The endowment model of investing is characterized by a significant allocation to alternative investments, such as private equity and real estate funds. One of the leading proponents of the endowment model, Yale University, recently reported an impressive since-inception internal rate of return (SI-IRR). Yale’s high SI-IRR has persuaded other investors to attempt to emulate the endowment model. But the author notes several interpretation difficulties associated with using the SI-IRR model that could produce highly misleading performance results. He suggests several alternative approaches to ameliorate these difficulties.

Practitioners generally realize that the internal rate of return (IRR) methodology implicitly assumes that intermediate cash flows are reinvested at the IRR. What practitioners may not realize is the dramatic effect this reinvestment assumption may have on reported SI-IRR. The author constructs a simple example to illustrate this point.

In the example, $100 million is invested in 1990, followed by a return of $500 million in 1995, which results in an approximate annual return of 38%. This investment and return are then followed by another $500 million investment in 2000. The author calculates the SI-IRR for three possible terminal values in 2010: $5 billion, $2.5 billion, and $250 million. Despite the wide difference in terminal values, the calculated SI-IRRs vary only slightly from each other: Case 1 ($5 billion) has an SI-IRR of 35%, Case 2 ($2.5 billion) is 33%, and Case 3 ($250 million) is 30%.

The author notes that the closeness of the SI-IRRs is a result of the fact that it is assumed that the 1995 cash flow is reinvested annually at the 38% rate, thereby generating a value that dwarfs any of the assumed actual terminal values. This research is valuable in that it highlights the dramatic effect the reinvestment assumption has on reported SI-IRR and reveals the limitations of SI-IRR as a measure of performance returns.

The author examines the private equity 10-year annualized return and the SI-IRR of the Yale Endowment Fund for 2000–2010 and notes three oddities in the data. First, the data show a 30.4% SI-IRR since 1973. As a rate of return, this figure is unrealistically large because it would multiply wealth by 24,000 times over a period that long. The second oddity in the data is that, despite the volatility in the markets during 2000–2010, the SI-IRR did not change much. The last oddity is that the reported 2010 SI-IRR is almost 4% more (30.3% versus 26.6%) than the 2000 SI-IRR combined with the 10-year annualized return as of 2010.

The author constructs a hypothetical track record for 1973–2010 to demonstrate one way to verify the reasonableness of SI-IRRs. In this example, he constructs cash multiples as the quotient of the sum of all cash distributions and cash investments. Despite the fact that this hypothetical example has an SI-IRR of 30%, it has a cash multiple of only 1.5, which the author notes is not consistent with a 30% SI-IRR. Of course, cash multiples have their own limitations because they ignore the time value of money. The author suggests two alternative approaches to using SI-IRR: a modified IRR using an assumed annual reinvestment rate, such as a default rate as recommended under the GIPS^® standards (e.g., 8%), for the interim period between investments rather than using the SI-IRR earned during the earlier investment period or a net present value approach.

Yale’s reported success has encouraged others to attempt to emulate the endowment model. The author has performed an important service by pointing out that the model’s success may not be as attractive as first appears.