Firms' COVID-19 Pandemic Exposure and Corporate Cash Policy: Evidence from China.

The COVID-19 pandemic adversely impacted economic activity, decreased corporate revenues, and magnified cash flow fluctuations. We study how Chinese listed firms' COVID exposure influences their cash holdings. A firm's COVID exposure is measured by its excess stock return responses to globally newly infected cases while controlling for market return. Firms increase (decrease) cash balances when their stock returns fall (increase) with COVID severity due to precautionary motives. Firms cannot predict the evolution of the pandemic, which impacts demand and supply and the cash conversion cycle. The deteriorating business condition also increases external financing costs with non-state-owned, low-growth, small, and firms without overseas businesses facing higher financial frictions. Furthermore, firms with good corporate governance tend to pre-empt operational uncertainty by increasing cash holdings. The increased cash holdings translate to more R&D expenditure but lesser capital investment. Our results remain robust to placebo tests, using excess cash and alternative COVID exposure measures.

Introduction

The COVID-19 pandemic has triggered a worldwide crisis that significantly curtailed global economic activity (Bernanke, 2020) and caused massive stock price movements (Ding et al., 2021), resulting in an acute increase in market risk aversion that matches the global financial crisis.1 In addition, countries worldwide have adopted disease containment measures that destabilised the world's supply chain, impacting industries' demand and supply.2 Facing this unprecedented shock, firms' corporate revenues have been adversely affected, and their cash flow fluctuations magnified. To this end, this paper investigates how firms manage their corporate cash holding when exposed to the pandemic.

Firms' exposure to the pandemic can have two opposing effects on cash holding. On the one hand, the precautionary motive of cash holding predicts that firms will likely increase cash holding when exposed to the pandemic. As epidemic diseases are unpredictable, managers cannot predict the evolution, severity, and duration they last.3 As a result, firms’ exposure to the pandemic could negatively impact the supply of intermediate goods and the demands for final goods,4 affecting the duration of the cash conversion cycle. In response, firms might increase their cash balances to hedge against cash flow fluctuations.

On the other hand, the stock market turmoil caused by the pandemic and the deteriorating market conditions could heighten uncertainty and increase financial frictions for corporate managers. Uncertainty increases the real option values of investment, making firms more cautious when investing (Bloom et al., 2007); thus, firms may not hold as much cash for investment. However, firms envisage that external financing costs in future states will rise, leading to inefficient investment caused by insufficient liquidity.5 Therefore, in anticipation, firms may increase their cash holding when exposed to the pandemic. The overall effect of firms’ exposure to the pandemic on their cash holding stemming from precautionary motives, uncertainty and financial frictions is hard to predict theoretically; hence this is addressed empirically.

For two reasons, we study a large sample of 3924 listed firms in China for 2020Q1-2021Q1. First, the coronavirus outbreak started in Wuhan, China, in December 2019, followed by a lockdown of Wuhan in January 2020. The initial COVID-19 lockdown threatened Chinese enterprises in the first quarter of 2020; tens of millions of small and medium-sized enterprises (SMEs) were shut down (Dai et al., 2021). China's economy has been steadily recovering through economic recovery policies and pandemic-related restrictions.6 As the world's second-largest economy, China presents an interesting case study to empirically test the overall effect of firms' exposure to the pandemic on their cash holding. Secondly, China has continued to enforce the COVID-Zero policy with draconian lockdowns and strict quarantine measures (Fang et al., 2020). Travel bans and lockdown measures have impacted the tourism and hospitality industries the most. In addition, the retail sector has also been adversely affected, with many shopping malls subject to COVID-19-related restrictions, including a reduction in store operating hours, temporary closures, and cancellation of events. Under China's COVID-Zero policy, these measures may continuously impact firms' operations and cash holding policies.

Recognising that not all firms are exposed to the pandemic equally and negatively, we measure the firms' exposure to the pandemic (i.e., COVID exposure, henceforth) by the slope coefficient of a linear regression of the firm's risk-free adjusted stock returns on the pandemic severity measure while controlling for market returns. For firms that are negatively (positively) impacted by the pandemic, the slope coefficient is negative (positive) and statistically significant.7 We then perform the baseline regression of firms' cash holding on firms' COVID exposure with various control variables as previously suggested in cash holding determinants literature. The results reveal a negative association between the firms' COVID exposure and cash holding. Firms whose stock returns respond negatively to pandemic severity increase corporate cash holding. However, firms whose stock returns respond positively to pandemic severity decrease their cash holding. Our results remain robust after controlling for firm and quarterly time-fixed effects. The results support precautionary motives and financial frictions that increase cash holding amongst firms with a harmful COVID exposure.

To understand how firms' COVID exposure influences their cash holding, we examine their financial frictions and corporate governance. Specifically, firms' financial frictions are captured through the firms' state ownership status, growth, size and overseas business diversification. We find that their cash holding increases when the pandemic severity adversely impacts their stock returns for non-state-owned firms, low-growth and small firms, and firms without overseas business. These results also suggest that firms increase cash holding for precautionary motives, especially when financial frictions peak. In addition, we find that firms' COVID exposure significantly influences their cash holding when firms exhibit good corporate governance (i.e., firms have high board independence, no CEO-Chairman duality and low institutional ownership). These results suggest that the link between firms’ COVID exposure and cash holding could stem from operational uncertainty during the pandemic, leading firms with good corporate governance to take pre-emptive measures by retaining cash to hedge against the effects of pandemic shocks on cash flows.

Finally, we shed light on how firms use cash holding in the pandemic. We investigate whether the increased cash holding translates to increased investment for firms whose stock returns are negatively impacted by the pandemic. The high cash holdings during the pandemic are associated with less capital investment consistent with the prediction of the real option value of an investment in times of high uncertainty. Instead, the higher cash holding is associated with more R&D expenditure.

