Three Essays on Business Analytics: time-series causality, panel data analysis, and design of experiments

Abstract

This dissertation includes three empirical studies focusing on the applications of Business Analytics in the context of financial markets and the online advertisement industry. The first essay examines the existence of a causal relationship between various prediction markets and global financial markets time series. This essay uses over 27 different countries and regions' financial market data (Dow Jones Global Indexes) and uses the Toda-Yamamoto causality test. Preliminary results indicate that prediction markets may be used to predict some global financial markets. From a managerial perspective, our result quantifies the connection between some countries' economy, as measured by a financial index, and the political events captured by the prediction markets we consider.

The next two essays focus on the online advertising industry's business policies. The second essay uses a panel data analysis to compare the effect of two different IP protection policies, Monetize and Track, on YouTube music channels' viewership. This research provides insights for content owners on how IP protection policies on user-generated contents (UGCs) affect their YouTube channel viewership, and on how UGCs impact their ability to maximize profit. The third and final essay proposes a new data-driven statistical framework (DDSF) to determine what ad formats maximize a company's revenue generated from online advertising. The developed DDSF is applied in a real-world experiment. The experiment results help our YouTube industry partner determine what ad formats to run on their videos in order to trade off two key performance indicators of interest.