It's not really a "theory" is it? If you look at the Thiel-Small equations, you have that the sensitivity of speakers being determined by a number of parameters, one of which is Sd, or radiating area. However, the sensitivity is proportional to the square of Sd, but only linear with respect to Mms (moving mass). So you're doubling the Mms, and doubling the Sd but _squaring_ the Sd parameter in the sensitivity equation, so you get, effectively, another 3dB of SPL's for the same amount of work.
Some equations:
Vas = 1.4 * 10^18 * Cms * Sd^2
Sensitivity = 112.2 * 10 log (9.78 * 10^(-10) * Vas * Fs^3 / Qes)
Where Cms is the mechanical compliance of the suspension (a number depending upon the rigidity of your spider, surround, etc.)
That's my understanding...
b