Disclaimer before I add my take: I am not a corporate finance practitioner and do not work on budgeting decisions at a large corporation. Rather, I make budgeting decisions for the quant fund I manage. As thus my points may not reflect the point of view of a corporate finance practitioner but nonetheless I believe my points are equally applicable and having had numerous conversations with corporate treasurers of banks but also non financial corporations I believe I have an idea about the thought process going on inside someone who makes budgeting decisions.
I like to present a theoretical view and then the practitioner's view:
- I do not want to go into the details of pitfalls of IRR or NPV but overall the advantage of using IRR is that it produces a single number, it leads to results without making too many assumptions. However, IRR suffers from deficiencies in regards to intermittent cash flows. Also, IRR can result in multiple solutions.
- On the other hand NPV may require more assumptions. Especially tricky is which rate to chose to discount each of the cash flows. However, in its complexity lies also its accuracy: NPV allows you to discount cash flows at different discount rates. In corporate finance the discount rate can be interpreted in various ways. Some of such interpretations are "hurdle rate", weighted cost of capital, risk adjusted cost of capital, or financing cost. They are not one and the same and thus assumptions have to be made which exact discount rate to chose. However, despite lots of literature claiming NPV to be complicated I find the theoretical construct very logical because it results in a present value of future cash flows which is what anyone interested in such computations should ultimately be interested in.
Here couple links that describe the comparison between IRR and NPV more in detail:
INTRODUCTION TO CAPITAL BUDGETING
Time Value of Money and Investment Analysis
Capital Budgeting & Project Appraisal
Comparing Net Present Value and Internal Rate of Return
This is as far as theory is concerned.
- As IRR involves an easier calculation it is the preferred method of use. Management likes simple numbers even if the assumptions are faulty but within the ballpark. IRR has traditionally been used so most practitioners prefer not to shake up things and rather stick to the status quo. Nobody wants to be challenged to answer why they chose different discount rates for each future cash flow and how they arrived at each of those rates. The truth is that any such rates and future cash flows are inherently assumptions that involve a margin for error. When upper management makes decisions they want to be told by their subordinates whether a project produces an IRR that is higher than their overall financing cost for that overall group, region, brand,...most of those guys do not like to hear that a large degree of guess work was at play.
NPV is by far the more logical way to arrive at a project's expected return. Not only can any type of intermittent cash flows be embedded but each cash flow can be discounted at a different rate, and it should be that way! If a project produces different types of future cash flows (for example, news paper subscriptions: Positive future subscription revenues embed a different type of risk for long-term subscribers vs the subscription revenue earned from new subscribers, hence the future cash flows originating from such subscriptions should be discounted at different rates.). NPV produces far more accurate results despite its multiple assumptions because it reflects a more realistic treatment of future cash flows. Another example are budgeting costs for a trading business of one specific desk. A trading desk may trade in different asset classes, each asset class being of different risk grade. Thus, the financing cost are different. While a bank or hedge funds itself externally on just one or very few rates (realistically not just one rate) internally financing rates are entirely different: An equity trading desk will be charged very different financing rates than a commodities or bond trading desk because they all produce future cash flows of different risk grades. Even one single desk has different traders, some taking wild swings but larger expected cash flows while others are more risk averse and steadily but more slowly build pnl. That should in theory all be considered. Of course that is not the case because a trade-off has to be made between efficiency and accuracy. But I think it is a good reflection of what really goes on within managers involved in capital budgeting.
However, I like to highlight another aspect which is the political component: Sometimes capital budgeters are forced to arrive at a specific number so they basically need to tweak inputs to arrive at the desired outcome rather than the other way around. Managers' motivations are rarely perfectly aligned with long-term business growth, not even with shareholder or stake holder interests. If one's bonus depends on how many new projects have been proposed and approved by upper management then it tremendously helps to present "juicy propositions" that may not rely on realistic numbers to push the project to the approval stage. Especially for long-term projects most managers care very little what happens 5 years down the road if they look to cash in big time on their next 3 years' worth of bonuses. Sadly, especially in financial operations this is rather the norm than the exception and this I think is born out of the fact that labor mobility is very high between financial corporations. Also financial practitioners generally have a very short-term memory. If one fails, he/she can take some time off or maintain a low profile only to knock on the next door several months or a year down the road and still get a very fair shot. That also happens in normal corporations but I would guess at a much lower rate. One reason for that could be that CEOs of normal corporations are more credited but also held more accountable for the success/failure of the business while financial corporation's CEOs generally get away with bad performance as they can blame market conditions for such challenges. The world is what it is and as long as especially the current American business model with extreme short-term focus on quarterly results, dividend payments, bonuses, the ridiculous amount of short-term staff evaluations (lots of research has shown that it greatly diminishes staff morale and efficiency to evaluate and monitor each other on an almost constant basis and that the cost of such constant evaluations are higher than the perceived benefits) prevails, things most likely won't change much. As long as shareholders and consumers maintain a short term focus, so will corporations and its management.
Summary, in theory IRR suffers from simplistic assumptions and but it is easier to come up with a single number while NPV is far more accurate but involves more assumptions and is more complex to structure. Practitioners need to balance efficiency vs accuracy, however, keep in mind that there are also political undercurrents involved in the choice of technique. Not everything in business comes down to simple number crunching.