An Empirical Investigation of Asset Price Bubbles Using The Local Martingale Theory of Bubbles and Multiple-Factor Return Models

Abstract

Asset price bubbles have been ubiquitous throughout history in nearly every global market. Given the numerous negative impacts caused by bubbles bursting, policy-makers are looking for ways to detect bubbles in real-time. In this thesis, asset price bubbles since the Great Recession of 2008-09 are studied using two new bubble detection methodologies advanced by Robert Jarrow, Younes Kchia and Philip Protter. The stochastic process approach is based on the local martingale theory of bubbles while the multiple-factor return model extends the Fama-French and Carhart four-factor model by introducing bubble risk as systematic, following the work of Anderson and Brooks (2014). These methods attempt to break from the traditional tests that are shown to be inconclusive due to the joint hypothesis problem. Motivated by the technology sector driving the recent bull market to historic highs, these methodologies are applied to the FAANG stocks and Microsoft to determine whether any of them have been in a bubble in the past decade. The empirical findings suggest that Netflix stock was consistently in a bubble between 2009 and 2018 while the other five stocks were not. This analysis will hopefully guide investors to better detect asset price bubbles in real-time and understand which stocks are “overvalued” by examining the intrinsic properties of the processes that influence asset prices.