Many federal tax systems use formula apportionment to assign the taxable profit of large businesses to the federal subunits where typically the subunits’ own tax rates are applied. The formulas (including the one recently proposed by the EU Commission as well as the one agreed upon by the OECD/G20 Inclusive Framework) are remarkably similar, yet little is known how to rationalize them, i.e. from which normative criteria they are derived. To fill this gap, I take an axiomatic approach and derive a generalized system of formula apportionment from three criteria: fixed coverage, positive responsiveness and external independence. I demonstrate that any formula apportionment system that satisfies the three criteria suffers from the same distortion (unless all local tax rates are identical). The generalized system comprises existing real-world systems (as observed in federations like the US, Canada and Germany) as special cases, but allows for a degree of flexibility that has the (so far unrealized) potential for surplus increasing reforms.
