Comparative Analysis of Pure Temperature and Pure Index based

Tailed Portfolio Risk Capital Estimates

Research

This paper administers the volatility estimates in case of portfolio and risk exposures for temperature (climatic) data with index movements on monthly basis. The portfolio risks and risk capital assessments are made with the help of more popular method adopted by corporate called VaR at both individual and portfolio levels. The selection of weather was kept for Delhi, Chennai and Mumbai for simplest reason that the present research is aimed for weather-sensitive industries and their businesses which are affected by weather movements. Also, both Equally weighted moving averages (EQMA) and Exponentially weighted moving average (EWMA) methods are adopted to draw volatility estimates and then a comparison is made in between the performance of EQMA and EWMA methods with respect to temperature and index data series on one side, and the performance of individual and portfolios risk performance between temperature and index series on the other. This gives emphasis on the second objective of the paper to analyze the possible hedging opportunities among temperature and indexes, and between temperatures across cities.

One of the foremost contribution of Modern Portfolio Management is the advent of stochastic modeling techniques including Value-at-risk estimates which in principal explains what is the possible loss within a given investment horizon and the same technique is further used extensively in Portfolio risk assessment and management. With this aspect in mind, the present paper will provide a detail framework on two important methods of time-series modeling viz. Equally weighted moving average (EQMA) and Exponentially weighted moving average (EWMA) models for assessment of Pure temperature series and Pure index series and to check whether at the portfolio risk and portfolio risk capital (based on Dirty backtesting-pointbacktest method) provide some clue about the performance of these two asset series and essentially can a long-term hedging possibility arise between the two. For this, a more robust Tailed based approach, underlining parametric and non-parametric estimates will be taken into account.

Introduction

EMWA, EWMA, Confidence intervals, Tailed measures, Non-normality, Risk Capital

Keywords

Abstract

by Rohit Malhotra

Literature Survey

According to Clements, Doolan, Hurn and Becker (2012) while selecting the forecasting model it is forecast precision or the value they generate which primarily remain the two soul objectives of the model. A correct forecast will usually provide a portfolio variance which doesn't create the problem but minimizes it as a solution. Further Clements et.al (2012) in his research work used an approach called EXP model where the decay rate is calculated by using the maximum likelihood function. Under multivariate GARCH family models, use of Constant conditional correlation (CCC) by Bollerslev and Asymmetric Dynamic conditional correlation (ADCC) were used. Later in the paper, the use of Orthogonal GARCH (OGARCH) was utilized. The concluding remarks of the paper says that loss functions with maximum likelihood performed better also as Portfolio variances which may be effective in one period and turns ineffective in the next period and so on.

Pesaran, Schleicher and Zaffaroni (2008) used the mean-variance optimization problem using the VaR constraint, under which conditional volatilities which are essentially non-Gaussian are used. The models used were conditional constant correlation (CCC) by Bollerslev (1990), the orthogonal GARCH model by Alexander (2001) and Dynamic

FIIB Business Review. Volume 1, Special Issue, October - December 2012