On SO(2,1) symmetry in general relativity