 |
|
|
|
|
|
|
|
|
 |
 |
 |
 |
 |
 |
 |
 |
|
|
|
||||||||||||||||||
|
GCR 99-780 - Estimating Social and Private Returns from Innovations Based on the Advanced Technology Program: Problems and Opportunities
12. EXTENDING AND IMPLEMENTING THE MODELIn Section 10, we cautioned that forecasts of social rates of return at the project proposal stage or development stage are likely to be quite inaccurate. No matter what model is used, this would seem to be true. After all, if you have only a hazy idea of what it will cost to complete and commercialize a certain R&D project (or set of such projects), and if you have even a hazier idea of what the outcomes will be, it obviously is very difficult to obtain reliable estimates of social rates of return from these projects. Moreover, these problems are exacerbated by the fact that the ATP program is focused on high-risk projects (which are ones where the forecasting problems are most difficult) and the knowledge spillovers are expected to be substantial. NIST may, of course, decide to have such forecasts constructed, internally or externally (or both), using the most serviceable techniques at hand. The closer to commercialization the projects are in most of the potential application areas, the more dependable the forecasts are likely to be. Allowing time for the innovation to be subjected to a market test and basing the studies at least partially on actual market performances is preferable. Based on the Foster and Nathan studies, as well as our own, the analytical framework described in Sections 5 and 6 is likely to be as serviceable as any, although it is hardly a panacea. In fact, it is a very simple, straightforward analytical technique that is based largely on principles dating back a good many generations (to Alfred Marshall and beyond). The trick is to get an adequate understanding of what each innovation is (though discussions with the relevant parties), what its effects were, and how to model these effects faithfully. Imagination and respect for reality are more important than technical flourishes. But this does not mean that this simple model cannot and should not be extended.12 For example, if an innovation results in significant spillovers of information that are directly responsible for other innovations in other industries (not merely imitation by the innovator's rivals), an attempt should be made to include this in the calculations. Since none of the innovations in our sample was of this type, we did not have to worry about this possible complication in our 1977 study, but ATP expects that many of the innovations it funds will be of this type. Another kind of extension of this basic model was carried out by Trajtenberg (1990) in his study of CT scanners, which may prove useful in the study of some ATP innovations. One could develop in advance and try to implement a uniform social rate of return model with extensions covering a broad spectrum of innovations. A more realistic approach is to use the model described above as a flexible framework for examining individual cases as they arise, having gained an understanding of the appropriate business realities of the case. Then one can extend the model accordingly, in those cases where it is necessary.Return to Table of Contents or go to next section. Date created: June 15, 2006 |
 |
ATP website comments: webmaster-atp@nist.gov / Technical ATP inquiries: InfoCoord.ATP@nist.gov. NIST is an agency of the U.S. Commerce Department |
 |