In the midst of the unprecedented economic challenges in the US elicited by the 2020 Novel Coronavirus (COVID-19) pandemic, the intricate relationship between dividend announcements, government grants, and public firms’ stock returns gained popularity in the media. As the US government passed both the CARES Act and the PPP program¹, aimed at supporting firm operations and keeping workers employed, the mass media started increasingly reporting on the public firms receiving grants. These firms were observed to allocate funds towards dividend payouts to shareholders instead of utilizing government support for bolstering firm operations. Eventually, some of them were forced to return the COVID-19 relief loans². After publicly revealing firms’ motives towards their dividend payout policies, a drop in firms’ stock price is not surprising. A reason behind this is that not only will the dividend announcements impose direct costs on the firms in the form of public scrutiny, and perceived corporate ethical irresponsibility, but it will additionally alter how market participants, stakeholders, and the broader public see these firms in future (Cororaton & Rosen, 2022; Carberry et al., 2018) .

Within these lines, this study initially quantifies the current estimate of the economic impact of the dividend announcements upon the stocks listed on NYSE and NASDAQ Composite. It specifically investigates the economic impact of the dividend announcements made by the listed firms receiving government grants aiming at smoothing the negative effects of the pandemic on their operations. The study finds that the Cumulative Abnormal Returns (CARs) are negative, yielding ≈ 6.5%, prior to and up to ten days after the announcements. These findings simply imply that firms that receive COVID-19 government grants and announced to pay dividends were negatively seen in the eyes of the market participants. This further leads to investigate the statistical significance of other factors triggering the CARs of the grant receivers. Regression analyses reveal that capital markets do indeed negatively react to the dividend announcements of the firms receiving government support. Interestingly, yet somewhat as expected, the proxy for media coverage shows evidence that the negative effect towards the CARs intensifies with higher media coverage. This finding is consistent with Pacheco and Francis (2020) showing that the coverage highlighted the idea that public firms have other sources of financing, making it inappropriate for them to use the government support in the first place. This finding also supports the reasoning behind the coefficients of the firm size variable being negative. In addition, firm leverage and liquidity levels are found to have a significant impact on the returns, as well as it is firms’ ESG performance, which may be reinforced during periods of higher economic uncertainty (Heba, 2022) .

As it appears, the immediate relation between the COVID-19 government grants, shares’ returns, and firms’ dividend announcements has not yet been addressed in the past literature (articles that come close include: Xu et al., 2023 ; Li et al., 2022 ; Sterenczak & Kubiak, 2022 ; Kluzek & Schmidt-Jessa, 2022 ; Cejnek et al., 2021 ; Huang, 2020 ; Harjoto et al., 2020 ; Krieger et al., 2020 ; Baker et al., 2020 ; Baker & Wurgler, 2016 ). This study contributes to the aforementioned literature by examining the effect of investor sentiment on the financial markets by quantifying three confounding effects of the dividend announcements: 1) that dividend announcements made by COVID-19 government grant receivers trigger a negative sentiment effect among the stock market investors, 2) the effects towards the CARs differ among firm size, leverage, liquidity, and firm’s ESG performance, and 3) that the media coverage intensifies the effect of the announcements. A policy-related contribution would be that a study of public firm grant receivers’ behavior would generate novel policy insights for the design of future emergency lending programs.

This study observes a comprehensive set of 1218 newspaper-published articles incorporating information on dividend announcements made by 812 (NYSE Composite and NASDAQ Composite listed firms) COVID-19 grant receivers in the period from January 2020 to January 2022. To ensure that the source of the article is reliable and reaches large masses, “The Wall Street Journal”, “The New York Times”, and “The USA Today” have been considered³. The newspaper articles are identified via the ProQuest search engine using the keywords: COVID-19 government grants, dividend announcement, and dividend payout during COVID-19. Around 84% of the relevant articles come from “The Wall Street Journal”, and the rest is from “The New York Times”, and “The USA Today” altogether.

