Forecasting Risk: Realized Quantile Approach

We are looking at risk all wrong, according to Fabian Rinnen. “Historically, economic time series models have been unable to predict risk satisfactorily, causing worldwide problems—notably during the recent financial crisis,” he says. “I am combining these models with realised quantiles to yield more meaningful forecasts.” In addition to assessing financial risk more accurately, Rinnen’s model may be able to detect macroeconomic risks such as unemployment, exchange rate and inflation.
This information would help institutions like central banks plan for the future more effectively. In the context of climate change, his methods can predict short- and medium-term temperature increases. “My goal is not only to improve financial risk assessment, but also to give policymakers risk forecasting tools to minimise public exposure,” he concludes.
My research focuses on models risk using the statistical foundations established since the 1970s in econometric time series analysis combined with the advantages that quantiles have in this context. In the past the econometric methods I apply have been used to model movements in the mean of a variable of interest. While forecasts obtained from these regressions are meaningful and important in practice, the need for accounting for risk becomes evident when recalling the crisis that was caused by neglecting risk in the financial markets in the recent past.