This paper studies the timing of trade liberalization in a small open economy, in which capital is sector-specific in the short run and is costly to move in the long run. In our two-period model with endogenous adjustment cost, if the intersectoral capital is specific initially and is mobile in second period, 'gradualism' will be better than 'shock'. The existence of adjustment cost may affect the speed of liberalization, but it does not imply a case for gradualism. When the cost of capital adjustment is infinite, a 'big bang' will be the welfare-maximizing choice. Furthermore, even if the reallocation of capital is costless, gradual liberalization will also be the optimal trade policy. Numerical simulations are also provided.