we dream, we create, we change, we love

How to Lose $50 Billion

leave a comment »

At the end of their 2006 fiscal years, S&P 500 companies that sponsored defined benefit pension plans held some $1.441 trillion in pension assets. That is, that was the amount for the S&P 500 set that existed as of 12/31/2007. For the S&P 500 set that had existed as of 12/31/2004, the end-of-year 2006 pension asset number was over $50 billion higher, at $1.492 trillion.

Where’s the pension asset leakage coming from? Several times each month, the make-up of the S&P 500 set is changed, dropping one company in favor of another. Now although S&P’s selection criteria quite likely does not even look to employee benefits as even one of its many factors, in practice it has not been uncommon during recent years for the exiting company to be a sponsor of defined benefit pension plans, while the incoming company has only defined contribution plans. The pension assets of the departing companies obviously don’t disappear from the pension universe, but they’re gone from the S&P 500 set.

Presto, by using the most recent calendar year end for our S&P 500 set, we’ve “lost” over $50 billion that our numbers would have included if we’d kept the same set we were using 3 years earlier. The chart on this post shows the progression of the pension assets as of the end of the 2006 fiscal year for each of the four S&P 500 sets as of calendar year ends for 2004 through 2007.

All of the previous graphs I’ve posted during the past week used – and unless otherwise specified, all future graphs will use – the single S&P 500 set as of 12/31/2007 for the data for all years. Thus for example, if I show asset growth from 2003-2007, the numbers represent the pension assets during that period for the fixed set of companies in the 12/31/2007 S&P 500. If instead I were to pull the 2006 asset numbers using the 12/31/2006 S&P 500 set, the 2005 asset numbers using the 12/31/2005 S&P 500 set, and so on, then the asset growth over the graphed period would have appeared slightly more muted than it has been for the fixed set of companies. (I’ll illustrate that point with further graphs in future blog posts.)

Which approach is “right” – to use one S&P 500 set for all past years of data? or to use each past S&P 500 set that corresponds to each past year of data? Both, actually. As long as one knows which method is being used and what the numbers are then telling us. Imagine, if you will, if the recent S&P 500 experience were to continue indefinitely into the future, successively switching out DB pension plan sponsors for DC sponsors until only one DB sponsor remains, let’s say only GM, for instance. I know, at the hypothetical extreme that seems extraordinarily unlikely, even with all the recent pension plan freezes and closures and terminations, but the extreme hypothetical does illustrate what is going on even in the ongoing marginal situation. So for an as if, let’s say that all the other pension plan sponsors were bought out by private equity, like Chrylser or Georgia Pacific. And during the period that it takes to do so – let’s say 13 more years, through the end of 2020 – let GM’s pension assets rise from the 2007 level of $117 billion up to $250 billion.

OK then, if we were then to plot out all the pension assets from now through then based on using the S&P 500 set as of 2020, we would get a valid picture: the growth of pension assets for GM through those years from $117 billion to $250 billion, so that assets during the 13 years would have increased an average of about 6% per year, close to the average rate of asset increase for the S&P 500 set for 2007.

But conversely, if we use each interim S&P 500 set for each of the past fiscal yearends, we would get a different story that would be valid in its own right: the decline of pension assets held by the entire set of S&P 500 companies, from about $1.54 trillion at the end of 2007 to our hypothetical $.25 trillion 13 years later, a decline of about 13% per year.

The growth of pension assets for GM in that hypothetical is no more or less valid than the decline of pension assets for the historical progression of year-by-year S&P 500 sets, just as the growth in pension assets for the investor in the single company GM is as important to that investor as would be the decline in pension influence in the S&P 500 set over time would be to the investor of the total market. Both perspectives tell a story, and each of those stories is valid in its own right.

One last word on this point: Notice that our $50 billion was not actually “lost,” merely is not being included in our existing set anymore. The $50 billion number represents companies that used to be in the S&P 500 set that have not liquidated or been swallowed up by other companies, and that had not terminated their pension plans by the close of 2006. A third – and again, equally valid – picture might watch the entire pension universe from what might be closer to the PBGC’s perspective, where we would not only reflect the movement of companies in and out of the S&P 500 set, but further would reflect the recent broader trend away from DB plan sponsorship entirely.

Returning to our hypothetical, this third picture might be like noting that perhaps not all of the pension asset data was lost to private equity buyouts that left the pension plans intact (in which case the 6% increase for GM might actually still reflect the overall pension universe’s experience, even as the lone DB survivor in our hypothetical 202 S&P). Rather, for our third picture, let us further hypothesize that some of the decline from out current $1.54 trillion to that future $250 billion came from pension plan terminations, which would in fact have removed those assets from our pension charts entirely. Of course, the first two perspectives continue to remain valid – for instance, GM’s asset increase in that hypothetical would still be worth the notice as such, as an increase for the then-existing S&P DB plan sponsors – but for the overall picture of the pension plan universe, we would need to recognize the departure of assets via the plan terminations.

More on these ruminations in future posts.

actuary cat feed subscribe to aftermath’s actuary category

bumper sticker [] - actuary

Written by macheide

2 March 2008 at 7:14 am

Posted in άctuary


Please log in using one of these methods to post your comment: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: