Paper: Analyzing airlines stock price volatility during COVID-19 through internet search dataSoudeep Deb
Journal: International Journal of Finance & Economics. (Early view version is available in – http://dx.doi.org/10.1002/ijfe.2490.)
The pandemic has affected millions of lives across the world. An immediate consequence has been observed in the social and economic dynamics of different countries as well. In particular, many regulations, such as complete or partial lockdown, closure of academic institutions, holiday extension and work from home system have been imposed because of the highly contagious nature of the disease. These regulations are in turn affecting the stock market. The airline industry has been suffering, especially because of lockdowns and this fact is the focus of the paper.
The author shows that the ongoing pandemic has an unprecedented severe effect on the airline industry, which is in fact similar to the great financial crisis of 2007-08. Next, the author analyzes the stock price movements of three major airline companies, using a new approach which leverages a measure of internet concern on different topics. His method quantifies people’s reaction to sudden news by using their internet presence and search history. In the proposed approach, Twitter data and Google Trends are used to create a relevant set of predictors which are shown to have significant effect (in the Granger causality sense) on stock market movements. These predictors then lead to an appropriately modified generalized autoregressive conditional heteroskadasticity (GARCH) model. The proposed model is used to analyze and forecast stock price volatility of airline companies. The findings establish that the researcher’s approach can successfully use the effects of internet concern for different topics on the movement of the stock price index and provide good forecasting accuracy.
He also finds that the internet concern for trending topics (in this case, news related to COVID-19) has a more significant effect on the stock price index than general topics. Model confidence set (MCS) procedure further shows that the short‐term volatility forecasts are more accurate for this method than other candidate models. An attractive feature of the proposed method is its flexibility in incorporating more information. For instance, a more detailed analysis of Twitter data can in fact be used to formulate a time‐dependent set of topics. The proposed method can provide a useful analytics tool for market participants, policymakers, and market regulators to understand the movement of stock market based on the internet concern for different topics.