Is cash the panacea of the COVID-19 pandemic: Evidence from corporate performance.

This study investigates the impact of COVID-19 crisis on corporate investment and financing policies. Using a difference-in-difference approach, I find while firms suffer from a real negative shock from the pandemic on average, firms with an abundant cash reserve prior to the crisis outperform firms without. Consistent with the precautionary motive behind corporate cash holdings, this paper demonstrates the effect of cash holdings is meaningful to mitigate adverse effect of the aggregate market. My finding also highlights the difficulty in estimating the optimal cash level when rare market condition is considered.

Introduction

The global pandemic of the coronavirus has spread at frightening rates. What is equally alarming is the severe impact COVID-19 has on the economy.2 This unprecedented pandemic poses a negative supply shock to firms’ production capacity due to government lockdowns. The damaging effect is aggravated by an even larger scale of a demand shock, as affected jobless workers lose income and reduce their consumption of goods and services (Brinca et al., 2020; Guerrieri et al., 2020).

In this study, I examine whether corporate cash holdings can mitigate the adverse effect of the COVID-19 pandemic on firms’ performance. It has been well acknowledged that U.S. corporations hold much more cash than they used to. On the one hand, the advantage of keeping a liquid balance sheet is vital for a business to stay alive in a bad time. In theory, financing frictions combined with deteriorating market conditions can lead to credit rationing and exacerbate refinancing risk, making externally obtained capital more expensive than those generated internally. (Stiglitz and Weiss, 1981; Holmstrom and Tirole, 1998; Froot et al., 1993). I refer to this as the Precautionary Hypothesis. Empirical evidence lends strong support to the precautionary or risk management motive of cash holdings.3 In a similar vein, product market competition literature suggests that firms with an ample amount of cash can deter rivals’ entry and use aggressive pricing strategies to eliminate opponents’ market shares (Bolton and Scharfstein, 1990; Fresard, 2010).

On the other hand, there are agency costs associated with excessive cash holdings to shareholders. (Jensen, 1986; Stulz, 1990). Under this framework, excess cash holdings give a manager an incentive to extract rent from shareholders and cause distortions to a firm's optimal corporate policies. I refer to this as the Distortion Hypothesis. A number of empirical studies demonstrate the downside of excessive cash holdings.4 Ultimately, whether having abundant cash is beneficial to firms becomes an empirical question.

In a natural experiment setting, I employ a differences-in-differences (DiD) approach to compare firms’ performance before and after the COVID-19 pandemic as a function of their cash holdings one year prior to the crisis, controlling for firm fixed effects and observable firm characteristics. My empirical study delivers two sets of results. First, I find that the pandemic has an adverse effect on firms’ performance, where performance is measured by investment, operating performance, financing, and payout. This finding is consistent with the evidence that COVID-19 exerts a real negative shock to firms’ profit. More importantly, I find that firms with more cash holdings prior to the crisis perform significantly better than firms with less cash in using all four measures of firm performance, consistent with the Precautionary Hypothesis. My findings are robust to additional industry by year-quarter fixed effects as well as standard errors clustered by both firm and year-quarter (Petersen, 2009).

My study offers two important contributions to the literature. I show that the precautionary motive of corporate cash holdings is beneficial to firms in bad times when credit tightens and access to external finance is infeasible. While prior literature has considered the role cash holdings plays in the 2007-2008 financial crisis on corporate policy, the focus has largely been on the impact on corporate investment (Almeida et al., 2012); Campello et al., 2010; Duchin et al., 2010). This study is the first to test the effect of cash holdings on a broader set of corporate performance measures during the COVID-19 crisis, corroborating prior findings. This paper highlights the risk management function of corporate cash holdings, which is meaningful to mitigate the adverse effect of the overall market.

I also add to the growing literature on the impact of COVID-19 on firms (e.g., (Achary and Steffen, 2020); Brunnermeier and Krishnamurthy, 2020; Fahlenbrach et al., 2020). A concurrent study by Tawiah and Keefe (2020) documents similar findings of the effect of cash holdings on corporate investment during the pandemic. My results compliment (Tawiah and Keefe, 2020) findings, and also shed light on other important corporate aspects, such as operating performance, financing, and payout.

