Hands up if you’re a media commentator that has demanded the government undertake a cost benefit analysis (CBA) of the National Broadband Network?
Keep your hand up if you have any idea at all about how to actually do one?
If we were in a large hall right now filled with the nations media chirpers – there wouldn’t be a raised hand in the entire room. The reason for this is pretty simple – if you actually know how economic cost benefit analysis works, you understand that it is completely incapable of producing even a marginally meaningful result on a project like the NBN.
Demands for a CBA on the NBN – as if it were some definitive document whose result is pivotal to the project’s implementation – speaks more to the ignorance of those making the demand than it does about anything actually related to the value, viability or otherwise of the NBN itself.
Save up to 50% on a year of Crikey
Choose what you pay, from $99.
It’s not a matter of opinion, it’s a simple matter of methodology, and to understand why we need to understand uncertainty.
There are two types of cost benefit analysis. The first is effectively an accounting exercise utilised by (usually) private firms where the *only* costs and benefits they are concerned about are their own. It’s a way to determine whether the benefits to the firm that would be delivered from a given investment over some specified period of time, would be greater than or less than the costs incurred by the firm from that investment over that time.
It estimates the net private benefit that a given investment opportunity would be expected to deliver and ignores the costs and benefits that people and organisations other than the firm would experience as a result of that investment choice.
A good example here is a private firm looking to build a shopping centre. If the benefits to the firm of building a new shopping centre outweigh the costs to the point where the rate of the return on that investment is larger than they could achieve by investing in some alternative – the new shopping centre would go ahead. But what doesn’t turn up in this sort of CBA are the benefits accrued by the *public* from the shopping centre (such as reduced travelling time to obtain goods and services as well as employment). Nor do the costs borne by the public from the shopping centre turn up in this type of CBA such as increased traffic congestion around the shopping centre, increasing the commuting time for those that have to drive past the new development on their way to work. The CBA for the firm is only interested in the costs and benefits that apply to the firm and not the costs and benefits that are externalised onto others.
Another issue that commonly arises with this type of firm level CBA is the length of the future time period over which all costs and benefits are measured. As a result of modern shareholder and broader market expectations having a larger focus on the relative short term performance of firms and their investment activity , what we find is that CBA analysis increasingly has to justify an investment over a much shorter period of time than it once did.
So an investment that might have run at a net loss for a decade before earning squillions from then on (earning a huge net private benefit over a long period of time), is harder to justify at the firm level these days (harder to justify to shareholders and the broader market) than an investment that is expected to run at a net loss for a much shorter period before earning substantially less than squillions from then on (turning a net private benefit quicker, but one which is smaller in magnitude than the longer term alternative).
As a result of this shorter term, firm level focus – firm level CBAs are generally time constrained, forced to use a shorter interval of time over which to judge any cost-benefit differential.
Some commentators are making the mistake of believing this type of CBA on the NBN is appropriate – they are wrong. The NBN is a piece of public policy, not an exercise in measuring private net benefit.
The second type of cost benefit analysis – the type actually used for analysing public policy – is a much more complicated economic beast. Rather than looking at the net private benefit of an investment, it looks at the net benefit to human welfare. It looks at all costs and all benefits of a given investment, including the internalised costs and benefits of the investment on the organisation undertaking it (in this case NBN Co) as well as the externalised costs and benefits born by others (in this case, Australia as a whole).
The other difference between this type and the firm level type CBA is that the cost benefit analysis of public policy isn’t time constrained – it can measure costs and benefits over the entire lifetime of a project, unencumbered by the type of noisy bitching and moaning that comes from pimply faced 25 year old market analysts in ticket clipping firms obsessed with next quarter’s financial results.
However – it is this lack of time constraint which ultimately makes any CBA on the NBN a meaningless exercise, simply because time is both risk and uncertainty, but not risk and uncertainty as we may ordinarily understand the terms.
At its most basic, uncertainty is stuff we simply don’t know.
Risk, on the other hand, is the uncertainty that we still don’t know, but which we can reasonably quantify within some probabilistic framework – an educated guess, if you will, that will have varying degrees of robustness (depending on how clever we all are).
Let’s take the weather for example. We could say that there is an 80% probability of rain tomorrow. This is a risk – we don’t actually know if it will rain tomorrow, but we can give a probabilistic estimate of the likelihood of it raining based on our reasonably robust weather models. So if there were, say, ten days in a given month where we gave a forecast of an 80% probability of rain occurring the next day, it would actually rain on around 8 out of those 10 “next days”.
But what is the probability of you coming home tomorrow from work, only to find that your mail is wet?
We know there’s an 80% probability of rain, but we don’t know if we’ll get any mail tomorrow, we don’t know if the Postie will push it all the way into the letterbox, we don’t know if the wind will be blowing in such a way that our mail would get wet regardless.
This is a an example of uncertainty – an event that we can’t quantify nor even guess at, even though we’re pretty sure it will happen at some time in the future.
