The present paper suggests a new way to carry out IV estimation with many instruments. Our suggestion is to cross-sectionally average the instruments and use these averages as instruments. We provide a theoretical and Monte Carlo analysis of this approach.

We propose a way to test the New Keynesian Phillips Curve without having to estimate its structural parameters. We find that it does not exist as a combination of the structural parameters consistent with the US data. This might be due to the failure of the joint hypothesis of rational expectations.

We extend GLS detrending procedure to testing for unit roots against STAR and SETAR alternatives. Monte Carlo simulations and applications to DM/Yen real exchange rates demonstrate that GLS detrending-based nonlinear unit root tests are more powerful than OLS detrending-based counterparts.