IRR is like fish, when someone gets hold of it, it slips away. Hard to seize, hard to terminate – with incredible survival instinct, it tries to jump out of any bucket where it has been secluded.
There so much academic evidence to call the IRR dead but, hold on not yet, it is still the private equity industry’s GIPS (global investment performance standard) – alive and kicking.
Its most recent twist that has come about is in a post titled “IRR Can Be Useful – If You Know How To Use It“, whose concluding paragraph includes the following clue:
“Investors that are more focused on receiving money back quicker or need to meet certain requirements may look to invest in fund managers or strategies that tend to generate higher IRRs.”
It looks that the IRR is then good at signalling when money comes quickly – but not necessarily how much money, as the post duly explains.
So it seems the real value of the “global investment performance standard” is actually “speed measurement”.
This is not new though. Messrs Kocis, Bachman, Long and Nickels in their 2009 book “Inside Private Equity: The Professional Investor’s Handbook” identify the topic but unfortunately leave it in the Appendix F: Advanced Topics (page 235) where they define the notion of Duration of Performance:
TV = PI (1+IRR) ^ (Duration)
Duration = ln(TVPI) / ln(1+IRR)
Good try, IRR. Still, as you try to survive, you help us prepare the pan where you’ll inexorably end up frying. Because math does not lie:
- You can stand for annualised performance only for the period identified by the Duration (as calculated above – that will be called IRR Duration);
- The IRR Duration can’t signal (the calendar time) when you will actually materialise as performance – the simplification of the formulas above compress all events (in particular PI, i.e. the paid-in capital) as if they were happening from time zero;
- Unfortunately you bias the formula of the IRR Duration because – again! – the implicit reinvestment assumption makes “high performance run faster” and even IRR Durations cannot be properly compared among themselves (unless there are only bullet / zero coupon transactions in the analysis).
You’re done fish. The fact that you, IRR, are useful to calculate the IRR Duration of the TVPI does not make you any better.
And if you pretend to be around longer as a credible performance standard, you’ll start to smell fishy!