They reference the Report on Pension Charges in Ireland 2012 by Dept Social Protection which mentions an average figure of 2.18% and then reference a note to this Dept Social Protection document by the Trinity School of Business which claims the 2.18% figure underestimates the total fees significantly and provides some detail of where the original document went wrong. The Trinity note doesn't come to a firm number, but it does reference studies in the UK which saw figures of 3.2%.It assumes that the "typical" annual charge is circa 3% without having any evidence to back it up, at all.
I suspect the author of this new paper took 3% as a round number somewhere between the low 2.18% from Social Protection and the high 3.2% mentioned in the Trinity note. Considering they went on to reference 3% nearly 90 times in the document, it seems quite sloppy not to have provided a clear statement as to why they chose that figure.
It's a strange document. Some significant effort clearly went into writing it, but some of the language is very casual and neglecting to provide more context for that 3% figure seems like an obvious error. I guess it wasn't ready for publishing yet.
Bit of a missed opportunity because the substantive point that "Under the current system, the individual pension investor is encouraged to play a role in making important decisions which affect their pension. In practice, once the individual has signed up they are substantially at the mercy of an industry that is primarily interested in its own welfare rather than the outcome of the investors." is entirely valid and needs to be dealt with.