Data and empirical design

Sample selection

My sample consists of quarterly data on publicly traded firms from Compustat. The sample includes firm observations with fiscal quarters that fall between 2018Q4 and 2020Q4. I define the post-COVID-19 pandemic period when a firm-quarter observation appears after the fourth quarter of 2019. This approach splits the pre-and post-crisis evenly by calendar quarter, and helps to average out any seasonal patterns in the quarterly data (Shin and Kim, 2002). I exclude financial and utility firms (SIC codes 4900-4999 and 6000-6999). I further exclude firms with non-positive sales and total assets, following Harford et al. (2014). These selection criteria yield a sample of 30,822 firm-quarter observations.

Variable construction

Because the focus of this study is on the impact of cash holdings on firms’ performance, I include four broad categories of corporate performance as my dependent variables: investment, operating performance, financing, and payout. For each of the corporate performance variables, I include multiple proxies to mitigate the concern of measurement error. I measure investment with capital expenditure, research and development expense, sales, general and administrative expense, net working capital excluding cash, and total inventories. I measure operating performance using the return on equity, return on asset, Tobin's Q, asset growth, and market share. For financing activity, I proxy it using changes in short-term debt, changes in long-term debt, and equity issuance. Finally, I measure a firm's payout using total payout, repurchases, and dividends. All measures are scaled by total assets.

The independent variable of interest is cash holdings, which I define as the ratio of cash and short-term investment to total asset. To control variations in a firm's characteristics, I include size (log of total assets), cash flow, leverage, and Q, following Derrien and Kecskes (2013). All continuous variables are winsorized at the 1st and 99th percentiles to reduce the influence of outliers.

Empirical design

Following Duchin et al. (2010), I employ a DiD approach that considers the COVID-19 crisis as an exogenous shock and compares firms’ operating performance conditional on their cash holdings one year prior to the crisis:where i indexes the firm, and t indexes the year-quarter. θ is firm-fixed effect. After is an indicator variable that takes the value of one if the calendar quarter falls after 2019Q4. A firm's cash position is measured only once within the sample – at the end of 2018Q4. Firm fixed effects subsume the level effect of cash, and standard errors are clustered at the firm level. This identification strategy assumes that the COVID-19 crisis is a truly exogenous event that is unrelated to a firm's characteristics and year-before cash positions are not positively correlated with unobserved within-firm changes in the business fundamentals. β1 captures the added effect of cash holdings conditional on the crisis.5

Table 1 provides summary statistics of the variables used in the analyses. The average cash position measured one year ahead of the COVID-19 crisis is 22.0% of a firm's total assets. This is comparable to the magnitude reported in Duchin et al. (2010). I do not impute any missing values, but the main results remain quantitatively similar if I use a balanced sample.

Table 1

Summary Statistics.

	Mean	Std. Dev.	25th Percentile	Median	75th Percentile
Cash reserve
Cash Holdings	0.220	0.249	0.037	0.114	0.321
Controls
Size	6.388	2.611	4.677	6.640	8.190
Q	1.939	1.431	1.036	1.454	2.344
Cash Flow	-0.037	0.179	-0.028	0.011	0.026
Leverage	0.261	0.255	0.048	0.211	0.391
Investment variables
Capex	0.009	0.013	0.002	0.005	0.028
R&D	0.019	0.046	0.000	0.012	0.019
SG&A	0.087	0.122	0.022	0.049	0.102
NWC	-0.112	0.679	-0.101	-0.009	0.091
Inventory	0.091	0.119	0.000	0.044	0.135
Performance variables
ROE	-0.024	0.426	-0.057	0.011	0.048
ROA	-0.017	0.131	-0.015	0.019	0.034
Q	1.939	1.431	1.036	1.454	2.344
Asset_growth	0.045	0.229	-0.032	0.005	0.049
Market Share	0.004	0.011	0.000	0.000	0.002
Financing variables
∆ST_debt	0.014	0.087	0.000	0.000	0.001
∆LT_debt	0.018	0.084	-0.005	0.000	0.018
Equity Issuance	0.064	0.184	0.000	0.000	0.008
Payout variables
Total payout	0.006	0.015	0.000	0.000	0.006
Repurchases	0.004	0.011	0.000	0.000	0.001
Dividends	0.003	0.006	0.000	0.000	0.002

This table reports summary statistics for the main variables used in the analyses. All firm-quarter observations are between the fourth quarter of 2018 and the fourth quarter of 2020. Financial and utility firms are excluded. Quarterly measures that are reported on a year-to-date basis are adjusted by subtracting from the current quarter's measure for fiscal quarter 2,3,4.

