Message Font: Serif | Sans-Serif

No. of Recommendations: 4
What are the odds that you will live to age 80? 90? 100? Alternatively, what are the odds that you will need to fund your retirement at least until then? There are many ways to estimate one’s lifespan. But one way to obtain an impartial estimate is to convert a period-life chart (such as is found at the Social Security website) into a betting table, as has been done below. http://www.ssa.gov/oact/STATS/table4c6.html#fn2
`		Male 	FemaleAge		Odds	Odds79		51.0%	64.6%80		48.0%	61.9%81		44.9%	59.1%82		41.7%	56.1%83		38.4%	52.9%84		35.1%	49.6%85		31.7%	46.1%86		28.3%	42.5%87		25.0%	38.8%88		21.8%	35.0%89		18.7%	31.1%90		15.7%	27.3%91		13.0%	23.6%92		10.5%	20.0%93		 8.3%	16.7%94		 6.4%	13.6%95		 4.8%	10.8%96		 3.5%	 8.4%97		 2.5%	 6.4%98		 1.7%	 4.7%99		 1.1%	 3.4%100		 0.75%	 2.41%101		 0.48%	 1.66%102		 0.30%	 1.11%103		 0.18%	 0.72%104		 0.10%	 0.45%105		 0.06%	 0.28%106		 0.03%	 0.16%107		 0.02%	 0.09%108		 0.01%	 0.05%`

Yes, females are going to live, on average, 3-4 years longer than males. But the important question either needs to answer is this. “How much risk am I willing to accept?” If you fund your retirement to a confidence level of one standard deviation, or 68%, you’re making a 2:3 bet that you won’t live past age 85 (male) or age 89 (female). If you lose that bet and do live longer, but meanwhile have run out of money for failing to having created the income-streams you need, you’re in the bread-lines. So that’s the tradeoffs you’re making. It’s easy to fund retirement to 1 StdDev, but the risk of failure is quite large. Funding retirement to 2 StdDevs is tougher, but doing so hugely reduces the risk of financial failure. Funding retirement to 3 StdDevs might be overkill. But such a person should sleep well at night. Reformatting the previous table gives us the following one.
` Odds of Failing If Fund Retirement Only Until Various Ages 	        AgesOdds	   Male	    Female	~32%	    85	     89	~5%	    95	     98	~0.5%	   101	    104	`

If food stamps are taken as a proxy for ‘bread-lines', 1 in 7 people in this country of any age are already in them. And due to the failure of most retirees to adequately fund their retirement, more will be joining them soon. So, what might be a solution? First, they shouldn’t have quit their day job. Second, they need to stop over-estimating their investments-gains and to stop under-estimating their lifespan and the impact of inflation. At an unrealistically-low inflation-rate of 4%, their expenses will double in 18 years and quadruple in 36. At a probably realistic rate of 5%, their expenses will double in 14.4 years and quadruple in 28.8. At a more conservative estimate of an inflation rate of 6%, their expenses will double in 12 years and quadruple in 24.

Now run this thought-experiment. Create a retiree, have him or her embark on their journey at age 65, and then estimate how soon they will have gotten themselves into a squeeze where their expenses are rising faster than they can find income from their investing to meet them. Now comes the tough part. In an investing environment that is projected to offer only modest returns going forward, how does a current retiree maintain his/her present income, much less increase it? For bond-investors, two courses of action are traditionally proposed. Extend maturities, or accept greater credit-risk. But there is actually a third path, and it’s the one I embarked on with this morning’s purchase. Instead of requiring that each new investment's return remain at prior levels (which isn’t going to happen), one can simply increase the amount of money one puts to work. (I.e., \$5k at 10% is the same at the grocery store as \$10k at 5%.)

To see how this might play out, let’s create a fictitious retiree. Let’s make the retiree male, age 65, willing to project a 6% inflation-rate, is current managing a mid-sized portfolio that offers a 10% return, and is carrying \$50k-\$100k in cash that offers nothing. Furthermore, his incomes and expenses look like this:
`SSI	                \$20,000 	COLA indexed to CPIPension	                \$30,000 	no COLA	Investing-Income	\$50,000 	no COLA							Present-Expenses	\$25,000 		Current-Ratio	          4.0					Inflation-rate            6%		Future-Expenses	       \$100,000 		(in 24 years) 			Future-Ratio	          1.0	`

As his bonds come due, he’s going to find it all but impossible (until genuine economic recovery does occur) to put the money back to work at the same rate of return. But if every maturing dollar is matched from his stash of cash and then put back to work at half the rate of return, then his income-stream remains unimpaired, the credit-risk profile of his portfolio doesn't rise, and he gets himself to age 89 with only a 1 in 5 risk of outliving his cash-flows, at which time he can begin to spend down his \$1 Million plus balance-sheet to get him the next seven years to age 95, at which time his odds of living have dropped to 1 in 20, or a bet he’s willing to make for not being likely losing it.

