IRR vs. TWR, or why can’t we all just get along By Jesse Reyes, QuartileOne, LLC

After leaving graduate school so many years ago (at the risk of dating myself as this was

right before the advent of modern PC-based spreadsheets) I was convinced that the

most dangerous weapon in the hand of a newly minted MBA was the HP-12C calculator

with its nifty IRR/NPV function, which at the time was a real game changer. While the

tools financial analysts have at their disposal have progressed significantly, that fear

has evolved to really mean that the IRR is probably the most dangerous weapon in the

arsenal of the typical MBA.

I’ve been in the performance measurement and benchmarking business for over 20

years, during which the state of the performance analytics field has become increasingly

complex and academically esoteric. To many it has evolved into quasi-religions, each

one with priests, ideology and schools of thought feeding into a debate that is often

rancorous, vociferous and polarised.

The two schools that have evolved actually result from arguments which are probably

50 years old and revolve around the issue of IRRs (money weighted returns) and geometric

mean returns (time weighted returns).

I wish I could say I am totally agnostic in this debate, but I have probably been an IRR

apologist for some time (after all my first born daughter’s initials are I.R.R.). And I am

one of the inaugural members of a group of like-minded acolyte dedicated to all things

IRR and who meet several times a year to sort out the latest research on the topic. I do

however see the value of both measures and feel that the issue isn’t so much the ideology

but the appropriateness of their application.

If you google ‘IRR’ you usually get not only the exexpected ‘how to calculate’ entries, but

also lots of articles and links to white papers on how the IRR is bad and how it can be

fixed. Just about every year there is another paper that tries to modify the IRR to fix

some of these flaws and just in the last six months I have reviewed three different

approaches proposed to fix the issues with the allegedly schizophrenic IRR. But like a

bad relative, it may be that we must have to learn to accept and love the IRR and realise

that the flaws are inherent and can’t be fixed easily, but the measure can never be

abandoned to the streets – we just have to take care in how we use it.

I will provide a bit of background on the evolution of the issues. I cannot do justice in

such a short treatise to all the various methods/measures/modifications that have been

proposed and there is not enough room to go through all the mathematical intricacies that are inherent but I hope to provide a survey of the state of the world and how we

got here and see if I can provide a bit of guidance on navigation. I leave the reader to

other resources on the calculation and other metrics.