Essays on Empirical and Computational Analysis of Limited Attention

Abstract

This dissertation consists of three chapters on the limited attention and incomplete information. The first and the second chapter deal with the random consideration model empirically and computationally, respectively. The third chapter proposes a hypothesis testing procedure in an environment where an econometrician is not provided with sufficient information.<br/> The first chapter examines an investor's attention allocation choice when information includes both depth and scope. By explicitly introducing these two aspects into the model, anomalous pricing patterns in the Korean racetrack betting market are explained. The results show that two types of attention allocation strategies coexist in the Korean racetrack betting market: generalists, who seek the scope of information, and specialists, who seek the depth of information. Here, I show that the different choice of attention allocation strategies can be explained under a standard rational framework.<br/> I propose a fast and reliable computation algorithm for the random consideration model in the second chapter. I develop a simple representation of the model to reduce the computational burden. I show that the new algorithm outperforms existing methods in terms of accuracy and efficiency.<br/> The last chapter deals with a Wald test in an environment where the design matrix is rank deficient or near singular. I propose a robust Wald test procedure that is attainable and that has a smaller type-I error than that of the classical Wald test procedure. I show that the robust Wald test procedure alleviates the phenomenon that a null hypothesis is overly rejected if the econometric model is high dimensional.<br/>