Table 1.

A chronological summary of the literature on COVID-19 and financial markets.

Study	Data, methodology and research scope	Main findings
Haroon and Rizvi (2020)	EGARCH model was employed for 23 sectoral indices for the US from Dow Jones from 1 January 2020 till 30 April 2020.	A strongly positive relationship between news coverage and market volatility. However, price volatility had little and moderate effects.
Chen et al. (2020)	Hourly Google search queries on coronavirus-related words were proxied for the sentiment from 15 January 2020 to 24 April 2020. In addition, the Vector Auto-Regression (VARs) was employed with the Bitcoin market.	This study found that an increase in fear of coronavirus is likely to negatively predict the Bitcoin returns and higher trading volume. Furthermore, it indicates that investors perceive Bitcoin as a conventional financial asset rather than a safe-haven asset during market distress.
Buckman et al. (2020)	A brief policy note with the newly developed Daily News Sentiment Index that provides real-time data from 1980 to 2021 was released.	This study found the in-line results of news sentiment and the COVID-19 news coverage. A high correlation with consumer sentiment was found.
Sun et al. (2021)	Coronavirus-related news for 14 events (CRNs) and economic-related announcements for 10 events (ERAs) was used for China, Hongkong, Korea, Japan, and the U.S, from December 2019 to February 2020. Furthermore, the event-study approach and regressions of three-factor models were taken to examine the relationship between this sentiment and stock performance in medical portfolios.	Both indices do not cause irrational behaviours in medical portfolios but they exhibit the positive relationship with five markets’ medical portfolios. This study also found stronger effects on institutional investors than individual ones.
Valle-Cruz et al. (2021)	The Twitter data (the content of COVID-19) and important worldwide financial indices were used to examine whether they exhibit the relationship or not. This study used the fundamental and technical financial analysis combined with a lexicon-based approach on financial Twitter accounts. There are two sub-periods (for H1N1: June to July 2009; COVID-19: January to May 2020) for this research.	The market reacted after 0 to 10 days for coronavirus tweets and 0 to 15 for the H1N1 posts. The data source of The New York Times, Bloomberg, CNN News and Investing.com exhibits a high correlation between investor sentiments and equity market behaviour.
Sun et al. (2021)	The sentiment index, retrieved from GubaSenti established by the International Institute of Big Data in Finance for the investor sentiment, was measured by analysing the textual meaning on the largest media platform in China. An event study and regression to define whether sentiment impacts the abnormal returns or not. The Data frame covers the period from 25 July 2019 to 31 March 2020 with 71 industries in China.	Comparing the usual circumstance, the effects of investor sentiment is stronger in the pandemic period. Furthermore, firms having high PB, PE and CMV, low net asset, and low institutional shareholding are more pronounced to the impacts of investor sentiment on stock returns.
Lyócsa et al. (2020); Lyócsa and Molnár (2020)	Google search terms were used to measure the fear and panic feelings of investors over the period from December 2, 2019 to April 30, 2020. This study employs a simplified version of the heterogeneous autoregressive (HAR) model for 10 stock market indices.	This study indicates that this investor sentiment index can be a predictive factor in stock price variation around the world.
Fassas (2020)	Using variance risk premium analysis to measure risk-aversion behaviour, this paper aims to calculate the willingness-to-pay of market participants to hedge the variation before and after COVID-19. The data period stretches from April 2011 until May 2020 in three advanced economies and the methodology is TVP-VAR methodology to capture the connectedness.	This study found that the COVID-19 strengthened the risk-aversion connectedness among these markets.
Smales (2021)	The extended study of Google search terms as proxies for ’investor attention’ in G7 and G20 economies from January 2020 to June 2021. This paper also used the robustness check with the ’FEARS’ index by Da et al. (2015) to see how this attention influences the stock markets.	There is an association between GSV (Google Search Volume) and the financial market returns. This effect is more pronounced to volatility and weaker effect in the government bond yields, where the institutional investors mostly participate in. The retail investors paid more attention to the FEARS terms.
Mazumder and Saha (2021)	This study proxies the fear by constructing the equally weighted index of both newly infected cases and deaths over the period from January-2019 to July-2020. The set of IPO firms’ characteristics were employed for regression to see how the IPO firms perform during the COVID-19 pandemic.	The IPO firms exhibited higher returns in 2020; however, they decrease when increasing fears. Compared to the existing firms, the IPO companies are more sensitive to COVID-19 shocks.
Cepoi (2020)	Using six indicators (The panic Index, The Media Hype Index, The Fake News Index, Country Sentiment Index, The Contagion Index, media coverage Index) for panel data over the period 3 February 2020 to 17 April 2020 in six countries, this study explores the asymmetric relationship between news and stock returns.	There are heterogeneous effects of news on different types of markets (inferior, superior, and middle class). Furthermore, gold is not the ’safe-haven’ asset during the COVID-19 pandemic.
Salisu and Akanni (2020)	The global fear index (GFI) for the COVID-19 pandemic was constructed by reported cases and death cases for OECD and BRICS countries since the COVID-19 outbreak.	This study found that GFI has predictive power on stock returns. Furthermore, the “asymmetric” effects of macro (common) factors improve the quality of forecasting power.
Xu et al. (2020)	This study constructed the new sentiment index with more accurate and critical news than state-controlled media from the study of You et al. (2018) for the period from 1 January 2019 and 30 August 2020. The methodology in this study is the regression to see how this news index influences the stock returns in China.	The public attention (infection scale), as well as this news, play an important role in stock market response to firm-specific information. Particularly, the Chinese stock markets are more sensitive to firm-specific information after the COVID-19 outbreak.
Smales (2020)	Google search terms were used to construct the investor sentiment across 11 industries in the US market from 31 December 2019 to 31 May 2020. The authors constructed the regression to examine whether the investor’s attention negatively influenced stock returns.	The author documented the heterogeneous effects of investors’ attention on the stock returns (for example, consumer staples, healthcare and IT having better performance in COVID-19 and gaining more attention).
Aloui et al. (2021)	By using the high-frequency domain, this paper draws the data from October 7, 2005 to September 25, 2020 with the methodology of the continuous wavelet transform. This paper also takes into account the American Association of Individual Investors (AAII), St. Louis Fed Financial Stress Index (FSI) and the volatility index (VIX) for constructing the investor sentiment.	This paper contributes empirical evidence that the investor sentiment links to Islamic stocks and bonds over timescales and investment horizons.
Salisu and Vo (2020)	Using the dataset of 20 countries with keywords in Google searching terms ”health news” in the period of 30th of March 2020 starting from 1st of January 2020, this paper explores the role of sentiment index, proxied by the attention to ”health news” on equity markets. This study also uses a variety of methodologies such as pool regression and the forecast evaluation of the predictor.	The newly constructed index has significant predictive power on stock return. The asymmetric effects improve the quality of prediction. The results hold robust for in-sample, out-sample, outliers and heterogeneity.