High-dimensional Index Volatility Models via Stein's Identity

  • 2019-03-14 16:40:28
  • Sen Na, Mladen Kolar
  • 0

Abstract

We study estimation of the parametric components of single and multiple indexvolatility models. Using the first- and second-order Stein's identity, wedevelop methods that are applicable for estimation of the variance index in ahigh-dimensional setting requiring finite moment condition, which allows forheavy-tailed data. Our approach complements the existing literature in alow-dimensional setting, while relaxing the conditions on estimation, andprovides a novel approach in a high-dimensional setting. We prove that thestatistical rate of convergence of our variance index estimators consists of aparametric rate and a nonparametric rate, where the latter appears from theestimation of the mean link function. However, under standard assumptions, theparametric rate dominates the rate of convergence and our results match theminimax optimal rate for the mean index estimation. Simulation resultsillustrate finite sample properties of our methodology and back our theoreticalconclusions.

 

Introduction (beta)

None

 

Conclusion (beta)

None