Let's assume that the ratio of inexperienced riders to experienced riders has increased over the past 10 years.
It stands to reason that during the motorcycling slump of the 1990s, a large percentage of the registered motorcycles belonged to older, more experienced riders who were dedicated enough to stay with the sport even as support and interest wained. Let's also assume that all motorcycles registered since 1996 were registered by newbies.
Can we correlate the increase in new riders with the increase in accidents? Would this explain the ~35% rise in fatalities even as motorcycle training, gear, and equipment has improved?
I think changing average experience in the riding population--and consequent changing average risk--explains part of the decrease in the fatality rate from mid-80s to mid-90s, and virtually all of the subsequent increase.
Hurt and MAIDS both found that risk decreases with experience, as you would expect. Another confirmation can be found in an evaluation of the California Motorcyclist Safety Program. But I don't have a way to show quantitatively the correlation we both think exists.
Go back to the first chart in this thread, which shows regs, deaths, and rate. It appears that registration decline and growth is the force that drives average risk--at least that makes more sense than the converse. When the riding population fell, the n00b percentage must have fallen too. So average experience would increase and average risk would drop. Then when registrations began to climb, the n00b percentage followed, average experience fall, and risk increased.
Viewed that way, the puzzle becomes the rather mild rate increase since 1997. Paralleling a steady drop in regs from the early '80s was a similar decrease in the rate. But then when sales really took off, the rate increased more slowly, and is still nowhere near its '80s highs.
That disparity can be explained, as I see it, by the changing age distribution. N00bs in the 1980s were teens and twenties, the highest risk age groups. Today the average rider is a 40-something and many n00bs are in their 50s, both historically low risk groups.
If, as it now appears, the motorcycle market is headed for a slowdown, this hypothesis would predict a drop in the fatality rate in the next few years, possibly to a new all-time low. Regs will stabilize, if not fall, there will be fewer n00bs, average experience will increase, and average risk will decrease.