Broadband and the Missing Cost-Benefit Analysis


This piece [over the fold] appeared on ABC Online today. It was inspired by some Twitter interaction with Malcolm Turnbull following his SMH piece today. And yes irony about OzEmail investors is (a little) intended.

Broadband and the missing Cost-Benefit Analysis

The Drum, 15th September, 2010.

In 1492, Christopher Columbus sailed the ocean blue, didn’t fall off the edge of the earth and returned with enough prospects to do it again another three times. The project was funded by a combination of public and private funds. Whatever may have been its goals, there is little doubt that it earned a “commercial rate of return” for all concerned and arguably at least for Spain, a decent public return too.

But think about what might have happened here had hard-nosed economists and accountants had a go at evaluating the project’s worthiness. There was a high probability of failure with a complete loss of all funds. There was also a good chance that whatever was discovered was of little use. And then there were ‘pie in the sky’ outcomes of which it would be impossible to quantify properly. It is little wonder that the whole venture was touch and go and it took Columbus years to put it together. 500 years or so later some accountants tried to replicate the cost-benefit analysis and found it hard to do.

There is some similarity between Columbus’s First Voyage and the current debate over the NBN. Both are expensive. Both are requiring a mixture of public and private funds making commercial returns an issue. And in both cases, there is a considerable degree of uncertainty over the potential benefits. To be sure, the NBN will not necessarily fall of the edge of the earth but one of the scenarios is that wireless or a new technological breakthrough comes through and makes it a ‘white elephant.’

The problem here is that when there is uncertainty of this order in play, trying to crunch the numbers and yield a meaningful result is difficult. And what is more, it is going to depend on assumptions and maybe depend on them in a dramatic fashion. It is child’s play to dampen expectations and enthusiasm about the potential upside of the NBN. All you have to do is look at what people use broadband for now (with most bandwidth being video downloads) and there is no case for public investment.

Interestingly, if we played this game ten years ago, we would be able to do the same thing. I recall then that people identified video downloads (so called on-demand movies and television) as something that would come when we moved off dial-up and onto basic broadband (i.e., the slower stuff we have enjoyed since). The basic argument was that consumers already had good options through broadcast television, cable television (which was pretty much just beginning) and video libraries and just wouldn’t pay to watch videos on their computer screen. This means that all of the investment in broadband since would surely have been in error.

In many respects, those assumptions about demand were correct. But they were dramatically incorrect because they failed to conceive that non-copy-protected, relatively low quality video output could absorb so much time and fuel broadband demand. The point here is that no cost-benefit analysis with reasonable assumptions and sanity checks would have included the YouTube scenario. And yet here we are.

I don’t really want to speculate on whether there is a hidden YouTube in our future that justifies the NBN. But I do want to point out two things. First, around the world, there are few examples of firms and governments who have invested significantly in infrastructure in telecommunications who have ended up bankrupt or regretting it. Second, the hidden YouTube could equally be something that drives us to prefer wireless rather than wired networks or perhaps a network configuration that the NBN may not be considering.

My point is that the Government should be held to task on articulating what the benefits of the NBN are. My concern is less that those benefits be quantified but that they are actually public benefits. That is, I don’t think YouTube is worthy of government subsidy but I think education and health may be. But there is precious little discussion of all of that and trying to actually put in policies that deliver them. Asking for finances detracts from the real economic issue of delivering a social dividend.

Malcolm Turnbull likes to talk about opportunity cost: what else could we spend billions of dollars on? I agree that that is a good question. But I also want to ask him: what is the opportunity cost of trying to get the government to cough up numbers that will surely be of little value relative to the real job of working out why the Government is using the NBN to detract from the many other issues in digitisation and competition in telecommunications that remain in the dark? Mr Turnbull: ask Columbus what he is going to do after he discovers America.

Joshua Gans is an economics professor at Melbourne Business School. He blogs on these topics at Here is his recent advice to Malcolm Turnbull.

11 Responses to "Broadband and the Missing Cost-Benefit Analysis"
  1. A tautology is not an argument. Anything follows from a tautology. And anything follows from Columbus discovered America without a cost-benefit analysis, therefore we should do x!

  2. Matt, you’ve missed the point. The point is simple – CBA is only useful if you can can get reasonable estimates of the chances of different outcomes.  There’s nothing tautological about that statement.

    In economics, it’s the distinction between risk and uncertainty.  Risk is where you know enough to make  reasonable probability estimates.  Uncertainty is a true “Black Swan” (eg what’s the probabilty aliens will destroy the world next week?).  Tools like CBA, Black-Scholes option formulas, etc are pretty good at managing risk, but hopeless at managing uncertainty.

  3. That’s a fair point Derrida but it’s not how Joshua framed the argument.
    Can we just drop the rubbish about telegraph lines and railroads and discuss it rationally?
    In any case, I do not agree that the NBN is a case of Knightian uncertainty. I don’t see why we can’t make reasonable estimates of what demand, price and cost is likely to be. Indeed, the government spent $25 million doing just that.
    But can we release the KPMG-McKinsey financial model. Clearly the NBN should not be on the govt’s books as a $43 b asset. The report itself shows that it will not make a commercial return so how can the government justify keeping it on its books at cost?
    Then we will know the size of the implicit govt subsidy and we can then discuss whether the spillovers are worth such a subsidy?

