Time Scale Macroeconometrics

The central focus of this dissertation is to develop robust econometric methods to identify and learn about the effects of time horizon on economic relationships. The first chapter provides a broad overview of some of the disparate methods used in the literature to identify such effects and, where appropriate, attempts to interpret them in a unified filtering framework. The second chapter briefly summarizes wavelet methods -- a central tool used in the framework I develop throughout the rest of the dissertation. In chapter two I also establish some useful asymptotic properties of wavelet transforms. The third chapter introduces a multiresolution regression (MRR) model based on nonparametric wavelet methods. I derive statistical properties and asymptotic behavior of the MRR estimator -- establishing that the MRR is robust to a broad class of scale misspecification. I conclude the third chapter by using a simple MRR model to demonstrate that common assumptions of high frequency noise in portfolio return models are inappropriate. The final chapter expands on the MRR framework by introducing dynamic MRR models in the context of cross sectional asset pricing. I find considerable evidence for scale structure in the cross section of portfolio returns and show that a single financial variable can aggregate multiple sources of risk over different horizons, yet the market prices these risks associated with different time horizons separately. By directly incorporating scale structure into asset pricing models, asset pricing performance dramatically improves while offering inference on the time horizon dynamics of risk channels.