Development of a model for predicting money laundering rate

Abstract

The article suggests model for predicting the level of money laundering on the basis of data from the Ministry of Internal Affairs of the Russian Federation on the state of economic crime in Russia since the beginning of 2011. Using a seasonally integrated autoregressive moving average (SARIMA) model, it compares different regression models for the research tasks (linear regression, logistic regression, autoregressive and SARIMA). The necessity of taking into account seasonal regularities in the structure of money laundering was underlined, and the SARIMA model with the lowest deviations from the actual values was chosen. The necessity of taking into account seasonal regularities in the structure of money laundering was underlined, and the SARIMA model with the lowest deviations from the actual values was chosen. The article presents the results of data analysis using the method of least squares, calculating the mean squared error (MSE). High accuracy of short-term forecasts was noted: the deviation from the actual number of cases is about three cases (with the average number of cases being 68 over the last 10 years). The forecasting model can be recommended for implementation in the analytical complexes of financial monitoring and supervisory authorities.

Keywords

• financial monitoring
• anti-money laundering
• economic crime
• crime forecasting
• statistical analysis