The baseline regression results remain robust when performing a battery of robustness tests. Should firms' COVID exposure stems from unobserved firm characteristics associated with high cash holding in the pre-pandemic period, our results may be subject to endogeneity bias. Therefore, we conduct a placebo test examining the link between firms' COVID exposure and pre-pandemic cash holding. We do not find a significant association between them, suggesting that firms' cash holding policy is based on their COVID-19 pandemic exposure. Second, we study firms' COVID exposure and excess cash to study the level of cash holding adjustment in the pandemic period. Using the predicted value of cash holding in the most recent pre-pandemic quarter (i.e., 2019Q4), we define a firm's excess cash as the difference between its real cash holding and the benchmark. Firms whose stock returns fall (increase) with pandemic severity hold higher (lower) excess cash, corroborating our baseline results.

Next, we obtain a more stringent measure of firms' COVID exposure by incorporating the Fama-French three-factor model, the Carhart four-factor model, and the Fama-French five-factor model, in addition to the market model. Based on this stringent firms' COVID exposure measure, we re-estimate the baseline regression and find the results remain qualitatively unchanged. Finally, to alleviate concerns that firms' COVID exposure is not endogenously determined by firms' pre-pandemic financial conditions, we undertook t-tests for equality of firms' COVID exposure between high and low firms' pre-pandemic characteristics and corporate governance, comprising cash holding, return on equity (ROE), return on assets (ROA), financial constraint, institutional ownership, board independence, and CEO-chairman duality. Firms' COVID exposures are not significantly different from each other between any pair of high and low firms' characteristics, indicating that firms’ COVID exposure is unlikely to stem from pre-pandemic financial conditions and corporate governance measures.

This study contributes to the literature in the following ways. First, we advance our understanding of the drivers of corporate cash policies. Prior studies document that firms with more considerable aggregate risk tend to hold more cash (Acharya et al., 2013), and non-operational cash protects against future negative cash flow shocks in recessions (Lins et al., 2010). We show that firms' exposure to the COVID-19 pandemic has predictive power on corporate cash holding during the pandemic crisis. This study also sheds new light on how firms’ COVID exposure may influence corporate cash policies.

Second, like many studies on the ongoing COVID-19 pandemic, our study has shown significant policy relevance. Firms' COVID exposure will likely influence their refinancing frictions, ultimately influencing corporate strategy. The Chinese government can learn about the companies' concerns and design appropriate policies to ease the firms' refinancing costs. For instance, the government could formulate preferential policies for bank loans, tax cuts, and fee reductions according to firms’ pandemic exposure to help them maintain sufficient cash holding for business operations and overcome increased financial frictions. In addition, government subsidies or bailouts should be granted to firms adversely affected by the COVID-19 pandemic. The conclusions drawn from this research are also relevant and applicable to many economies facing stringent restrictions and travel bans to control the disease spread, which in turn affect corporate cash flows.

Third, our contribution lies in constructing a firm-level pandemic exposure based on individual firms' stock price responses to the severity of the pandemic spread. Hassan et al. (2021) use a new word pattern-based method to measure U.S. firms' primary concerns over epidemic diseases spread communicated in their quarterly earnings conference calls. In the absence of data on firms' quarterly earnings conference calls, our method is sufficiently general to appropriately capture the extent of a firm's exposure to the COVID-19 pandemic over time. This approach, as we show in this paper, yields a robust COVID exposure measure that enables the study of corporate policy responses during the pandemic.

The rest of the paper is structured as follows. Section 2 describes the data, sample, and variables. Section 3 shows our baseline findings together with robustness tests. Section 4 explores how firms’ pandemic exposure influences corporate cash holding. Section 5 provides evidence on the link between pandemic period cash holding and investments. Finally, we conclude the paper in Section 6.

Data, sample, and variable construction

Data and sample

We use the China Stock Market and Accounting Research (CSMAR) database to retrieve stock price and firm characteristics data. In addition, we utilise the pandemic data from the Center for Systems Science and Engineering at Johns Hopkins University to obtain global and country-level daily new confirmed COVID-19 cases. Our initial sample includes all firms listed on the Shanghai and Shenzhen Stock Exchanges from 2020 to 2021. After excluding firms in financial sectors and special treatment (ST) firms and winsorising all continuous variables at the 1% and 99% levels, in our final sample, we have 3,924 firms with 18,058 firm-quarter observations.

Variable construction

Our key variable of interest is the firms' pandemic exposure, which measures the sensitivity of a firm's daily stock return to the COVID-19 pandemic spread. The spread of the COVID-19 pandemic is defined as the increased newly confirmed global COVID-19 cases. For each day from January 22, 2020 to March 31, 2021, we use the country-level COVID-19 daily reported new confirmed cases and aggregate the number from each country to obtain the global number. Motivated by the literature that measures exchange rate exposure (Jorion, 1990), we calculate a firm's COVID-19 exposure using the regression8 shown in Eq. (1). In each quarter, we use the daily data to run the following regression:where is a firm's daily stock return (over the risk-free rate), is the increased number of confirmed global COVID-19 cases, and is the daily value-weighted market return. The coefficient is the firm i's COVID Exposure captured by the firm's stock return response to the evolving pandemic severity after controlling for market return. A positive (negative) COVID Exposure, denoted by (<0) represents the firm's stock returns respond (un)favourably to the pandemic severity in terms of increased confirmed cases. For a typical firm, COVID Exposure is predicted to be harmful because of deteriorating market conditions due to the pandemic and as the stock market reacts adversely to the increasing number of COVID-19 cases. However, for firms that operate in pharmaceutical or vaccine development industries, their COVID Exposure () could be positive, indicating that firms are rewarded based on business activities that alleviate pandemic uncertainty.

Summary statistics

The descriptive statistics for our main variables are presented in Table 1 a. Panel A reports the summary statistics for our dependent variable of interest, i.e., cash holding (ratio of cash to total assets). Panel B presents the summary statistics for our primary explanatory variable, COVID Exposure. Panels C and D present the descriptive statistics of the control variables. Panel A shows that our sample firms' median value of cash holding is 11.2%. The 90th and 10th percentiles of cash holding are 3.3% and 29.6%, respectively, indicating significant differences in cash holding across our sample firms. The results in panel B show that the 10th and 25th percentiles of COVID Exposure are negative, while the 75th and 90th percentiles are positive, suggesting that there is significant heterogeneity in the firms’ stock returns responses to the daily new confirmed COVID-19 cases worldwide. The correlation matrix of the main variables is reported in Table 1 b. It shows that none of the correlations between explanatory variables is high enough to cause collinearity concerns for our multivariate analysis.

