Nonlinearity and Time-varying Dependence in Money Markets Zeynep Senyuz (with Emre Yoldas, Bernd Schlusche and Selva Demiralp) Abstract: We develop a comprehensive empirical framework to assess the source and severity of stress in short-term funding markets and to shed light onto their time-varying dynamics. Our model accounts for nonlinearity in the mean, the volatility, and the correlation of money market rates. We estimate threshold vector error correction models for each pair of rates that forms one of the commonly monitored spreads and identify varying levels of funding stress in different money market segments. We then model time-varying volatilities and correlations of the rates using dynamic conditional correlation models. The model estimates are updated in real-time as more data become available to market participants and policy makers. Our results show that spread thresholds that determine regimes have become significant in key money market segments at the onset of the crisis. Dislocations led to lower rate correlations in the aftermath of the crisis, although such correlations are still sizable.