The Grattan Institute’s report on solar PV, written by Tony Wood and David Blowers, has received plenty of adverse reactions. That’s not surprising, considering its sensationalist presentation, tendentious arguments and dubious analysis; in all these respects, it’s well below the standard I’d expect from Grattan.

I’m a little reluctant to pile on further, but I don’t think the criticisms I’ve seen have really focused on the fundamental problem in the Grattan analysis. Although the summary suggests the report is about subsidy schemes designed to promote renewable energy and particularly solar PV, this is highly misleading. The core of the Grattan analysis relates to the electricity distribution network. The analysis is both conceptually unsound and empirically weak.

The approach used in the Grattan report is to compare the cost of rooftop solar PV with that of (mainly) coal-fired electricity, adjusted for a carbon cost of $30/tonne. No allowance is made for distribution costs, and any difference is regarded as a subsidy. That’s defensible, though dubious, in relation to feed-in tariff schemes (all of which have now been closed), but the big numbers in the Grattan analysis come from applying the same analysis to self-generation. According to the report, someone who installs PV and uses the electricity themselves, saving the retail price (say 25c/kWh) is being subsidised by the amount of the difference between the retail price and the wholesale price of the electricity they would otherwise have bought (say 5c/kWh, for a subsidy of 20c/kWh).

This reasoning is bizarre, to put it mildly. Grattan’s reasoning is equally applicable (almost) every time you turn off an appliance or lightbulb: by saving 25c/kWh, you are robbing society of 20c.

The underlying argument is that electricity distribution charges are levied on a uniform basis per kWh, but the costs of the electricity distribution network depend, to a large extent on peak demand. This mispricing reflects the mess that was called “electricity market reform”, dating from the 1990s, and has nothing to do with solar PV or renewables. The real subsidy here is to air-conditioning, which produces a demand peak in the late afternoon and early evening. Stretching a bit further, you could say that electricity mispricing subsidises hot dinners and Who Wants to be a Millionaire?, which produce electricity demand at or near the evening peak.

The Grattan argument is that consumers who reduce their electricity use at times other than the evening peak, for example by installing solar PV, or simply by being careful, are thereby shirking their obligation to fund the costs of meeting peak demand. Apparently, consumers are supposed to make decisions based on the prices that would prevail under ideal pricing rules, rather than those they actually face.

But, even granting all these assumptions, the Grattan analysis is both theoretically and empirically unsound. The errors are such as to make their estimated subsidy to solar worthless, even accepting their analytical framework.

The empirical error is to assume, on the basis of misleading and inadequate data, that solar PV makes no contribution to meeting peak demand. The data on which this assumption is based is derived from just five days, and one state: the second Tuesday in October, in south-east Queensland from 2009 to 2013. For this data set, they find increased use of solar power at midday, reducing demand, but no change in the peak demand, which occurred at 6pm, when most people are at home, settling down for dinner, and when air-conditioning demand is strong.

This is a startlingly convenient choice of observation. It happens that sunset in Brisbane in early October takes place around 5.50 pm, just in time to rule out any contribution from solar PV. In choosing this date to request data from Energex, it may have escaped the Grattan team’s attention that:

  • Brisbane is close to the easternmost point in Australia, implying an early sunset;
  • Brisbane is the most northerly state capital, implying less seasonal variation in daylight times;
  • Sunset is later in December than in October; and
  • Queensland does not have daylight saving.

If the Melbourne-based researchers had sought data from local sources, for December, they might well have found that the peak demand occurred well before sunset (8.40pm).

It is, of course, true that, given the pricing distortions noted by Grattan, many solar panels are located sub-optimally from a social viewpoint. But it is absurd to claim that this constitutes a subsidy.

Important though the empirical error is, it is less significant than the theoretical error. In treating the entire avoided wholesale cost as a subsidy, Grattan assumes that the optimal distribution charge for off-peak use is zero, and that the entire charge should be imposed at peak usage times.

This is an elementary confusion of marginal and total effects. If such a policy were adopted, the distribution charge at the current peak time would have to be increased to something like$5/kwh. But of course, with such a pricing structure, the peak would turn into a trough, when only the most essential uses of electricity continued.

A more reasonable assumption would be that the peak charge should be something like double the current value. Assuming that the peak charge applied to 10 per cent of total usage, that would permit a reduction of 10 per cent in charges at other times, or around 2c/kwh. But with that pricing structure in place, the social value of solar PV would be around 23c/kWh, at least four times the value imputed by Grattan. Recalculated on that basis, the alleged subsidy would disappear almost completely.

To sum up, the headline finding of the Grattan report is totally wrong, even ignoring all the criticisms that have been made of the analytical framework. The report should be retracted and rewritten.