Table 1a

Descriptive statistics.

Variables	# of obs	Mean	STD	10%	25%	Median	75%	90%
Panel A: Dependent variables
Cash to asset	18,058	0.142	0.112	0.033	0.062	0.112	0.189	0.296
Excess cash	18,058	0.010	0.259	−0.094	−0.057	0.107	0.177	0.275
Capital Expenditure	3,739	0.048	0.046	0.004	0.012	0.031	0.061	0.103
R&D	3,406	0.027	0.023	0.002	0.010	0.021	0.032	0.050
Acquisition	2,882	0.119	0.263	0.001	0.009	0.040	0.116	0.272
Panel B: COVID Exposure
COVID Exposure	18,058	−0.005	0.016	−0.024	−0.013	−0.003	0.002	0.012
ThreeFactor_COVID Exposure	18,058	0.000	0.014	−0.016	−0.006	0.000	0.006	0.016
FourFactor_COVID Exposure	18,058	0.001	0.014	−0.015	−0.005	0.000	0.007	0.017
FiveFactor_COVID Exposure	18,058	0.000	0.019	−0.015	−0.006	0.000	0.006	0.016
Panel C: Firm characteristics variables
Firm size	18,058	22.290	1.314	20.810	21.350	22.090	23.020	24.100
Leverage	18,058	0.424	0.210	0.152	0.258	0.412	0.569	0.706
ROA	18,058	0.019	0.042	−0.011	0.002	0.013	0.034	0.066
Tobin's Q	18,058	2.038	1.530	0.988	1.182	1.555	2.235	3.519
Sales growth	17,590	0.125	0.687	−0.517	−0.238	0.021	0.284	0.746
Firm age	18,058	3.047	0.271	2.708	2.890	3.091	3.258	3.332
Total risk	18,058	0.030	0.011	0.017	0.022	0.029	0.036	0.045
Cash Flow	18,058	0.016	0.057	−0.045	−0.015	0.010	0.044	0.088
SOE	18,054	0.291	0.454	0.000	0.000	0.000	1.000	1.000
Panel D: Corporate governance variables
Board independence	14,187	0.380	0.058	0.333	0.333	0.364	0.429	0.429
CEO chairman	14,220	0.305	0.460	0.000	0.000	0.000	1.000	1.000
Institutional ownership	17,898	40.630	24.410	6.593	19.550	41.240	60.150	73.380

Notes: This table presents descriptive statistics for the sample from January 2020 to March 2021. Definitions of all variables are provided in Appendix A. Financial firms and ST firms are removed from the sample. Panel A reports statistics on COVID Exposure, while Panel B presents dependent variables of interest. Panel C reports the firm-characteristic variables (firm controls), while Panel D reports the corporate governance variables used in the study. The mean, standard deviation, and several percentiles are shown for each variable.

Table 1b

Correlation coefficients.

Variables	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)	(13)
Cash to asset (1)	1
COVID Exposure (2)	−0.007	1
Firm size (3)	−0.174	−0.002	1
Leverage (4)	−0.370	0.007	0.437	1
ROA (5)	0.240	0.019	0.024	−0.316	1
Tobin's Q (6)	0.218	−0.013	−0.326	−0.212	0.197	1
Sales growth (7)	0.034	0.039	−0.020	0.026	0.065	0.036	1
Firm age (8)	−0.128	−0.007	0.180	0.179	−0.101	−0.071	0.006	1
Total risk (9)	0.019	−0.005	−0.199	−0.073	−0.030	0.167	−0.160	−0.161	1
Cash Flow (10)	0.151	0.015	0.066	−0.110	0.450	0.105	0.087	−0.019	−0.085	1
SOE (11)	−0.027	0.004	0.359	0.211	−0.083	−0.157	0.007	0.209	−0.174	−0.060	1
Board independence (12)	0.006	−0.001	−0.033	0.007	−0.018	0.054	−0.001	−0.049	0.043	0.000	−0.050	1
CEO chairman (13)	0.063	−0.007	−0.181	−0.125	0.070	0.105	−0.004	−0.165	0.121	0.019	−0.307	0.105	1
Institutional ownership (14)	0.037	0.010	0.446	0.155	0.046	−0.040	−0.004	0.061	−0.137	0.055	0.398	−0.072	−0.200

Notes: This table shows the correlation matrix for the main variables used in this study.

Main findings

The baseline regression

We estimate the following baseline model in Eq. (2) to examine the relation between COVID exposure and corporate cash holding:where is the cash holding (cash scaled by total assets) of firm i in quarter t. The key explanatory variable, , is the measure of firms' COVID pandemic exposure. represents a group of control variables. Following Gao et al. (2013) and Liu et al. (2015), we control for firm characteristics that influence a firm's cash holding. These control variables comprise firm size, age, leverage, return on assets (ROA), sales growth, Tobin's Q, cash flow, total risk, board independence, CEO duality and institutional ownership. Definitions of all the variables are provided in Appendix A. Firm and time fixed effects are also included to control for the influence of time-invariant firm characteristics and business cycles.

Table 2 reports the results of our baseline regression. We observe that a lower COVID exposure is associated with higher cash holding. Specifically, in column (1), we regress cash holding on COVID exposure with quarterly time and firm fixed effects. In column (2), we incorporate a set of control variables in the regression as documented in cash holding literature. The coefficients on COVID Exposure are negative and statistically significant at the 1% level in both specifications. For firms with positive (negative) COVID exposure (i.e., their stock returns respond positively (negatively) to the pandemic severity), they reduce (increase) their cash holding. In terms of economic magnitude, the coefficient on COVID Exposure is −0.10 in column (1), suggesting that a 1% increase in positive (negative) COVID Exposure would lead to a 0.1% decline (increase) in the cash holding ratio. The coefficient on COVID Exposure is approximately 0.07% in column (2) after controlling for firm characteristics. Given our sample's average firm size of 4.791 billion yuan, the 0.1% change in cash holding ratio is equivalent to an average increase or decrease of 4.791 million yuan in cash holding depending on the sign of the COVID Exposure.9

Table 2

COVID exposure and corporate cash holdings.

