Bruce Mountain, 08 Jun 2015
The Grattan Institute recently released a report “Sundown, Sunrise. How Australia can finally get solar power right”. By the end of the second paragraph we have been told that rooftop photovoltaics (PV) have made Australia almost $10bn poorer and that “governments created a policy mess that should never be repeated”. The authors have not minced their words.
The Australian and the Australian Financial Review followed-up a day later with editorials saying that they had been banging this drum for years and now, at last, they have been proved right.
Others have been less complimentary. John Quiggin takes issue with the analytical framework and calls for the report to be withdrawn and re-written.
We thought we would leave their analytical framework (not that we agree with it) and simply check if their sums add up. The main thing we focussed on was their calculation of the “costs to society” of rooftop PV systems installed between 2009 and 2014, from which their controversial $10bn conclusion.
The calculation that we understand they have done is to set the discounted costs of PV against their discounted benefits, to give the net present value of the difference. For costs they include the initial outlay and then subsequent maintenance expenditure. For benefits they value emission reductions and generation that is avoided. We think their analysis does not seem to withstand scrutiny for several reasons:
Firstly, their calculations discount future benefits at 5% real per annum. Discounting at 5% real is implausible for economic “cost to society” analysis such as theirs. Infrastructure Australia recommends that the Public Sector Comparator Discount Rate should be the nominal risk free rate, which should be based upon a long-term government debt instrument. This is 3%, nominal. Even lower rates in social cost studies are easy to find. The authors said that following our questions they “made inflationary adjustments to all future transactions”. This turns a 5% real discount rate into a 5% nominal rate, but must also mean a much lower net present value from the amount that is in their published report.
Second, their total capital outlay discounted at 3% nominal instead of 5% real is, we estimate, around $17.5bn (they said $16.6bn discounted at 5% apparently). Using the same data sources they reference we think the outlay is $12.3bn.
Third, as part of their maintenance cost estimate they assume inverter replacement after 10 years assuming this costs 15% of the initial capital outlay. Future inverter costs are independent of the cost of systems installed many years earlier. But using the correct aggregate capital outlay produces a plausible estimate, which we have used.
Fourth, they have done a 15-year calculation. This is not plausible. We have used 25 years, which is typical of the lifetime warranty that solar panel manufacturers offer.
Fifth, in their report they recognise that PV production at the time of simultaneous network peak demand provides a benefit by reducing demand on the grid. In this respect, PV does exactly the opposite in terms of demands on the grid to what air conditioners do. But their report does not account for the network benefit of PV because the authors say it is difficult to quantify. Grattan have not found this difficulty in quantifying the impact of airconditioners, which in a previous report they said added network costs of $1200-$1550 per system. We agree it’s hard to be certain on the network benefit of PV (and network cost of airconditioners). But if Grattan can overcome the difficulty of counting the network cost of airconditioners they can find a way to do the same in counting the network benefit of PV. In research with my colleague Paul Szuster (Australia’s Million Solar Roofs: Disruption on the Fringes or the Beginning of a New Order?) we have estimated network benefits of rooftop PV of between $0.9 and $2.1bn present value. In our re-run of the authors’ calculations we have taken the mid-point of this range.
Correcting the analysis, the authors’ conclusion of a net cost of $9.7bn becomes a net benefit of $1.2bn.
Interpretation of the authors’ report in the media has been further confused in respect of the $3.7bn that the report calls a “cross subsidy” that households with rooftop PV will be receiving from other electricity users (because PV households buy less from the grid). However we now understand that this was intended only to refer to the current revenue cap arrangements whereby network businesses get from Paul what they lose from Peter. The reference to “cross subsidy” was not meant to suggest that they believe that this is appropriate. It would be helpful if Grattan clarified its stance on this.
The authors have made an unequivocal claim that there was “a policy mess that should never be repeated” and that Australia is about $10bn the poorer for this. As far as we can see, even leaving aside the concerns about their analytical framework, their numbers don’t add up. This is easy to confirm or refute if the authors put their data and calculations in the public domain as might be expected for any public report but even more so for one that makes such unequivocal claims.