Estimating the Hungarian New-Keynesian Phillips curve

This paper estimates the hybrid New-Keynesian Phillips curve (NKPC) for Hungary with different techniques. Because of weak instruments, single-equation GMM estimations yield very unreliable results. More robust methods show that basically no statistical inference is possible as to the true specification of the inflation dynamics. However, a more efficient simultaneous-equations method, the full information maximum likelihood (FIML) estimator provides identified parameters. Furthermore, coefficient estimates on the driving variable are positive and significant for the first time, lending much-needed empirical support in favour of the New-Keynesian model. Inflation appears to be determined to an equal extent by past inflation and forward-looking expectations. Structural analysis yields realistic estimates for the frequency of price adjustments and suggests that the dominant price setting behaviour is backward-looking