Dep. Variable =	Cash holdings
	(1)	(2)
Covid Exposure	−0.100***(0.000)	−0.073***(0.005)
Firm size		0.038***(0.000)
Leverage		−0.131***(0.000)
ROA		−0.106***(0.000)
TobinQ		−0.005***(0.000)
Sales growth		0.001**(0.015)
Firm age		0.019(0.451)
Sigma_ret		0.116*(0.052)
CF		0.303***(0.000)
Quarter FE	Y	Y
Firm FE	Y	Y
Observations	18,058	17,590
Adj R2	0.033	0.097

Notes: This table shows the estimated results of regressing cash holdings on firms' COVID exposure and firms' characteristics. Each regression contains firm fixed effects and quarterly time fixed effects. Standard errors are clustered by firm, and the p values are reported in parentheses. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

The consistently significant coefficients on COVID Exposure in both regressions underline the importance of firm-specific COVID-driven uncertainty in shaping corporate cash policies. In addition, we also observe the sign of the coefficient estimates on control variables are consistent with the literature. Large firms and firms with greater sales growth, stock return volatility, and cash flow hold more cash during the pandemic, while firms with higher leverage, ROA, and Tobin's Q hold lower cash during the same period.

The results in Table 2 show that a firm's COVID Exposure significantly negatively affects its cash holding during the pandemic. Next, we conduct four groups of tests to address the endogeneity concern: the placebo test, excess cash holding analysis, COVID exposure measures with alternative factor models, and difference test for COVID exposure in firms with various pre-pandemic conditions.

Robustness tests

Placebo test for endogeneity bias

The results presented in the previous section may be fraught with endogeneity concerns as firms' COVID exposure may stem from unobserved firm characteristics, and firms' decision to hold less (more) cash with negative (positive) pandemic exposure may exist even before the pandemic period. To address this concern, we conducted a placebo test to study the association between firms’ COVID exposure and their pre-pandemic cash holding (i.e., their cash holding in 2018 and 2019).

The regression results of the placebo tests are shown in Table 3 . We find no evidence that firms' COVID exposure is associated with corporate cash policies before the pandemic. In particular, the coefficient on COVID Exposure is positive in 2018, while it becomes negative with the sample in 2019. However, both cases result in an insignificant coefficient on COVID Exposure, independent of controlling for a host of firm-specific variables. In brief, firms with negative COVID exposure hold more cash during the pandemic. This relationship indicates that COVID exposure unlikely stems from unobserved firm characteristics that lead to changes in cash holding. The placebo test also implies that managers will likely alter their cash holding strategy based on the new information embedded in firms’ COVID exposure.

Table 3

COVID exposure and corporate cash holdings (placebo test).

Dep. Variable =	Cash holdings in 2018		Cash holdings in 2019
	(1)	(2)	(3)	(4)
COVID Exposure	0.043(0.171)	0.040(0.175)	−0.048(0.111)	−0.042(0.136)
Firm size		0.004(0.737)		0.010(0.403)
Leverage		−0.070***(0.007)		−0.029(0.345)
ROA		−0.082***(0.005)		−0.100***(0.001)
Tobin's Q		0.002(0.445)		0.002(0.403)
Sales growth		0.002*(0.086)		0.001(0.319)
Firm age		0.013(0.631)		0.033(0.218)
Total risk		0.003(0.968)		−0.134*(0.099)
Cash Flow		0.407***(0.000)		0.389***(0.000)
Quarter FE	Y	Y	Y	Y
Firm FE	Y	Y	Y	Y
Observations	12,691	12,469	13,450	13,212
Adj R2	0.026	0.106	0.031	0.107

Notes: This table shows the results of the placebo test using corporate cash holdings in 2018 and 2019. Quarterly time and firm fixed effects are considered in all regressions. Standard errors are clustered by firm, and the p values are reported in parentheses. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

Excess cash

In this section, we concentrate on firms' excess cash holding rather than their cash holding. Following Gao et al. (2013), we define excess cash as the difference between a firm's actual cash holding and its predicted cash holding before the pandemic. We adopt the cash holding in the last quarter of 2019 as our pre-pandemic benchmark. Next, we estimate how firms in a pandemic would act regarding cash holding policy if the COVID-19 pandemic did not occur. In particular, excess cash also reflects firms' cash holding adjustment during the pandemic period.

Table 4 presents the estimation results with excess cash. Again, we observe consistent evidence that firms with negative (positive) COVID exposure is significantly related to lower (higher) excess cash holding. Specifically, the estimated coefficient on COVID Exposure is −0.047 (−0.043) under the simple (multivariate) regression, suggesting that a 1% decline in COVID Exposure leads to a 0.047% increase in excess cash holding. More importantly, this effect is statistically significant, although smaller than our baseline regression, where a 1% decline in COVID exposure leads to a 0.100% rise in cash holding. These results imply that firms adjust their cash holding policy during the pandemic.

Table 4

COVID exposure and corporate excess cash holdings.

Dep. Variable =	Excess cash holdings
	(1)	(2)
COVID Exposure	−0.047**(0.042)	−0.043**(0.044)
Firm size		0.030***(0.005)
Leverage		−0.083***(0.000)
ROA		−0.037**(0.033)
Tobin's Q		−0.003**(0.027)
Sales growth		−0.000***(0.000)
Firm age		0.008(0.748)
Total risk		0.062(0.285)
Cash Flow		0.223***(0.000)
Quarter FE	Y	Y
Firm FE	Y	Y
Observations	16,574	16,514
Adj R2	0.030	0.092

Notes: This table shows regression results of excess cash holdings on firms' COVID exposure and firm characteristics. Quarterly time and firm fixed effects are considered in all regressions. Standard errors are clustered by firm, and the p values are reported in parentheses. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

COVID exposure with alternative factor models

We test to determine that the negative association between firms' COVID exposure and corporate cash holding is not sensitive to the factor model used to estimate firms’ COVID Exposure. We re-estimate the baseline regression by replacing the COVID Exposure of the market model with COVID Exposure based on the Fama-French three-factor, the Carhart four-factor, and the Fama-French five-factor models.

