Stability Watch Views

"Stability Watch" is a report from Rmetrics offering views on the stability of historical economic indicators and financial time series. The report helps everybody to watch non-stationarities, instabilities and stressed scenarios in historical time series.

The reports may also help, even if not guaranteed, to detect during early stages indications that severe changes and instabilities may occur in the near future. The aspects which are considered by the Stability Watch report are (i) Value, (ii) Volatility, (iii) Multiresolution, and (iv) Robustness or Stability.

We apply modern statistical approaches to the analysis of the time series. Time Series analysis with fat tailed innovations, robust modeling with GARCH models, phase space embedding approaches from chaos theory for time series pattern analysis, identification and extraction of unstable pattern which can be used in stress testing. Other approaches are concerned with time/frequency multi-resolution analysis with wavelets, robust covariance estimation including minimum covariance determinant, shrinkage und random matrix theory approaches, and spline smoothed divergence indicators, amongst others.

The Value View shows the time series values on a linear or logarithmic scale, gives the probabilities of a structural change or break for each data point using a Bayesian Monte Carlo approach, identifies likely outlying points by phase space embedding and robust covariance analysis, and displays ranges of World (US) recessions.

The Volatility View shows the time series volatilities, identifies stressed or unstable reward patterns by phase space embedding and robust covariance analysis, and displays also the ranges of World (US) recessions. Volatility estimates are displayed and outlying points are identified by robust volatility estimation due to leveraged interquartile range GARCH models with lambda distributed innovations

The Multiresolution View offfers a time/frequency wavelet analysis allowing to consider and interpret the data on different resolutions and scales.

The Stability View uses a phase space embedding approach of data sequences. Robust covariance analysis of these sequences allows the detection and identification of outlying values in mean and variance. The results are represented as divergence plots of the probabilities by smoothed splines.

More Views? Read more ...

Any Questions? Do not hesitate to contact us.

(c) Rmetrics Association, 2010