This paper argues that the 'competitive liberalisation' of national governments of the past several decades reflects national governments' expectations of larger trade impacts from regional economic integration agreements (EIAs) than typical ex ante economic models have suggested. Moreover, we show that previous (typically cross-section) ex post empirical evaluations of the effects of EIAs on trade have seriously over- or underestimated the effects, partly due to ignoring the (endogenous) self-selection bias of country pairs into EIAs. Accounting for this bias, we find that economic integration agreements in the Americas have had much larger impacts on trade over the period 1960-2000 than previously found and the ex post estimates are less fragile than those in earlier cross-section analyses. The results shed further light on understanding the causes and consequences of the growth of regionalism in the world. Copyright 2007 The Authors Journal compilation 2007 Blackwell Publishing Ltd .