I was studying return measures such as NPV and IRR from Damodaran's "Applied Corporate Finance" and one thing that he continuously mentioned was that NPV is biased towards projects with longer lives while IRR is biased towards projects with small initial investment. While the second point seems valid given that IRR is a percentage based scaled measure and should prefer projects with small investments, I haven't been able to wrap my head around the NPV's bias.
Is the author talking about some inherent statistical bias that comes from discounting the cash flows? If yes, then how does this bias comes up?
There is no statistical "bias". It is just that NPV and IRR can provide different rankings to projects. For example consider a project with a small initial investment and positive cash flows after that. When the initial investment approaches zero the IRR approaches infinity but NPV does not. Here you can see how the IRR criterion is more favorable to projects with small initial investments.
Assume you have a project with positive cash flows far ahead. Assume the NPV is positive under some discount rate. IRR is defined as the discount rate under which the NPV of the project would be zero so this must be higher than the discount rate you used to obtain a positive NPV. The effects of this higher discount rate compound when discounting cash flows in the distant future so this penalizes such projects more heavily.