Does a Horizontal Section Exist in Our Phillips 'Curve'? Peru 2005-2017.

The paper investigates whether the existence of a relatively horizontal section in the Phillips curve can explain a possible flattening linear aggregate version for the case of Peru. The paper starts with the linear semi-structural model for regional Phillips 'curves' as proposed by Fitzgerald & Nicolini (2014), where the identification of the output-gap parameter in the aggregate 'curve' is feasible in spite of monetary policy's taking compensatory actions in an optimal way. The paper then proposes a nonlinear version of this regional model that allows a continuous change of the inflation-output gap relationship depending upon the variation range of the output-gap, i.e., the regional Phillips curve may encompasses a relatively horizontal section. The empirical evidence (i) disproves the hypothesis that a flattening Phillips relationship exists (equivalently, the hypothesis that the output-gap parameter is zero) in the linear models as well as in the nonlinear model that assumes regional heteroskedasticity (e.i., a flattening Phillips relationship does not exist), and (ii) only the nonlinear model that assumes regional homocedasticity supports the existence of a relatively horizontal section (e.i., a flattening Phillips relationship exists).