Thoughts on Starting a Pension (or SS) Early

By Alan Silverstein, Fort Collins, Colorado. Email me at
Last update: January 17, 2023



(There's a surprising amount to say about this subject. This webpage, and some related ones referenced here, have grown into more of a shotgun blast than a rifle shot, sorry.)

Here are some thoughts about when to begin taking pension (or SS = Social Security) payments (distributions) earlier (at a discount) versus deferring until you are older (and thus get a higher rate for life).

This applies to pensions or similar vehicles that offer you the option to start early, usually within some range of ages. Normally there's a permanent penalty for starting your payments early. This is similar to how you can start (or defer) your Social Security (SS) benefits during the age 62-70 range for an approximately 5-6.7%/year decrease (or increase) for life in the benefit amount (more on that later) -- before COLA (cost of living adjustments) that mostly take inflation out of the picture.

When considering this choice, an interesting question is: "How long must I live (and continue to withdraw) to break even if I start my benefits early, versus waiting until the pension's full retirement age?"

The "break even" (BE) point is the age at which you have the same total dollars received whether you start early payments or defer them. (Nominal dollars, ignoring inflation for now.) Note well that if you start early (cutting your payment amounts for life), but live past your BE age, then you might wish you hadn't started early because you'd be better off (in total dollars received) if you'd waited. This is kind of backwards from the usual meaning of "break even."

The reduction per year for starting benefits early is a kind of "discount rate". Consider two extreme cases:

  1. If the discount rate is 0% there's no point in deferring benefits (any further). This is the situation once you pass the age at which the pension plan starts to pay full benefits (such as 70 for SS).

  2. If the discount rate is 100% you simply cannot start obtaining benefits yet. This is the situation before you reach the age where you can begin receiving anything from the pension (such as 62 for SS).

Note that what SS calls "full retirement age" (FRA) falls in the middle of the 62-70 age range; exactly where depends on when you were born. Many rules and rates change at your SS FRA, but it's not really your FRA for retirement benefits insofar as a discount rate continues beyond this point up to age 70. Although (and this turns out to be important, as discussed later), framing the boost in lifetime monthly benefits for deferral after SS FRA as an annual increase (at an 8%/year rate) is different than framing it as a discount for starting before age 70 (at a somewhat lower discount rate when calculated backwards).

See also: Social Security Discounts and Break Even Ages, and SS (Social Security) EMTR (Effective Marginal Tax Rate), and the latest edition of this excellent book: Get What's Yours (Revised & Updated): The Secrets to Maxing Out Your Social Security.

And if all this isn't enough, see also: Choosing An Appropriate Discount Rate For Retirement Planning Strategies, a typically excellent, but deep and abstract, analysis of this issue by Michael Kitces. One take-away for me is that the relative discount rate versus other alternative investments should also be considered!

Math (BE = 1/D)

Here's my thinking about pension discounts starting with simple straight-line math (nominal dollars). This ignores for now many complicating factors including inflation or COLA (real dollars), survivor benefits, lump-sum payout options, etc, that are discussed later in this article.

  F = annual total payments at full retirement age (FRA), if you wait long enough
      (the actual amount F doesn't seem to matter, as you'll see below)
  X = years (whole or fractional) that you might start early
  Y = years after your FRA until your BE if you do start early
  D = annual payment discount rate for starting early

For example, for SS D is approximately 0.05 - 0.067 (5-6.7%, but more about this later). My wife qualified for a pension plan where D was 0.03 (3%). That plan's FRA was age 65, but she could start 10 years early at age 55 by receiving 3% * 10 years = 30% less (monthly, for life) than the FRA benefit amount. Note that her pension plan didn't state a discount rate of 3%, but it was clear from their table of payout alternatives.