Main empirical analyses

Investment

Table 2 presents estimates from my baseline regressions using a set of corporate investment variables as the dependent variables. The coefficients on After are negative and significant at the 1% level across all specifications, indicating that COVID-19 leads to a significant reduction in firms’ investment. The coefficients of interest, β1, are positive and significant in all columns except for R&D.6 A sufficient amount of cash reserve can mitigate the adverse effect caused by deteriorating market conditions. To gauge the economic significance, for example, one standard deviation increase in the pre-COVID cash reserve leads to an increase of Capex by 0.1 percentage point in column (2), which is comparable to the magnitude documented in Duchin et al. (2010). Across all specifications, the vector of control variables does not subsume the explanatory power of the interaction term. This provides additional evidence on the validity of the DiD setting.

Table 2

Pre Covid-19 cash holdings and corporate investment.

Specification:	Capex		R&D		SG&A		NWC		Inventory
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)
After	-0.363***	-0.354***	-0.114*	-0.112*	-0.693***	-0.712***	-2.839***	-3.778***	-0.891***	-0.794***
	(-20.34)	(-19.84)	(-1.76)	(-1.88)	(-8.68)	(-12.25)	(-4.50)	(-5.86)	(-11.91)	(-11.00)
After × Cash holdings	0.343***	0.368***	-1.465***	-0.766***	0.629**	0.509*	6.951***	4.083**	1.598***	2.091***
	(7.26)	(7.75)	(-5.53)	(-3.56)	(2.09)	(1.87)	(3.66)	(2.29)	(8.44)	(10.28)
Size		-0.091**		-3.029***		-7.598***		23.156***		-2.021***
		(-2.07)		(-14.63)		(-25.10)		(8.94)		(-8.60)
Q		0.071***		0.106		0.868***		-2.102***		-0.003
		(3.64)		(1.46)		(9.40)		(-4.97)		(-0.04)
Cash Flow		0.179		-6.583***		-9.801***		48.738***		-0.965**
		(1.05)		(-9.05)		(-9.16)		(8.83)		(-2.41)
Leverage		-0.224***		0.798**		0.069		25.801***		-0.451
		(-2.85)		(2.00)		(0.15)		(3.44)		(-1.29)
Fixed Effects	F	F	F	F	F	F	F	F	F	F
Obs	28,981	28,800	16,999	16,838	27,461	26,252	30,176	29,928	30,140	29,897
Adj R2	0.576	0.578	0.796	0.847	0.889	0.930	0.880	0.905	0.946	0.948

This table presents estimates from the firm-level quarterly OLS regression: Investment = α +  β1Cash*After + β2After + γControl + θ + ε. Measures of corporate investments: capital expenditure in columns (1) and (2); research and development in columns (3) and (4); sales, general, and administrative expense in columns (5) and (6); Net working capital excluding cash in columns (7) and (8); total inventories in columns (9) and (10). All investment measures are divided by total assets and expressed in percentage points. Other variables are defined in Appendix A. F denotes firm fixed effects and I × T denotes industry by year-quarter fixed effects. Standard errors are corrected for heteroskedasticity and clustered at the firm level, t-statistics are reported in parenthesis, with ∗, ∗∗, and ∗∗∗ represent statistical significance at the 10%, 5%, and 1% level, respectively.