Table 5 presents the estimation results with COVID Exposure obtained using alternative factor models. We find that COVID Exposure continues to be negatively and significantly associated with firms' cash holding. In particular, a 1% decline in firms' COVID exposure leads to a 0.046% rise in its cash holding under the Fama-French three-factor model scenario while controlling for a host of control variables. In comparison, this effect becomes 0.038% and 0.042% under the Carhart four-factor model and the Fama-French five-factor model, respectively. These results indicate that the model to estimate firms’ COVID exposure is unlikely to be the dominant force behind the vital link between COVID exposure and corporate cash holding. In brief, our baseline results remain unchanged with the COVID exposure definition using alternative factor models.

Table 5

Robustness tests with alternative COVID exposure measures.

Dep. Variable =	Cash holdings
	(1)	(2)	(3)
ThreeFactor_COVID Exposure	−0.046**(0.037)
FourFactor_COVID Exposure		−0.038*(0.075)
FiveFactor_COVID Exposure			−0.042*(0.058)
Firm size	0.030***(0.003)	0.030***(0.003)	0.030***(0.003)
Leverage	−0.083***(0.000)	−0.083***(0.000)	−0.083***(0.000)
ROA	−0.037**(0.030)	−0.037**(0.030)	−0.037**(0.030)
Tobin's Q	−0.003***(0.008)	−0.003***(0.008)	−0.003***(0.008)
Sales growth	−0.000***(0.000)	−0.000***(0.000)	−0.000***(0.000)
Firm age	0.016(0.559)	0.016(0.564)	0.016(0.563)
Total risk	0.122*(0.055)	0.124*(0.052)	0.123*(0.054)
Cash Flow	0.231***(0.000)	0.231***(0.000)	0.231***(0.000)
Quarter FE	Y	Y	Y
Firm FE	Y	Y	Y
Observations	17,590	17,590	17,590
Adj R2	0.091	0.091	0.091

Notes: This table shows the regression results of cash holdings on alternative firms' COVID exposure (i.e., ThreeFactor_COVID Exposure, FourFactor_COVID Exposure, and FiveFactor_COVID Exposure) and firm characteristics. The alternative COVID exposure measures are estimated based on the Fama-French three-factor model, the Carhart four-factor model, and the Fama-French five-factor model. Quarterly time and firm fixed effects are considered in all regressions. Standard errors are clustered by firm, and the p values are reported in parentheses. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

The t-test for equality of COVID exposure across firms

Although the COVID-19 pandemic outbreak is an exogenous shock for the real economy and stock market, one may still be concerned that firms' COVID exposure may be endogenously linked to firm characteristics. These characteristics are related to firms' pre-pandemic financial conditions, such as financial constraints and profitability, or influenced directly by firms’ pre-pandemic cash holding. For example, Ding et al. (2021) find that the decline in the stock market during the COVID-19 pandemic is milder among firms with better pre-pandemic financial conditions.

To alleviate this concern, in this section, we first divide firms into high and low groups according to the median of firms' pre-pandemic financial conditions and corporate governance measures. Specifically, we use the median of the following variables in 2019 for the high and low group division: cash holding, ROE, ROA, KZ-index (a measure of financial constraint)10 , institutional ownership, board independence, and CEO-chairman duality. Next, we conducted t-tests for mean equality of firms’ COVID exposure between the high and low groups. By doing so, we seek to determine whether COVID exposure is significantly different for firms with different pre-pandemic characteristics.

Table 6 provides the t-statistics and their associated p-values for firms’ COVID exposure equality between the two groups. In column (1), the t-test results are insignificant for all COVID exposure measures (i.e., COVID Exposure, ThreeFactor_COVID Exposure, FourFactor_COVID Exposure and FiveFactor_COVID Exposure) between the two groups of firms classified according to their pre-pandemic cash holding. The results indicate that the means of COVID exposure measures are not significantly different between firms with high versus low pre-pandemic cash holding. In columns (2) to (7), we divide firms based on other pre-pandemic characteristics and find similar results. Specifically, our results show that the means of COVID exposure measures are not significantly different between high versus low ROE firms, high versus low ROA firms, high versus low financial constraint firms, high versus low institutional ownership firms, high versus low board independence firms, and firms with versus without CEO-chairman duality.

Table 6

t-tests for equality of COVID exposure between groups sorted by pre-pandemic firm characteristics.

	t-test for equality of COVID exposure between two groups sorted by different pre-pandemic firm characteristics
	Cash Holding	ROA	ROE	Financial Constraint	Institutional Ownership	Board Independence	CEO-chairman Duality
	(1)	(2)	(3)	(4)	(5)	(6)	(7)
COVID Exposure	−0.8608(0.3894)	−1.1806(0.2378)	−0.1919(0.8478)	0.9513(0.3415)	−0.6379(0.5236)	0.5719(0.5674)	0.6795(0.4969)
ThreeFactor_COVID Exposure	−0.2371(0.8125)	−0.3135(0.7539)	0.9035(0.3663)	0.6413(0.5213)	−1.0435(0.2967)	−0.0269(0.9785)	0.9830(0.3256)
FourFactor_COVID Exposure	−0.5501(0.5823)	−0.2918(0.7704)	1.1798(0.2381)	−0.0610(0.9514)	−0.3662(0.7142)	−0.3014(0.7631)	0.6263(0.5311)
FiveFactor_COVID Exposure	−0.2847(0.7759)	−0.5179(0.6045	0.8796(0.3791)	0.4494(0.6532)	−0.8457(0.3977)	−0.1313(0.8956)	1.0756(0.2821)
# of Observations of High Group	7,476	7,230	7,239	7,776	7,771	8,246	10,916
# of Observations of Low Group	8,134	8,380	8,371	7,834	7,839	7,364	4,694

Notes: This table shows the t-test results for the equality of COVID exposure across two groups of firms sorted by pre-pandemic firm characteristics, including cash holdings, ROA, ROE, financial constraint (KZ index), institutional ownership, board independence, and CEO-chairman duality. Firms are sorted into two groups based on the median of each firm characteristic in 2019; then, firms' COVID exposure means are estimated for each group from the first quarter of 2020 to the first quarter of 2021. The p values are reported in parentheses. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

Overall, the empirical results from a set of t-tests indicate that firms’ COVID exposure measures are unlikely to be influenced by their pre-pandemic financial conditions or corporate governance measures.

Channels through which firms’ COVID exposure influences cash holding

We attempt to understand how firms' COVID exposure influences corporate cash holding policies by conducting two groups of tests. Specifically, we explore firms' financial frictions or agency conflicts as channels for COVID exposure to affect firms' cash holding. Accordingly, we first investigate firms’ financial frictions as important mechanisms in subsection 4.1; then, we test the agency conflict mechanism by considering a set of corporate governance measures in subsection 4.2.

Financial frictions

Financial frictions might force firms to abandon profitable investment opportunities, especially companies with much information that is asymmetric to external providers of capital. To evaluate firms' financial frictions as a mechanism for COVID exposure to affect corporate cash holding, we examine how our main results vary according to firms’ state ownership status, growth, firm size and overseas business diversification.

We present in Table 7 the regression results for different groups of firms. Specifically, we report in columns (1) and (2) the results using state-owned and non-state-owned firms, respectively. We find that firms' cash holding are only statistically negatively associated with their COVID exposure in non-state-owned firms. Similarly, we find that only low-growth and small firms are significantly affected by COVID exposure on their cash holding (columns 3–6). Finally, we analyse the variation of our baseline results by considering firms with and without overseas business. The effect of COVID exposure on cash holding for firms without overseas business is approximately three times larger than that for firms with overseas business, with COVID Exposure coefficients of −0.077 and −0.026, respectively (columns 7–8). Moreover, the effect is only significant in the former case, suggesting that overseas business diversification can mitigate the impact of firms’ COVID exposure on cash holding. In brief, corporate cash holding are significantly negatively associated with COVID exposure only in firms with high financial frictions and those without global business. The results suggest that financing friction is an important channel to explain our main findings.

Table 7

COVID exposure, corporate financial frictions and cash holdings.

Dep. Variable =	Cash holdings
	SOE	Non-SOE	High growth	Low growth	Large firm	Small firm	Overseas business	Non overseas business
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
COVID Exposure	−0.062(0.133)	−0.072**(0.024)	−0.070(0.112)	−0.092**(0.034)	−0.049(0.105)	−0.092**(0.024)	−0.026(0.662)	−0.077***(0.009)
Firm size	0.036***(0.004)	0.039***(0.000)	0.041***(0.001)	0.030***(0.002)	0.064***(0.000)	0.039***(0.007)	0.091***(0.002)	0.031***(0.000)
Leverage	−0.149***(0.001)	−0.127***(0.000)	−0.152***(0.000)	−0.102***(0.001)	−0.166***(0.000)	−0.099***(0.001)	−0.240***(0.000)	−0.116***(0.000)
ROA	−0.103**(0.018)	−0.105***(0.000)	−0.125***(0.000)	−0.083***(0.008)	−0.137***(0.000)	−0.078***(0.002)	−0.376***(0.000)	−0.062***(0.009)
Tobin's Q	−0.001(0.692)	−0.006***(0.000)	−0.005***(0.009)	−0.007***(0.001)	−0.003(0.132)	−0.006*(0.056)	−0.005**(0.049)	−0.004**(0.016)
Sales growth	0.002**(0.024)	0.001(0.130)	0.001(0.417)	−0.009*(0.059)	0.002**(0.024)	0.001(0.340)	−0.001(0.568)	0.002***(0.008)
Firm age	0.020(0.646)	0.019(0.544)	0.087**(0.028)	−0.000(0.992)	0.016(0.607)	−0.002(0.962)	−0.042(0.402)	0.030(0.329)
Total risk	0.008(0.935)	0.153**(0.042)	0.159*(0.054)	−0.005(0.957)	0.051(0.460)	0.097(0.327)	−0.002(0.986)	0.130*(0.065)
Cash Flow	0.327***(0.000)	0.293***(0.000)	0.338***(0.000)	0.275***(0.000)	0.290***(0.000)	0.325***(0.000)	0.466***(0.000)	0.289***(0.000)
Quarter FE	Y	Y	Y	Y	Y	Y	Y	Y
Firm FE	Y	Y	Y	Y	Y	Y	Y	Y
Observations	5,208	12,378	8,794	8,796	8,897	8,693	3,325	14,265
Adj R2	0.115	0.092	0.114	0.075	0.122	0.101	0.185	0.090

Note: This table shows the results of regressions investigating how firms' COVID exposure affects corporate cash holdings through different financial friction channels. SOE and non-SOE, High and Low, Large and Small, and Overseas and Non overseas represent the subgroups of state-owned and non-state-owned enterprises, high-growth and low-growth firms, large and small firms, and firms with and without overseas business, respectively. Standard errors are clustered by firm, and the p values are reported in parentheses. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

Corporate governance

In this subsection, we test agency conflicts as a potential channel for COVID exposure to affect firms' cash holding. Suppose a negative COVID exposure aggravates the agency conflicts leading to higher cash holding. In that case, we expect the association between firms' COVID exposure and cash holding to be pronounced in poor corporate governance firms. On the other hand, we conjecture that good corporate governance would strengthen the firms' COVID exposure and cash holding association if the firms’ harmful COVID exposure represents more operational uncertainty that motivates them to hold more cash. Accordingly, we adopt three proxy measures of corporate governance: board independence, CEO duality, and institutional ownership.

Table 8 shows the empirical results by regressing corporate cash holding on COVID exposure across firms with different levels of corporate governance. In columns (1)–(2), we estimate the regression model using subsamples of firms with high and low board independence, respectively. Firms' COVID exposure only significantly affects cash holding when board independence is high. Similarly, the results in columns (3)–(6) reveal the significant COVID exposure and cash holding association holds when firms have no CEO-chairman duality and low institutional ownership. These results suggest that firms' COVID exposure is more likely to capture information about firms’ future operational uncertainty instead of a proxy for agency conflicts leading to changes in cash holding.

