Research
Working Papers
"What is the effect of the corporate marriage of Disney and Pixar on their films’ image quality? " (Job Market Paper, Latest Draft)
This paper estimates the impact of Disney’s acquisition of Pixar on the image quality of Disney’s animated feature films. Image quality is one of the explicit measurements for the product’s key attributes. By improving image quality, Disney reduces the cost of technology that animation makers use. Better image quality, therefore, signifies that another innovation has been created to make technology cheaper and more competitive. Although visual attributes in the animated films are the critical factor for the decision making of the firm’s production, previous literature describes them as unobservable. This paper uniquely adopts a modern image quality assessment technique –Blind/Referencelss Image Spatial Quality Evaluator (BRISQUE)– used in engineering literature to quantify image quality. In this technique, statistical properties are used to extract features of the images, and this information is identified through the algorithm. Those predicted features are used to compute the image quality of animated films. To find the impact on quality improvement following the merger, this paper conducts a causal analysis using the Synthetic Control Method. Still, it is hard to know which variables should be included to find the optimal synthetic controls. In this study, the best set of possible predictors is chosen by applying the out-of-sample (OOS) model selection technique. In the OOS approaches, the pre-treatment period is split into two parts. The first training set is used to build control units among all possible models. The second testing set is then used to evaluate the performance of each model, and the least root mean squared prediction error is selected as the best optimal set of alternative predictors. Our empirical findings from the SCM imply that the merger between Disney and Pixar has improved the image quality of Disney’s animation since the transaction in 2006.
"Model Selection for the Synthetic Controls"
This paper answers the question of how to select the best set of predictors to find synthetic controls. Although the set of predictors is crucial in the Synthetic Control Method, there is no clear guidance on which predictors to be included in the estimation. In the empirical analysis, some literature uses either the mean of the pre-treatment outcome or all the pre-treatment outcome to resemble the outcome in the absence of the treatment. We apply the out-of-sample (OOS) forecasting technique to find the best set of predictors. For OOS, the pre-treatment period is split into two parts: the training set and the testing set. The training set is used to build the synthetic controls for each possible model. The predictive power is computed using the testing set. The smallest root mean squared prediction error is chosen to identify the optimal model. Our analysis is based on the simulation and empirical experiments. Results show noticeable differences in the performance of the model selection technique and the original method. The results from the simulation suggest that the model selection technique systematically yields more accurate estimates, provided that the model has been correctly identified. In particular, empirical results suggest that the selected model from the OOS performs better than the original method. The magnitude of the effect after the treatment shows different results compared to the original method. Researchers should avoid the specification searching through conducting the model selection for the Synthetic Control Method.
Work in Progress
" Model Averaging for Out-of-sample Forecasting Predictive Performance with an application to the box office opening weekend "
"Asymptotic optimality of model selection for the Synthetic Control Method"
Research Awards
Sep 2020, Bacon Family Economics Scholarship, University of Colorado-Boulder
Feb 2013, Best Paper in Graduate student, Economics Joint Conference, Korean Economic Association
Publications
Choi YE, Ha HY, Lim JY, Lee EK, “Revisit the effect of the prenatal medical care use on the birth outcome of newborn baby,” Hitotsubashi Journal of Economics, Vol.56, No.2, 2015.