Operating performance

The Table 2 results suggest firms with abundant cash reserve invest more during the post COVID-19 crisis. A natural question to ask is how do we interpret the finding? Under Distortion Hypothesis, a rise in corporate investment can be viewed as a form of inefficient overinvestment rather than the alleviation of underinvestment. To distinguish the two hypotheses, I focus on firms’ operating performance. Regardless of how I measure performance, Table 3 displays that firms experience a reduction in performance after the crisis, but the negative effect is mitigated by a sufficient pre-COVID cash reserve. Inconsistent with the Distortion Hypothesis, I find firms with large cash holdings are performing significantly better, which emphasizes the advantage of maintaining a large cash reserve. Importantly, the effect of cash holdings on market shares in column (9) & (10) demonstrates that firms “benefit from cash holdings especially from the financial turmoil by gaining a leading position in their product market” (Fresard, 2010).

Table 3

Pre Covid-19 cash holdings and performance.

Specification:	ROE		ROA		Q		Asset Growth		Market Share
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)
After	-0.025***	-0.019***	-0.008***	-0.007***	-0.167***	-0.149***	-0.021***	-0.017***	-0.000***	-0.000***
	(-3.34)	(-2.59)	(-6.33)	(-8.39)	(-4.23)	(-4.10)	(-6.44)	(-5.15)	(-5.81)	(-4.97)
After × Cash holdings	0.067***	0.062**	0.038***	0.014***	0.490***	0.885***	0.063***	0.035**	0.001***	0.001***
	(2.81)	(2.51)	(7.11)	(4.47)	(2.64)	(5.43)	(4.36)	(2.46)	(5.13)	(6.01)
Size		-0.058***		0.042***		-1.085***		0.339***		-0.000***
		(3.05)		(9.36)		(-4.19)		(27.51)		(-5.20)
Q		0.027***		-0.004**				0.019***		-0.000***
		(3.31)		(-2.53)				(3.32)		(-4.10)
Cash Flow		0.395***		0.367**		-2.715***		0.275***		0.000***
		(4.74)		(18.91)		(-6.47)		(8.96)		(3.96)
Leverage		0.056		-0.123*		0.017		-0.029		-0.002***
		(1.32)		(-1.71)		(0.05)		(-1.40)		(-8.69)
Fixed Effects	F	F	F	F	F	F	F	F	F	F
Obs	30,709	30,446	29,648	29,427	30,814	30,550	30,219	29,951	30,822	30,550
Adj R2	0.138	0.146	0.754	0.867	0.850	0.862	0.063	0.212	0.837	0.837

This table presents estimates from the firm-level quarterly OLS regression: Performance = α +  β1Cash*After + β2After + γControl + θ + ε. Measures of performance: return on equity in columns (1) and (2); return on assets in columns (3) and (4); Tobin's Q in columns (5) and (6); asset growth in columns (7) and (8); market share in columns (9) and (10). Other variables are defined in Appendix A. F denotes firm fixed effects and I × T denotes industry by year-quarter fixed effects. Standard errors are corrected for heteroskedasticity and clustered at the firm level, t-statistics are reported in parenthesis, with ∗, ∗∗, and ∗∗∗ represent statistical significance at the 10%, 5%, and 1% level, respectively.

Financing

The previous two tables provide a great support for the Precautionary Hypothesis. In Table 4 , I focus on the effect of internal slack on firms’ financing activity. Except for changes in long-term debt, firms experience a reduction in financing activity after the COVID-19 pandemic. In column (1) & (2), I find that while the crisis reduces the level of short-term debt financing on average, firms with large cash holdings have significantly more short-term debt than firms with low cash holdings, which is consistent with the view that lenders take refinancing risks into consideration when issuing short-maturity debt (Harford et al., 2014). In column (3) & (4), I discover the opposite effect: firms take more long-term debt in the post-crisis period, while cash-rich firms have relatively less long-term debt. Unlike short-term debt that has to be repaid in a fairly short period of time and is subject to a higher refinancing risk, long-term debt provides affected firms enough liquidity to overcome the adverse shock. My empirical results support Almeida (2020)’s finding that firms take long-term debt to mitigate negative liquidity shock induced by COVID-19 pandemic. The liquidity constraint may not be binding if firms already stock up sufficient cash reserve. I find while the After dummy is positively significant, the interaction term with cash holdings is negative, supporting the notion that internal slack is less costly than external debt. Lastly, I find firms are less likely to issue new equities after the crisis on average, and this is especially true for firms with large cash reserves. Since equity is usually viewed as the last resort in the presence of financing friction, this finding is consistent with the pecking order theory.

