Abstract. We construct two new tests of calendar effects, apply them on 12 stock indices during 1996-2016, and compare the results with that using Hui and Chan (2016)'s method. The results show that the January and Halloween effects are significant for the four western generalized equity indices for small moving-window sizes. Furthermore, the securitized real estate indices show a greater difference in the overall calendar effect between the three methods than the general equity indices do. This study has an implication that a certain sector of the market is riskier than the whole market.