Table 8

COVID exposure, corporate governance and cash holdings.

Dep. Variable =	Cash holdings
	High board indepedence	Low board indepedence	Duality chairman	Non-duality chairman	Low institutional ownership	High institutional ownership
	(1)	(2)	(3)	(4)	(5)	(6)
COVID Exposure	−0.169***(0.000)	0.036(0.383)	−0.023(0.686)	−0.097***(0.005)	−0.079**(0.027)	−0.058(0.121)
Firm size	0.068***(0.003)	0.040***(0.001)	0.059***(0.009)	0.047***(0.001)	0.030**(0.017)	0.062***(0.000)
Leverage	−0.135***(0.000)	−0.132***(0.000)	−0.167***(0.000)	−0.120***(0.000)	−0.110***(0.000)	−0.121***(0.000)
ROA	−0.178***(0.001)	−0.206***(0.000)	−0.249***(0.000)	−0.158***(0.000)	−0.106***(0.002)	−0.124***(0.000)
Tobin's Q	−0.008***(0.000)	−0.001(0.454)	−0.006**(0.019)	−0.004**(0.017)	−0.005**(0.013)	−0.005***(0.006)
Sales growth	−0.000(0.754)	0.002**(0.035)	0.002(0.208)	0.001(0.429)	0.001(0.183)	0.002**(0.025)
Firm age	0.002(0.952)	−0.011(0.776)	0.015(0.769)	−0.012(0.705)	−0.015(0.672)	0.007(0.853)
Total risk	0.154*(0.084)	−0.032(0.717)	0.043(0.712)	0.077(0.303)	0.037(0.658)	0.233***(0.005)
Cash Flow	0.410***(0.000)	0.374***(0.000)	0.380***(0.000)	0.397***(0.000)	0.316***(0.000)	0.303***(0.000)
Quarter FE	Y	Y	Y	Y	Y	Y
Firm FE	Y	Y	Y	Y	Y	Y
Observations	7,371	6,479	4,140	9,710	8,728	8,704
Adj R2	0.140	0.117	0.108	0.136	0.094	0.107

Note: This table shows the results of regressions investigating how firms' COVID exposure affects corporate cash holdings through different corporate governance channels. High board and low board, duality and non-duality, and high institutional and low institutional represent the subgroups of low-board-independence firms and high-board-independence firms, CEO duality and non-duality firms, and high institutional ownership firms and low institutional ownership firms, respectively. The p values reported in parentheses are based on standard errors clustered by firms. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

Pandemic period cash holding and investment

Having established how firms' cash holding are affected by their COVID exposure, we further explore how firms allocate the change in their cash holding or excess cash in capital investment and R&D expenditures. Specifically, we adopt four proxy measures of firms' investments: capital expenditure, R&D expenditure, acquisition expenditure, and their sum. We run the regression of each investment proxy on firms’ cash holding or excess cash, together with the same control variables in our baseline regression.11 The results are reported in Table 9 .

Table 9

Pandemic period cash holdings and investment policies.

Dep. Variable =	Capital Expenditure		R&D Expenditure		Acquisition Expenditure		Total Investments
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
Cash	−0.020***(0.005)		0.013***(0.001)		0.005(0.870)		0.021(0.598)
Excess cash		−0.020***(0.002)		0.002(0.475)		0.033**(0.042)		−0.016(0.627)
Firm size	0.001*(0.082)	0.001(0.135)	−0.000(0.638)	−0.000(0.698)	−0.029***(0.000)	−0.028***(0.000)	−0.029***(0.000)	−0.029***(0.000)
Leverage	0.005(0.276)	0.010**(0.038)	−0.003(0.236)	−0.004**(0.034)	−0.000(0.995)	−0.005(0.805)	0.019(0.422)	0.018(0.436)
ROA	0.101***(0.000)	0.100***(0.000)	0.019**(0.023)	0.022***(0.008)	−0.217**(0.018)	−0.219**(0.015)	−0.178*(0.080)	−0.172*(0.092)
Tobin's Q	0.003***(0.000)	0.003***(0.000)	0.003***(0.000)	0.004***(0.000)	0.035***(0.000)	0.035***(0.000)	0.048***(0.000)	0.048***(0.000)
Sales growth	−0.001**(0.015)	−0.001**(0.012)	0.000(0.873)	0.000(0.593)	0.004*(0.082)	0.004*(0.077)	0.002(0.410)	0.003(0.358)
Firm age	−0.021***(0.000)	−0.021***(0.000)	−0.005***(0.000)	−0.005***(0.000)	−0.018(0.151)	−0.018(0.151)	−0.055***(0.001)	−0.056***(0.000)
Total risk	0.132**(0.039)	0.142**(0.026)	0.106***(0.000)	0.097***(0.000)	0.292(0.226)	0.306(0.206)	0.367(0.243)	0.343(0.274)
Cash Flow	0.090***(0.000)	0.091***(0.000)	−0.012*(0.095)	−0.010(0.165)	−0.086(0.177)	−0.093(0.143)	0.015(0.843)	0.024(0.749)
Quarter FE	Y	Y	Y	Y	Y	Y	Y	Y
Firm FE	Y	Y	Y	Y	Y	Y	Y	Y
Observations	13,131	13,118	11,895	11,886	13,202	13,186	9,829	9,820
Adj R2	0.118	0.120	0.295	0.292	0.169	0.170	0.211	0.211

Note: This table shows the results of regressions investigating the impact of firm cash holdings (excess cash holdings) on their investment expenditures, including capital expenditure (Capital Expenditure), R&D expenditure (R&D), acquisition expenditure (Acquisition), and the sum of them (Total Investments). The p values reported in parentheses are based on standard errors clustered by firms. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

The results in Table 9 indicate that with increasing cash holding, firms are likely to decrease rather than increase capital and acquisition investments. However, the impact on acquisition expenditure is insignificant. However, increasing cash holding is followed by higher R&D investments. These results suggest that firms have inadequate investment opportunities due to more uncertainty during the pandemic and that R&D has become a vital cash outlet. Taken together, increasing cash holding is related to a decline in total investment, even though the effect is not statistically significant. A similar pattern emerges when we replace cash holding with excess cash holding. In brief, these results suggest that firms use the increased cash holding driven by their COVID exposure mainly for R&D expenditure. In further analyses not reported here for brevity but are available from the authors upon request, we observe that the decrease in capital investment resulting from increased cash holding largely occurs in firms with negative COVID exposure.

