Detecting Accounting Fraud with Various Models: the Evidence from the USA

Diploma work studies the relation between financial values, ratios of a company and probability of earnings manipulation and fraud in the process of composition of financial statements. The empirical data for the research is extracted from releases of U.S. Securities and Exchange Commission (SEC), that is responsible for enforcing the federal securities laws, rules, including regulations in accounting sphere. Fraud detection technics are implemented for 299 non-financial companies under SEC regulation. The time boarders for the investigation are 1996-2015 years. There are several regression models for estimation the probability of accounting fraud. In the dissertation Messod D. Beneish, Maria L. Roxas, Obeua S. Persons and Patricia M. Dechow at al models were tested and all of them except Obeua S. Person showed enough good results. The author investigates main characteristics of firms from the sample, compare results of four models and make suggestion for using the models in order to analyze financial statements with more accuracy.