To construct the main variable DivPay, Refinitive Workspace is used to supply the data on COVID-19 government grants, and firms’ dividend announcements and payments. Data on COVID-19 government grants is available for the years 2020 and 2021. DivPay equals to one if a firm in the sample announces a positive dividend payment to its shareholders after receiving a COVID-19 related grant, and DivPay equals zero if a firm announces no dividend payments to its shareholders for the period when a COVID-19 related grant is effective. Of the total number of firms in the sample, 59 firms announced a positive dividend payment, despite receiving COVID-19 related government grant, and 753 announced no dividend payouts, i.e. ceased dividends, while receiving COVID-19 related government support.

Next, to evaluate whether announcing to pay dividends while receiving a government grant has an impact on companies’ CARs during the COVID-19, the study exerts the value-weighted CARs as calculated by the EventStudyTools⁴ (Wolf et al., 2014) .

Lastly, firm fundamentals data is collected from Refinitive Workspace (total assets, long- and short-term financial debt, cash/assets, ESG score)⁵, and the media coverage proxy data, the Google Trends Index⁶, is collected from Google Trends.

To evaluate the economic impact of the dividend announcements the event-study method is used upon companies’ stock returns through the one-factor value-weighted market model (e.g. Ichev, 2021 ; Carberry et al., 2018 ; Donadelli et al., 2016 ; Peress, 2014 ; Kaplanski & Levy, 2010a, 2010b )⁷. The event-study results would be biased due to the overlapping windows if all published articles on firms’ dividend announcements were considered as independent, hence the “first announcement” selection criterion is used⁸. It is important to note that the selection criterion gives special attention to dividend announcements that happen around the same time as firms’ earnings announcements. This is crucial because the time proximity of these events could cause potential confounding effects influencing the estimates, introducing bias. This procedure results in 812 non-overlapping events.

To evaluate the statistical impact of the dividend announcements made by the receivers of COVID-19 government grants on the CARs, the following regression is estimated (Berlinger et al., 2022; Donadelli et al., 2016) :

$\begin{array}{c}{\text{CAR}}_{i,t}={\gamma }_0+{\gamma }_1{\text{DivPay}}_{i,t}+{\gamma }_2{\text{Size}}_{i,t}+{\gamma }_3{\text{Leverage}}_{i,t}+{\gamma }_4{\text{Cash}/\text{assets}}_{i,t}\\ +{\gamma }_5{\text{ESGscore}}_{i,t}+{\gamma }_6{\text{GoogleTI}}_{i,t}+{\phi }_t+{ϵ}_t\end{array}$ (1)

where ${\text{CAR}}_{i,t}$ is the cumulative abnormal return of firm i on event window t, ${\gamma }_0$ is the regression intercept, ${\text{DivPay}}_{i,t}$ is a dummy variable denoting firms announcing a positive dividend payment after receiving a COVID-19 related grant⁹, ${\text{Size}}_{i,t}$ is the natural logarithm of total assets, ${\text{Leverage}}_{i,t}$ is firm’s long- and short-term financial debt scaled by total assets, ${\text{Cash}/\text{assets}}_{i,t}$ is the cash-to-total assets ratio, ${\text{ESGscore}}_{i,t}$ is the ESG score of the firm, and ${\text{GoogleTI}}_{i,t}$ is the Google Trend Index which is used as a proxy for media coverage. ${\phi }_t$ denotes the industry controls.

This study documents that announcing positive dividend payouts to the shareholders while using COVID-19 government support is negatively recognized by the financial market. Tests reveal that size, leverage, cash-to-asset ratio, ESG performance, and media coverage play significant roles in companies’ financial performance during the COVID-19 period.

This study is limited to the ability to study the short-term effects of the dividend announcements on the stock markets in the US simply because of the lifespan of the government grants (grants are available only from January 2020 to January 2022) provided for the firms. In addition, it is consciously limited to the use of newspapers as a credible mass media source to create a sample of reported announcements. Another limitation is that the analysis does not differentiate among various types of investors. However, it is worth documenting these effects as market participants may find publicly listed firms’ dividend payments during times of uncertainty particularly provocative, hence they potentially re-consider their existing investment strategies.

Slovenian research agency, program No. P5-0161.

