The relevance of the research topic lies in the fact that, on the one hand, stakeholders need reliable information contained in accounting financial statements to make managerial decisions. Models for detecting the financial statement manipulation can minimize the risks of making wrong decisions based on the data of these reports. On the other hand, the application of such models are excellent tools within the framework of anti-corruption practice when presenting the performance indicators of an economic entity in the ESG reporting. The aim of the research is to analyze the applicability of the existing foreign models created to detect financial statement manipulation in the Russian economy, and to develop the main directions for their improvement. The objectives of the study are to identify the shortcomings of existing reduced regression models from the point of view of Russian business, as well as to verify their effectiveness on the example of the largest agricultural enterprises with annual revenues of more than 2 billion rubles engaged in the breeding of farm animals and poultry. The result of the research is the assessment of falsification of financial statements on the example of the largest enterprises of the agro-industrial sector using three reduced regression models adapted by Russian scientists. As a result, the authors concluded that not all the adapted models used in this paper provide an accurate prediction of falsification of companies' reporting. Also, taking into account the changes in the Russian legislation in the field of non-current assets accounting, the applicability of the reduced models becomes controversial. In addition, the main directions for improving the methodology for assessing the financial statement manipulation are: the development of models that take into account the industry specifics; the possibility of using non-financial indicators that directly correlate with changes in financial indicators that may be falsified.