Following the methodology proposed by the recent literature on this topic, we will first specify a gravity model of trade that includes multilateral resistance terms (MRTs), as proposed by Anderson and van Wincoop (2003), and we will obtain unbiased estimates by controlling not only for country-and-time effects, but also for country-pair fixed effects, as proposed by Baier and Bergstrand (2007). Next, as suggested by Head and Mayer (forthcoming), a multinomial PML is also applied to solve the zero trade issue and the presence of heteroskedasticity. The main contribution of the paper is twofold. First, this is to the best of our knowledge the first attempt to obtain ex-post unbiased estimates of trade creation and trade diversion effects in ACFTA taking into account the endogeneity bias of an FTA. Second, we will estimate the model not only using aggregated trade data, but also disaggregated data for four different sectors: agricultural goods, manufactured goods, chemical products and machinery products. The reason for doing so is to ascertain whether or not the trade effects in this region differ by commodity