Metodologias de previsão de risco do desfazer de bolhas financeiras

: análise empírica

Abstract

In this thesis we discuss the definition of bubble, the distinctive marks of several episodes where they have occurred, and the econometric tests for bubbles. We innovate by using the Phillips et al. (2015) test for multiple bubbles at unknown dates, which is dominant over others. We also innovate in applying the test to the bubble hypothesis in the SCDS market in the Euro periphery, during the crisis. The analysis suggests that the new method is capable of picking up multiple bubbles, whereas the prior SADF test wasn’t. It also confirms that the new test is capable of detecting bubbles while still in formation, without the burst period being in the sample. Near real time detection of bubbles is now available for regulators, allowing to prevent the negative results of the burst. The leveraged SCDS market has had episodes of excess reaction. We have now, established rigorously that SCDS spreads may exhibit explosive behaviour, both in the upward phase, and in the burst, as in the Cyprus example. The bubble periods identified for the markets chosen match relevant events. We do not argue about a rationale for bubbles. Rather we claim to have shown that they exist. We have also concluded that regulation choices are sometimes inadequate. The naked SCDS ban in the EU did not prevent subsequent bubbles, nor did it mitigate existing ones. Banking regulation in the EU desserves further care, as the Cyprus bail in has prolonged the bubble in the country’s SCDS spreads.

Date of Award	14 Jul 2015
Original language	Portuguese
Awarding Institution	Universidade Católica Portuguesa
Supervisor	Carlos Manuel Ferreira dos Santos (Supervisor)