Conclusion

The COVID-19 pandemic has enormously impacted economies and the stock market. In turn, the operational uncertainty embedded in firms' stock prices reveals essential information for managers to quickly respond and adjust their cash holding policy. In this study, we construct firms' COVID exposure by regressing firms' daily stock returns on increased COVID-19 infection cases while controlling for market returns. In addition, we examine the link between firms’ COVID exposure and corporate cash holding for a large sample of listed firms in China.

We provide robust evidence to show that firms' COVID exposure is significantly and negatively associated with cash holding. Firms whose stock returns are negatively (positively) impacted by the COVID severity measure tend to increase (decrease) their cash holding. Placebo tests exclude the concerns that the COVID exposure stems from unobserved firm characteristics that lead to high cash holding in the pre-pandemic period. Confirming evidence of the significant link is supported by robustness tests that consider excess cash and employ different models to measure firms’ COVID exposure.

We also explore how firms' COVID exposure influences their cash holding. We show that operational considerations are the main drivers of cash policies during the pandemic. The empirical evidence shows that the relationship between firms' COVID exposure and cash holding only exists in firms with high financial frictions (i.e., non-state-owned, low-growth, and small firms) and firms without global business for diversification. We also consider the channel of agency conflicts. Our empirical results do not support the agency conflicts channel. Instead, the more pronounced negative link between firms’ COVID exposure and cash holding shows up in firms with good corporate governance. The results indicate that the change in cash holding policy stems from precautionary motives under operational considerations, particularly for firms with good corporate governance that seek to benefit shareholders.

In a set of additional analyses, we examine how firms react to cash or excess cash holding by investigating their investment policies. We observe that increased cash holding during the pandemic is followed by less capital investment but more R&D expenditure. This result indicates that firms face inadequate capital investment opportunities and are cautious about investing in a highly uncertain business environment to the extent that R&D investment has become the main cash outlet during the pandemic.

Overall, our study suggests that firms' COVID exposure is an important characteristic that includes information about future uncertainty and financial frictions. This firm characteristic provides a precautionary motive for managers to shape firms' cash holding policy during global crises. From the government's perspective, it is necessary to formulate preferential policies in bank loans, tax cuts, and fee reductions according to the different firm's pandemic exposure to help them maintain sufficient cash holding for business operations and increased financial frictions. Firms with negative COVID exposure tend to suffer more from the pandemic and face more significant financial frictions than before the pandemic; thus, government subsidies or bailouts benefit these firms. In addition, to mitigate a decline in investment efficiency for firms with sufficient cash flow, the government should use big data platforms to share more information to help them find better investment opportunities. Firms with positive COVID exposure should be encouraged to reduce their cash holding to save opportunity costs. Our findings also inform entrepreneurs and policymakers by granting policy support to ease the financing cost for firms with negative COVID exposure.

Declaration of competing interest

We hereby declare that there is no known competing financial interests or personal relationships that could have appeared to influence the work reported in the paper entitled “Firms’ COVID-19 Pandemic Exposure and Corporate Cash Policy: Evidence from China”.

Variables	Definition
Panel A: Dependent variables
Cash holdings	The ratio of cash and equivalent assets to total assets.
Excess cash holdings	The difference between a firm's actual cash holdings and its predicted cash holdings before the pandemic period scaled by total assets.
Capital Expenditure	Total capital expenditure scaled by total assets.
R&D Expenditure	Total R&D expenditure scaled by total assets.
Acquisition Expenditure	Total acquisition expenditure scaled by total assets.
Total Investments	Total investment expenditure scaled by total assets.
Panel B: COVID sensitivities
COVID Exposure	COVID Exposure is a coefficient that captures the sensitivity of a firm's stock returns to the daily new confirmed COVID-19 cases worldwide. It is estimated by the regression model which regresses stock returns on the pandemic and market factors within each quarter.
ThreeFactor_COVID Exposure	ThreeFactor_COVID Exposure is an alternative measure that captures the sensitivity of a firm's stock returns to the daily new confirmed COVID-19 cases worldwide. It is estimated by the regression model, which regresses stock returns on the pandemic factor and the Fama-French three factors within each quarter.
FourFactor_COVID Exposure	FourFactor_COVID Exposure is an alternative measure that captures the sensitivity of a firm's stock returns to the daily new confirmed COVID-19 cases worldwide. It is estimated by the regression model, which regresses stock returns on the pandemic factor and the Carhart four factors within each quarter.
FiveFactor_COVID Exposure	FiveFactor_COVID Exposure is an alternative measure that captures the sensitivity of a firm's stock returns to the daily new confirmed COVID-19 cases worldwide. It is estimated by the regression model, which regresses stock returns on the pandemic factor and the Fama-French five factors within each quarter.
Panel C: Firm characteristics variables
Firm size	The natural logarithm of the book value of total assets.
Leverage	Total debt divided by total assets.
ROA	Operating cash flow divided by total assets.
Tobin's Q	Market value/replacement value.
Sales growth	(Operating income of the current quarter-Operating income of the last quarter)/Operating income of the previous quarter.
Firm age	The natural logarithm of the number of years since a firm's incorporation.
Total risk	The standard deviation of daily stock returns over the current quarter.
Cash Flow	Operating cash flow scaled by total assets.
SOE	SOE takes a value of one if the ultimate controller is the state and zero otherwise.
Panel D: Corporate governance variables
Board independence	Independent board of directors to total board of directors.
CEO duality	Dummy variable which equals one if the chairman and CEO are the same person, 0 otherwise.
Institutional ownership	Shares owned by institutions scaled by total shares outstanding.