Bubble Detection in Financial Markets - A Survey of Theoretical Bubble Models and Empirical Bubble Detection Tests

Abstract

Asset price bubbles have fascinated economists for decades. In consequence, the literature on bubbles and their detection is abundant, with many researchers taking very opposite positions on the topic, however. This survey gives a structured overview of the two branches of research that have received the most attention in bubble research. First, we describe the theoretical models that have been developed to model bubble phenomena. These can be divided into rational bubble models and behavioral bubble models. Second, we provide a structured overview of empirical methods for the detection of rational bubbles. We focus in particular on recently developed bubble detection methods, namely recursive unit root tests, fractional integration tests, and regime-switching tests. These tests are predominantly advanced stationarity and cointegration-based tests and as such are not based on the fundamental factors of the assets but rather focus on the time series of the asset prices. As a result, they avoid testing a joint hypothesis of the presence of rational bubbles and the validity of the model used to determine the asset’s value. Furthermore, they are capable of detecting multiple, periodically collapsing bubbles. While a consensus both on the appropriate theoretical bubble model as well as on the most applicable empirical bubble detection test has not been reached, especially empirical research has made significant progress. Different bubble detection tests now increasingly find overlapping evidence of rational bubbles when used to analyze the same time series. Nonetheless, many results are still inconclusive and bubbles remain an interesting avenue for further research.