Grant: Discovery Grants Program - Individual

The objectives of the proposed research is to develop new inference methods for the Compound Autoregressive (Car) models, and to explore selected applications of L-moment-based estimation methods in time series. The Car model is a conditional Laplace transform model for nonlinear times series. It provides a convenient framework for inference on processes with non-tractable conditional density functions, that are unknown or have too complicated analytical expressions. In finance and economics, the conditional Laplace transforms often appear as functional parameters of interest, especially in the term structure of interest models, asset pricing models and optimal portfolio management models. For these, and other potential future applications of the Car model, the availability of reliable inference methods is very important. The proposed research will address this issue, and provide simple estimation methods and testing procedures that can be easily implemented by researchers and practitioners. The L-moments are widely used in statistics for estimation of models for cross-sectional data in hydrology, meteorology and other fields. The proposed research will investigate a time series application of L-moments. In particular, it will examine the properties of a new method-of-L-moment estimator for dynamic quantile models. In the framework of cross-sectional data, the project will examine the use of L-moments for the measurement of fund performance. The proposed L-performance measure will improve upon the Sharpe performance measure with respect to tail sensitivity and robustness.