A carregar...
Publicação
Predicting extreme equity returns with binary response models
| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 1.17 MB | Adobe PDF |
Autores
Orientador(es)
Resumo(s)
In this paper a parsimonious methodology for estimating the probability of observing an extreme negative movement in monthly stock returns is proposed. It uses Extreme Value Theory to define an extreme return and exploits dynamic probit models based on (Kauppi and Saikkonen 2008) which are expected to improve the performance of the regression. The results are convincing, as the dynamic feature indeed enhances the models’ performance. Moreover, successive extreme returns are observed, confirming the fact of extremal clustering in the tails of the distribution.