A few weeks ago I argued that if the US wanted Australia to increase defence spending, then rotating troops through Darwin and announcing a 're-balance' that would result in more US military capability being moved to Australia's neighbourhood was precisely the wrong way to do it. Such moves would have the opposite effect of actually encouraging free-riding. Why spend more on defence when the Americans will do it for us?
Two follow-up points on that, the first relating to the graph I just saw in Mark Thomson's recent post on Australian defence spending as a percentage of GDP. If my point about incentives is right, then the corollary ought to be that a reduction in US defence commitment to the region should have the opposite effect on Australia, encouraging us to do more. Yet the table suggests that Nixon's Guam Doctrine announcement in 1969 seems not to have stimulated Australian defence spending. And the rise in the 1980s actually coincides (roughly) with the Reagan era, when US defence spending was also on the rise. 
The second point is that this sustained decline in defence spending as a percentage of GDP may not indicate declining military capability. As John Sides recently wrote on The Monkey Cage, such an assumption may overlook the declining costs of military capability:
...as defense technologies improve and states develop economically, they are able to sustain or increase their militarily capabilities with less expenditure and personnel. The anecdote of Singapore is telling, where military spending as a share of GDP fell from 5.1% in 2002 to 4.2% in 2007. During this time period, the state added 6 frigates, 20 fighter jets, 20 attack helicopters, 96 main battle tanks, and scores of precision-guided munitions into its arsenal.
This analysis suggests that the cross-national declines in military expenditures and force sizes have less to do with political demilitarization, and more to do with the increasing technological efficiencies of defense markets.
But is military technology really getting cheaper? Other evidence suggests not.