Returns on Public Pensions: What Rates Should We Assume?
It seems that Andrew Biggs, at the American Enterprise Institute, is taking issue with my argument on the rate of return that public pensions should assume on the portion of their assets held in stock. (The rate of return on the asset is the same issue as the rate of discount applied to future liabilities. I use rates of return just because it makes the discussion easier to follow.)
To start, we should be clear on what exactly is at issue. Andrew and I are not debating the expected rate of return on stocks. Both of us agree that the pension funds are at least close to the mark in their assumptions on stock returns. Rather, Andrew feels that their return assumption does not correctly account for the risk associated with stock returns. He notes that the higher return on stocks comes in exchange for higher risk. Since the pension obligation is an absolute commitment, he argues that pensions should assume a risk-free rate of return on their assets.
My contention is that because a state or local government is essentially an infinitely lived entity, it need not be as concerned about the variance in returns as individuals. Therefore state pension plans can make their projections based on the expected value of their stock holdings.
It is striking that Andrew and I are now in a situation where I am arguing for assuming higher returns on stock than he is. In our first exchanges the roles were reversed. In the context of Social Security privatization Andrew was arguing that we could project that the stock held in individual accounts would get the historic 10 percent nominal rate of return. I argued that based on the bubble-inflated stock prices of the time (2000), the projected returns should be roughly half this rate.
While we disagreed on the number, both of us were arguing that we should use expected values, at least for a first assessment of what individuals should expect from their accounts. Now Andrew seems to be arguing that not only individuals, but governments, should discount future stock returns for risk. It's fine that Andrew seems to have changed his mind on this issue -- intelligent people do that when they get new information -- but I still find it interesting. I look forward to having Andrew as an ally the next time there is a push for privatizing Social Security that relies on the higher returns that people can get from holding stock in individual accounts than from their Social Security contributions that are invested in government bonds.
I don't know exactly what caused Andrew's change in perspective, in my case there is a very simple explanation: the stock market bubble deflated. I had always based my projections of future stock returns on projections of profit growth (generally assumed to be equal to GDP growth) and the dividend yield (which includes share buybacks). When the price to earnings ratio is high, as it was in the late 90s, then the dividend yield is low. On the other hand, when the price to earnings ratio is relatively low, the dividend yield is high. A low price to earnings ratio also has the advantage that there is less room for stock prices to fall barring a cataclysmic collapse of the economy.
The chart below shows price to earnings ratios for economy as a whole. (The value of corporate equity is taken from Federal Reserve's Flow of Funds Table L. 213, Line 23, after-tax corporate profit use the average share of profits in GDP over the last 10 years. The profit number comes from the National Income and Product Accounts, Table 1.12, Line 15.) I prefer a broad measure to the S&P 500 both because I prefer to use government estimates of profits, rather than private estimates, and also because pensions invest in private equity funds that will include companies that are not part of the S&P 500.
Source: see text, above.
The basic arithmetic is straightforward. A PE of 15 implies that earnings are just under 7 percent of the share price. Companies typically pay out about two-thirds of earnings as dividends or share buy backs, which translates into a dividend yield of just over 4.0 percent. The projected real rate of growth is around 2.5 percent (a bit higher over the next few years), which gets us to a 6.5-7.0 percent real return. For a fund assuming 3.0 percent inflation, that translates into the nominal 9.5-10.0 percent yield that most assume for the portion of their funds held in stock.
Note that such returns were certainly not possible in the middle of the last decade and certainly not at the peak of the bubble in 2000, when Andrew and I had our first exchange. I in fact did argue that pensions were being overly optimistic in their return assumptions at that time. It would have been helpful if Biggs and the others now complaining about return assumptions had expressed their concern before the stock market plummeted and state governments faced fiscal crises.
But Andrew and I don't fundamentally disagree on the expected values, where we differ is on how the funds should view the risk associated with stock returns. The point that I have made is a simple one, different agents have differing abilities to tolerate risk.
This is why we have the insurance industry. For most people, having their house destroyed by a fire is a distrous financial event since their house is their major financial asset. However, an insurance company can spread this risk over tens of thousands of homeowners. Therefore people are willing forego money in order to protect themselves from the risk that their house will burn down.
I am arguing that states and their pension funds are similar situated relative to the risk of a drop in the stock market. What exactly would be the consequence of the bad scenario -- there is a substantial drop in the market for a sustained period of time -- for a pension fund that is invested 70 percent in equities (including private equity) and 30 percent in bonds and relatively liquid assets? In almost any plausible scenario, they could easily meet annual payouts from the bond and liquid asset holdings in the fund. Furthermore, a drop in the trend price to earnings ratio would mean that the fund would be able to get higher returns on the money invested at the market trough, as wwas true for those who bought in 2009.
Back during the Social Security privatization debate we developed the "No Economist Left Behind Test," which challenged those assuming high returns in the stock market to produce a set of dividend yields and capital gains that produced their returns. None of the privatizers could do this simple act of arithmetic since it implied absurd price to earnings ratios. Those who think it is plausible that the stock market will on a sustained basis provide markedly lower returns than what the pension funds assume might want to take this test. They will find that they get implausibly low price to earnings ratios or incredibly bleak projections for economic growth.
If they think these incredibly bleak growth projections are plausible then we have far greater worries than underfunded pensions. In short, in all but complete disaster scenarios, states should easily be able to bear the risk associated with market returns, given current price to earnings ratios.
Finally, some have argued that it pays to be cautious and therefore assume the lower risk free rate of return even though we actually expect a higher rate. Assuming that pension funds did this and continued to invest in equities, it would create a situation in which we would fund to a much higher level, then find that better than projected returns would allow us to pay less into the pension funds in future years.
While it might be nice to have lower required payments in future years, this comes at the cost of higher payments in the present. That means some combination of higher taxes and less money spent in other areas now, so that we can have lower taxes and more spending eslewhere in future years. Of course we could do this with every area of spending -- we could accumulate money to prefund education and the cost of maintaining our fire departments. We don't generally think this is a good way to go. First, future generations will be on average richer than we are today, so the equity in this picture is not obvious. More importantly, if we tax our children's parents more just to accumulate reserves or cut what we spend on their education, we probably are not helping them, even if it might mean slightly lower taxes in 20 years. In other words, it makes sense to get the numbers right.
There is also the possibility suggested by Andrew and others that pension funds would stop investing in equities altogether. This would actually lower the returns on public pension funds. This means that taxpayers will have to pay higher taxes, since they will have to make larger contributions to pensions for workers to enjoy the same benefit. (The pension benefit is a trade-off for lower wages, so if the benefit is cut, it would mean that wages would have to be increased, so there would still be a higher tax burden.)
In short, the issues here are not really that complicated, they are mostly questions of simple arithmetic. Unfortunately economists tend not to be very good at arithmetic (that's why almost all economists missed the housing bubble), so we may be going back and forth on this one for a while.