Now BE (break even) is the point in time when receiving less monthly or annual income (multiplying F by 1 - (X * D)) for a longer time (X + Y) equals receiving more monthly or annual income (FRA full pension amount) for less time (Y):

  F * (1 - (X * D)) * (X + Y) = F * Y

The F's fall out (it's the same math no matter the actual benefit amounts), and this equation reduces as follows:

  (1 - (X * D)) * (X + Y) = Y            # divide by F, meaning F > 0.
  X + Y - (X * X * D) - (X * Y * D) = Y  # expand terms on the left.
  X - (X * X * D) - (X * Y * D) = 0      # subtract Y from both sides.
  X = (X * X * D) + (X * Y * D)          # add all but X to both sides.
  1 = (X * D) + (Y * D)                  # divide by X, meaning X > 0.
  (1 / D) = X + Y                        # divide by D, meaning D > 0.
  Y = (1 / D) - X                        # subtract X from both sides.

Some interpretations of the last two equations above:

I find some of this a little surprising, but the math seems clear. And if I have it right, you should either start as early as you can, or else wait all the way until FRA. Why would you start in the middle? ...Unless as in my wife's case you overlooked the option to start earlier, and began as soon as you figured it out. Or perhaps if you originally chose to defer (gambling on living long past your BE), then are diagnosed with a condition that shortens your life expectancy -- or any other life event occurs where you need more money sooner -- you'd abandon your plan to wait any longer.

More interpretations:


SS Discount Rates (In Particular)

Regarding SS discount rates in particular, here are some nicely done SS webpages I found:

Retirement Planner: Full Retirement Age
Retirement Planner: Delayed Retirement Credits

I learned from these pages that the SS system is chock-full of non-linearities, step-functions, and breakpoints that make mathematical modeling harder although not impossible. That's probably why people prefer to just do spreadsheet simulations for their own situations.

Ignoring spousal issues completely (note well), I found that for a single person:

When approaching age 62 it's probably wise to model various scenarios for your own birth year, hence FRA and D, to see what is the BE in each starting age, at least for each one-year step (although apparently you can start during any month).

I think per my linear model (Y = (1/D) - X), a D that rises with smaller X (like SS does versus age 70 at least up until SS FRA) tends to make the BE more constant, not allowing it to slip out one year-per-year of additional deferral. In other words you are penalized less for initially choosing to defer but then changing your mind along the way. (Does this sound right?) Note however that with SS brackets and step functions, your BE isn't literally constant.

Remember that BE is the point where if you live that long after starting early withdrawals, you might say: "If I'd known, I would have deferred to full FRA to do better from here on out." Deferral only makes sense if you are pretty sure of living at least to the BE point, and if you don't need or want the money sooner, that is, can afford to wait.

However one more SS wrinkle is that you can start your payments retroactively after your SS FRA. So while (I think) they did away with the pre-FRA "do-over" at some point in time (pay back all receipts so far and we'll restart you at the higher rate), apparently you can still get a free "do-over" if you defer past the SS FRA and then, say, get a diagnosis of a life-threatening condition. (Meaning your expectations of making it to the BE point are reduced, hence starting sooner is more attractive.)

Anyway there are definitely other, variable factors (like spousal or personal circumstances) that could figure into your own SS decisions. In particular if you don't need the money immediately but you want to have it in your hands, you could spend more now from your non-SS kitty and "bank on" the increased (by deferral) SS payments later. This can even result in paying lower total income taxes during the rest of your life! (Web search for esoteric details, or find an updated copy of the book, "Get What's Yours: The Secrets to Maxing Out Your Social Security".)

Also if you don't need the money sooner, you can defer SS as a form of increased "longevity insurance" instead of buying an equivalent annuity (probably at a higher cost). Actually SS already is a kind of longevity insurance, should you accidentally make it to age 62+ (grin), but by deferring beyond that point you boost your later stipends. Compare the total cost of deferral (for however many years you choose) and the increased payouts for life, against the cost and payouts for buying a simple deferred annuity.

More About SS (Technicalities)

Yet more thoughts about SS early-start discount rates and break-even times:

I've heard it suggested that deferring SS benefits beyond age 62, if you can afford to do so, is a no-brainer because of the high discount rate D (although they don't use that term). My reaction is, "well, are you feeling lucky?" To win this game, you must outlive many other people in your cohort (or at least your personal BE age), because nominally D is set to be "actuarially flat." Your increased benefits for waiting represent lost benefits promised to others who didn't live as long. (If SS didn't adjust for inflation with COLA then a portion of the increase would also represent inflation and/or nominal investment returns, but that's irrelevant here.)

