This paper evaluates the empirical validity of the New Keynesian Phillips Curve (NKPC) model of rational expectations. We employ an instrumental variable (IV) projection method to approximate inflation expectations, and show that the inference based on this approach can differ significantly from the one based on rational expectations. More importantly, using an IV test for serial correlation in the GMM context, we find that the error term in the stylized NKPC model is significantly serially correlated. To compensate for the serial correlation problem, we propose an extended framework which can be easily rationalized in terms of sticky price setting of backward-looking firms. Empirical results show that further lags of inflation are needed in the hybrid specification of the NKPC in order to rule out serial correlation in the Euler equation.