Under the model, the returns for Olympic stock increase with the DVWR returns; because the scale of the equation increases with increasing DVWR, the errors are likely also to increase, making the model heteroscedastic. Apart from a loss of efficiency in estimating the coefficients (which is not relevant here since we assumed that the coefficients were estimated without error), the effect is to overstate the precision of the Olympic stock price regression prediction when DVWR is in the high range; as a result, the model is even less powerful in that range than it appears to be. However, since the prediction interval, as calculated, is so wide, the fact that it should in some contexts be even wider is of little importance.