key: cord-0840289-8tr8m979 authors: Aharon, David Y.; Baig, Ahmed S.; DeLisle, R. Jared title: The impact of government interventions on cross-listed securities: Evidence from the COVID-19 pandemic date: 2021-07-05 journal: Financ Res Lett DOI: 10.1016/j.frl.2021.102276 sha: 3f8a2885896d18ab2e165b2fd3624ec793c26f35 doc_id: 840289 cord_uid: 8tr8m979 This paper examines the impact of COVID-19 related governments' interventions on the volatility and liquidity of American depository receipts (ADRs). Using a wide dataset of 387 ADRs from 34 countries around the globe, we provide an examination of the effect of economic and non-economic interventions on the quality of these cross-listed securities. Our results suggest that closures, restrictions, as well as containment health steps implemented during the outbreak period of the pandemic, seem to deteriorate the ADRs' liquidity and stability. The negative impact holds for different control variables and regression specifications and is not subsumed by the inclusion of the daily confirmed cases as a proxy for the severity of the pandemic. The information documented here may assist financial market participants in their risk management. The findings could also be important for policymakers for their preparedness plans in case of future crises. There is a rapidly growing stream of literature examining the impact of the COVID-19 pandemic on financial markets. Since the outbreak of the pandemic in early 2020, sizeable efforts have been devoted to exploring the influence of the pandemic on a variety of assets classes, inter alia, the equity markets (Corbet et al., 2020; Engelhardt et al., 2021 , Zaremba et al., 2021a , the debt markets (Zaremba et al., 2021b (Zaremba et al., , A. 2021c , commodities (e.g., Lin and Su, 2021; Sifat et al., 2021) and cryptocurrencies (e.g., Sarkodie et al., 2021; Wasiuzzaman and Rahman, 2021) . As the pandemic deepened, governments implemented various health-related programs, economic support plans, and social restrictions in an attempt to ease the adverse impacts of the COVID-19 spread. Subsequently, an evolving strand in the financial literature examines the potential impact of these interventions on key variables such as the returns, volatility, and liquidity in equity markets (e.g., Albulescu, 2021; Baig et al., 2021d; Harjoto et al., 2020; Zaremba et al., 2021d) . Existing cross-country analyses predominantly employ an aggregate approach by utilizing country-level market indices as their main focus of study. Thus, the firm-level effects of COVID-related government intervention across countries remains largely unexplored. This study aims to start filling this void in the literature by examining American Depositary Receipts (ADRs). While ADRs are traded in the US, the underlying stocks in our sample come from 34 countries around the globe. ADRs offer several advantages over market indices, the first of which is the advantage of a firm-level cross-country investigation. The use of ADRs provides a unique design that controls for different market structures, currencies, and other country effects. It allows us to exploit the cross-sectional variation in government interventions across countries, in addition to the time-series variation used by prior studies as the main methodology. Prior studies also demonstrate using ADRs alleviates several endogeneity issues (Blau et al., 2014; Baig et al., 2019; Baig et al., 2021c) . Lastly, but no less importantly, the use of ADRs listed in US markets mitigates a potential non-synchronous trading bias that may arise in exploring international capital markets, or liquidity issues, which are a common feature in emerging markets. There are several potential channels through which government interventions can influence ADRs' performance. According to the supply of stock returns hypothesis (Diermier et al., 1984; Harjoto et al., 2020; Ibbotson and Chen, 2003) , the performance of the domestic capital market reflects economic activity. Therefore, any disruption to sustainable business activity might affect the real economy, and consequently, will be translated to a response by investors. The outcome of these disruptions can be related to an increased uncertainty originating from government interventions. According to the theoretical model by Pastor and Veronesi (2012) , government interventions change the perception of the economic status quo and might induce two types of uncertainties. The first type of uncertainty concerns the future consequences of the type of intervention imposed. The second is ambiguity concerning future policy, i.e., whether similar actions are expected to encounter in the future. As capital markets are overwhelmed with uncertainty during COVID-19 (Baker et al., 2020) , the implementation of interventions may even worsen investors' ambiguity about the real consequences for the economy and consequently trigger undesirable effects on volatility and liquidity. This suggests closures, quarantines, mobility restrictions, and any type of government intervention that poses uncertainty for sustaining economic activity, will potentially be reflected in increased volatility and poorer liquidity levels. Using data from University of Oxford's COVID-19 Government Response Tracker, we test this hypothesis by examining the impact of various government interventions on the volatility and liquidity of ADRs. The main results of our study are as follows. First, after controlling for the specific characteristics of ADRs as well as country and industry fixed effects, restrictions such as stay-at-home requirements and workplace closings are related to higher volatility and lower liquidity in ADRs. A one percentage increase in the stringency index is associated with at least a 0.082 (0.144) percentage increase in ADRs' bid-ask spread (volatility). Secondly, this relation also holds for all types of indices reflecting different types and expressions of intervention intensity, such as economic, health, containment, and general response indices. We find that the spread and volatility of emerging market ADRs are most sensitive to the economic type of government interventions. Lastly, the adverse impacts of government interventions are not subsumed by the country-specific rate of documented COVID infections. These novel results are consistent with previous pandemic intervention-related studies (e.g. Zhang et al., 2020; Baig et al., 2021d) and more generally with studies suggesting that government restrictions potentially affect the stability and efficiency of financial markets (Pastor and Veronesi, 2012; Blau et al., 2014; Blau, 2017; Baig et al., 2019; Baig et al., 2021c) . Our findings are of importance to policymakers who wish to understand the effects of government interventions, as well as financial market participants that maybe affected by such intervention. We retrieve the required data from various sources. First, closures, economic, health, and general interventions indices, as well as country-level COVID case data, are retrieved from the Oxford database. 1 Specifically, we extract the Stringency_Index, Economic_Index, Containment_and_Health_Index, and Government_Response_Index. Table 1 details the information about each Index. Second, we obtain the daily ADR price-and volume-related information from the Center for Research in Security Prices (CRSP). For each ADR we manually identify the corresponding home country from Bloomberg. Finally, we use the International Monetary Fund (IMF) definition for emerging markets. We use the raw data to construct the following variables 2 : Range_Volatility is the daily ADR volatility calculated as the difference between the natural log of intra-day high and low prices. Spread is the daily bid-ask spread computed as the difference between ask and bid prices of ADRs scaled by the mid-point. Amihud_M is the daily Amihud (2002) price impact measure computed by scaling the absolute return by the dollar volume scaled up by one million. Size the daily market capitalization computed as the product of price and shares outstanding. Price is the daily closing ADR price. Turnover is calculated as the daily trading volume scaled by the number of shares outstanding. ΔS&P500 is the daily return of the S&P 500 Composite Index. EM (Nasdaq) is a dummy variable that takes on a value of 1 for ADRs from emerging markets (ADRs listed on NASDAQ), zero otherwise. Cases (Deaths) is the daily number of COVID-19 infected cases (fatalities) in the respective country. Our final dataset is an ADR-day panel consisting of 387 ADRs from 34 countries and spans from October 1st, 2019, to March 30th, 2020. This provides us with six months surrounding the arguably first confirmed COVID-19 case. The three months pre-COVID-19 window provides us a control sample to draw efficient comparisons and inferences vis-à-vis the COVID-19 period. Table 1 summarizes the statistical properties of variables of interest. About 41% of our sample is listed on NASDAQ, and 59% listed on the NYSE or AMEX exchanges. Additionally, almost 57% of the ADRs' underlying stocks are from emerging markets. Additionally, Appendix A2 presents detailed summary statistics by country. China provides the most ADRs in the sample and has the most stringent government response and containment actions, while Hong Kong has the most intensive economic interventions. 3 To ensure the robustness of the regression models, we utilize a series of univariate and multivariate regressions. We employ several variations of specifications and combinations of control variables, industry fixed effects, 4 country fixed effects, and robust standard errors clustered at the ADR-level. In our first set of tests, we investigate the relation between the government intervention indexes and the bid-ask spread using variations of the following regression specification 5 : Table 2 presents the results for our main independent variable of interest, LN_INV, the natural log of a certain intervention index. The left, middle, and right sides of the table report the results obtained for the Stringency, Economic, and Health indexes, respectively. 6 The adverse relation of all types of interventions is clear, as our estimations imply that all types of interventions are associated with increased ADRs spread. The findings are robust under alternative models. Hence, for brevity, we discuss the results for models [3] and [5] for each index. In model [3], we include industry and country fixed effects along with the controls variables. 7 In model [5] we include industry fixed effects and the controls excluding country fixed effects as we include a dummy variable for emerging markets and its interaction with the respective intervention index. Model [3] yields a positive coefficient (0.041%, p-value<0.01) on the natural log of stringency We obtain the four stringency and policy indexes namely the Stringency Index, the Economic Support Index, the Government Response Index, and the Containment Health Index from the Oxford COVID-19 Government response tracker. Stringency_Index represents the strictness of 'lockdown style' policies. Economic_Support_Index captures government measures including income support and debt relief programs. Government_Response_Index reflects an overall government response. The Containment_Health_Index captures 'lockdown' type policies and actions including testing policy, contact tracing, urgent healthcare, and vaccine investments. Range_Volatility is the daily ADR volatility calculated as the difference between the natural log of intra-day high and low prices. Spread is the daily bid-ask spread computed as the difference between ask and bid prices of ADRs scaled by their mid-point. Amihud_M is the daily Amihud (2002) price impact measure computed by scaling the absolute return by the dollar volume scaled up by a million. Size the daily market capitalization computed as the product of price and shares outstanding, it is presented in billions. Price is the daily closing ADR price. Turnover is calculated as the daily trading volume scaled by the number of shares outstanding. ΔS&P500 is the daily return of the S&P 500 Composite Index. Emerging_Market (EM) is a dummy variable that takes on a value of 1 for ADRs from emerging markets and developing economies based on the World Economic Outlook classification of the IMF, zero otherwise. Nasdaq is a dummy variable that takes on a value of 1 for ADRs listed on NASDAQ, zero otherwise. Confirmed_Cases (Cases) is the daily number of people infected with coronavirus in the respective country. Confirmed_Deaths (Deaths) represents the daily number of people who lost their lives due to coronavirus-related complications in the respective country. 4 The industry fixed effects are based on Fama and French 48 industry classifications. 5 We also investigate the impact of governments' interventions on other liquidity aspects such as price impact. We repeated our examinations of model (1) without Amihud (2002) Illiquidity measure as a control and our results are virtually unchanged. We also use Amihud (2002) as our dependent variable and draw similar inferences. 6 Additional results for the Government Response Index are included in Appendix A3. 7 We note that Turnover is a dimension of liquidity. However, we control for turnover in our regressions because recent literature suggests that stock trading activity and intensity increased during the pandemic (Baig et al. 2021b, Chiah and Zhong, 2020; Ozik et al., 2021) while liquidity decreased (Baig et al., 2021a; Eaton et al., 2021) . We therefore ensure that we control for this aspect while drawing inferences on the bid-ask spreads. This table provides the results from the variations in the estimation of the following OLS regression specification. The dependent variable is Ln_Spread which is the natural log of Spread. The main independent variable of interest is Ln_INV, which refers to the natural log of the Oxford Stringency, Economic Support and Containment_Health Indexes, respectively. For remaining variable definitions please refer to Table 1 . Robust T-stats corresponding to standard errors clustered at the ADR level are reported in parenthesis. ***, **, and * reflect statistical significance at 0.01, 0.05, and 0.10 levels, respectively. -3.682*** -4.167*** -3.618*** -2.779*** -2.728*** -3.502*** -4.014*** -3.403*** -2.625*** -2.608*** -3.689*** -4.177*** -3.631*** -2.758*** -2.713*** (-48.523) (-16 This table provides the results from the variations in the estimation of the following OLS regression specification. The dependent variable is Ln_Volatility which is the natural log of Range_Volatility. The main independent variable of interest is Ln_INV, which refers to the natural log of the Oxford Stringency, Economic Support and Containment_Health Indexes, respectively. For remaining variable definitions please refer to Table 1 . Robust T-stats corresponding to standard errors clustered at the ADR level are reported in parenthesis. ***, **, and * reflect statistical significance at 0.01, 0.05, and 0.10 levels, respectively. index (Ln_INV). Economically, a one percent increase in home country stringency is associated with a 0.041% increase in ADRs spread. Model [5] , shows similar results. The coefficient is 0.033 (p-value<0.05) is still both statistically and economically significant. However, as shown by the negative and statistically significant coefficient estimate on the EM × Ln_INV interaction term, the effect of this type of intervention is muted in emerging markets. Similar results are obtained for the impact of the Economic Support Index, with a positive coefficient of 0.085 [Model (3), p-value<0.01]. In economic terms, a one percent increase in home country economic support is associated with about a 0.085% increase in the illiquidity of the corresponding ADRs. Model [5] , confirms the previous results an impact of 0.082 (p-value<0.05). Also, the emerging market ADRs are generally more illiquid by 0.182% (p-value<0.05), while the coefficient on the interaction term EM × Ln_INV is statistically insignificant, suggesting a similar impact of economic interventions on the liquidity in emerging markets as in developed markets. Regarding emerging markets, Harjoto et al. (2020) document negative returns in MSCI indices in response to Fed stimulus. However, they find the negative response is more pronounced in emerging markets. We document that the adverse response in market liquidity to economic government intervention is similar in developed and emerging markets. Lastly, the results of the Containment_and_Health index impact are essentially similar to the aforementioned findings. Meaning, there is an adverse impact of the health interventions on ADRs spread. For the remaining control variables, Amihud_Measure and Nasdaq dummy are positively related with Ln_Spread while Turnover and the Natural logs of Price and Size are negatively correlated with Spread. Similarly, there is a positive and statistically significant coefficient of 0.056 (p-value<0.01) on the Government Response Index (Appendix A3) suggesting an adverse impact of home country government interventions on the liquidity of ADRs. Additionally, similar to social government interventions (measured by the Stringency index), the effect of health interventions on liquidity is muted in emerging markets. Next, we examine the relationship between interventions and the ADR-level Range_Volatility, which follows the Alizadeh et al. (2002) measure, with the following regression specification: (3), p-value<0.01] on the natural log of stringency index. Thus, in economic terms, a one percent increase in home country stringency is associated with a 0.151% increase in ADRs volatility. In Model [5] , the coefficient decreases to 0.144 (p-value<0.01), controlling for emerging markets. Emerging market ADRs are more volatile vis-à-vis developed markets by 0.364% (p-value<0.01). The interaction term EM × Ln_Stringency is significantly negative (-0.113, p-value<0.01 ). Meaning, the positive relation between social government intervention and volatility is milder in emerging markets. For the remaining controls, Turnover and Nasdaq dummy are positively related with Ln_Range_Volatility while Amihud_Measure, S&P500_Return and the Natural logs of Price and Size are negatively correlated with the range volatility measure. As for the impact of the Economic intervention, we find a positive and statistically significant coefficient of 0.266 in Model [3]. Economically, a one percent increase in economic support is associated with 0.266% increase in volatility of the cross-listed securities. In Model [5] , the coefficient becomes weaker in magnitude to 0.176 (p-value<0.01) but remains both statistically and economically significant. Emerging market ADRs are generally more volatile by 0.289% (p-value<0.01). Similar to previous results, the coefficient on the interaction term EM × Ln_INV is statistically insignificant suggesting indifferent impact of economic interventions for emerging and developed markets. Lastly, the impact of Health measures is positive and statistically significant across all models. This suggests an adverse impact of home country government interventions on the volatility of the corresponding cross-listed securities. The Government Response Index results suggest a similar interpretation (Appendix A3) that conforms to the former findings obtained for the overall government response index, the economic support index, and the stringency index. The overall government response index hint that governments' measures are significantly associated with volatility and the Bid-Ask spreads. In economic terms, Model (3) suggests that a one percent increase in the overall government response index is associated with 0.151% (0.266%) increase in the Bid-Ask spread (volatility) of the cross-listed securities. Notably, the effect of Economic Interventions seems to have the major influence on ADRs volatility and liquidity among all types of interventions and markets. Overall, the results lend further support to the idea that COVID-19 related government responses were associated with an increase in the volatility and liquidity of the ADRs. Taken together, the evidence highlights the effects of government interventions on the stability of cross-listed securities that seem to hold across both the developed and emerging markets. This study presents the first attempt to examine the impact of COVID-19-related government policy interventions on cross-listed securities. Using comprehensive data from ADRs, we show that government interventions across countries are associated with an increase in ADRs volatility and widening their Bid-Ask spreads. These effects are robust when controlling for a wide variety of characteristics. Our results are in line with the prior studies and are also consistent with studies that suggest that regulatory restrictions impact the quality of financial markets (Pastor and Veronesi, 2012; Blau et al., 2014; Blau & Thomas, 2015; Blau, 2017 ). It appears that Governments' interventions could act as a double edge sword in the sense that, while their intended goal may be positive (e.g. pandemic containment and improving social welfare), their influence manifests negatively in the ADR market. This evidence may be of importance to government policymakers and regulators who wish to understand the spill over effects of government intervention. Financial market participants may also incorporate this evidence into their portfolio allocations as governments intervene in social welfare in various ways. COVID-19 and the United States financial markets' volatility Range-based estimation of stochastic volatility models Illiquidity and stock returns: cross-section and time-series effects COVID-19 Intensity Across U.S. States and the Liquidity of U.S. Equity Markets Do Retail Traders Destabilize Financial Markets? An Investigation Surrounding the COVID-19 Pandemic. An Investigation Surrounding the COVID-19 Pandemic Free trade and the efficiency of financial markets Price clustering and economic freedom: the case of cross-listed securities Deaths, panic, lockdowns and US equity markets: the case of COVID-19 pandemic Economic freedom and crashes in financial markets Economic freedom and the stability of stock prices: a cross-country analysis Economic Freedom and Financial Market liquidity: Evidence from ADRs The unprecedented stock market reaction to COVID-19 Trading from home: the impact of COVID-19 on trading volume around the world The contagion effects of the COVID-19 pandemic: evidence from gold and cryptocurrencies Zero-Commission Individual Investors, High Frequency Traders, and Stock Market Quality COVID-19: stock market reactions to the shock and the stimulus Long-run stock returns: Participating in the real economy Does COVID-19 open a Pandora's box of changing the connectedness in energy commodities? COVID-19 Pandemic improves market signals of cryptocurrencies--evidence from Bitcoin, Bitcoin cash, Ethereum, and Litecoin The COVID-19 pandemic and speculation in energy, precious metals, and agricultural futures Performance of gold-backed cryptocurrencies during the COVID-19 crisis The quest for multidimensional financial immunity to the COVID-19 pandemic: evidence from international stock markets Term Spreads and the COVID-19 Pandemic: evidence from International Sovereign Bond Markets Volatility in international sovereign bond markets: the role of government policy responses to the COVID-19 pandemic COVID-19, government policy responses, and stock market liquidity around the world: a note Financial markets under the global pandemic of COVID-19 Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.frl.2021.102276.