| Economic risk factors and their impact are not constant over time. For example, the liquidity of financial markets, the risk of bankruptcy, and the effectiveness of economic policies all vary significantly over time and over different states of the economy. Not accounting for potential time-variation may lead to an incorrect understanding of economic mechanisms and to flawed policy advice. Koopman, Lucas, Schwaab (2009b) illustrate the importance of this for the 2008 financial crisis. They show that models with time-varying bankruptcy probabilities would have predicted the large rise in defaults substantially better than models with constant probabilities or models with external conditioning variables. The incorporation of time-varying parameters in empirical economic models, however, is not widespread. This is mainly due to the complexity of the corresponding econometric techniques. This complex approach used in the current literature obstructs the broader adoption of these models and impedes our further understanding of the magnitude and origin of time-variation in economic relationships. The proposed research agenda develops a new methodological framework for time-varying parameter models that is (i) as flexible and general as the current (econometrically complex) methodology, (ii) statistically rigorous, yet (iii) sufficiently tractable to be actually used by a wide(r) community of applied researchers. Such a new model class has a wide potential research impact. The idea builds on recent findings by Creal, Koopman, Lucas (2008). This proposal develops the statistical foundations of the new approach. The methodology is applied to investigate time-variation in a number of complex economic settings involving the stability of financial markets and the assessment of risk. Further conceptual of generalizations of the methodology are also explored to open up new areas of research. In this way, the proposed research agenda contributes to pushing the potential instability of economic models more to the center of the policy debate. |