A cost benefit analysis can incorporate the former but not the latter. It can and must incorporate risk – or reasonably quantifiable uncertainty that can be positive or negative – but it cannot incorporate ordinary uncertainty, simply because ordinary uncertainty is, by its very nature, completely unknown.
Which gets us to the NBN – a project with an expected life frame of 50 years.
Most of the cost side is actually pretty simple. Once we get the final estimated cost of building the project, off we go (we’ll deal with opportunity cost a little later)
The benefit side of the NBN CBA equation however is rather different. We could price in some benefits where the risk (remembering that risk can be both positive and negative) is reasonably quantifiable. For instance, we could probabilistically quantify the value of the benefit that would derive from telecommuting over the short term, by pricing in the reduction in traffic congestion that such telecommuting would deliver. We could use evidence from other countries to assist us with estimating that magnitude, as well as our own experience in Australia with existing trends in telecommuting as a function of high speed broadband availability.
However, the further out in time we go, the less robust (i.e. the more meaningless) those estimates become. Risk starts becoming uncertainty. For instance, how do we price in telecommuting benefits 25 years from now – a point in time that is only half way through the projects life – when we have little idea of how our transport systems will be operating in 25 years?
This is the problem that everything on the benefit side of the CBA equation faces. While the costs are reasonably known and relatively short term (because they are upfront), the benefits are either all dominated by uncertainty through the entire life of the project, or are quantifiable risk in the short term, but revert to unquantifiable uncertainty over the medium term and into the long term.
To highlight the nature of the problem – imagine doing a CBA on the post-war Post Master General’s roll out of what became the bulk of our existing copper telephony network. PMG spent 42 million pound to roll that out – about 10 billion in today’s dollars.
If a CBA was undertaken on the project back then, one of the benefits that could not have been included in the mix was the internet because it was yet to be invented (let alone fax, or basic data exchange before that etc) – even though if you have a DSL connection, there’s a good chance you are still using the original PMG cable system to read this very post. Even with the application of a standard discount rate applied to the value of the economic activity that the PMG cable side of the net has generated over the last 15 years in Australia (and fax and basic data exchange before that), it would still have been a not insignificant benefit in any original CBA done on the project if the creation of the internet and its economic consequences could have been foreseen at the time.
But it couldn’t have been included at the time because it couldn’t have been foreseen. It was pure uncertainty of the exact same type that dominates the benefit side of any CBA we could do today on the NBN.
As a result of the uncertainty involved, even the most miraculously brilliant and insightful CBA would simply tell us that over the life of the NBN project, Australia would be expected to experience somewhere between a small net loss to human welfare as an absolute worst case scenario, through to a best case scenario where the net benefit to human welfare is some magnitudes of order greater than the value of the investment itself, with a ceiling on that benefit that is simply unquantifiable.
A CBA on the NBN would effectively tell us less than nothing.
No amount of monte carlo simulation jiggery pokery to estimate risk distributions using the most innovative pricing mechanisms the world has ever seen, can overcome that simple fact – as it is uncertainty, not risk, that dominates the benefits. It is the unquantifiable things we don’t know, rather than the quantifiable things we can pretend to think we know, that dominates the entire benefit side of the equation.
When people tell you that we need a CBA on the NBN to determine whether it should go ahead, they are either being deliberate mischief makers (like, say, Malcolm Turnbull who would understand the reality, but needs to run political lines) or they are just simply ignorant – and that includes the media commentators and some surprising arsehats from the financial sector who bloody well should know better.
The other great furphy on the NBN is the parading around of arguments about the opportunity cost – the “alternative” things that the NBN money (the current 43 billion estimate of combined public and private investment) could be spent on. The problem here is that people are trying to have their opportunity cost and eat it too.
The opportunity cost isn’t what else we could spend the money on – it is the benefits that could be derived from spending the NBN money on alternative investments MINUS the cost of replacing the copper network with some other functional alternative.
The copper network is on its last legs. It is a decrepit, broken thing attempting to provide something it was never designed for, whose performance has already peaked and which will become increasingly degraded and prohibitively expensive to maintain over time.
The net you get over copper today is as good as it will ever be.
When we look at the historical demand for bandwidth in Australia, we see that the performance the copper network can theoretically deliver at its absolute maximum (which is not what nearly all of us actually receive mind you, since we don’t live next door to an exchange and the network has become degraded) is just about to diverge from our historical trend in data demand. There was a nice little chart in an NBN Co presentation that highlights the issue (click to expand)
The copper network will not be able to meet our future demands because our current demand is already at the maximum physical capability of the copper. It will need to replaced. There is no ifs, buts or maybes here – the copper cannot meet our demand into the future.
That makes the opportunity cost of the NBN the benefits we can derive from alternative investments MINUS the cost of replacing the copper network – which paints an entirely different economic picture altogether.
Especially when the most technologically efficient replacement to the copper network is the very thing that is being proposed – fibre to the premises.