I recently posted about the return assumptions used by state pension funds. Most seem to use around 8% which seemed a bit high. In order to back out the expected return on equities that an 8% overall return implies, you have to make an assumption about the asset allocation of the fund. My colleague, Doug Pearce kindly provided Federal Flow of Funds data which shows the actual composition of state pension funds (in aggregate). This data is here. The state fund data is item L119. I took the liberty of putting the data in the graph below. (Note the 2010 data is for the second quarter).
We can see that on average, states had about 70% of their allocation in stocks. Going back to my original analysis, if we assume an 8% expected return on the portfolio and 3.9% risk free rate, we can solve for the expected return on the equity part.
8% = R(TBond)*wbond + R(Equity)*wequity
8% = 3.9*0.3 + R(Equity)*0.7
Solve for the R(Equity) = (8 - 3.9*0.3)/0.7 = 9.75%
An assumption of a 10% return for equities seems more reasonable and roughly in line with the back of the envelope calculation I did last week. But it doesn't end there.
Another colleague, Robert Clark (who does research in this area) suggested I take a look at the work of Joshua Rauh. Josh Rauh is a Professor at Northwestern who has written several excellent papers on the whole topic of state pension funds. In particular, his piece (with Robert Novy-Marx) in the Journal of Economic Perspectives is very readable and highly recommended. I've linked here to a version on his website.
Novy-Marx and Rauh paint a grim picture of state pensions. First, they point out that an asset allocation that is 70% equities is far too risky. Remember that we are talking about an 8% expected return here. The big word is "expected". In finance, an expected return means that on average we might expect to earn 8%, but we also expect a wide range of other possibilities. For large stocks in the US this range could easily be +/- 20% for 2/3 of the time. Of course, it is argued by many that in the long run, stocks should earn their expected return even if there are short term ups and downs in the market. But Novy-Marx and Rauh argue that the duration (average maturity) of pension fund liabilities is about 15 years. Considering that we are in year 11 of a flat market, this means that many funds need the next 4 years to be pretty spectacular to make that 8% average.
Second, and perhaps even more disturbing, is that state pension funds use their assumed expected return to present value the liabilities of the fund. This makes no sense as the liabilities are virtually risk free to the extent that the state has promised to pay them and is unlikely to renege. This is more than a mere accounting artifact, as by using a too high discount rate, states are, in effect, undervaluing their pension liabilities. They are making what they owe look much smaller.
Novy-Marx and Rauh attempt to quantify how much state pension funds are underfunded just based on the use of the wrong discount rate. The results are quite shocking. They estimate that the liabilities of state funds to be about $5.17 trillion as of the end of 2008 compared to the $1.94 trillion in actual assets. Thus there is a funding shortfall of $3.23 trillion. This equates to a short fall of about $161,500 per plan participant.
Finally, the authors point out that just because personal retirement accounts might hold a high level of equities doesn't mean that state pension funds should do the same. A state pension fund has to meet the clearly defined annuity obligations to its participants. A personal retirement account (like a 401-k) is really just a means on transferring wealth through time. Although it is worth noting that Zvi Bodie strongly advocates very heavy allocations to risk free securities in personal pension portfolios also.
The conclusion? States, and taxpayers are in trouble. No doubt we'll keep kicking this massive debt down the road, but it isn't going to go away.
Footnote to North Carolina residents - our state pension fund is in better shape than most. For one, they only use a 7.25% return assumption.