My colleague Richard Hayes is working on a project to analyze various methodologies that could be used for assessing the benefits of a national broadband network (a companion project exists on the cost side). Richard recently described key aspects of his project on ZdNet’s Twisted Wire program.
The main thing I’ve learnt from that podcast is that an accurate and precise measure of the NBN’s benefits will be difficult to calculate. There are two constraints, the first being the availability of data and the second being our difficulty in estimating externalities across economic sectors. For example, one approach would be to estimate a discrete choice model, asking people to choose between hypothetical bundles of broadband options. This would provide an estimate of their willingness to pay for specific characteristics. The approach would require data that does not currently exist, and even if such data were obtained (e.g., through surveys), it is unclear people can accurately assess their utility for some broadband-related goods/services that do not yet exist. A broader approach involves using a Computable General Equilibrium model which would yield an economy-wide estimate of the NBN’s impact on activity, but is especially difficult to implement where there are lots of interdependencies (such as with broadband). I also learned from the podcast that some benefits are easier to quantify than others, especially those that are already in use by large existing organizations.
It’s not entirely clear what this implies. However, it seems to me we can learn from parallel situations of how R&D projects are managed within large firms. Perhaps, we should stop looking at the NBN as an all-or-nothing investment. It is perhaps not realistic to do a complete analysis and match incremental costs to incremental benefits ex-ante. However, by breaking the project up into stages (geographically or by some other criteria), one could postpone the decision of whether to do later stages until additional information is obtained. Consider the example of Google’s decision to build a fiber broadband network for communities in the US. It would be difficult for Google to value the overall benefits of this network ex-ante. But that hasn’t stopped it from trying out this “experiment” with a few communities initially with the possibility of scaling up later. Shouldn’t we take a similar approach with the NBN?