This week, my colleagues at Crikey have been looking into the data advanced by the copyright lobby about piracy.
In recent months, the content industry has shouted about piracy using two reports supposedly showing the vast economic impact of illegal downloading in terms of lost income for media industries and fewer Australian jobs. One of them, entitled The Impact of Internet Piracy on the Australian Economy, was not even released to the general public (Crikey managed to find a copy on Wednesday). Bernard Keane analysed it yesterday, finding it “calculated piracy cost figures with virtually no reference to what is happening in Australia” and ”uses projected growth rates to expand the cost figures massively”.
Today, I’m going to have a look at the other report commissioned by the Australian Federation Against Copyright Theft, entitled Economic Consequences of Movie Piracy: Australia. This one was released to the general public, and purports to be more rigorous than previous studies of this nature.
“The approach taken in building up these estimates offers a conservative view of piracy and does not treat every pirate view as a lost sale,” the report claims. “As such, readers should consider these as indicators that piracy is at the very least causing this level of harm.”
Unfortunately, this report’s methodology is also — at the very least — questionable. Electronic Frontiers Australia has already made some pretty valid criticisms of the research, but today I’m going to concentrate on the economic methodology used by the report’s author’s, Ipsos and Oxford Economics.
To start with, let’s examine their laughable “Annex 1” in the full report. This purports to explain how Australian Bureau of Statistics input-output tables are used to generate a final figure for total piracy impact in terms of lost sales and job losses. That’s important because it is these figures that AFACT has been using to argue that movie downloading represents a “$1.37 billion loss to [the] Australian economy”.
I’d like to say I carefully checked the report’s methodology for its econometric accuracy. Unfortunately, I can’t — because the authors at Oxford Economics and Ipsos don’t publish their equations; nor do they publish their raw data.
Just as an exercise, I downloaded the ABS input-output tables and attempted to match the ABS data to the AFACT report. It’s impossible. The data tables in the AFACT report which might allow that kind of scrutiny are missing.
What Annex 1 does tell us is that Oxford Economics and Ipsos have made all sorts of behind-the-scenes calculations to do with the exact value of the multipliers they use and the precise allocation of various ABS industry data to various categories of their assumptions. But they don’t tell us how these figures were arrived at.
To get a flavour of the opacity of the modelling, here’s their full explanation of two of the the multipliers they use:
“Type II multipliers of 2.5 (Gross Output) and 1.1 (GDP) were estimated. This covers activity in the Australian motion picture exhibition, production and distribution industries as well as TV VOD, internet VOD, downloads of motion pictures and the retailing of these motion pictures.”
There is no further explanation of how the numbers of “2.5” and “1.1” were “estimated” and no equation that shows us what they multiply. Hence, it is literally impossible to verify, cross-check or otherwise scrutinise these figures. Indeed, the full report contains no true methods section.
You couldn’t get this stuff published in a peer-reviewed economics journal, and it’s time journalists were smart enough to check this sort of data before they unquestioningly trumpet its findings to their readers. So, while the Ipsos/Oxford report is a slightly better effort than that produced by Sphere Analysis, it’s still opaque, unverifiable and unreliable.