Stock returns show dangerous tail-dependence, challenging traditional financial assumptions.
The joint distribution of stock return-pairs shows strong connections in the lower tail and weak connections in the upper tail. This means that stock returns are not normally distributed. A test was conducted using a bivariate GARCH model on European stock indices, revealing that the assumption of normal or student-t dependence is not supported. Instead, an asymmetrically tail-dependent distribution was found to be more accurate.