Finance 101 teaches us that we should only invest in positive NPV projects, and avoid negative NPV projects, and that the value of the firm is the sum of the NPV of all its projects, net of any company costs not otherwise incorporated into the project valuations. In one number, we account for all factors of production, and for the trade-off between short term and long term. That's the theory.
And it's perfectly sound. If you want to create value--at the project or the firm level--there is no better guide than to maximize NPV.
What NPV is missing
We can't know the future. That's no problem for a measure that is entirely forward-looking if we are at a decision point on a new investment. It is highly problematic for a forward-looking measure being used for backward-looking evaluative purposes.
What makes NPV unusable as a historical measure of performance is the biases such a use would encourage in the projection of cash flows and the estimates of discount rates that are the heart of an NPV analysis. The managers being evaluated would be the only folks capable of making those projections and estimates, It would be akin to having people grading their own work. NPV-based evaluations were tried in several Enron divisions. It predicably led to inflated values supporting bad investments.
When it works
NPV has the distinction of being the best forward-looking estimate of the value for a project or a company while being unworkable as a backward-looking, period-to-period performance measure.
