The unemployment rates of then-communist Central European countries were unnaturally low because of manipulated data and latent, "intra-firm" unemployment. As the transition of these states is now more than two decades ago, the question of how the NAIRU evolved since then arises. Because of the "sure" structural break at 1989, the concept of time-varying NAIRU is highly plausible, therefore this paper aims to perform a time-varying NAIRU estimation using different model setups and estimation methods.Our departing model for NAIRU estimation is a simple, backward looking Phillips curve, where inflation is regressed on its own inertia, demand pressures and exogenous supply shocks, this is called the "triangle" model by Gordon (1997). The model can be augmented to have forward looking dynamics by employing inflation expectations, becoming the New Keynesian Phillips Curve framework as price stickiness is accounted for through the forward looking elements. The demand pressures in both versions are described as the unemployment's deviation from its time-varying natural rate, so the NAIRU can be estimated by solving a latent-variable state-space model. The default method of doing so is using the well known Kalman filter (via Maximum Likelihood), but we also employ its lesser known alternative, the flexible least squares (FLS) method. It has been shown that FLS is a restricted version of Kalman filtering, but the in exchange for the restrictions we can drop all the probability assumptions, which can be highly desirable in some settings.In Hungary, the unemployment rate fell sharply during the nineties, but increased again in the last decade. Our expected results are however, that the natural rates in these countries were decreasing during both decades, as the transition economies approached its new steady states.
