The high cost of high expected returns

  • Using asset returns to discount liabilities has important behavioral implications
  • Choice of a discount rate should not impact investment strategy


Pensions and endowments might create issues for themselves by relying on high expected returns


How do you think the following conversation with your mortgage banker would go? You give him a call and say that according to your most recent statement, you owe the bank a hundred thousand dollars. However, you have decided to invest all the assets you own in equities, junk bonds and this hot new hedge fund that everyone thinks will return 20% per annum. Therefore, instead of one hundred thousand dollars, you feel that you owe the bank only sixty thousand dollars.


The most likely outcome of this conversation would be your mortgage banker hanging up and flagging you as a high risk borrower.


As absurd as this may sound, some of the largest borrowers in the world are playing this exact game. Pension plans are generally allowed to select what rate they would like to discount their future obligations in order to determine liabilities. Selecting the appropriate discount rate has real economic and political decisions tied to it, and  choosing a discount rate that makes liabilities look smaller plays into our human behavioral biases.


Theoretically, the present value of liabilities ought to be calculated using a risk-free interest rate for each point in time that the cash flows are expected to occur. This value of liabilities would represent how much money you would need to set aside today and not have to worry about the “what ifs”, except perhaps potential changes to actuarial assumptions. This number might even resemble something similar to what an insurance company might be willing to accept to take on similar future obligations.


However, discounting using a risk-free rate can make the result look ugly for entities such as public pension plans, which are allowed to use their expected return on assets as the discount rate. As an illustration, if we use data from an excellent paper on pubic pension plans by Joshua Rauh and Robert Novy-Marx the funded status for public pension plans in the US would fall from about 62% to about 44%.


In dollar terms, this is material – Using Rauh and Novy-Marx’s data for state pension plans from 2009, that almost doubles the stated shortfall of $1.2 trillion to $2.48 trillion!


Pension Assets and Underfunding

Underfunding increases materially if using a Treasury Rate. Source: Josh Rauh and Robert Novy-Marx



The behavioral benefit/pitfall of using expected returns


I was having drinks a few months ago with the CIO of a large insurance company, and he recounted his first experience with behavioral economics back in business school. One of his professors would be routinely inundated by requests to review marks awarded for tests and papers, despite reminders that the final grade would be awarded on a curve. To make his life easier the professor came up with a creative grading system. Instead of marking the papers out of 100, he would mark them out of 117. Almost immediately, the number of post-marking requests he received went down dramatically. The student who would earlier come to his office complaining about a 70, was now satisfied with a 82!


This type of human behavior is one of the typical human biases that Dan Ariely refers to in his book Predictably Irrational. However, when applied to the world of pensions, it can result in potentially flawed decision making.

The funded status of pension plans is defined as the ratio of assets over liabilities. A funded ratio close to 100% therefore means that the pension plan has sufficient assets to meet its liabilities, and provides a sense of comfort. However, what if that sense of comfort is false? What if we have changed the base, and awarded a 100 out of 117, lulling decision makers to make incorrect funding, benefit and investment decisions?


For example, if you look at the hypothetical illustration below, a plan that is 68% funded using a risk-free rate of 3%, appears over 100% funded at a 7% discount rate. Similar to the business school students, even though economically we have done nothing different, a 68 does not look as good as a 100, and we don’t like that answer.


Screenshot 2016-05-01 14.42.57

Illustrative example of present value of future cash flows


So we end up looking for ways to make ourselves feel better. We start saying that well, this method is unrealistic, as we can definitely earn more than a risk free rate. If we invest in high quality bonds, we should be able to earn more than the risk free rate, even if some of those bonds default. So now we discount using a corporate bond rate, and feel a little better. We start putting away less money (i.e. make less contributions), maybe even start promising more benefits. Now for some pensions, even a corporate bond rate doesn’t seem to make them feel comfortable. Then they start stretching even further, and start saying that they can get better investment returns, so lets discount at the returns that I expect to earn on my investments.


However, expected returns are not certain, and there is a chance that those expected returns are not achieved. Generally, higher returns are accompanied with higher risk. Therefore higher the expected returns, the higher the chance that we don’t achieve the returns, and even potentially lose money. However, the more risk we take, the more returns we can say that we expect to earn (note – its not more returns, but more expected returns) – and if we have greater expected returns, we just magically reduced our liabilities.


As a result, we have moved from a low funded status number that we didn’t like to a higher number that we like. But in the process, all we have done is fool ourselves. I am not saying that we may not be able to achieve the 10% annual return on equities, or a 20% return on alternatives – skilled investment managers and CIOs may have the ability to generate these returns. However, assuming that these returns are certain, and using these expected returns to discount liabilities distorts the picture.

There any several scenarios where the expected returns may not be achieved. That is the nature of risk. And by shifting the goal posts (i.e. the funded status), we end up hiding the risk.


Hiding the risk has important implications.


Hiding the risk has important implications. For example, contributions to pension plans are determined using funded status – Therefore if you have fooled yourself into thinking you are well funded, you will make a lower contribution. It can also work the other way – if you think you are well funded, you may determine you don’t need to take too much risk, and end up investing in less risky assets than what you might otherwise.


A more stable exercise might be to stick to a standard method of valuing liabilities, lets say using a risk-free discount rate. Funded status becomes the result of Value of Assets / Value of Liabilities. The pension plan can calculate how much better / worse off that number they would be based on scenarios or expected return assumptions that they want to use for their assets. Effectively, you would separate your asset return assumptions from liability discounting.

Pension plans do not necessarily need to change their investment strategy as a result of this shift in measurement. If they are still confident that their investments and asset allocation can result in the returns that they expect, they should continue to make those investments and follow their chosen investment strategy. Changing the discount rate on a liability calculation should not shake their confidence in their investment return generating capabilities.


In an interview with Sal Khan of Khan Academy, Eric Schmidt of Google spoke about how Google conducts meetings – they first put the facts on the table, and then they start the debate on what to do given the facts. If we obfuscate the facts that are presented, its hard to move forward and have a quality debate.


Therefore, lets put the numbers as they stand on the table, and then make our investment decisions following that.


Further reading:
  1. Public Pension Promises: How Big Are They and What Are They Worth?
  2. Cash flows, returns and discount rates are probabilistic

Disclaimer: All views expressed in this article are that of the author and do not necessarily reflect the views of his employer or any of its affiliates. The author may be associated as an investor or as an advisor with certain companies mentioned in this article.

Also published on Medium.

3 thoughts on “The high cost of high expected returns

  1. I would be interested in the application of this reasoning to the post office. My understanding is that the congress has required the US post office to prefund its retirement in a way that no other business or even other branch of government is required to. Is that a massive over funding making their losses look bigger than they are or is it a good thing?

Leave a Reply

Your email address will not be published. Required fields are marked *