In the article on The fundamental problem with using NPV in project evaluation. The author presents three key reasons on why managers prefer using IRR over NPV.

Regarding assumptions of the two models, some are explicit, and some are implicit! One of the implicit assumptions involves what happens to cash flows that come in how (what rate) are they invested for the remainder of the project? What does each of the models assume, and what is more realistic?

Article:

While most MBA graduates are indoctrinated with a dogmatic faith in the supremacy of NPV as a tool for evaluating investment opportunities, the reluctance of managers to use NPV in their evaluations is often surprising. What is even more suprising is that they swear by a technique which is fundamentally similar though widely derided as inferior in academic circles, the Internal Rate of Return (IRR). In this passage, I seek to indentify a few key reasons why I feel managers prefer using IRR over NPV. 1. Flawed comprehension - it is perhaps an indictment of the rigour in most MBA programs offered by universities that many managers long into their careers are ignorant of what exactly NPV is supposed to measure. NPV is a measure of the incremental value that a project adds to what can be achieved by investment in existing avenues. Put in another way, most managers think that a negative NPV implies a project that is 'not profitable', which is patently false - a negative NPV means a project is 'not profitable enough'. The difference between the two is subtle but significant. A project may have uniformly positive cash flows all through its life yet show a negative NPV because the rate of return on equity demanded by investors may be too high, and that brings us to the next point; 2. Dubious assumptions - A project evaluation begins with a forecast of the cash flows that can be expected from a project. While that is often a difficult task riddled with ultimately discretionary assumptions, at least with the IRR, that's where the modeler's judgement ends. With the NPV however, there is a further estimation that needs to be made, which is hardly straightforward nor definite. And that is the return on equity from similar projects or equivalently the equity Beta. This is difficult enough for projects similar to those being undertaken in stable sectors, imagine the level of ambiguity in undertaking this estimation for a project that is different from any undertaken so far. There is a fundamental rule in measurement theory that more the number of variables to be measured / estimated, more the error in the final answer. While with IRR the effect is confined to assumptions for estimating cash flows, for NPV calculations it is further compounded by errors in taking the equity beta. Another analytical manipulation that leaves managers scratching their heads (if at all they are aware of it) is the complicated requirement of keeping the D/E constant during the life of the project and the attendant problems of estimating project betas for projects whose financing is different from the firm's. This requires a tedious re-calculation of betas based on the new D/E ratios, which by itself are difficult to estimate. 3. (F)Utility of NPV - an NPV calculation fundamentally measures the returns given to investors who have sufficient liquidity to transfer funds between firms in equal risk classes. For most part, investors have no clue about similar risk profile sectors and projects, hence for managers, the metrics of measurement are only those of adding to the worth of investors in an absolute sense without worrying about returns available in equally risky alternatives. While a negative NPV may destroy value in the sense that monopoly pricing leads to a net social loss of value, it may be sufficient to keep managers, CEOs and investors happy with the absolute returns they get. In such a system, the IRR which simply measures the absolute returns that accrue from a project is a simple and ultimately more meaningful metric for evaluating a project. Most managers find the concept of hurdle rate for more intuitive and useful for identifying projects that are worth undertaking - they just compare the IRR with this bare minimum benchmark and if the IRR is higher, then the project may be proceeded with otherwise the proposal goes to the business plan graveyard. Most of the textbook problems that are used to demonstrate the inferiority of the IRR method occur rarely in the careers of most managers and given the above stated problems with the NPV technique, it is no wonder that most managers prefer the simplicity of the IRR rule.