¹Coronavirus Aid, Relief, and Economic Security (CARES) Act provided about $2 trillion in fast and direct economic assistance for American workers, families, businesses, and industries. Under the CARES act, the Paycheck Protection Program (PPP) was meant to support (with about $658 billion) businesses with the resources they need to maintain their payroll, hire back employees who may have been laid off, and cover applicable overhead. Source: https://home.treasury.gov/policy-issues/coronavirus/about-the-cares-act , and https://home.treasury.gov/policy-issues/coronavirus/assistance-for-small-businesses/paycheck-protection-program .

²Among the many cases, AutoNation which is the US largest auto dealership chain, and restaurant chains Ruth’s Chris, and Shake Shack agreed to return their government loans after a public outcry. Source: https://www.nytimes.com/2020/04/26/business/coronavirus-small-business-loans-large-companies.html .

³Source: https://fullintel.com/top-media-outlets/the-top-10-us-daily-newspapers . Note, as social media source of information cannot always be verified and traced back to the original author, the social media as a source of information is disregarded. Moreover, many social media articles during the COVID-19 pandemic are found adjusted, deleted under suspicious circumstances or even faked after initially receiving reliability criticism.

⁴ https://www.eventstudytools.com/ .

⁵For consistency among the data items, calendar-quarter based periods on Refinitiv Workspace’s classification is used. Firm descriptives are as follows assets in $ mil amount to 215.32; number employees equal to 548.89; debt/assets % equals to 34.5, and cash/assets % equals to 20.0.

⁶The Google trends index takes values from 0 to 100. It represents the search activity relative to its highest point during a specified period of time. In this case, data is collected relative to the event windows considered.

⁷The Gujarati technique (Gujarati, 1970) is used to uncover the shifts in the market model estimate, setting 100 days in the estimation period. Event windows [−10, +10], [−5, +5], and [−1, +5] are chosen as such in order to capture the initial public information effect stemming from the dividend payment chronology process (Berk & DeMarzo, 2019) .

⁸ Bernard (1987) shows that overlapping of event windows results in cross-correlation of the stock returns. The bias recorded is downward-directed and defined as $\frac{C_{kk}}{E\left[{\hat{C}}_{kk}\right]}=\frac{{\sigma }^2}{E\left[{\hat{\sigma }}^2\right]}\times \frac{{I^{\prime }}_k{\left(X^{\prime }X\right)}^{-1}\left(X^{\prime }AX\right){\left(X^{\prime }X\right)}^{-1}{I}_k}{{I^{\prime }}_k{\left(X^{\prime }X\right)}^{-1}{I}_k}$, where C is the correct covariance matrix of the OLS-based coefficients, $I_k$ = (K × 1) is a vector with kth element equal to one, and other elements equal to zero, and X is a matrix of independent variables. In this paper, the selection criterion takes the articles, i.e. events, in a chronological sequence, where it starts with the first event in the sample, ignores all events showing up in the following 10 or 5 days-depending on the length of the event window [−10, +10], [−5, +5], and [−1, +5]. In addition, the selection criterion continues with the second event ignores the events in the proposed range, and so on. This procedure results in 812 non-overlapping events.

⁹This study is limited to not being eligible to include lagged variables (e.g. Fama & French, 2002 ) as the lifespan of the grants is very limited.

¹⁰CARs significance is demonstrated through a range of parametric and non-parametric tests. In Patell Z-score test, each security’s abnormal return is normalized by its estimation period standard deviation to limit the potential impact of stocks with high return volatility. StdCsect Z accounts for event-induced volatility and serial correlation (Boehmer et al., 1991) . The generalized sign test (Gen Sign Z) shows the proportion of positive and negative CARs against an assumed 50% split under the null hypothesis of no reaction to the event (Cowan, 1992) . Lastly, a post-estimation test to examine whether the difference in coefficients between the panels is jointly equal to zero, or significantly different is presented. Test statistics is the ratio of the departure of the estimated value of a parameter from its hypothesized value to its standard error.

¹¹Conclusions are robust upon controlling for the size of the dividends paid out. Namely, the 59 firms announcing to pay dividends do not change their dividend payout policy over the period of observation.