Moody’s flawed estimate on public pension liabilities
As the debate continues over public pension funding levels, we have this headline from the¬†Financial Times¬†this week: ‚ÄúUS States need $980 billion to fill pension gap, says Moody‚Äôs.‚ÄĚ This is not exactly news. A number of studies, including ones from the Pew Trust and the Public Fund Survey, have identified a massive shortfall for public pension funds. In fact, the Pew Trust said that the shortfall in 2010 was $1.38 trillion, so perhaps we should be applauding state legislatures for improving the gap since then.
The shortfall numbers in these studies, to put it simply, are all over the place. There are many variables that go into these models, but the main factor that causes variation is the expected rate of return on the assets in the plans. The official assumed return on the assets that are held in trust to pay pension liabilities is 8 percent, according to the Public Fund Survey. Fiddling with this projected rate of return can cause swings in the amount of unfunded liabilities. The Moody‚Äôs study uses an unconventional assumption. According to the¬†Adjustments to state pension liabilities document:
Accrued actuarial liabilities will be adjusted based on a high-grade long-term taxable bond index discount rate as of the date of valuation (of the fund).
Rather than using the average historical investment return rate of 8 percent, Moody‚Äôs uses a return on a taxable bond index. This return would be no higher than that on a basket of high-rated corporate bonds ranging from 4.4 to 6.2 percent, the FT says. The problem with using this rate is that we have been in a five-year period of zero-interest rate policy while the Federal Reserve has artificially suppressed interest rates to promote financial stability and spur economic growth.
If we look back over the last 54 years in the Fred chart above, we see that Aaa and Baa corporate bond rates have had higher yields than what Moody‚Äôs is using. The historical corporate bond rate is in fact much closer to the rate used by government actuaries. The fact is that using any suppressed bond yield in an easy money period will make future projections unreasonable.
Moody‚Äôs makes this exact point on page 4 of the Adjustments to state pension liabilities document:
Because interest rates are currently at an historic low, the market approach to measuring liabilities results in much larger current total liabilities than those reported using the conventional governmental approach.
In conjunction with our use of market asset valuations, which currently fall below smoothed asset valuations that have not yet fully factored in the impacts of 2008-2009 equity declines, this leads Moody‚Äôs adjusted net pension liabilities to be much greater than actuarial unfunded liabilities. The approach also introduces greater volatility into the measurement of the adjusted net pension liability.
This approach might work if interest rates have not been so low for so long. Moody‚Äôs model is so volatile that it risks overstating the liabilities of state governments.
The FT article bangs on governments for not accepting the Moody‚Äôs corporate methodology:
The Moody‚Äôs research is the first to look state-by-state at public sector pension schemes in the context of overall state finances which relies on a metric of future investment returns increasingly adopted by the private sector internationally.
The methodology is important because the US public sector pension plans have been very slow to adopt what are increasingly recognized as realistic accounting standards when it comes to showing the actual costs of retirement promises.
As the FT points out, state governments continue to use historical returns to guide their future projections. Models are useful, but they have limitations. Moody‚Äôs model, in periods of artificially suppressed interest rates, has enormous flaws.
Journal of Accountancy: GASB vote places unfunded pension liabilities on government balance sheets