola'd annunities and concludes that an older retiree could increase his/her SWR by 1% or more, at the expense of giving up some/all of their portfolio to the insurance company. The author does not recommend cola'd annuities for younger (example = 45 yo) retirees. To the extent that delaying SS is in effect purchasing a cola'd annuity, the article has some relavance. http://www.retireearlyhomepage.com/inflannu.html

If you want to see how a COLAed annuity might increase your SWR check out the example I gave in this old post http://early-retirement.org/forums/i...6091#msg136091 and the discussion in the rest of that thread. BTW if you think SS is a heated discussion just look at the discussions on annuities.

[New Thinking]

Quote:

Pay particular attention to the legislative articles that Dalton links to. And before the same misinformation that floated the last time Daltons article came into play, NO, he did not later admit that he screwed up any calculations and NO, he did not retract any portion of his article.

CFB - I don't know where you get your information, but I can tell you that Dr. Dalton admitted his mistakes in the CPA article to me personally. I have proof but would not post without his blessing. He did not take the increases into account by inflation from age 62 to FRA to age 70 and these do cut down the "cross-over ages" by a few years. Many scholars have made this mistake in the past.

[cute fuzzy bunny]

He submitted a letter to the editor and noted that while his article created a furor, that he stood by it. In particular, he noted that he DID inflation adjust the amounts appropriately.

So, I dont think so.

With regards to the 25x...no, you're not going to live to be 100, but why not tie another arm behind the alternatives back just to be sure you get the 'right' number. And substituting an annuity for the money market isnt a big step up either.

Although I'd trust the vanguard annuity to pay the quoted amounts far more than anticipating that the SSA will be paying you the full quoted amount of SS 38 years out, which is where you'd be if you were 62 and looking at a 30 year run after age 70 for comparison purposes. I'm quite sure near term numbers are pretty safe. Every other dashboard indicator says that wont be the case 38 years from now. Too many retirees, too few workers, not enough money to go around.

Quote:

Originally Posted by vimwick then why when i took 3 minutes to run the numbers (as cfb suggested) did firecalc indicate it was better (higher revenue stream) to delay?

If you feel comfortable with the inputs you provided, then you should delay receiving benefits. However, I didnt see a link posted to look at your results. Perhaps something was out of whack.

I have seen situations where people position the input to firecalc as being a decision made immediately, and with a very small portfolio where the failure rate is low early in all the runs due to the immediacy issues, but the same run calculated ten years in advance or using a reasonable sized portfolio tips the balance the other way.

So the results can be 'gamed' unintentionally. Which is why I recommend that people visit this 'decision' as part of their financial planning NOW and not wait until they're 62. Much of the potential "benefit" of income stream substitution and higher portfolio ramp up returns is neutralized at that point.

Not sure I'd promote retiring at 62 with only 500k and social security either. I'm quite sure that doesnt end well given these 100, 107 and 120 year life spans you guys all think you're going to get to...whether you take late social security or not. Probably good if you want to live in a foreign country and suck the morning dew from the leaves under the tree you're sleeping under but...

[Independent]

Quote:

Originally Posted by Bikerdude I was under the impression that FC added inflation only after you had reached the year of your SS payment. Are you saying FC adds inflation from 2007 to SS at a later date? I've been putting the inflated amount from the SS web calculator for future dates. If you're right that's double dipping.

I got my results from putting in $24,000 of SS starting in 2007, and then $32,000 of SS starting in 2011. I didn't inflate the $32,000 before I entered it (i.e. I did not say "let's assume inflation is 3%, so the $32k will increase to $36,016 by age 66, so I need to tell FireCalc $36,016 starting in year 2011.)

I still get a higher "safe" spending plan by delaying.

Dory is the authority on FireCalc. I didn't see this issue in the FAQs in FireCalc.

But I'm pretty sure I'm doing it right. When I did my two runs, I downloaded the Excel files that show the portfolio balance at the end of each year. Then I "checked" the calculation by rolling forward from one year to the next, using the SS benefits above. Note that if you do that, all the values are shown in "constant" dollars (inflation has been backed out). I did not get "perfect" matches, but I wasn't off by anything as big as 12.5%.

[cute fuzzy bunny]

Dory has previously reported that firecalc inflation adjusts every number entered, so the social security numbers you enter should be the ones from the SSA web site or report you receive.

Look at the great grampa of social security threads...

http://early-retirement.org/forums/i...hp?topic=142.0

[ERD50]

Quote:

Originally Posted by Cute Fuzzy Bunny With regards to the 25x...no, you're not going to live to be 100, but why not tie another arm behind the alternatives back just to be sure you get the 'right' number.

You seem to be hung up on this long life span thing, using it as a 'proof' that people are twisting the numbers.

I can make the decision to plan for an unlikely event. That is different than saying I *expect* that event to occur.

I don't *plan* or *expect* to live to 100. But, I cannot rule it out, and I don't want to be in the position that I am dependent on my kids (who would be in their 60s and 70s) for support in those final years. So, to the extent that I can prepare for it, I will.

