We examine the social efficiency of alternative intertemporal permit trading regimes. Banking with a 1-to-1 ratio and with a non-unitary intertemporal trading ratio (ITR) are compared with each other and with the no-banking permit trading regime. The more industry-wide shocks vary, and/or the more they are negatively correlated across time, the more efficient is a bankable permit regime. When the slope of the benefit function is greater than the slope of the damage function, banking with ITR=1+r is more efficient than a no-banking regime. Banking with ITR=1 can be more efficient than a no-banking regime. However, whether ITR=1 or ITR=1+r is better depends on the covariance structure of the shocks and the benefit and damage functions.