A Roll Call article today looks at my positions on Social Security personal accounts and public employee pensions, finding a seeming contradiction:
As a top aide to President George W. Bush, Andrew Biggs argued for allowing workers to funnel payroll taxes into stocks instead of the Social Security trust fund backed by Treasury bonds. But Biggs has now emerged as a leader in prodding public pension funds to use a new gauge — based on Treasury bonds, not stocks — to evaluate unfunded liabilities.
Put more simply: “Biggs denies any change of heart of about the value of stocks as retirement investments,” the article says. I also deny beating my wife, but who’s going to believe that?
This claim of hypocrisy or bias has come up a number of times, including as part of my work with the Society of Actuaries Blue Ribbon Panel on public pensions. And at first sight, it seems pretty compelling: I advocated personal accounts investing in stocks, and even argued that account holders would have done well in the most recent market downturn. Yet today I’m arguing that public pensions should assume that their equity investments will return no more than Treasuries. What gives?
In most personal account plans, an individual could choose to invest part of his Social Security taxes in an account, with the option to hold stocks if he wished. The issue then was how to describe the benefits he might receive from this account. SSA’s actuaries focused on the ‘expected benefits’, based on assumptions regarding typical returns for the assets held by the accounts. Even then, SSA provided illustrations of higher or lower returns, which goes much farther than public pensions do today in illustrating the effects of market risk. At the time I supported this approach.
But over time other approaches became more prominent. The CBO used “Monte Carlo” simulations to illustrate the full range of possible outcomes for account holders, which I supported. And CBO also argued, strongly enough to convince me, that if a single “point estimate” was to be provided, benefits should be calculated using a bond return to account for the fact that stocks are riskier than Social Security is. In other words, so that accounts didn’t appear to be a free lunch. But to be clear, these were descriptive issues: the costs of the plan to the budget wouldn’t be affected at all by how you chose to think about investment returns.
But in the mid-2000s came several personal account plans, sponsored by then-Rep. Jim DeMint and Rep. Paul Ryan, in which the government would have guaranteed accounts against market downturns. And a guarantee, as I’ve argued in my public pension work, is a lot different than a mere “expectation.” If the worker’s account balance wasn’t sufficient to pay his full promised benefit, the government would make up the difference. For the individual it was no-lose: you get at least as much as Social Security promises, and probably more. Better yet, SSA’s actuaries argued that such guarantees would be relatively cheap. Thus, a personal accounts plan could guarantee participants a Social Security benefit at least as high as current law at lower cost than the current program. That’s a budget score that personal account proponents like me should have jumped at.
Except that it was wrong. That government guarantee is very similar to a “put option” – a financial product that gives the holder the right to sell a stock for some minimum price in the future – and, I argued, it needed to be priced as options are, using the so-called Black-Scholes formula. And once you do, you find that such guarantees are really pretty expensive, so much so that these guaranteed accounts plans would cost more, not less, than the current program. And in 2006 I co-authored an article showing why that was true.
Similarly, when a public employee pension guarantees participants a given level of benefits but funds those benefits with risky assets, there’s an implicit put option – provided by the taxpayer – that would top up the pension fund if its investments failed to generate their projected returns. And that’s precisely the logic I used in a 2011 journal article on public pension liabilities: once you count the contingent liabilities that risky pension investments place on taxpayers, the plans are a lot more expensive than you think. Unfunded liabilities that are reported at less than $1 trillion dollars become something north of $4 trillion.
That 2011 paper also shows that you don’t need to actually price the implicit put options that guarantee public pension investments. Through a principal known as put-call parity, pension liabilities calculated using this options pricing approach are mathematically identical to simply discounting the pension’s liabilities at a riskless rate of return.
This shows something important: that, contrary to Dean Baker’s somewhat-disingenuous claims in the Roll Call article, my arguments for discounting public pension liabilities using a Treasury bond rate don’t in any way assume that pensions invest in Treasuries. I can generate the exact same liability numbers based on the pensions’ own investment portfolios of stocks, bonds, real estate and alternative investments. All I need to do is calculate, based on the characteristics of those portfolios, the cost of guaranteeing that they’ll produce the returns they promise.
So the conclusion isn’t simply that there’s no inconsistency between my positions on Social Security accounts and on public employee pension accounting. It’s that my views on pension accounting are derived directly from my views on personal account plans, in particular plans that would guarantee account holders against low market returns. In both cases, the taxpayer is taking on a big liability. Yet, in both cases, the taxpayer would be kept in the dark about it. That’s what I've been fighting against.