We are using an Sawtooth ACBC module + HB estimation to built an excel based market simulator. We are using the logit rule to transform utilities in market shares (choice shares). Given that ACBC part-worths are scaled quite extreme we find that the logit scale factor (alpha) cannot be scaled too large (we want to adjust the logit rule to be almost maximum utility in order to cope with the IAA problem). Excel seems to be unable to work with scale factors higher than 1,5.
Do we have to admit that an ACBC + logit is probably more exposed to IAA issues than say an CBC + logit?