I don't expect my house to burn down, yet I have insurance against it. I don't expect to get sued, but yet, I have umbrella insurance as some protection against it.

Quote:

Although I'd trust the vanguard annuity to pay the quoted amounts far more than anticipating that the SSA will be paying you the full quoted amount of SS 38 years out,

IMO, this is a worthwhile debate (but in a separate thread - please!). As I said before, I think we should understand the underlying assumptions before layering other scenarios on top of those assumptions.

Quote:

I have seen situations where people position the input to firecalc as being a decision made immediately...

What is wrong with this? IMO, it is the most sensible approach. You can't act on the decision until you are at age 62. So it makes sense to model it as if you are age 62, ready to decide. I am looking at it now, just to get an understanding of the principles and decision points. But no way I would 'decide' prior to age 62. Things may change between then and now. For me this is an exercise, for people near age 62 I hope it is of some immediate value.

Quote:

Not sure I'd promote retiring at 62 with only 500k and social security either.

Depends on your spending level. I've seen a wide range of numbers on this board. Some people seem to be happy with an income that I would consider 'dew sucking' already. The default in Firecalc is $30,000 spend. So if we changed that would be accused of 'gaming' the inputs? So earlier, you said you wanted to see a FireCalc run support a SS delay. Now that you have been shown them, you say those inputs are unreasonable?

-ERD50

[FIRE'd@51]

Quote:

Originally Posted by ERD50 Fire'd@51, I think we went around on this a bit on the previous thread, but let me look at it a bit differently this time. What if we try to value the SS difference as you did? Not by how much it reduces a portfolio that we are only willing to draw 4% from, but by looking straight at the 'cost' of the additional payments received for SS. Like this: Assume SS would be $10,000 for a 62yo. $17,600 if delayed to 70yo. The 70yo gets an added $7,600. But he had to 'pay' for it by not getting the $10,000 for 8 years. So that is $80,000 plus add in a compounded real rate of return of 4% on that increasing balance should get you about $95,800. So the added $7,600 SS annuity cost him $95,800, or a SWR of about 7.93%, or a multiplier of 12.61. Is that correct?

I think your methodology is very plausible. It probably comes close to the "correct" multiplier. It also allows one to estimate the fees built into the Vanguard (AIG) product - looks like about 1% per year, although their pool of annuitants probably have a longer-than-average life expectancy since 70 year-olds in poor health presumably don't tend to buy annuities.

A "theoretical" way to estimate this without any reference to the return on early SS payments might be to take the current real interest on TIPS (about 2.5%) and plug it into a financial calculator as follows:

PMT = 7600

FV = 0

i = 2.5%

N = 13 years (the life expectancy of a 70 year-old from Spanky's table)

PV (calculated) = 83,500

SWR = 7600 / 83500 = 9.1%

multiplier = 11 (not far from your result)

[ERD50]

Quote:

Originally Posted by Cute Fuzzy Bunny Dory has previously reported that firecalc inflation adjusts every number entered, so the social security numbers you enter should be the ones from the SSA web site or report you receive.

I need to look at this more closely. I have a non-cola pension that I can take in 3 years or delay and take 2X in 13 years. It looked to me that FireCalc was inflating it between now and when I take it, but I need to review the excel output again to be sure. This *may* be a 'bug' in FireCalc. I'm out the rest of the day, or I would test it now.

-ERD50

[New Thinking]

Quote:

He submitted a letter to the editor and noted that while his article created a furor, that he stood by it. In particular, he noted that he DID inflation adjust the amounts appropriately. So, I dont think so.

Well here is the letter that you refer to and I am sure you read it as it is on the web. But he does not say he put the inflation in appropriately, he implies that he did not..I guess others can read it and make their own conclusions.

"The author responds:
The effect on a spouse regarding the decision of when to begin Social Security benefits is an important point. My calculations consider only one person, not a husband and wife together. When a spouse is involved, the choice becomes more complicated and requires a more sophisticated analysis. I appreciate the insight offered by the readers’ comments.
Mr. Saglinbeni’s estimate (that it would take 14 years of collecting Social Security beginning at age 65 to equal the benefits one would receive if benefits began at age 62) is consistent with my analysis in the article. I estimated that it would take 14 to 15 years to reach equality if the normal retirement age is 66 (i.e., 81 years less 66 years). Certain assumptions (such as COLAs), however, can significantly influence this estimate.
Although the reduction in benefits for working and collecting Social Security was modified in the Clinton administration, it was not eliminated. The rule still applies to all who receive benefits before their normal retirement age. The rule is correct as stated in the article."
Thomas M Dalton, PhD, CPA
University of San Diego, San Diego, Calif.

[haha]

Thanks for posting his letter. I was not likely to search the web for it, because I would rather trust analysis than debate the bona fides of gurus.

Still, it is interesting as an example of how someone might stretch or even ignore what is printed in order to make a point.