  4. Before a cost benefit analysis is launched – to determine whether a given thing is worth doing or not – shouldn’t we undertake a comprehensive national inspection and audit of the physical state of our existing internet infrastructure?

    After all, it consists of metal wires buried in the ground half-a-century ago in many cases, in some places for longer.

    Unless we first know the true physical condition of the infrastructure we are reliant upon, any cost-benefit analysis will begin with the assumption that the existing infrastructure is in good shape all round and can be expected to be so for decades to come. After half-a-century of exposure to water, earth movement and soil chemistry, I think it is reasonable to suspect that this is not the case.

    I don’t want to get into a debate about whether or not alternative technologies driven by the market will be able to adequetly step in to replace the ageing copper with little or no public investment needed but I do think this is something that needs to be determined first.

    If we never take a thorough look under the bonnet of our old car, we really don’t now how long we do or don’t have before it needs to be replaced by a newer model of some description.

  5. One of the issues that complicates any discussion of cost-benefit on the NBN is the governments’ choice of a “platinum” option, i.e. fibre optic backbone to the home, of broadband for all. I am guessing that the marginal costs of doing so are driving the $43billion price-tag. What is the trade-off between a defined level of broadband capacity and a universal public good, e.g. is it 1 Gbps for the platinum version or is a 1 Mbps cable service good enough for most households. (Might determine the pay-off between FTTN and FTTH, for example.)

    I think the 1492 analogy is not very helpful to the argument. Why not use recent history with respect to universal fixed line phone coverage and Australia Telecom. Can we derive useful information from that experience. For example did enabling all households with a personal (copper line) phone increase productivity or was it a pure public good. Did creating the public network enable businesses to be the driver of productivity? Or some combination of all of the above? Build it and they will come?!?

  6. My broadband costs $60 for 4Gb ($15/Gb), and local/std phone calls (Virgin Broadband).
    On the NBN, iiNet is charging ( $60 for 100Gb ($0.60/Gb) peak (plus 100Gb offpeak) includng local/std calls.  If NBN costs $5000 per household (ie $43B/9M homes),
    then how long until the new lower price per GB pays back my $5000 NBN investment? The NBN is 25 times cheaper per Gb on my plan at an investment cost of $5000. At normal usage (4Gb/mth), it will take  90 months (347 GB) to pay back the $5000. But if I start to use 10 times as much data, does that mean I will earn back the 90 months rate in only nine months. I will get 25 times more data for the same price. How much value can I get from that extra data (Gb)? Is that value worth $5000 investment?
    Of course not every household pays our rate per Gb? What about a household which pays $3/Gb ($90 for 30Gb, inc phone calls), or another that pays $40/Gb ($30 for 2Gb, plus $50 phone inc calls)? If these households can increase their data usage ten fold, does that mean the payback period for the $5000 investment is under twelve months? What about houses without computers? What about 20% of households who aren’t interested in broadband? Can their lack of interest be offset by households who get more than $5000 benefit from the NBN?
    Cost per Gb on the NBN is halving every twelve months. This is the real economic incentive for the NBN. Price per GB is heading towards zero, plus fixed installation costs. How can we rebuild the economy to have it driven by a resource with a zero marginal cost? What kind of business do you want to be in that has zero marginal cost? Selling atoms, or selling bits….?
    Investment tends to exploit the cheapest resource: labour, oil, iron, coal and now broadband. With cheap plentiful Gb, we will likely find many new things to do with a Gb, that can substitute for other costs, such as, transport costs.

  7. There is a growing body of knowledge that indicates that Christopher actually used a Chinese map that had, almost 100 years before, explored and documented that part of the world…the risk would have been trusting the Chinese map…which is far more modest a risk than thinking you were sailing off the end of the world.  A very poor argument when discussing private versus public good. To support the NBN means you believe that there has been catastrophic market failure in the provisioning of internet/broadband services over the last 10 years, despite multiple major suppliers and the ACCC managing wholesale pricing from Telstra and that the only alternative is a government monopoly to rebase the economics and service levels of the entire sector.  Naive doesn’t begin to describe this assessment of the Labor govt and its wealth redistribution mechanism called the NBN.

  8. I agree with the general point, but I’m not sure  I follow the youtube analogy.  They predicted we’d use it for downloading video and they were right?  And it wouldn’t mean a priori that the investment was unjustified, just that publicly subsidised investment wouldn’t be.
    In any event, at the time the investment was justified on clearly documented exponentially increasing traffic. Network operators didn’t particularly care about what it was used for.

  9. Fibre to the home has been attempted by TransACT in the ACT. It didn’t revolutionise our health/education systems, and economics limits the data/speed pricing despite much higher theoretical speeds being possible.
    Having spent some time working on e-Health, the barriers to a digitised health network are not technical ones. Intellectual property of medical records, the AMA protecting its members, and the inefficient pricing of medicare are the real issues.
    The problem with the NBN is the same problem as judging a car by its top speed. Speed and data are an issue, but mobility is more important once minimum speed/data is sated. Once mobility has been addressed, I would imagine that speed/data would again become an issue in a cyclical fashion.
    The tyranny of distance is forever prevalent in Australia. A mix of wireless with a fibre backbone, and improved pipelines to the USA and Asia would be a smarter plan.

  10. It doesn’t matter where the fibre’s going to, it’ll be fast enough to transmit and receive video, which are the assumed prerequisites for eMedicine.

%d bloggers like this: