key: cord-1034483-qv20y91f
authors: Hoang, Khanh; Arif, Muhammad; Nguyen, Cuong
title: Corporate investment and government policy during the COVID-19 crisis
date: 2022-03-18
journal: International Review of Economics & Finance
DOI: 10.1016/j.iref.2022.03.005
sha: 8bf50f5aa861cfc406fbaa1a8ae028e777fa1cbf
doc_id: 1034483
cord_uid: qv20y91f

We investigate the impact of the US government response to the COVID-19 pandemic, including stringent social measures and economic support packages, on corporate investment. The empirical results show that despite the overall decreased investment due to the economic impact of the pandemic, the government response to COVID-19 and economic supports have a positive effect on corporate investment after subtracting the impact of the pandemic on firm-level investment. We find that the impact of economic support packages on corporate investment is stronger than that of health containment policies. Further analyses show that the effect is weak in firms with higher levels of political risk and investment irreversibility, while being more pronounced in firms with higher technology intensity. Our findings provide fresh insights into the firms’ reaction to the government policies during the pandemic and suggest that both social measures and economic support are vital to restoring corporate investment as well as the economic recovery process.

The COVID-19 pandemic has sparked an unprecedented crisis around the world with respect to the fact that the virus is highly contagious and deadly. As responses to the pandemic, governments around the world have undertaken strong measures to prevent the spread of the novel coronavirus, including social distancing, school closure, business closure, limited public gathering, contact tracing, mass testing, travel restrictions, quarantine, and lockdowns. The effectiveness of these containment measures against the outbreaks of infectious diseases is confirmed internationally. However, the public is witnessing the economic impacts of COVID-19 as the long shadow of an economic recession is expanding across the globe.

As one of the countries with the largest number of infected cases and deaths by COVID-19, the United States (US) is experiencing one of the most devastating public health and economic crises ever in history. The former US President, Donald Trump was criticized for being slow to absorb the scale of the COVID-19 risk and to react appropriately 2 . A survey from

The University of Chicago Harris School of Public Policy in September 2020 shows that 78 percent of Americans blame their government for the COVID-19 crisis 3 in the US. Ironically, the outbreak of COVID-19 in the US has occurred and become severe during the presidential election year 2020. This coincidence induces even more uncertainty from three dimensions: political, public health, and economic crises, thus driving US firms' business risk to an extreme level. As political uncertainty is higher during election years, firms with higher degrees of political risk are more likely to retrench their hiring and corporate investment to protect themselves from increased uncertainty (Hassan et al., 2019; Campello et al., 2020) . Similarly, under economic uncertainty and changes in macroeconomic policy, firms tend to delay their investment until they find it a safer time to do so (Kang et al., 2014; Gulen and Ion, 2016; Kim and Kung, 2017) . Mass reduction in investment at the firm-level would negatively affect job creation and economic growth (Adelino et al., 2017; Coibion et al., 2020) , thus stimulating policymakers to react to secure social welfare and economic outlook of the nation during this difficult time. Because disasters at the global scale and severity of the COVID-19 pandemic have not been seen in a century (Goodell, 2020) , it is a unique natural experiment to study how governments and businesses respond to an exogenous negative shock that is blended with a politically-complex business environment. We find it compelling to investigate how the government responses to the COVID-19-induced crisis influence the investment decisionmaking of US firms. Are the government responses to the COVID-19 crisis and economic support policies effective in alleviating the negative impact of the health crisis? How investment revertability, technology intensity and firm-level political risk play a role in moderating the influence of the government policies on firms' investment? This paper aims to answer these questions.

The US government implements a wide range of economic countermeasures to the COVID-19 crisis, including the USD 2.2 trillion fiscal stimulus bill approved by the House Democrats in early October 2020 4 , and the new economic stimulus offer of USD 1.8 trillion made by the US President Donald Trump on October 9, 2020 5 . Before that, the government has declared numerous economic stimulus packages and immediate actions to support its citizens and the economy during the outbreak, such as interest rates cut 6 , debt contract relief, and several fiscal aids for the economy 7 . The sizes of the quantitative easing packages declared by the US government in 2020 surpassed the total value of all quantitative easing since the Global Financial Crisis 2008, demonstrating the government's acknowledgment of the economic severity of the coronavirus crisis. As most US states reopened from the lockdown in mid-2020, economic activities in the US were expected to resume in a new normal, despite the number of new positive cases and deaths by COVID-19 are still on the rise.

We take into consideration the corporates' reactions to the COVID-19 news, including news on the new confirmed COVID-19 cases and deaths by the coronavirus in the US. There is a well-known fact that the COVID-19 pandemic has caused large-scale disruption to financial systems and economic activities around the world, especially in countries heavily affected by the pandemic. Recent studies investigating the economic impact of COVID-19 usually employ an event study setting that uses a set of dummy variables to represent the events 4 See https://www.ft.com/content/06a7a77e-7b96-43e3-9c63-4967a03ad997 5 See https://www.nytimes.com/2020/10/09/us/politics/trump-covid-stimulus-pelosi-republicans.html 6 See https://www.theguardian.com/business/2020/mar/15/federal-reserve-cuts-interest-rates-near-zero-prop-upus-economy-coronavirus 7 See https://www.wsj.com/articles/rbc-lays-out-700-billion-stimulus-effort-for-u-s-economy-11602183581 J o u r n a l P r e -p r o o f of the coronavirus outbreak and the windows around it. This approach is effective if they can well control for the government response to the disease and other confounding factors, otherwise, the findings may be contaminated. Following this conjecture, we separate the impact of the government response to COVID-19 on corporate investment from the overall impact of COVID-19, using the approach of Gulen and Ion (2016) from their influential study in policy uncertainty literature. We obtain a cleaner measure of corporate investment after accounting for the impact of COVID-19, then use it to examine how corporate investment reacts to the government's actions and policies against the COVID-19 crisis.

The empirical results yield four significant findings. First, we document, on average, a positive reaction of corporate investment to the government responses to COVID-19. This suggests that the government's stringency in combating COVID-19 and the economic supports have built up a certain degree of confidence of US firms to invest during the crisis. Although the general level of firm-level investment of US firms is still lower than the pre-COVID-19 periods 8 , our finding implies the US firms are more inclined to invest when the government employs more responsive policies, including stringent coronavirus containment measures and economic supports. We attribute this to the higher expectation of businesses for economic recovery given the government policies and economic supports during the outbreak. We suggest that government policies to response to COVID-19 help alleviate the adverse economic impact of the pandemic on corporate investment. The finding remains qualitatively unchanged after a battery of sensitivity tests.

Second, firms in industries with higher degrees of investment irreversibility generally invest less than their counterparts in response to the government's efforts to fight the pandemic. This is in line with those of Gulen and Ion (2016) and Kim and Kung (2017) that firms with more investment irreversibility tend to be more cautious in making investment decisions under increased uncertainty at the macro-level. As the payoff increases due to uncertainty induced by the COVID-19 pandemic, the opportunity cost of investment increases, thus firms have to consider the option to delay investment until more is known (Guiso and Parigi, 1999; Gulen and Ion, 2016) . Our results suggest strong government response to the COVID-19 crisis plays a crucial role in encouraging corporate investment in both more and less irreversible firms 8 Except for the Healthcare, Medical Equipment and Drug industries as shown in our mean difference tests in Table 3 . The industry classification follows the categorization of Fama-French industries.

J o u r n a l P r e -p r o o f during the pandemic. However, we find that the impact varies cross-sectionally across the two groups.

Third, we find that firms with higher technology intensity react more positively to government policies during the pandemic. Specifically, firms with higher degrees of technology intensity invest more as a response to government policies and economic supports regarding COVID-19 crisis. The intuition of this finding emerges from the argument that firms with higher technology intensity likely face less impact from uncertainty (Czarnitzki and Toole, 2011; Vo and Le, 2017) , so they can adapt to the new economic situation faster and have more flexibility to adjust than their counterparts. This finding suggests the importance of technologies in firms during the COVID-19 crisis, and possibly other similar uncertainty shocks.

Fourth, the analysis shows that the positive effect of the government responses to COVID-19 and economic supports on corporate investment is more pronounced in firms with lower political risk and vice versa. Intuitively, firms might retrench their investment during election years because higher political risk has a negative impact on investment at the firmlevel (Julio and Yook, 2012; Hassan et al., 2019) . The US election year 2020 exhibits a high degree of political uncertainty due to its unique situation surrounded by the COVID-19 crisis, the Black Lives Matter protests, the Capitol riots, and the ongoing US-China trade war. Therefore, it is understandable that firms with more political risk to react less positively to the government's policies during the pandemic compared to other firms.

This study is related to the recent literature in economics and finance studying the impact of exogenous events on corporate investment (Julio and Yook, 2012; Kang et al., 2014; Gulen and Ion, 2016) , however, extends to the analysis of how the government response to an extreme negative shock would influence the incentives to invest of firms. Our study has several significant contributions to the literature on government policy during extreme uncertainty.

First, we show the importance of governments taking strong measures, both socially and economically, to combat the COVID-19 crisis and alleviate the adverse impact of the pandemic on corporate investment decision-making. By using a measure of corporate investment separated from the impact of COVID-19 health news, we demonstrate the positive effect of the government stringency and economic supports in the fight against COVID-19 on corporate investment, despite the fact that the US government underreacted to the COVID-19 risk in early 2020. Second, our study provides evidence of how the impact varies in the cross-section, J o u r n a l P r e -p r o o f thus providing a new understanding of the mechanisms by which the government response influence corporate investment regarding the coronavirus crisis. Our findings enrich the literature on the economic impacts of the pandemic worldwide. Third, we show how irreversible investment, technology intensity, and firm-level political risk are economically important to influence the impact of the government response to the pandemic during heightened political and economic uncertainty in the US. Fourth, our study provides practical implications for corporate strategy, policymakers, and governments across the world in decision-making or considering different measures and supports to combat the new waves of the virus or a similar pandemic to come.

The paper is organised as follows. Section 2 provides the motivation and literature review. Section 3 discusses the variables, research models, and data. Section 4 presents the empirical results and discussion. Section 5 concludes the study.

The economic impact of COVID-19 is a strand of literature that attracts great attention. Under the impact of the pandemic, economies worldwide experience sudden negative demand shocks (Goodell, 2020; Hassan et al., 2020) , sharp drop in oil product prices (Mensi et al., 2020; Rajput et al., 2021 ), surges in unemployment (Campello et al., 2020 Coibion et al., 2020) , and social lockdown that heavily affected economic activities. At the firm-level, businesses experience sharp shortfalls in stock prices and revenues (Fahlenbrach et al., 2020) , exhausting corporate cash reserves (Vito and Gómez, 2020), and higher bankruptcy risk in large corporations (Wang et al., 2020) . The impact seems to vary cross-sectionally, with some firms have a stronger immunity to the pandemic compared to others (Hoang et al., 2021; Ding et al., 2021) .

Not until the COVID-19 pandemic, the fact that uncertainty hinders corporate investment has been well investigated for several decades. For instances, Lucas and Prescott (1971 ), Abel (1983 ), Pindyck (1993 ), Dixit (1995 , Guiso and Parigi (1999) We study corporate investment under the COVID-19 crisis to address the economic impact of the devastating pandemic and the consequences of the government response to the pandemic. Amid the extreme uncertainty induced by the COVID-19 crisis during the US election year 2020, how the US government reacts to the pandemic is of utmost importance as it forms the future outlook of the economy, both domestically and internationally. Strong and quick responses to protect the citizens and the economy will build public trust and boost confidence in the prospects of the economy. Prior studies show that higher expected economic growth is associated with more financial investment and corporate investment opportunities (Chen, 1991; Gulen and Ion, 2016) . If firms believe in the long-term effectiveness of the government policy regarding the health crisis, then we would expect them to invest and adapt to the new state of the market during COVID-19. If firms are afraid that the worst of the crisis is yet to come, then they would continue to delay investment or even resort to disinvestment.

In this scenario, the efforts of the government seem to be insufficient to encourage corporate investment. Based on this conjecture, we expect a positive impact of the government response to the COVID-19 crisis on corporate investment. Our expectation is in line with the market reacts positively to government interventions during the COVID-19 pandemic (Ashraf, 2020).

The government responses to the COVID-19 crisis (i.e., health containment and economic support policies) have a positive impact on corporate investment during the pandemic.

There are some potential channels through which COVID-19-induced uncertainty affect corporate investment. The first channel is investment irreversibility, as addressed by Dixit (1995) , Gulen and Ion (2016) , and Kim and Kung (2017) . The intuition is that corporate managers can delay investment projects under uncertainty if those particular investment projects can be delayed, and more importantly if those investments are difficult to be reversed when needed to. The reversibility of investment lies in its asset liquidation value. Investments with lower asset liquidation values under uncertainty are considered less reversible and associated with more costly asset redeployment (Kim and Kung, 2017) . According to Almeida 9 For example, the Gulf War (1991), the 11/9/2001 terrorist attack in the US, or the Brexit referendum in 2016.

J o u r n a l P r e -p r o o f and Campello (2007) and Gulen and Ion (2016) , firms in highly cyclical industries likely have higher degrees of investment irreversibility. Compared to other firms, such firms are more vulnerable to negative demand shocks. As such, they would have lower asset liquidation values, as the best potential buyers of these assets (e.g., their peers or firms in the same industry) and also suffer from the same demand shocks. Therefore, investment irreversibility might be a channel through which uncertainty affects corporate investment. As such, firms with higher degrees of investment irreversibility are likely to invest less than their counterparts under the impact of COVID-19-induced uncertainty. This understanding forms our second research hypothesis as follows:

Hypothesis 2: The impact of government responses to the COVID-19 crisis on corporate investment is weaker for firms with higher degrees of investment irreversibility, and vice versa.

The second channel is technological intensity. Intuitively, firms with stronger technological foundations may adapt better than their counterpart in shifting business models during the pandemic. Literature on technology intensity show that higher corporate technology intensity associates with better and less volatile future performance (Pandit et al., 2011) , less negative impacts of market uncertainty (Czarnitzki and Toole, 2011), higher productivity (Baumann and Kritikos, 2016), more successful equity financing (Aghion et al., 2004) , and higher firm value (Greenhalgh and Rogers, 2006) . Therefore, firms tend to invest more in research and development activities as a preemptive strategy when they face higher uncertainty (Vo and Le, 2017) . The COVID-19 pandemic changes the world in an unprecedented way (Goodell, 2020) and undermine corporate immunity like no previous crises ever did (Cheema-Fox et al., 2021; Hoang et al., 2021) . However, amid the long COVID-19 macroeconomic shock stand the role of technology in enhancing corporate resilience (Bai et al., 2021) . The use of innovative technologies seems to have a positive impact on corporate immunity during the pandemic (Papadopoulos et al., 2020). As the COVID-19 crisis provides opportunities to firms to innovate (Seetharaman, 2020), firms with higher technology intensity tend to adapt to the new economic situation faster and have more flexibility to adjust than other firms, all else equal. Therefore, we predict that: Because the outcome of the presidential election is uncertain until November, firms that are sensitive to political risk or dependent on government spending are likely to delay investment (Gulen and Ion, 2016; Hassan et al., 2019) . On the other hand, firms with less political sensitivity may find it costly to invest under uncertainty, as the cost of capital increases during heightened uncertainty at the macro-level (Xu, 2020) . The year 2020 is undoubtedly the year of uncertainty for US firms, both politically and epidemically, thus it amplifies the impact of firm-specific political risk and may discourage firms from making investment decisions.

Following this conjecture, we propose the fourth research hypothesis as follows:

Hypothesis 4: The impact of government responses to the COVID-19 crisis on corporate investment is weaker for firms with higher degrees of political risk, and vice versa.

As there is no data on the economic damage of the COVID-19 pandemic available anywhere at the time this research is conducted, we use the numbers of infected cases and deaths by COVID-19 as the direct measures of how severe the pandemic is. The daily data is then aggregated into quarterly data to match with the frequency of corporate data. We take the natural logarithm of one plus the number of new confirmed COVID-19 cases in the US during a quarter as the proxy of how contagious the disease is . Similarly, we also add one to the number of new deaths by COVID-19 and then log-transform the total to obtain a proxy of how deadly the virus is (COVID-19 DEATHS). Since the first case of COVID-19 in J o u r n a l P r e -p r o o f the United States was reported on January 21, 2020, we treat all daily data points before that date with the value of zero for both COVID-19 CASES and COVID-19 DEATHS.

To measure the stringency in the US government response to COVID-19, we use the Stringency Index and the Economic Support Index proposed by Hale et al. (2020) . The two indexes were constructed to quantify the governments' responses to the outbreaks of the novel coronavirus from two dimensions, including the stringency of the responses to prevent the spread of COVID-19 (Stringency Index) and the support to the economy (Economic Support Index).

Specifically, the Stringency Index (Hale et al., 2020) consists of eight indicators of social distancing and lockdown, including school closure, workplace closure, cancelling public events, restrictions on public gathering, public transport closure, stay-at-home requirements, domestic movement restrictions, and international travel restrictions. To fit the data frequency with corporate data, we aggregate the daily Stringency Index into a quarterly measure by taking the average of the daily Stringency Index, then take the natural logarithm of one plus the quarterly index as the variable-of-interest in our study (GSTRINGENCY) . GSTRINGENCY is used to investigate the impact of the pandemic social distancing and lockdown measures on corporate investment of US firms. The Economic Support Index (Hale et al., 2020) indicates whether the government provides income support and debt or contract relief as the policies to support households and the economy. Similar to the case of the Stringency Index, we also compute the quarterly Economic Support Index and then log-transform it after adding one. The new variable (GESI) is to evaluate the impact of the government's economic support on corporate investment of US firms. As there were no COVID-19 cases confirmed in the US before January 21, 2020, and the US government did not have any economic support and health containment measures implemented before January 21, 2020, so Hale et al. (2020) assign all the data points of GSTRINGENCY and GESI prior to that date in January 2020 with values of zero. We further apply this treatment applies to the government response data (GSTRINGENCY and GESI) for the whole study period. Therefore, the value zero of GSTRINGENCY and GESI indicate the zero-COVID-19 period in the US.

Following Gulen and Ion (2016) and Chen and Wang (2019), the common corporate investment proxy is the ratio of capital expenditure on total assets (CAPEX). However, we doubt that CAPEX is a valid measure to explore the impact of the government response to the J o u r n a l P r e -p r o o f COVID-19 crisis on corporate investment for two reasons. First, the effect of government response on corporate investment might be contaminated by the impact of the COVID-19 pandemic if we use CAPEX in our model. As the negative economic impact of the pandemic on corporate operations is well acknowledged (Baker et al., 2020; Campello et al., 2020; Goodell, 2020) , the effect of the government response is likely overwhelmed by the disastrous impact of the virus on the economy as a whole. We assume that firms would react to the pandemic first, then taking the government policy into consideration for investment decisionmaking. This assumption is reliable because the social distancing, lockdown, and economic support policy were only implemented after the spread of COVID-19 went out of control. As we expected strong government responses have a positive effect on corporate investment, such an effect should be submerged in the destructive impact of COVID-19. Consequently, there might be a potential serious measurement error existing in this study if we use the raw CAPEX variable to investigate the relationship between the government response and corporate investment. Second, the devastating impact of the COVID-19 is likely a negative driver of corporate investment in the United States via various known and unknown mechanisms, thus a general measure of corporate investment may inadvertently capture the overall economic impact of COVID-19. Using CAPEX as the dependent variable to investigate the impact of government policies on corporate investment during the pandemic, which are undoubtedly determined by the developments of COVID-19 outbreak, would expose the empirical model to a serious endogeneity problem.

We borrow the idea from Gulen and Ion (2016) to overcome this endogeneity issue. In their study, Gulen and Ion (2016) got rid of the contaminating component of the US policy uncertainty variable by extracting the elements of the US policy uncertainty orthogonal to the Canadian policy uncertainty. Their rationale behind this approach is that a policy shock in Canada might also affect the US because of the tight link between the two economies. The authors run a regression of the US policy uncertainty variable on that of Canada, controlling for several confounding factors at the macro-level. The residual from this regression is defined as the measure of policy uncertainty free of uncertainty shocks affecting both the US and Canada. Based on this method, we conduct a regression of CAPEX on the COVID-19 variables as follows:

where , is corporate investment proxy of firm i during quarter-year t. and represent the firm and calendar quarter fixed effects. In this setting, the residual , is the measure of corporate investment after accounting for the impact of the COVID-19 outbreak.

In other words, , should have been purged of the COVID-19 health news affecting investment decision-making of US firms. By applying this procedure, we obtain a measure of corporate investment after extracting the direct impact of COVID-19, namely Res_CAPEX. We use this new variable to investigate how corporate investment under the COVID-19 crisis reacts to the government's policies.

For robustness check, we compute a different specification of CAPEX using the data of COVID-19 confirmed cases and deaths at the state level instead of the total confirmed cases and deaths as in Equation (1). Similarly, we obtain the residual from the regression of Equation

(1) using COVID-19 data at the state-level (S_Res_CAPEX). Moreover, while capital expenditure is a common proxy for capital investment in the finance literature, non-cash asset growth is referred to as the total investment of a firm (McLane and Zhao, 2014). We apply the same procedure and obtain the third proxy of corporate investment after subtracting the impact of COVID-19 from non-cash assets investment, namely Res_NCAINV.

We use the following model to estimate the impact of government response to COVID-19 on corporate investment:

where _ , is the corporate investment proxy of firm i during quarter-year t as discussed in the sub-section 3.1.3. Table   1 . We also control for firm and calendar quarter fixed-effects ( and , respectively) to control for potential confounding factors at firm-level and seasonality following Gulen and Ion (2016).

Standard errors are double-clustered by firm and quarter-year to alleviate the concerns of heteroskedasticity and serial-correlation in our regression.

We collect quarterly financial data of US firms from the Bloomberg database from 2002:Q1 to 2020:Q4. The use of the long data period is to capture the average investment at the firm-level across a longer period of time and avoid the bias arising from short-horizon data samples. (2012) show that political uncertainty in election years affect corporate investment cycles. Conversely, MTB, CASH, PRISK, and NWC negatively correlate with firm-level investments.

Res_NCAINV. Figure 1 Table   4 ).

Column 5-8, Table 4 More importantly, the significant coefficients of GESI are higher than those of GSTRINGENCY, indicating that economic support policies have a stronger impact on corporate investment of US firms relative to that of health containment and social lockdown policies.

[Please insert Table 4 here]

Findings from the regression of the control variables are also worth mentioning.

Looking at the results of full model regression specifications in columns 4 and 8, Table 4 , it is notable that firm size (SIZE), financial leverage (LEVERAGE), cash holdings (CASH), and firm-level political risk (PRISK) exhibit a significant negative impact on corporate investment regardless of the corporate investment measured used, indicating that larger firms, firms that maintain high liquidity, firms using a higher portion of debt, and firms with higher political risk invest less compared to their counterparts. Profitability (ROA), sales growth ( Overall, the results from Table 4 

We conduct several tests to confirm the robustness of our empirical results. First, we use alternative variable measurements to reperform the analysis, including using S_Res_CAPEX and Res_NCAINV as the dependent variable instead of Res_CAPEX. We also use the size of quantitative easing (in billion US dollars) as an alternative for GESI in Model

(2). Such unconventional monetary policies have been implemented not once in the US, but four times in total including the economic support packages during the COVID-10 pandemic, thus fitting well in our data. Second, we use different study periods to re-estimate the baseline model. We select four alternative periods: the post-H1N1 period from 2011:Q1 to 2020:Q4, the Trump administration period from 2017:Q1 to 2020:Q4, and the short period from 2019:Q1

to 2020:Q4 to test whether our empirical results are sensitive to choices of the study period.

Next, to draw a cleaner sample that is free from market-level trends arising from other J o u r n a l P r e -p r o o f uncertainty shocks, we independently exclude the Global Financial Crisis (2008-2010) period and re-estimate the models.

Third, another concern about the economic impact of COVID-19 is that the pandemic has an unequal impact on various sectors, with certain sectors suffering more or less than the others. During the COVID-19 crisis, firms from certain sectors may receive government aids earlier than others and thus recovering at faster pace. Therefore, the impact of government responses to the pandemic on corporate investment might vary in the cross-section of sectors.

We use a simple method to control for these variations. We generate a fixed effect recording the firm's GICS sector 11 and substitute the firm-fixed effect with it, then re-estimate Model (2) to test whether our findings remain after controlling for sector-level variations.

Furthermore, we employ the Driscoll-Kraay to control for the potential cross-sectional dependency caused by the simultaneous impact of the COVID-19 pandemic on all aspects and sectors of the economy. Cross-sectional dependence in corporate investment could be a potential source of estimation error given the unprecedented financial and economic contagion observed during the pandemic crisis (Akhtaruzzaman et al., 2020). To test whether our findings are sensitive to the autocorrelation of residuals, we re-estimate Model (2) using the Newey-West estimator. The estimation results are reported in Table 5 . In summary, the results of robustness tests are in line with our baseline estimations, which suggest that variable measurements, sampling periods, and regression methods do not determine our findings regarding the impact of government response on corporate investment.

Instead, our results confirm that there exists a positive net effect of government response to COVID-19 on corporate investment.

To provide a better understanding of how the government response to the pandemic could motivate corporate investment, this section seeks to explore the channels of the relationship between government responses to COVID-19 and corporate investment via three channels: investment irreversibility, technology intensity and political risk.

In this section, we investigate the role of investment irreversibility in the newfound relationship between government policy during COVID-19 and corporate investment. To measure investment irreversibility, we follow the approach of Almeda and Campello (2007) and Gulen and Ion (2016) to calculate the correlation between each firm's quarterly revenues and the US's Gross National Product (GNP) for each quarter of our study period. Consequently, we compute the industry-average correlation for each 2-digit SIC industry from the firm-level correlations. The rationale for this measure is that firms in highly cyclical industries likely experience lower asset liquidation values under macro-level negative demand shocks (Shleifer and Vishny, 1992), implying that their investments are less reversible under uncertainty.

Hence, we define firms with higher investment irreversibility are firms in industries with revenues-GNP correlations lower than the cross-sectional median, and vice versa. Table 6 presents the results that show the impact of government response on corporate investment in the case of firms with higher/lower investment irreversibility.

[Please insert Table 6 here]

The results reveal that corporate investment of firms with higher investment irreversibility react less positively to the government's policies during COVID-19, relative to their counterparts. Specifically, the coefficients of GSTRINGENCY (columns 1-2) and GESI we attribute this type of corporate behaviour to the payoff between the cost associated with redeploying assets and investment opportunities.

In summary, the results support Hypothesis 2 and show evidence of cross-sectional variations in the impact of government response to COVID-19 on corporate investment of US firms.

In this section, we test Hypothesis 3 to find out whether the positive impact of government responses to COVID-19 on corporate investment is stronger for firms with higher technology intensity. We follow previous studies in the literature (Bah and Dumontier, 2001; Aghion et al., 2004; Padgett and Galan, 2010; Honoré et al., 2015) to measure technology intensity using R&D intensity (R&D_INTENSITY) that equals corporate research and development expenditure scaled by total sales, however, both in the preceding period. We sort firms by quarter-year and by the median of R&D_INTENSITY, then re-estimate Model (2) on the subsamples of cross-sectionally high-and low-R&D intensity. Following our hypothesis development, we expect the impact found in the primary analysis to be more pronounced in firms with higher R&D_INTENSITY and vice versa. To summarize, our empirical evidence suggests that firms with higher technology intensity react more positively to government response to COVID-19 in terms of investment.

This finding suggests the importance of technologies in firms under the economic impact of the COVID-19 unprecedented shock. 12 We obtain similar results when using Res_NCAINV as the dependent variable in Model (2).

Given that 2020 is the presidential election year in the US, accounting for political risk becomes even more critical. To measure political risk, we use the firm-level political risk (PRISK)

proposed by Hassan et al. (2019) . The measure is constructed using textual analysis of quarterly earnings conference call transcripts of US firms that show the management's view on their firms' risks associated with politics. We divide firms into higher and lower political risk firms to ascertain the impact of political risk on the association between government response and corporate investment. We use the median of PRISK as the cut-off point to define higher and lower political risk firms. The high-and low-political risk firm groups are defined for each quarter-year. We then re-perform the analysis using the subsamples of high-and low-political risk firms. In line with our proposition, we expect the impact of government response to COVID-19 on corporate investment to be weaker in the high-political risk firm groups than in low-political risk firm groups. The estimations results are in Table 8 .

[Please insert Table 8 here] Table 8 shows that the government response in the shape of stringency measures and the economic support exhibit a more pronounced effect on corporate investment of firms with a lower level of political risk compared to the firms with a higher degree of political risk. A comparison of coefficients indicates that the impact of government response on corporate investment for the firms with a lower degree of political risk is higher than that of firms with a higher degree of political risk (0.0035 and 0.0050 compared to 0.0022 and 0.0030, respectively). 13 Our findings corroborate with the existing literature that indicates that higher degrees of political risk are more likely to retrench their hiring and corporate investment to protect themselves from increased uncertainty (Julio and Yook, 2012; Hassan et al., 2019; Campello et al., 2020) .

To summarise, the empirical findings support our Hypothesis 4 that the positive impact of government policy response to COVID-19 on corporate investment seems weaker in firms with higher political risk, suggesting that political risk may drive firms to react less positively to the government policies during the COVID-19 crisis.

13 These results are robust for even when we use Res_NCAINV as the investment proxy.

J o u r n a l P r e -p r o o f

The unprecedented and devastating impact of COVID-19 on the US public health and economy has driven us to the question of whether the government response, in terms of both social stringency and economic relief bills, have an impact on corporate investment during this oncein-100-year pandemic. The fact that the US health response to the pandemic is below expectation in the early stage of the COVID-19 outbreak, and that the economic response has prevailed with the economy grew at a record pace in the third quarter -increasing 7.4% over the third quarter and at a 33.1% annual rate 14 could exhibit any hint to how corporate investment reacts to the development of the pandemic. We provide the answer to the questions by using data on the US government response to the pandemic and show that the government responses to COVID-19, especially the economic support policies, have a positive effect on corporate investment.

The special feature in the US is that the pandemic occurs in the same year as the presidential election, which shows a higher level of political risk. Our further analysis reveals that the effect of the government response in terms of the stringency measures and the economic support packages on corporate investment is more pronounced on firms with a lower level of political risk. As firms with higher political risk have higher levels of risk aversion under increased uncertainty, they react more cautiously to the government's policies compared to other firms. Furthermore, regarding investment irreversibility, the empirical results indicate that corporate investment of firms with higher investment irreversibility react less positively to government policies during COVID-19 relative to their counterparts. Hence, those firms are less inclined to make long-term investments due to the prevailing uncertainties under the pandemic. Moreover, we emphasize the importance of technology intensity in motivating corporate investment as a response to government supports during COVID-19. This study complements our understanding of corporate behaviour during the COVID-19 pandemic as a unique natural experiment. Analysis of corporate investment at the time of the pandemic provides us with a straightforward evaluation of the effectiveness of the government's policy and actions in this uncertain time.

Finally, the findings of this study suggest that both stringent measures and economic support stimulus packages, at the same time, by a government to counter the economic and 14 See https://www.wsj.com/articles/us-gdp-third-quarter-2020-11603908566 J o u r n a l P r e -p r o o f social effects of COVID-19 could increase the businesses' confidence to invest and accelerate the pace of economic recovery. This provides practical implications to policymakers and governments who might face dilemmas during the decision-making process to control the next waves of COVID-19 or a similar pandemic in the future. The findings also have implications for corporate R&D as a preemptive strategy to be well prepared for future uncertainty shocks. Total investment in non-cash assets variable that equals changes in non-cash assets scaled by total assets. Bloomberg database COVID-19 CASES Natural logarithm of one plus the total accumulated number of COVID-19 cases in the United States.

Natural logarithm of one plus the total accumulated number of deaths caused by COVID-19 in the United States.

The residual of the regression of CAPEX on COVID-19 CASES and COVID-19 DEATHS at country-level, control for firm and quarter fixed effects.

The residual of the regression of CAPEX on COVID-19 CASES and COVID-19 DEATHS at state-level, control for firm and quarter fixed effects.

The residual of the regression of NCAINV on COVID-19 CASES and COVID-19 DEATHS at country-level, control for firm and quarter fixed effects.

The natural logarithm of the Stringency Index (Hale et al., 2020) recording the strictness of lockdown policies of the United States in response to the COVID-19 pandemic.

Ox-CGRT database This table reports the summary statistics and correlation matrix of variables. ***, **, and * denotes statistical significance at 1%, 5%, and 10%, respectively. This table reports the mean-difference test results of CAPEX between the COVID-19 period and the pre-COVID-19 periods using full sample and a sub-sample consisting of Q1 and Q2 Table 1 . Standard errors are clustered by firm and quarter. Robust t-statistics are reported in parentheses. ***, **, and * denotes statistical significance at 1%, 5%, and 10%, respectively. Robust t-statistics are reported in parentheses. ***, **, and * denotes statistical significance at 1%, 5%, and 10%, respectively.

J o u r n a l P r e -p r o o f Table 1 . Standard errors are clustered by firm and quarter. Robust t-statistics are reported in parentheses. ***, **, and * denotes statistical significance at 1%, 5%, and 10%, respectively.

J o u r n a l P r e -p r o o f 80 01Jan2020 09Jan2020 17Jan2020 25Jan2020 02Feb2020 10Feb2020 18Feb2020 26Feb2020 05Mar2020 13Mar2020 21Mar2020 29Mar2020 06Apr2020 14Apr2020 22Apr2020 30Apr2020 08May2020 16May2020 24May2020 01Jun2020 09Jun2020 17Jun2020 25Jun2020 03Jul2020 11Jul2020 19Jul2020 27Jul2020 04Aug2020 12Aug2020 20Aug2020 28Aug2020 05Sep2020 13Sep2020 21Sep2020 29Sep2020 07Oct2020 15Oct2020 23Oct2020 31Oct2020 08Nov2020 16Nov2020 24Nov2020 02Dec2020 10Dec2020 18Dec2020 26Dec2020

Stringency Index Economic Support Index

6403) ROA 0.0010 -0.0013 0.0227*** 0.0010 -0.0013 0.0227*** (0.9635) (-1.2055) (8.9402) (0.9629) (-1.2087) (8.9394) MTB 0.0003*** 0.0003*** 0

We are grateful to Carl Chen (the Editor), three anonymous reviewers for their constructive comments and suggestions that significantly improved our manuscripts. We authors are politically neutral and do not support any political faction or politician in the United States. The findings of our study solely based on the empirical evidence and existing literature in the fields of corporate finance and government policy. There is no potential financial, political, or personal interest that may bias our work. The remaining errors are ours.

This table reports the regression results of Model (2) using alternative variable measurements, sample choices and estimators to control for crosssectional dependency and serial correlation of residuals. Standard errors are clustered by firm and quarter. Robust t-statistics are reported in parentheses. ***, **, and * denotes statistical significance at 1%, 5%, and 10%, respectively.