Wage indexation and the monetary policy regime

New Keynesian dynamic stochastic general equilibrium (DSGE) models typically assume a fixed, and policy invariant, degree of wage indexation to past inflation (Christiano, Eichenbaum and Evans 2005, Smets and Wouters 2007). This assumption has, however, been rejected by empirical evidence for the United States (Holland 1986; Hofmann, Peersman and Straub 2012).

This paper examines whether wage indexation varies across monetary policy regimes. We estimate the reduced-form empirical New Keynesian wage Phillips curve of Galí (2011) on a panel dataset covering 24 OECD countries between 1960Q1 and 2011Q4, and allow the degree of wage indexation to vary according to the monetary policy regime. Four regime types are considered: those without a target, and regimes with an inflation, money growth and/or exchange rate target. To control for labour market institutions, we include the degree of labour market coordination and union density as explanatory variables.

We find that wage indexation varies across monetary policy regimes, i.e. wage indexation  in an inflation  target regime is both  statistically insignificant  and  significantly  lower than  indexation  in a regime without  a target. In contrast, regimes with money and exchange rate targets have statistically significant indexation levels, and these levels are not significantly different from a regime without any formal quantitative target.  

Overall, our results question the structural nature of hard-wiring a fixed degree of wage indexation in standard DSGE models. They further show that the monetary policy dependence of price indexation found by Benati (2008) can be extended to wage indexation. From a policy standpoint, our findings suggest that counterfactual policy simulations and the analysis of optimal monetary policy based on modern macroeconomic models are potentially misleading.