Tuesday, August 19, 2014

Conceptual Errors about Risk Trading, Part 1: Defined Benefit Pensions

James Hamilton has a good post on the issues confronting defined benefit pensions.  The San Diego County Employees Retirement Association is facing a shortfall in plan assets.  As a reaction to that, they have approved an investment policy that allows higher risk.

This issue goes to the heart of a prime intractable problem of the modern free world - finance.  In a world of abundance, extensive human capital formation, and longevity, consumption smoothing is necessary, which means, we all have to be financiers.  Finance is a difficult and complex topic, so it is infeasible to expect most people to develop a specialized understanding of the problems of finance.  But, the introduction of agents into such an important and personal issue carries its own problems.

When I am asked about the choice between letting savings ride in a defined benefit (DB) pension or taking the cash value and investing it individually, I sometimes suggest that, in worst case scenarios, DB pensions have plausible deniability.  In other words, in 20 years, when your pension comes up 30% short, do you want your spouse to blame you for it, or do you want to team up with your spouse and blame the pension manager?  I only mean that to be slightly flippant.  Blame avoidance has real utility.

There are at least three levels of risk for a saver, over time:

1) Transitory volatility of returns and asset values.
2) Portfolio-specific permanent negative outcomes.
3) Economy-wide changes in returns and asset values persistent enough to affect lifetime saving results.

DB pensions can, and generally do, solve problem number 1, since distributions are diversified among many beneficiaries over time.  DB pensions could, hypothetically provide some value for problems number 2 and 3, but, because of conceptual errors about risk trading, the way DB pensions are managed may worsen these problems.

One problem is that modern labor and debt conventions create wide expectations of context-specific certainty.  Workers tend to receive $x every hour or every week; bondholders usually receive a set coupon amount for every payment.  Ownership interests receive a premium for accepting all of the near-term volatility.  The problem with normal contracts is that nobody can guarantee all of the risk, so laborers and creditors tend to have binary risk exposure, with the downside being unemployment or default.  This is a much larger problem when it comes to pensions.  In this case, the savers are equity holders, but the terms of DB pensions mimic the near-term certainty expectations that labor agreements have led many people to expect - a broad promise of certainty that becomes catastrophic in unusual outcomes.  The demand for certainty is mostly met with accounting fiction.

And, it appears that pension funds, like the SDCERA, manage their portfolios according to this local-certainty/binary outcome expectation.  They are adjusting their risk profile in order to match a hard target.  This seems suboptimal.  It's clearly not the way someone would manage their own portfolio.  Surely it would be better to allow benefits to adjust with fund outcomes, but principal-agent issues make this difficult.  We wouldn't be able to separate type 2 and type 3 causes, and we tend to want to blame the pension funder or manager for the shortfall.  But forcing expectations of certainty into a context that is inherently risky will always lead to some set of negative outcomes, and hindsight will tend to be brutal in these cases.

We have a bias for certainty in benefits that we don't apply to contributions.  As in DB pensions, this is reflected in the setup of Social Security.  The strength of this bias is clear if we compare the tendency of Social Security benefits to be untouchable while the scale of changes we put up with in contribution rates, program solvency, and implicit rates of return would all be well outside the bounds of acceptable private portfolio management standards.

So, extreme negative outcomes are covered either by taxpayers or by the current workers and shareholders of a corporation with an underfunded pension, which doesn't seem reasonable.  It seems like some sort of pooled defined contribution pension, diversified among managers would create value for savers without creating the moral hazards and accounting gymnastics of DB pensions, but I suspect that there is generally an under-appreciation of how much risk we tend to trade away in our economic lives, and how well-earned the premiums are for taking that risk.  And, this prevents cultural conventions in these arrangements from bending to more flexible arrangements.

I can understand why DB pensions remain in the public realm, where the confiscatory power of the state can back up failures, but it seems like there should be more movement to a pooled defined contribution solution between the typical individual DC programs that the private sector has moved to and the public DB programs that are widely recognized as time bombs.

No comments:

Post a Comment