We investigate clustering properties of quasars using a new version of our semi-analytic model of galaxy and quasar formation with state-of-the-art cosmological N-body simulations (Ishiyama et al. 2015; Oogi et al. 2015). We assume that a major merger of galaxies triggers quasar activity. We find that the quasar bias does not depend significantly on the quasar luminosity, similar to observed trends. This result reflects the fact that quasars with a fixed luminosity have various Eddington ratios and thus have various host halo masses that primarily determine the quasar bias. The quasar bias increases with redshift, which is in qualitative agreement with observations. Our bias value is lower than the observed values at high redshifts, implying that we need some mechanisms that make quasars inactive in low-mass haloes and/or that make them more active in high-mass haloes.