The paper is devoted to evaluation of the econometric method applied as a part of a variance screen in collusion detection procedure. Validation is based on ex-post analysis of Indian cement industry in the 1994 - 2009 time period and comparative study of the obtained results with factual evidences of collusion at that market. The method in question is based on MS(M)AR (p, q) Markov switching model specification. As a result of the research we could identify variability regimes consistent with theoretical motivation of the marker and detect collusion and competition phases partly consistent with historical evidences. However promising, method had some drawbacks applied to high frequency data in the context of variance screen. We proposed some solutions for further research to overcome it.