This is a terrible paper and I suspect it's going to fail to be reproduced. That won't stop it from being quoted heavily. > In our experiment, the sample of fi

This is a terrible paper and I suspect it's going to fail to be reproduced. That won't stop it from being quoted heavily. > In our experiment, the sample of firms under enhanced scrutiny is the set of Arthur Andersen (AA) clients after the AA demise following the Enron scandal. Their initial data sets (2) indicate that detected fraud is in the realm of 2-5%, and they then hypothesis from Enron clients that the actual fraud rate is astronomically high, by abusing a statistical model. They are using an outlier to form statistical representations on the whole.