Diagnosing Stock Market Bubbles with Advanced ADF Tests

Recent work on econometric methods for identifying bubbles in asset prices on the stock market have demonstrated the effectiveness of recursive procedures, which feature the ability to identify and date bubbles in real time. For this reason, recursive procedures are potentially useful as warning signals in passive investing, as well as as a way to monitor and prevent stock market collapses for central banks and fiscal regulators. Over the past 10 years, there has been a major evolution in the methods of diagnosing bubbles based on recursive procedures. This work is aimed at empirical verification of the latest developments in this methodology, namely the GSADF test, which is a generalized recursive augmented Dickey-Fuller test using a sliding window, which allows it, in theory, to identify with high accuracy all moments of inefficiency, i.e. bubbles, in the price of any asset on any time intervals of data. The simulation shows that the GSADF test does significantly improve the discriminative power of the standard Dickey-Fuller test. Based on real data on two stock indices – the Nikkei 225 and the NASDAQ, it was possible to confirm and accurately date the well–known episodes of the stock market collapse-the Japanese economic bubble of 1986-1991 and the dot-com bubble of 1995-2001 in the United States. And the results of the GSADF test on current market prices have demonstrated the presence of a bubble in the NASDAQ index over the past 7 months. Based on this, it can be argued that the GSADF recursive test is highly empirically effective for identifying financial bubbles in the early and late stages, and, of course, after their collapse.