In short, I don’t like buying bonds that don’t offer a real-rate of return after taxes and inflation. But I held my nose and went shopping this morning, because I refuse to buy stocks at current prices, and I refuse to sit in cash while waiting for the correction we all know is coming but might take much longer than any one expects. In short, rather than suffer the negative returns of “mattress cash” while waiting (probably futilely for yields to rise to more normal levels), I’m now buying decent-quality bonds of short to mid-term maturity, regardless of the fact that they fail to offer a real-rate of return after taxes and inflation. In other words, rather than lose an known (-6%) on mattress-cash, I'd rather cut that loss to (-1.5%) or so for the next 5-10 years. There’s not a lot of issuers to choose from. But there’s enough to need my needs, at least until I spend down my cash to more tolerable levels.

Charlie
No. of Recommendations: 6
Charlie,

Good post, but I think it could be a bit better in one aspect.

Studying the SSA's actuarial tables is certainly interesting. I think the analysis has to be taken at least one stage deeper than the odds table presented in your post. That list is the odds at the time of birth. Not many of us have much reason to work from that basis, since we are all variously older than that.

Lets take a male at age 60 as an alternate frame of reference for an example. The original table you linked to has a column that starts with a population of 100,000 and then, for each age, gives the part of that expected to reach that age. For our hypothetical 60 year old that is 85,227. THAT is the basis that person should calculate from. So lets look down that column for half of 85,227 (42613.5). That turns out to be somewhere between 81 and 82, rather than 79. Change to another hypothetical male who is 70 years old and his basis is 72,066 and half of that (36033) falls between 83 and 84. Calculate for a guy who has already made it to 80 and his 50/50 point is between 87 and 88.

(Yes, I know there is a "Life expectancy" column in the chart. I like the confirmation from performing an alternate calculation that proves to me I understand what I am looking at.)

Anyway, I think the analysis in the rest of your post (good stuff!) should be adjusted for each individual based on their own age rather than using calculations based on age zero, which leaves everyone short of their actual expectancy.

RH in CT
No. of Recommendations: 2
RH,

Thanks for responding.

For sure, there are many ways to estimate one’s life-span. E.g., one can use the ‘interview approach’. What were the ages of your parents/grandparents when they died? and causes? Do you smoke? Exercise? Have a heart condition? Etc. But the questionnaires are too easy to game. You know what answers you should give, and it’s too easy to obtain an unrealistic result.

At the other end of the spectrum, one can simply guess a number, which isn’t as bad an idea as it sounds. On average, those who think they will live a long life, do so by an average of seven years longer than those they think they are short-lived.

A third approach is to pull a number from an actuarial life table, such as is found at the SS website. But the number can be pulled from the table in several ways, one which is your method of taking the 'half-life' of the remaining cohort, given one’s present age. But that doesn’t tell me anything about the distribution of the right-hand tail, i.e., the likelihood of losing the bet that one won’t live longer than that projected average, nor by how much. That’s why I prefer the “last man standing approach”. If I don’t fund my retirement until I’m the last man standing, and I do live to be that last man standing, then I could end up in the breadlines. (OTOH, if I die earlier, my heirs receive a chunk of unexpected money.)

What is the hope of most people? That they don’t run out of money before they run out of life. But my bet is that they are going to lose that bet, because they under-estimate their lifespan, under-estimate inflation, and over-estimate their investing gains. (See Easterling's work on this topic. http://www.crestmontresearch.com/docs/Stock-Retirement-SWR.p...) When I run my numbers, I’m out of money at age 132 or something equally ridiculous. So what I was trying to do was to stress-test my approach. If retirement is only funded to 1StdDev of one's life-expectancy (however that life-expectancy is estimated), then approximately one in three times the person ends up in the bread-lines. Funding to 2 StdDevs cuts the failure-rate to about one in twenty cases. Funding to 3 StdDevs cuts the failure-rate to about one in two hundred cases, or something close to an acceptable risk.

Financial planners glibly do 20-40-year projections for their clients that have a 100% chance of having a chance of failing, nor can they quantify what that chance is, because, as Taleb has proven in his technical papers, one can't use the known sample of rare events to estimate their distribution. I prefer a “Safety First” approach for my retirement income-streams, the easiest of which is to have 4x-6x coverage, most of which is disconnected from market-pricing. Yes, for sure, “One nuclear bomb could ruin your whole day.” So, no financial plan is truly bomb-proof, and I do accept the possibility that I might fail for systemic causes (e.g., the US defaults on its debts and/or WWIII).

But I don’t intend to fail because I’ve made the avoidable mistake of under-estimating how long I might live. And I’ve got bad odds. My grandmom died at 96, her brother at 103, her two sisters at 105 and 109. If I’ve inherited those genes, then there’s no question that I need to fund myself to at least 3 StdDevs of my life-expectancy, if not four, which, fortunately, is already in place. So my problem isn’t to amass more capital than I already have, but to keep it moving forward against the time when it might be needed. But that isn't most people's problem. Their problem is far more basic. They have no idea how long they will live, nor how much money they will need in retirement, nor where it will come from. Their hope, of course, is for a return of the bull market of the '90s, or when that doesn't happen, that the government will bail them out of their failure to do basic financial planning.