Table 4

Pre Covid-19 cash holdings and financing activity.

Specification:	∆ST_debt		∆LT_debt		Equity Issuance
	(1)	(2)	(3)	(4)	(5)	(6)
After	-0.005**	-0.004*	0.006***	0.007***	-0.005*	-0.004*
	(-2.10)	(-1.69)	(3.19)	(2.99)	(-1.82)	(-1.78)
After × Cash holdings	0.009*	0.010*	-0.002*	-0.008**	-0.042***	-0.060***
	(1.89)	(1.94)	(-1.74)	(-2.31)	(-3.13)	(-4.47)
Size		-0.013*		0.033***		0.093***
		(-1.86)		(8.87)		(10.63)
Q		0.002		-0.003**		0.023***
		(0.73)		(-2.08)		(5.73)
Cash Flow		-0.079***		-0.020**		-0.057***
		(-4.70)		(-2.51)		(-3.15)
Leverage		-0.021		0.201***		-0.161***
		(-1.29)		(14.25)		(-9.47)
Fixed Effects	F	F	F	F	F	F
Obs	14,716	14,592	29,071	28,895	29,314	29,133
Adj R2	0.555	0.574	0.275	0.340	0.533	0.558

This table presents estimates from the firm-level quarterly OLS regression: Financing = α +  β1Cash*After + β2After + γControl + θ + ε. Measures of financing activity: changes in short-term debt in columns (1) and (2); changes in long-term debt in columns (3) and (4); equity issuance in columns (5) and (6). Other variables are defined in Appendix A. F denotes firm fixed effects and I × T denotes industry by year-quarter fixed effects. Standard errors are corrected for heteroskedasticity and clustered at the firm level, t-statistics are reported in parenthesis, with ∗, ∗∗, and ∗∗∗ represent statistical significance at the 10%, 5%, and 1% level, respectively.

Payout policy

In the final set of the analysis, I turn to firms’ payout policy. Because a firm's payout policy directly affects the distribution of profits to its shareholders, this finding provides implications to the shareholder values during the COVID-19 crisis. I find the after dummy is negative and significant across total payout, dividend, and repurchase measures in Table 5 . The coefficients of the interaction terms are positive and significant at the 1% level, confirming the precautionary role of cash holdings. My finding on a firm's payout policy is especially relevant to a contemporaneous study by (Krieger et al., 2020). While the study documents that firms cut dividends to preserve financial flexibility in response to the COVID-19 crisis, I show that firms with an abundant cash reserve do not experience such a drop. By distributing profits to shareholders and mitigating the negative market reaction associated with dividends reduction, the analysis above indicates the payout policy can be a specific channel through which cash holdings increase shareholder values.

Table 5

Pre Covid-19 cash holdings and payout.

Specification:	Payout		Repurchase		Dividend
	(1)	(2)	(3)	(4)	(5)	(6)
After	-0.003***	-0.002***	-0.002***	-0.002***	-0.001***	-0.001***
	(-13.42)	(-12.57)	(-12.25)	(-11.85)	(-8.60)	(-7.55)
After × Cash holdings	0.002***	0.003***	0.001**	0.002***	0.001***	0.001***
	(3.98)	(4.47)	(2.46)	(3.34)	(5.28)	(5.12)
Size		-0.003***		-0.002***		-0.001***
		(-6.88)		(-6.32)		(-3.37)
Q		0.001***		0.000		0.000***
		(3.98)		(0.68)		(7.04)
Cash Flow		0.003***		0.002***		0.001***
		(5.92)		(4.51)		(5.61)
Leverage		-0.002**		-0.002***		-0.001*
		(-2.23)		(-2.83)		(-1.88)
Fixed Effects	F	F	F	F	F	F
Obs	28,970	28,789	27,622	27,455	28,742	28,564
Adj R2	0.470	0.475	0.401	0.402	0.667	0.671

This table presents estimates from the firm-level quarterly OLS regression: Payout = α +  β1Cash*After + β2After + γControl + θ + ε. Measures of payout: payout in columns (1) and (2); repurchase in columns (3) and (4); dividends in columns (5) and (6). Other variables are defined in Appendix A. F denotes firm fixed effects and I × T denotes industry by year-quarter fixed effects. Standard errors are corrected for heteroskedasticity and clustered at the firm level, t-statistics are reported in parenthesis, with ∗, ∗∗, and ∗∗∗ represent statistical significance at the 10%, 5%, and 1% level, respectively.

