Bruce Mountain and Paul Szuster
12 Jun 2015
Yesterday the Grattan Institute released the Technical Appendix to their report “Sundown, Sunrise. How Australia can finally get solar power right” which they released almost three weeks ago. In their report, the authors calculated the societal cost of photovoltaics and concluded that “governments created a policy mess that should never be repeated.”
In our earlier article we suggested that the authors seemed to have miscalculated the societal cost and we suggested that it would be easy to confirm or refute this if the authors put their data and calculations in the public domain.
The Technical Appendix clarifies some aspects of their calculation although other parts such as how they calculated benefits, particularly avoided generation, are still opaque. However on the basis of the information provided in their Appendix we now feel comfortable to conclude that their calculations are not correct.
Specifically, in calculating the societal value of PV, the authors have not done a standard net present value calculation as might be found in any standard corporate finance text book (see for example Brealey and Myers’ Principles of Corporate Finance). Instead of discounting the capital outlays to the start of the period they have inflated the nominal outlays at 5% per annum to put them in what they are calling 2015 dollars. This makes no sense. Inflating nominal outlays at a discount rate does not deliver a 2015 dollar value of those outlays it simply produces an inflated value of those nominal outlays.
A standard treatment would convert the nominal outlays into a constant currency – typically the currency at the start of the period and then discount the constant currency outlays at the real discount rate. The present value can then be escalated at CPI to put it in currency of whatever chosen date. This standard approach gives a much lower value of the outlays in 2015 dollars than the approach adopted by the authors. For those wishing to explore further I have put a spreadsheet on the blog part of our website to demonstrate this standard approach and compare it to the approach that the authors have adopted.
We also note that the Technical Appendix charts the cost of PV installations, but this shows values particularly in the period from 2009 to 2011 that are not consistent with the cited sources.
Returning to discount period, the authors defend a calculation to 2030 and no residual value, on the basis that “we have heard a great deal about warranty problems”, and that the investment peak occurred in 2011/12 and hence the average life will be 18-19 years. Neither of these reasons are plausible justification for an effective average life of around 17 years. The author’s calculation of PV output is already reduced by 25% of installed capacity to reflect their assumption of performance and in addition they assume that 25% of panels are west facing, reducing output even further. As a result, the authors calculation of PV electricity production is even lower than assumed in the Warburton Review.
Standard panel manufacturers provide panel guarantees for 20-25 years and the authors make allowance for maintenance and inverter replacement in their calculation. A more realistic (and standard) approach would have been to adopt an effective life of 25 years with no residual value. Or discount just to 2030 (an effective average life of around 17 years) and then make a reasonable allowance for residual value for a proportion of the systems and parts of systems that will reasonably outlive this.
Returning to whether they have discounted at 5% real or nominal, the authors are adamant that they have discounted at 5% nominal, though they never said so in their report. Its absolutely clear that in their calculation of carbon benefits and capital outlays this is not the case, and we simply can not be sure for the rest until they publish their Excel workbook or simply the relevant parts of their calculations. Again we call on them to put this in the public domain.
Furthermore in the justification of their 5% (apparently nominal) rate the authors say that they chose this rate because households borrow at this rate. This reflects a misunderstanding of the appropriate discount factor to use in societal economic analyses such as they claim to have done. Societal economic studies, unlike project investment analyses, do not rely on the cost of capital to investors to determine discount rate. They rely on the cost of capital to society. Economists specialised in this field will argue whether the Social Rate of Time Preference or Social Opportunity Cost should be chosen, but either way a plausible upper limit on economy-wide discount rates in such studies will be the long-term government bonds rates, currently 3% nominal.
For completeness we also note that these calculation problems relate to their calculation of subsidies as well. They have inflated the value of small scale renewable energy certificates by 5% to put it in what they call 2015 dollars. This is wrong. It is just an inflated statement of the certificate payments. If they wanted to get a 2015 dollar value of the certificate payments they should have inflated nominal payments by the consumer price index.
Likewise they define the $3.7m reduction in network revenue from households with PV as a “cross subsidy”. They present this as an objective truth. Its not, it simply reflects their assumption that networks have an entitlement to get back revenue as if households had not installed PV. An alternative perspective is that networks have no such entitlement in which case the $3.7m is not a cross-subsidy to households with PV but excess revenue recovered by the networks: a subsidy, if you like, but from consumers to shareholders.
Finally, these comments all relate to calculation. Defective calculation will break a report but correct calculation does not necessarily make it. We again refer to Professor Quiggin’s critique of the analytical framework and draw attention to the fundamental concerns he has identified.
While we don’t think the authors’ analysis is robust and hence their conclusions are not sustainable, they deserve credit for broaching a topic worthy of much attention. The very rapid development of rooftop PV in Australia was stimulated (largely) by government policies that pulled (through feed-in tariffs and capital subsidies) and pushed (through inadequate network regulation and insufficiently competitive retail markets). Far more has been invested in PV than other generation technologies since the start of the National Electricity Market, and the majority of that investment has been by households not utilities. This is a seismic shift. It has brought a whole new level of dynamism to value chains hitherto characterised by monopoly and oligopoly. In the process it is creating wonderful new opportunities for households and businesses. Or as neatly put by one of the utilities, advertising their wares on the side of a tram: “don’t waste your roof, make energy fresh daily”.
A “policy mess never to be repeated”? Nah, wide of the mark.