Ha

[Peter]

Quote:

Originally Posted by HaHa Still, it is interesting as an example of how someone might stretch or even ignore what is printed in order to make a point.

Indeed. But then some people have developed a high level of skill in that technique ...

Peter

[vimwick]

Quote:

from cfb: If you feel comfortable with the inputs you provided, then you should delay receiving benefits. However, I didnt see a link posted to look at your results. Perhaps something was out of whack.

no point in inserting a link. you'd have to check every input. from advanced firecalc's default state, please enter the few parameters explicitly stated in the post. if i can take 3 minutes to run the numbers, surely you can take 1 minute to check the results

maybe there is something wrong with firecalc or my input. please check. if not, please concur that firecalc indicates that delay is optimal (in terms of maximum safe revenue stream) instead of continually insisting otherwise

again, thanks to all for this interesting and informative thread!

[SecondCor521]

I don't know if this will help any, but I thought I would throw it out there for consideration.

I am currently taking my last MBA class on the subject of financial modeling and decision making. The instructor has been discussing a framework for solving these problems that is based on the following areas:

1. Inputs. These are the "knowns" or givens of the problem. For example, SS full retirement age of the individual is something that is an input to the problem. Inputs would also include any assumptions one wanted to make, for example, the average SS COLA.

2. Objective function. This is what you're trying to maximize or minimize. In my own case, I would be trying to minimize my age at FIRE. Another objective function might be to minimize the risk of running out of money. Yet a third objective function might be to maximize the amount of money spent between ages 62 and 70.

Note that, depending on the objective function one chooses, other variables in the problem become fixed inputs. For example, my objective function would have an input of my annual after-tax spending rate, whereas the second objective function would have an input of retirement age.

3. Decision variable(s). These are the things you're willing to change in order to achieve your objective function. I think that everyone here is discussing the decision variable "what age to take SS?" (1)

4. Constraints. These are the "subject to" factors, or limitations that one imposes on the solution. One example of the constraints that I would have is that my year end portfolio balance would never be negative for the rest of my life. Another person might add a constraint that their SS income never be more than 25% of their total income (maybe they don't want to rely on SS too much and want to mitigate their risk that way).

5. The Model. This is what you build to solve your objective function by changing your decision variables, based on your inputs and subject to the constraints. This would be the part of the spreadsheet that presumably projects out over the next X years one's spending, portfolio balance, age, taxes paid, etc. In my own case the model would be based on Firecalc -- that is, I would want to model my historical portfolio survival rate based on historical stock returns and inflation. My own model might also incorporate IRS mortality tables. A simpler model might be built on a static portfolio rate of return and a fixed death age.

I would respectfully submit a few things. First, once you identify the above five areas, the solution to the problem is pretty straightforward; if I identify to you precisely my choices for the above five areas, we should both be able to come up with the same answer. (2) Second, the solution to the problem is very dependent on all of the above five areas; if you and I have different constraints, or different objective functions, for example, we'll almost certainly get different answers (or if we get the same answer it's just a coincidence). Finally, I think many people on this thread are arguing past each other because they're arguing what the outcome of the decision variable should be even though they have different inputs, objective functions, constraints, and models. By my first and second points, unless you agree on those other things first, arguing about what the result should be is senseless.

2Cor521


(1) I don't think it has been noted here yet -- maybe it has -- but one can take SS any time after the early SS age (62 in most cases, I think). The benefit amount is reduced on a monthly basis prior to full retirement age and increased on a monthly basis after full retirement age but stops increasing at age 70, I think. The point is that the decision variable really should be allowed to range from 62 years 0 months, 62 years 1 month, ... 69 years 11 months, 70 years 0 months, and not just 62 vs. full vs. 70.

(2) Provided each of us hasn't made any miscalculations. Given that these models can be pretty complex, the fact that people make mistakes in their calculations adds an additional level of difficulty to the discussion. It's really easy to double count cash flows, forget to change a relative cell reference to an absolute cell reference, have an off by one error in an NPV calculation, etc.

[vimwick]

nuts!

[Bikerdude]

Quote:

Originally Posted by SecondCor521 I don't know if this will help any, but I thought I would throw it out there for consideration. I am currently taking my last MBA class on the subject of financial modeling and decision making. The instructor has been discussing a framework for solving these problems that is based on the following areas:

OK guys, funs over we got to add 'structure' to our discussion.

Good post SecondCor521!

[OAG]

I have really enjoyed reading ALL of these SS threads AND injecting some minor input to them. But gosh does anyone make a decision and then learn to live with it (the decision) anymore? It is like looking for 6% versus 5.75% on a CD for 10K; the difference is not much. Granted SS differences can be larger especially if we assume we will live forever. No one will make the decision UNTIL at least age 62 (preceded by the applicable application processing timeframe) and then remember it is revocable at any point before death. Just pay back the money and reset the clock. Now I have blown my initial "live with it" decision.