Conclusion

The severity and economic costs of the ongoing COVID-19 crisis is unprecedented, and the U.S. government stimulus spending is already twice as much as it was during the 2007-2008 financial crisis. In this paper, I explore the effect of the COVID-19 pandemic on firms’ performance. I find that corporate investment, profitability, financing activity, and payout policy are negatively affected by the pandemic, consistent with prior findings of the related literature. Moreover, I show a strong positive relation between the corporate pre-COVID cash reserve and firm performance, suggesting the precautionary motive behind corporate cash holdings.

This paper sheds light on the importance of risk management in a corporate setting and reveals the value-enhancing perspective of the function. My results imply that the positive effect of cash holdings on firm performance is more pronounced when the overall market condition is bad. The hedging benefit could explain the time-series pattern of increased corporate cash holdings. Contrary to the “dark side” – managerial rent-seeking view to the corporate cash holdings – I demonstrate that the advantage of the cash reserve may not be well understood if the unforeseen COVID-19 pandemic has not happened. This highlights the difficulty in estimating the optimal cash level when rare market shocks are incorporated.

Cash reserve and control variables
Cash holdings	Cash and short-term investment scaled by total assets. Measured once per firm at the end of fourth quarter of 2018.
Size	Natural logarithm of the total assets.
Q	Market value of assets scaled by book value of assets.
Cash Flow	Income before depreciation scaled by total assets.
Leverage	Long-term debt scaled by total assets.
Investment
Capex	Capital expenditure scaled by total assets.
R&D	Research and development expense scaled by total assets.
SG&A	Selling, general, and administrative expenses scaled by total assets.
NWC	Net working capital excluding cash scaled by total assets.
Inventory	Total inventories scaled by total assets.
Performance
ROE	Net income scaled by common equity.
ROA	Operating income before depreciation scaled by total assets.
Q	Same as above.
Asset_Growth	One year change in total assets scaled by lag total assets.
Market Share	Total sales scaled by the sum of Fama French 48 industry's total sales in that year.
Financing
∆ST_debt	Change in short-term debt scaled by total assets.
∆LT_debt	Change in long-term debt scaled by total assets.
Equity Issuance	Equity issuance scaled by total assets.
Payout
Total payout	Sum of share repurchases and dividends scaled by total assets.
Repurchases	Share repurchases scaled by total assets.
Dividends	Dividends scaled by total assets.

Panel A: Investment	Capex	R&D	SG&A	NWC	Inventory
	(1)	(2)	(3)	(4)	(5)
After × Cash holdings	0.154***	-0.410**	0.427*	5.064***	1.376**
	(3.38)	(-1.97)	(1.85)	(2.77)	(2.89)
Panel B: Performance	ROE	ROA	Q	Asset_growth	Market Share
	(1)	(2)	(3)	(4)	(5)
After × Cash holdings	0.050**	0.011***	0.587***	0.021*	0.000**
	(2.00)	(2.91)	(3.83)	(1.94)	(2.39)
Panel C: Financing	∆ST_debt	∆LT_debt	Equity Issuance
	(1)	(2)	(3)
After × Cash holdings	0.013**	-0.008**	-0.045***
	(1.96)	(-2.09)	(-4.27)
Panel D: Payout	Payout	Repurchase	Dividend
	(1)	(2)	(3)
After × Cash holdings	0.002**	0.001*	0.001**
	(2.47)	(2.08)	(2.21)
Fixed Effects	F, I × T	F, I × T	F, I × T	F, I × T	F, I × T
Control Variables	Yes	Yes	Yes	Yes	Yes