The formula for the area moment of inertia for a rectangular cross-section is b * h^3 / 12. But it looks like when computing the value for angle-iron in this table:
http://www.engineersedge.com/standard_material/Steel_angle_properties.htm
... they have assumed that the piece is bent into a flat plate, so that both sides are aligned to resist the bending force. Of course, this is completely unrealistic. Only one face is typically resisting the bending force (the vertical face in most applications) and the other face simple keeps the vertical face from buckling. As I missing something here ?