After slamming the last Government for not doing a cost-benefit analysis on the NBN, Malcolm Turnbull has produced the goods. My view on cost-benefit analysis remains unchanged: if you have already decided what to do, a cost-benefit analysis is of no help except for vindication.

Nonetheless, there is another unintended consequence of doing one: the Minister in charge of it all has to explain it. Well I have to admit that I was impressed by Malcolm Turnbull’s “on the fly” professorial treatment at the Guardian. There aren’t many politicians who can explain optimisation while at the same time using the notion of ‘marginal utility’ correctly but there you have it.

What is nice about the cost-benefit structure is that it grounds the debate and makes assumptions clearer. For instance, Turnbull argues that on the supply-side costs of middle of the road technologies (that support fibre to the node) are falling (or improving in their quality) at a faster rate than technologies that support fibre to the premises. I’m not sure whether that is the case but if it is, then that is support for fibre to the node. Turnbull also does not assume that technology is changing the ‘utility’ function that drives the demand for broadband. However, as many will argue, this assumption would, based on experience, not be valid. What the analysis forces people to do, therefore, is explain precisely how demand conditions are changing rather than ‘this is more and I want more.”

But let’s turn to the report itself. I’ve skimmed it because, frankly, that is all the exercise deserves (given that it won’t impact decision-making and all). The first odd bit is that the benefits and costs are not relative to say not having done anything years ago but relative to what they are intending to do. That means that if that is the best, it will have the highest ‘net benefit relative to the reference case.’ But that doesn’t mean that it has a positive net benefit overall. To be sure, as a matter of economic decision-making, net benefits relative to the status quo is a good approach but in a document like this it triggers some scepticism.

Moving on, the report is useful in that it provides some assumption testing. For instance, the report concludes that at the current level of willingness to pay (WTP) for broadband (I’ll return to how this is measured below) then the current government policy has higher net benefits than the FTTP policy of the previous government. But as Turnbull’s little lecture tells us, if WTP grows then the optimum will shift. The report says that, in a decade’s time, the optimum will shift is WTP is 250% higher than it is today (i.e., it grows at about 13% per annum). Now, let’s think about what that means. It does not mean that people use two and a half times more broadband than they do today. Instead, it means that their WTP for very high speed broadband will be two and a half times more than today. 

To get a sense of that, consider what the equivalent thing would have been from the perspective of 2004. There was no YouTube so the demand for broadband speeds of today would have been driven by YouTube. My guess is that (based on current Telstra broadband prices) that WTP of the marginal consumer increased from close to zero to over $50 per month. Thus, the WTP growth criteria was easily met which is why we have high speed cable running past most urban households. The point is that it is not that hard to imagine a new product coming out that does, in fact, drive WTP growth of the order that changes the optimum network. (Virtual reality comes to mind as a possibility but, the difficulty here is that it is hard to tell). That said, the fact that 6 years ago, a government believed that demand in four years would be high enough to justify FTTP looks somewhat silly.

On to WTP itself. The report uses ‘choice modelling.’ This is a standard way of trying to extract consumer willingness to pay. The issue faced is that when it comes to some products, consumers may not fully understand what they are payiing for. So the methodology includes some random sample who are given a lesson in broadband speeds and capabilities. This allows them to derive a WTP estimate for broadband today. This is a much better approach than the craziness that we have seen in the past on broadband in Australia and, it looks pretty reliable as far as these things go. 

The problem is that (a) it cannot possibly take into account new products and (b) it is based on private values. The first problem is endemic to these exercises. As I pointed out, an analysis in 2004, would have likely got the optimal network spectacularly wrong and what the report does not really grapple with is that it may be wrong in the same way again in 2014. That said, we move beyond our ability to do objective analysis when we get into technological forecasting but it is worth emphasising that this issue doesn’t go away.

The second problem is more critical. We have to remember that this is largely a public investment. The relative comparator is that the investment is all private. That scenario is not really investigated (as far as I can see) and so it is surely the case that we want the public benefits from this government policy to be enumerated and not simply a calculation of private ones. I guess that is too much to expect in this world but it still remains that the NBN is a massive public outlay mostly benefiting the rich. If it could bring real competition that would diffuse the benefits to the wider population but I am yet to see a plan as to how that is going to emerge.