Stochastic modeling of security returns: Evidence from the Helsinki Stock Exchange

This paper documents the presence of linear and nonlinear dependencies in Finnish stock returns and models these dependencies using autoregressive conditional heteroscedastic methods. Three conditional distributions (normal, Student-t, and the power exponential) are explored. The statistical estimates and the corresponding diagnostic tests indicate that a GARCH (1,1) model with a power exponential condition distribution, which is characterized by an autoregressive mean, represents the data better than any of the other models examined.