Detection Of The Earnings Manipulation Among Chinese Public Companies And Its Application

This paper firstly gives a brief review of the researches on the earnings manipulationfrom the public firms and the application researches on the fuzzy set theory used inthe economic field. And then builds a new model to detecting the earningsmanipulation based on the combination of FST (Fuzzy Set Theory) andMulti-Objective Linear Programming. It is shows that the manipulator detecting ratiocould arrive at78.4%for the in-sample test and76.9%for the out-of-sample test.Furthermore, we introduce this model into the selection of the profitable mainlandpublic pharmaceuticals companies, to show the practical benefit of using this modelsfor the users. In the empirical research,3models based on the clustering analysis withdifferent indicators have been taken into consider while they used in calculating theROA (Return On Assets). The indicators or the information set of model I is onlybased on the financial accounting data, the indicators of model II are the indicatorsfrom model I add the efficiency indicator which could decrease the bias of ROAcaused by the different business industrial constitution for each firms. At last theindicators of model III come from model II and the earnings manipulation detectingmodel the paper aforementioned, which could tests whether the ROA showed infinancial statement is real or credible. As a result, the companies selected by theclustering model III which add the efficiency indicator and the ROA reality indicatorwould have a higher sustainable profitability in average. And moreover in thisempirical research, for the companies detected by the detection model this papersuggested,2/3of them get into profitability declining in the following year.