Premium
The Stochastic Structure of the Time–Varying Beta: Evidence from UK Companies
Author(s) -
Choudhry Taufiq
Publication year - 2002
Publication title -
the manchester school
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.361
H-Index - 42
eISSN - 1467-9957
pISSN - 1463-6786
DOI - 10.1111/1467-9957.00324
Subject(s) - mean reversion , bivariate analysis , unit root , econometrics , long memory , economics , beta (programming language) , stock (firearms) , structural break , autoregressive conditional heteroskedasticity , mathematics , statistics , computer science , volatility (finance) , engineering , mechanical engineering , programming language
The stochastic structure of time–varying betas from 15 companies in the UK is investigated. Time–varying betas are estimated by means of the bivariate MA–GARCH model. The stochastic structure is investigated by means of two fractional integration tests, the Geweke and Porter–Hudak and the Robinson tests, and a structural–break–oriented unit root test. Results show that time–varying betas are mean–reverting but only few have a long memory and thus are mean–reverting at a slow rate. This result is further backed by the structural break unit root test. These results contradict earlier studies, which fail to find a stationary beta. Stationary betas may imply that stock returns may be forecast in the long run.