This dissertation consists of three chapters that attempt to understand how forward-looking firms make costly decisions and their subsequent implications for the industry's performance. The first chapter "Sunk Cost and Entrant's Choice of Capacity" develops a dynamic model of strategic entry to study how the structure of sunk entry costs influences the entrant's scale decision and the long-run market outcomes. Contrary to the typical dynamic entry model, my model features that the structure of sunk costs shapes not the number of competitors but also the industry's scale distribution. I empirically assess this channel using a case study of a land-use deregulatory reform in the South Korean cinema chain industry. The deregulation is estimated to act as an entry subsidy, particularly appealing to larger-scale theaters. However, the industry suffers a 5.6 percent loss of discounted net profits due to intensified competition and increased expenses on fixed operating costs. The resulting implicit cost of the regulatory action is not uncovered by the typical model, as it obscures the shift in the distribution toward a larger scale. The second chapter "The Differential Effect of Exporting on Input Productivities" examines how the firm's export decision shapes its input allocation in the long run, focusing on the non-neutral technological changes. I particularly study whether entering the export market results in differential increases in input productivities at the firm level (non-neutral change). I develop a model that distinguishes between firm-level skilled and unskilled labor-augmenting productivities and material input prices. Applying the model to data on the Colombian apparel manufacturers, I find that exporting raises the skilled labor-augmenting productivity by a 7.2-percentage point more than the unskilled counterpart. In a counterfactual simulation in which exporting raises the two productivities equally, the mean differences in skilled-to-unskilled employee ratios between exporters and non-exporters are 50 percent smaller than the data counterparts. The result suggests that non-neutral productivity gain from trade is central in shaping the input allocation differences between exporters and non-exporters. The third chapter "Trade Dynamics of Heterogeneous Producers under Trade Cost Complementarity" estimates a dynamic model of the firm's joint export and import decision process. In the model, participating in trade improves within-period profits and future productivity. In addition, doing one trade activity facilitates the other by reducing the associated fixed/sunk costs. Employing a Bayesian MCMC estimator, I fit the model to Colombian chemical plant panel data from 1981 to 1985. Two findings stand out: (i) importing increases future productivity significantly while exporting does not. (ii) importing facilitates exporting by lowering the sunk costs of entering the export market, while exporting facilitates importing by decreasing the fixed continuation costs of importing. A counterfactual simulation shows that subsidizing the fixed costs of importing is the most effective among trade cost subsidy schemes in improving the average productivity and firm value.