Public pension scourge is at it again

Here’s a quiz any undergrad business major should be able to ace: Assume you invest $10,000 in an asset with an expected return of 10 percent, and another $10,000 in an asset with an expected return of 4 percent. What’s the expected annual return on your portfolio over a 30-year period?

Answer: 8.1 percent (because $10,000 x 1.0430 + $10,000 x 1.1030 = $206,928, and $20,000 x 1.08130 = $206,928)

But in a new working paper, Rochester University finance professor Robert Novy-Marx asserts that a pension fund manager following accepted accounting rules for public pension funds would assume an expected portfolio return of 7 percent in this situation (which he gets by averaging 10 percent and 4 percent). From this false premise, Novy-Marx draws outlandish conclusions about pension fund accounting, such as the claim that a pension fund with just $10,000 invested in the higher-yielding asset would appear to be better funded, all else equal, than one with $20,000 split equally between the higher- and  lower-yielding asset (because $10,000 x 1.1030 > $20,000 x 1.0730). Novy-Marx concludes that these rules give public pension fund managers a perverse incentive to “burn” the low-yielding bonds in order to inflate their plan’s funding status.

If this sounds absurd, it’s because it is. To begin with, you can’t just average the two rates of return as Novy-Marx does, because over time the portfolio becomes more weighted toward the higher-yielding asset. In practice, pension funds periodically re-balance in order to prevent a portfolio from becoming too heavily weighted toward risky assets, but they would have to re-balance continuously in order to reduce returns to 7 percent, which is unrealistic. In any case, Novy-Marx doesn’t even mention re-balancing, nor any other realistic pension fund practices in his paper. If he did, he’d also have to acknowledge that public pension funds assume stable, long-run returns that vary little across plans, clustering around 8 percent—less than the roughly 9 percent these funds have averaged over the past quarter century. Thus, they wouldn’t be affected by the kind of gaming Novy-Marx conjures up in this paper.

Novy-Marx’s claims are exasperating because the accounting method he prefers would actually create perverse incentives. Novy-Marx et al. believe that since pension liabilities are guaranteed (only partially, but that’s another matter), pension funds should be required to assume a nearly “risk-free” rate of return no matter the fund’s actual asset allocation. Thus, in Novy-Marx’s example, the assumed rate of return would be the 4 percent yield on nearly risk-free Treasury bonds even if the entire portfolio were invested in stocks with an expected 10 percent return (Novy-Marx doesn’t deny the existence of an equity premium).

It’s important to note that this wouldn’t encourage prudent investment practices any more than the doctrine of predestination eliminated sin. If anything, it might have the opposite effect—incite a desperate hunt for yield—as all pension funds would immediately appear drastically underfunded. It would not guarantee that the fund would earn a 4 percent return or better, since it wouldn’t require funds to invest in Treasuries or other low-risk assets. All it would do is make pension funds look bad and cause required contributions to spike, inciting a taxpayer revolt. It would also cause funded ratios and required contributions to vary for no logical reason, since Treasury yields fluctuate with monetary policy and market conditions that may have little or no bearing on pension fund adequacy.

Elsewhere, Novy-Marx has actually suggested that state and local governments with shaky finances should be allowed to contribute less to their pension funds because their higher borrowing costs—and the greater likelihood that they renege on pension promises—should translate to a higher discount rate on future pension liabilities. Though this illustrates where his logic takes you, Novy-Marx isn’t trying to promote fiscal irresponsibility.

(However, allies like Andrew Biggs of the American Enterprise Institute want to be able to assume high expected returns on assets in 401(k)-style plans while requiring public pension funds to assume low returns on the same assets.)

Novy-Marx’s latest sally is more an effort to provoke than to persuade. But he and his allies have already had a significant impact in the policy arena. The Government Accounting Standards Board has proposed valuing some pension liabilities using low municipal bond yields, a change that will likely result in significantly lower funded ratios and higher required contributions.

More generally, Novy-Marx and a small group of other economists have succeeded in attacking public funds for supposedly engaging in aggressive accounting and ignoring risk, deflecting attention from the real problem (in states where there is one) of elected officials neglecting to make required pension contributions. Astonishingly, they have done so without presenting any actual evidence that public pensions take on too much risk or inflate expected returns, but have rather harped on arcane accounting issues until enough people have concluded that where there’s smoke there must be fire.

  • Frank Keegan

    Here’s a quiz any 6th grader can pass: If you must gain 8% a year but lose 20% in any one year, you have to gain 46% just to stay even. That is what happened to pension funds in the Great Recession, except they did not bounce back. To pay promised pension obligations without service cuts and tax increases for the next 50 years, states and municipalities would have to make full actuarially required contributions (they have not, do not and will not,) and achieve investment returns that never have occurred in history. Oh, and there never can be another market downturn. Leveraging the catastrophe is the fact that cheating on pensions is but one of the tricks politicians have used to defer costs. Unfunded Other Post Employment Benefits costs are staggering. The true magnitude of our leaders’ misfeasance and malfeasance is difficult to comprehend, but you can be sure we will have to comprehend it in the coming decades. We face decades of hard times paying off the obligations they inflicted on us through lies and deceptions. The longer we delay acknowledging that and dealing with it, the worse it will be. Some states and municipalities have slipped over the edge of a fiscal event horizon and most others rapidly approach it. We must act now to pull back.

