The correct way to deal with this matter is to perfom a statistical hypothesis test: "That the difference between two and five year failure rates is statistically signifigant."
In my opinion, this is where our regulators are letting us down - badly. How long has CASA and its predecessors being collecting accident data, incident data, REPCON's and a host of other reports. Yet, predominantly decision making is based on hypotheticals from meetings in air-conditioned offices in Canberra.
Where do we see a learning effect? Where is there signs of continuous improvement? Surely by now we can refine regulations and practices in an evidence based manner?