EcoSta 2017: Start Registration
View Submission - EcoSta2017
A0620
Title: Testing stress Authors:  Manfred Kremer - European Central Bank (Germany)
Sulkhan Chavleishvili - European Central Bank (Germany) [presenting]
Abstract: Financial stress can impair the regular functioning of certain financial market segments or of the financial system as a whole, often associated with adverse macroeconomic consequences. Although the concept of financial stress is central for macroeconomic policy interventions, it still remains unclear how it can be rigorously measured. A novel statistical framework is introduced for defining and testing financial stress. The framework can be flexibly applied to an arbitrary number of time series measuring stress in different economic units (like corporations) or different financial market segments. Our semiparametric framework defines financial stress as the event that the relative rank of a realized time-series observation (or the vector of ranks in the general multivariate setup) exceeds a particular non-stress level. The estimation of ranks is cast in a simple linear regression framework, and the asymptotic properties of the estimator are used to develop a test statistic for the null hypothesis of no-stress. We interpret this test statistic as a meaningful and statistically well-founded financial stress index. We apply our statistical framework to data for the US financial system from 1973 to 2016 and estimate, among other things, the effects which shocks in our financial stress index may have on different measures of real economic activity.