  • Robert Novy-marx

    They aren’t my rules; they’re GASBs. GASB doesn’t let a
    pension plan assume a higher rate of return simply because the plan thinks it might
    take on more investment risk in the future, which seems to be what you’re
    advocating. GASB assumes a plan’s manager has made a thoughtful decision regarding
    asset allocation, and that the best indication of a plan’s future risk exposure
    is its exposure today. The appropriate way to calculate the expected return on
    such a portfolio is to take the asset-weighted average of the individual assets’
    expected returns, just as I have done. Your claim that for this to hold a plan “would
    have to re-balance continuously” is simply wrong. When talking about
    simple annual returns (as we both obviously are, given how we’re compounding),
    it requires annual rebalancing. I agree that something sounds absurd, because it is, but they are GASB’s rules, most certainly not mine.

  • Herb Whitehouse



    Competent professionals practicing in the world of portfolio
    investment management and portfolio asset allocation do not take a weighted
    average of asset class returns.


    The reason why can be quite informative as to whether the
    investment practices of most public sector pension fiduciaries and the GASB
    accounting standards themselves (including the ED) require an interpretation
    other than the sanctioning of burning money.  


    Now when I spoke to GASB in Chicago not even two months ago,
    I asked the board to make explicit what I see as inherent in the standards; namely,
    that investing in high return, but very risky and/or highly volatile
    investments, such as an investment portfolio with the10% expected return in
    your illustration would be inconsistent with the standards.


    Sophisticated pension fiduciaries – even in the public
    sector – order their investment policies to the goal of achieving reasonable
    funding policy objectives.  And those
    objectives always reflect the volatility of plan investments relative to the net
    expected cash flows of the particular plan. My point to GASB was that not all
    public sector fiduciaries or their advisors order their investment policies
    toward these objectives in an explicit way. 
    The danger that I see is not that this has resulted in either “burning
    money” or taking the kind of extreme investment risks that your example uses.


    I do see some risk at the margins where a public sector
    investment fiduciary has unsophisticated and/or unprofessional investment
    counsel coupled with actuarial advice that can push plan investments to be more
    risky in order to gain apparent reductions in liability and increases in
    funding ratios. But even in the most unsophisticated and unprofessional
    investment advisors will not move a portfolio allocation to one with a 10% expected
    return.  And GASB now has a wonderful
    opportunity to tighten its ED language to make explicit what is already
    inherent in its standards; namely, that anything close to your illustration is
    inconsistent with the standards.


    I want to thank you for your “burning money” illustration,
    not because it represents the reality of the GASB standards (it does not), and
    not because it is (and it is) a reduction ad absurdum argument proving the
    absurdity of many criticisms of the GASB standards; but rather, because it will
    help GASB to modify it language to more explicitly express principles that are
    already inherently in its standards. 


    Herb Whitehouse


    P.S.: The calculation of a portfolio expected rate of return
    depends on at least three capital market estimates: 1) the expected (arithmetic)
    return for an asset class or investment approach; the volatility of that asset
    class or investment approach; and the correlations of each asset class (or
    investment strategy) to every other asset class or investment strategy.  No competent professional in the pension world
    takes a simple weighted average of expected returns by asset class, e.g., 10%
    and 4%.  Using some common capital market
    assumptions for the three factors just mentioned I calculated the decrease in
    the 30 years geometric return from the combining of two portfolios, one with a
    10% arithmetic expected return and the other with a 4%.  The decrease is not the 30% that your
    approach suggests.  In fact, it is around
    a 20% decline. The reason is because volatility is a drag on geometric returns
    over the long term.

  • Andrew Biggs

    For what it’s worth, this post exhibits a common misunderstanding regarding the market valuation argument. Monique, you write:

    “Novy-Marx et al. believe that since pension liabilities are guaranteed
    (only partially, but that’s another matter), pension funds should be
    required to assume a nearly “risk-free” rate of return no matter the
    fund’s actual asset allocation. Thus, in Novy-Marx’s example, the
    assumed rate of return would be the 4 percent yield on nearly risk-free
    Treasury bonds even if the entire portfolio were invested in stocks…”

    The point of the market valuation critique is that the discount rate used to value a liability is independent of the expected return on assets used to fund the liability, which is the standard economic point of view. So it’s not that Robert argued you should assume that stocks will earn the same return as bonds; to claim that is to misunderstand or distort the argument.

    If pension benefits were unfunded, you’d obviously value them using a low discount rate to reflect their low risk. Setting aside money to pre-fund them doesn’t change that; the benefits are worth what they’re worth. How you fund DOES change how cost burdens are borne over time; investing in riskier assets lowers costs to current taxpayers but increases the contingent liability on future taxpayers. But the liability itself doesn’t change.   

  • Donald Fuerst

    Monique, I think you miss an
    important point. Even if one accepts your position that the return of the fund
    should be 8.1%, there remains an anomaly. If you discount the future cash flow
    at 8.1% the obligation is $16,915, but the obligation of the all stock fund is
    only $10,000. How can a payment of $175,000 in 30 years have a different value
    merely because assets are invested differently? Investments held to settle a
    future obligation do not affect the value of that obligation.

    On the other hand, Novy-Marx and Biggs seemingly fail to recognize that pension
    funds can earn an equity premium. A prudent plan sponsor that invests in a
    diversified portfolio chooses to take risk with the expectation that
    contributions might (or might not) be lower. The expected return on the
    portfolio is useful in determining contribution strategies, but does not
    provide a meaningful measure of the obligation. Novy-Marx might argue that the
    sponsor should not take that risk, but that ignores the reality that many
    choose to do it. Measuring the obligation both ways demonstrates how much they
    attempt to save by taking risk and provides at least an indication of the
    amount of risk they are taking. Neither method provides a full answer by