The pair approach applied to kinetics in restricted geometries: strengths and weaknesses of the method