Math Sux

There is a common joke, begun with Mark Twain (I believe) that there are "Lies, Damned Lies, and Statistics".  Around my house lately the common joke has become "Lies, Damned Lies, and Actuaries".  I have been puzzling over an analysis done by Milliman for PERS.  The document, on PERS' web site, evaluates about 20 different proposals for cutting the cost of the PERS system, and reveals the savings.  It frankly states it does not consider the costs of implementation or the legality of doing so.  It only reports the savings.  What it doesn't report is probably more important than what it does, but what it does puzzles the heck out of me.

I confess that I've never studied Actuarial Sciences, although I am pretty facile with math and especially financial math.  But I am befuddled by a couple of numbers that appear and reappear in various publications and reports, including the savings analysis.  Many, including the Governor, propose some sort of cap on the COLA distributed to PERS retirees.  The current cap is 2% of the benefit received each year, further limited by the actual changes in the CPI for the Portland Metro Area.  Current retirees get the lesser of 2% or the actual CPI change in the previous year.  So, no matter how you calculate this, the maximum savings that could be realized from a COLA cap at 0%, for simplicity, would be 2% of the current retiree payroll.  According to the December 4, 2012 document "PERS By The Numbers", total annual payments to retirees are $3.2 billion.  If we assume that all retirees are eligible for 2% COLA increases, each year PERS' payments would increase by $64 million, or $128 million over the biennium.  Of course, each time the benefits increase, the payroll costs increase, the COLAs go on.  But, while this is going on, members and beneficiaries die, new members are added to the retirement rolls, and so the PERS retiree payroll has probably remained within some reasonable bounds over time.  Thus, assuming a COLA cap of 0% (NOT WHAT IS BEING PROPOSED), the most savings that could be realized in a biennium is about $128 million give or take a couple of million.

According to Milliman's analysis, capping the COLA at 2% of the first $24,000 of annual benefits would yield savings for 2013-15 biennium of $810 million.  If you just read the previous paragraph and asked, "What the heck is going on?", you can join me in the confusion.  The COLA is NOT capped at 0%; the savings cannot be the full $128 million over the biennium.  So, how do you get savings of $811 million from capping a benefit that only costs $128 million every two years?  Interesting math wouldn't you say?  Actuarial math has always been a bit fuzzy to me, and this one leaves me more puzzled than is typical.  I suppose the answers lie in that wonderful concept of "time value of money".  I'm guessing that the actuaries calculated the cost of the COLA out over the full amortization period of cost recovery - 20 years, and then discounted the savings back to a net present value.  If you don't understand this, just think about it as taking all the savings you'd generate over 20 years and then reducing them to a single number that reflects the current cost of buying an annuity that would pay out the benefits in slow dribbles.  I'm sure I'll receive a snippy note from an actuary or from PERS explaining actually how they computed the number, but it all boils down to this.  The savings to PERS proposed by capping the COLA at 2% of the first $24,000 depend on front-loading them to the first year and then hoping all the rest of the assumptions work out over the next 20 years.  If any of the assumptions are seriously wrong, the savings will be far less and we could be back at this same point in another couple of years.  Also, the important assumption is that the cap will NOT BE indexed for inflation.

The short message this silly little exercise should provide is that even if this passes and is ruled constitutional, it is doubtful that either it will achieve the savings projected, or that this would be the end of the retiree tithing to pay for their own benefits a second time.  We paid for them once by working for our careers, and now we get to pay for them again by the Legislature slowly unwinding them over our retired lives.  

Here is another curious number.  The savings to employers from capping the COLA are estimated to be 4.4% of payroll.  So, again, ask yourself how a benefit that costs, at most, 2% of a fraction of the total employer payroll saves them 4.4% of their total payroll.  According to Mercer (Milliman) the total employer payroll for the 2013-15 biennium is estimated at $18.4 billion.  If we take 4.4% of that we get, magically, $810 million.  So again, we are left with the question of how reducing the COLA to 2% of $24,000 could save 4.4% of total PERS employer payroll, when retiree total payments themselves only constitute 18% of total employer payroll, much of which is already safely stored in the Benefits-In-Force reserve, from which retiree benefits come and does not come, in real time, from employers at all.

I write this post not to challenge the savings (actually I do challenge the savings), but to instead point out the curious and fuzzy math used by actuaries and, thence, by legislators to construct budgets built from smoke and mirrors.  

Another number, just for fun.  When the $24,000 first was posted in 2010 as part of a package of legislative concepts for reducing the costs of PERS, the number was announced as an "average" PERS retiree benefit.  While we have taken issue with that number and the derivation in other posts, others seem to be slow to catch on.  In a guest opinion, Tim Nesbitt, a Kitzhaber advisor and former Union honcho, defended the number as reasonable to capture a majority of PERS beneficiaries and prevent them from ever experiencing the cap directly.  When I protested this as inaccurate, Mr. Nesbitt responded to me that the $24,000 was "selected" as an amount that had been determined (by whom?) to be sufficient when combined with Social Security benefits to produce a decent income.  Never mind that on a $24,000 pension the Social Security benefits could vary all over the map.  Never mind that social security benefits can be drawn anywhere from 62 to 70, depending on when a member decides to take the money.  Never mind that it says nothing about how many people have to live on the pension and the social security, and never mind that medical care costs are going up at 4x the rate of inflation.  And never mind that $24,000 covers about 53% of PERS recipients ONLY because it largely excludes anyone who worked more than about 10 years in the system.  It sounds more like the $24,000 figure was selected on the basis of how much savings might result and how loudly they think that PERS recipients are likely to scream.  It is a calculated gamble that says, if we stay away from more than 50% of PERS beneficiaries, the remainder will have a hard time organizing and, even if they do, they'll just sound like a bunch of greedy pigs.

I think what I resent the most about this cynical ploy to deprive career civil servants and public employees - teachers, fireman, policemen, sanitation workers, road workers, and hundreds of other job classifications - of benefits earned, accrued, and promised for a lifetime of work, is the fact that the silly math games are rarely challenged and everyone takes them as an article of faith.  While I don't believe the actuaries are liars, I do believe that actuaries massage numbers to produce results yielding the most confusion, and are unlikely to be challenged.  I watch the PERS Board every other month just accept these numbers as given from God, rarely questioning how they are arrived at and whether they make any sense at all.  This all reminds me of the offensive Barbie doll that announced when you pulled her string:  "Math is hard.  Let's go to the mall."