| Pub. Date | : October, 2021 |
|---|---|
| Product Name | : The IUP Journal of Accounting Research and Audit Practices |
| Product Type | : Article |
| Product Code | : IJARAP251021 |
| Author Name | :Rama Krishna Yelamanchili* and Sager Reddy Adavelli** |
| Availability | : YES |
| Subject/Domain | : Finance |
| Download Format | : PDF Format |
| No. of Pages | : 17 |
Extensive research is going on in applied finance and econometrics on modeling the volatility of financial assets. Contemporary research uses most sophisticated methods to model the volatility. However, there exists two different schools of thought on the input variable, i.e., measure of volatility series. One school of thought advocates volatility measured using close-close price and the other school posits use of extreme values, that is, Open, High, Low and Close (OHLC). Little empirical work has been carried out on this topic in the Indian context. Using 1,421 daily observations of each of the three highly volatile sector-specific stock indices in India, we measure, compare and assess predictive ability of close-to-close estimator, Parkinson's and Garman and Klass's extreme value estimators over the period January 1, 2016 to September 30, 2021. The results of relative efficiency measures indicate superior performance of Extreme Value Methods (EVMs) over classical method. Furthermore, the results of 483 cross-sectional regression tests indicate the predictive ability of EVMs and report statistically significant t-statistics for a majority of the periods. It is also observed that EVMs have low standard errors when outliers are weeded out. Similarly, EVMs have t-statistic values and are statistically significant at 1% level during highly volatile market conditions. Our results support the argument that use of volatility measured with EVMs is efficient and suggest its use in modeling volatility of financial assets.
Volatility estimation is pivotal in empirical finance. Researchers and practitioners use volatility in prediction of stock returns, calculation of option price, estimation of speed of adjustment in stock returns, estimation of abnormal returns due to events and in many more instances. Efficiency is key in volatility estimation, because volatility may change over long periods of time; an extremely efficient method will allow to estimate volatility with a small number of observations. Methods based on range-based stochastic volatility estimation have arisen in