However, I've also heard that the SS D values were set decades ago when life expectancies were lower. Hence the "expected return" for deferring is positive for most people today.

Anyway I can now summarize as follows. This is output from a Perl script I wrote, for someone born in 1956:

SS discount/bonus rates when starting payments before age 70:
0.4167% (5/12) monthly discount before FRA-36 months
0.5556% (5/9)  monthly discount between FRA-36 months and FRA
0.6667% (8/12) monthly bonus for deferring after FRA
0.5155% equivalent monthly discount rate between this FRA and age 70
Note: The first three values are the same for everyone born 1943 or later.
The last value depends on your birth year => FRA => number of months from FRA
until age 70.  While the bonus rate is fixed at 8%/year, the effective annual
or monthly discount rate versus age 70 depends on the time span.

See how that works? There are two breakpoints or three brackets for determining your D:

  Dtot = f(X) based on the rules above, and then:
  D = (Dtot / X) for your chosen X (early start age)

For people not yet age 70 as of this writing (in 2016), your SS FRA is some variable in the range 66-67 depending on your birth year. Then you discount 5%/year for 62..(FRA-3) (that is 1-2 years depending on your birth year), then 6+2/3%/year for 3 years until FRA, then 6.06-6.45% (equivalent annual discount if you compare with the real FRA) for 3-4 years until age 70.

Given (Y = 1/D - X), you can see how with constant D (like my wife's pension plan), your BE slips out one year-per-year of deferral. But with the SS brackets, D rises (in two steps) as X shrinks, somewhat reducing this slippage although not literally making the BE constant -- as it could have been if they'd just defined it that way and used appropriate math to back-figure your resulting payment based on X!

So here's my real epiphany: For a given value of X and hence Dtot, you can figure an equivalent (D = Dtot / X) which is the average annual D for that value of X. Plugging that into (Y = 1/D - X) you get:

  Y = X * (1 / Dtot - 1) or in other words:
  Y = X * (1 / f(X) - 1)

This should hold up no matter what wacky Dtot = f(X) you choose, based on a table or any equation you like. The SS brackets are just one possible f(X). For example using birth years 1943-1954, "FRA" = 66.0, selected (annual) X => f(X) = Dtot values from the SS website (two different pages), for single beneficiary only:

  62  75.0% (relative to "FRA")
  63  80.0%
  64  86.7% (actually rounded to $0.01 not 0.1%!)
  65  93.3% (actually rounded to $0.01 not 0.1%!)
  66 100.0%
  67 108.0%
  68 116.0%
  69 124.0%
  70 132.0% (true FRA as I see it; your benefit maxes out)

Now to look at the whole picture for the 8-year range 62-70 (same for everyone no matter their birth year and nominal "FRA"), you'd have to first calculate your maximum deferral credit at age 70 (it drops as your FRA rises, to 124% with max FRA = 67), then scale down all the previous Dtot values accordingly.

Of course SS always quotes your benefit based on FRA, not based on age 70, at least in some person-specific contexts. Anyone choosing to view the whole age 62-70 range as one choice of X (versus true FRA = age 70) must beware to use their max (age 70) benefit as the starting point, either as told by SS or computed from their "FRA" value (also from SS) upgraded for their own "delayed retirement credits" schedule.

Yet More About SS (Plainer English)

(After I wrote this section, I later created a separate webpage that focuses more cleanly on the start/defer decision: Summary List of Reasons to Start or Defer Social Security.)

The decision when to start Social Security (SS), earlier (age 62 minimum) or later (age 70 maximum), is quite intricate when you try to view the entire forest and not get lost in the trees. Here is me trying to organize and share more thoughts, really applicable to any pension that lets you start early at a discount and not just to SS, but ignoring SS spousal issues that are even more complex.

All of this would be a lot simpler if we knew for sure how long we had left to live (and other future events too), but important decisions must be made based on just best guesses and most-likely scenarios.

A Few Thoughts on Annuities

Just a few random thoughts about taking any pension as an annuity rather than a lump-sum, or buying an annuity from an insurance company...