Charlie
No. of Recommendations: 0
Charlie,

A third approach is to pull a number from an actuarial life table, such as is found at the SS website. But the number can be pulled from the table in several ways, one which is your method of taking the 'half-life' of the remaining cohort, given one’s present age. But that doesn’t tell me anything about the distribution of the right-hand tail, i.e., the likelihood of losing the bet that one won’t live longer than that projected average, nor by how much.

My only point of disagreement was basing the distribution on age 0 rather than current age. I agreed completely with your insistence at looking at the distribution, not any one number. That's why I long ago took the SSA actuarial table you linked to into a spreadsheet and performed my own calculations. My spreadsheet shows my numbers for every age from 59 to 99 (with a graph). So, as of when I ran it at age 59, it showed I had a 25.3% chance of reaching at least 88. A one in four chance of being destitute doesn't appeal to me! Looking out to age 92 gave a 12.2% chance, which seems a lot less intimidating for me as nobody in my line got past mid-eighties. I may not have used standard deviation but I still got the full picture.

RH
No. of Recommendations: 0
RH,

Using my 'Last-Man-Standing' method (and the 2007 SS table) suggests your odds of reaching 88 are 21.8%. That's ball-park enough to your 25.3% estimate for you to worry, for me to worry. That was my only point. Most people are sure they won't live longer than the money they've budgeted for retirement, and in most cases, they won't. But the consequences of losing that bet are severe, like, they now have to go begging for food at a time of life when charity resources are going to be over-stressed due to so many other people having the same problem.

I think Americans have a very unrealistic view of retirement as being something they have a right to, rather than something they earn. If they hate their day job so much that they want to escape it, they should get a different one. And even when they do 'retire', a new one will be dumped into their lap, namely, the need to manage assets. When I finally realized just how much time effective money-management requires, I realized that I didn't retire. I merely changed jobs. The boss is nicer, and the hours are more flexible. But I'm still putting in as many hours per week as I averaged over the 30 years that I did project work as my means of support, namely, about 25 hours per week. Some years when I was working, we regularly put in 84-hour weeks, especially toward the end of the project. Then we might have a couple months off to fish, to plant gardens, to play with the kids, etc.

But far beyond any monetary benefits, investing is fun. There isn't a day that I'm not engaging new problems, conducting basic research at the same level and intensity as I did in grad school. But, also, 'enough is enough', and I demand that my afternoons and evenings be free, as well as any day or week I want to take off for other other things, like a good fishing trip or checking up on the grandkids. So in that sense, I am retired, and because I've done my financial planning, I know I can sustain that lifestyle indefinitely. My bet is that many people will find themselves having to go back to work or have to go begging. They make no effort to estimate worst-case scenarios and then defend against them. No defenses are fail-proof. But not getting their life-span right isn't a mistake they should be making. Last-Man-Standing planning sets a realistic upper-limit, namely, if you can't fund your retirement until you become the last man or woman still alive in your cohort, you're taking on avoidable risk.

The cost of minimizing that risk (of out-living one's retirement assets) is huge. Of that there can be no doubt. That's why so many people want to fudge their life-span number, so they can reduce their funding requirements. "Well, no one in my family lived past age such-and such. Therefore, I don't have to plan for living any longer, either." BRRZP! WRONG. That is a classic instance of misusing statistics. The sample size is too small to draw valid conclusions. This is an error those working in the field of behavioral finance document all too commonly. That's why I like the impartiality of an actuarial table, converted to a betting table. I can't fudge the fact that I could possibly live to age 108, but probably not much longer. That caps my responsibilities. Therefore, I know how much risk I have to take in my investing, or how much risk I can avoid. And if I guess wrong, Big Whoop! I had fun while I was accumulating some assets that my kids now get to spend, or if they're wise and if I've trained them properly, they'll roll them toward their own retirement, and yet another generation can enjoy the economic freedoms this country does make available to them that work to achieve them. Were my Dad still alive, he'd be gratified to see that his kids have done well due to his support and example, as have his grandkids, as will their kids in turn, an immigrant family that built a niche for themselves in the new world, one dollar at a time, one day at a time, one generation at a time.

Charlie
No. of Recommendations: 2
RHinCT,

You wrote, Studying the SSA's actuarial tables is certainly interesting. I think the analysis has to be taken at least one stage deeper than the odds table presented in your post. That list is the odds at the time of birth. Not many of us have much reason to work from that basis, since we are all variously older than that.

Thanks for posting that. I was just playing catch up on this board and read Charlie's post and was thinking the exact same thing.

While the average life expectancy of a man is 75, I hope to retire between the ages 55 and 58. Assuming 58, my average life expectancy is already up to about 80.5. If you go to the first deviation, I'm looking at potentially surviving to 87 (29 years). To cover the second deviation, I need to plan to live to 95 or 96 (37 to 38 years).

I honestly wouldn't consider anything beyond the second deviation; but even that is a pretty daunting target full of uncertainty...

And ironically the later you plan to retire, the older you should expect to die. It's a real-life example of the Monty Hall Paradox! ;-)

- Joel