ABSTRACT
We use a three-regime threshold regression model to assess the ability of the New Keynesian Wage Phillips Curve (NKWPC) to describe wage inflation in the US over the 1965–2018 period. Nonlinearity is clearly supported by the data and it easily resists an endogeneity correction. However, this correction exposes more clearly the shortcomings of the NKWPC as a successful description of wage dynamics in the extreme phases of the business cycles, when unemployment is either low or high. In both cases, it becomes completely flat.
Acknowledgments
We gratefully acknowledge the comments of an anonymous referee. Obviously, all remaining errors and omissions are our own.
Disclosure statement
No potential conflict of interest was reported by the authors.
Notes
1 We have estimated the threshold parameters using the sequential ‘one-at-a-time’ method proposed in Hansen (Citation1999) inspired in the change point estimation literature and analysed in Gonzalo and Pitarakis (Citation2002).
2 This benchmark bias coincides with that of the OLS estimator when the errors are conditionally homoskedastic and serially uncorrelated. However, unlike the Stock and Yogo test which is appropriate in that case, the Olea and Pflueger test is robust to violations of both these hypotheses.