This paper econometrically evaluates if there actually occurred a collusion in the negotiable certificates of deposit (CD) market during the period of the Korea Fair Trade Commission (KFTC)’s investigation. We propose a general mixture regression model to discriminate the collusion period from the competitive period. We apply our method to Korean CD market data from January 1, 2009 to May 23, 2019, and forecast the probability of collusion for each day. We find only a small portion - 163 days out of 2,579 days - of the whole sample is discriminated as a possible collusion. We also find that the banks did not issue the CDs on almost all dates discriminated as colluded in our empirical results. Our ﬁndings imply a strong possibility that the stickiness of the CD rates was induced by the depressed CDs market conditions rather than a collusion.