key: cord-0981750-2zvngttb
authors: Beirne, John; Renzhi, Nuobu; Sugandi, Eric; Volz, Ulrich
title: COVID‐19, asset markets and capital flows
date: 2021-08-11
journal: Pacific Economic Review
DOI: 10.1111/1468-0106.12368
sha: 6985767c23836c1b67d700c7d8ba452b322dfe5c
doc_id: 981750
cord_uid: 2zvngttb

This paper empirically examines the reaction of global financial markets across 38 economies to the COVID‐19 outbreak, with special focus on the dynamics of capital flows across 14 emerging market economies. The effectiveness of fiscal and monetary policy responses to COVID‐19 is also tested. Using daily data over the period January 4, 2010 to August 31, 2020, and controlling for a host of domestic and global macroeconomic and financial factors, we use a fixed effects panel approach and a structural VAR framework to show that emerging markets have been more heavily affected than advanced economies. In particular, emerging economies in Asia and Europe have experienced the sharpest impacts on stock, bond and exchange rates due to COVID‐19, as well as abrupt and substantial capital outflows. Quantitative easing and fiscal stimulus packages mainly helped to boost stock prices, notably for advanced and emerging economies in Asia. Our findings also highlight the role that global factors and developments in the world's leading financial centers have on financial conditions in EMEs. Importantly, the impact of COVID‐19 related quantitative easing measures by central banks in advanced countries extended to EMEs, with significant positive spillovers to EME stock markets in Asia, Europe and Latin America. Going forward, while the ultimate resolution of COVID‐19 may be expected to lead to a market correction as uncertainty declines, our impulse response analysis suggests that there may be persistent effects on bond markets in emerging Europe and on EME capital flows.

bond markets in emerging Europe and on EME capital flows.

COVID-19 has been a truly global shock. The pandemic and the resulting lockdowns have led to an unprecedented economic contraction and turbulences in financial markets, causing the largest ever outflow of portfolio capital from emerging market economies (EMEs). This paper is aimed at gaining an insight into the impact of COVID-19 on global financial market and capital flow dynamics. This approach also enables an assessment of the impact of COVID-19 in comparative terms across different advanced and emerging economy groups in Asia, Europe and elsewhere. In addition, in a similar way, our analysis assesses the effectiveness of fiscal and monetary policy responses to COVID-19 in advanced and emerging economies. More specifically, our approach is based on a panel regression and panel structural VAR approach with daily data over the period January 2010 to August 2020, whereby we assess the impact of COVID-19 on bond yields, stock prices, and exchange rates for a sample of 38 advanced and emerging markets. We also examine how equity and bond flows from a sample of 14 EMEs have been affected by COVID-19. Moreover, we isolate and test the role played by COVID-19 related fiscal stimulus packages and various monetary policy measures (conventional, quantitative easing and central bank swap arrangements) in supporting asset markets and capital flows across advanced and emerging economy sub-panel groupings. 1 We conduct a comprehensive empirical analysis to contribute to the growing literature on the financial market impact of COVID-19. First, our analysis incorporates three asset markets, as well as capital flows, in a consistent econometric set-up. Second, our analysis is on a global sample of countries, which is of crucial importance, particularly given that global markets are increasingly interconnected and that COVID-19 constitutes an example of a global exogenous shock to markets. Prevailing empirical studies to date have tended to focus on single country analyses, and on specific asset markets. Importantly, this set-up enables a determination to be made of the relative impact of COVID-19 across different advanced and emerging economy panels. Third, we use two alternative econometric techniques to draw our conclusions. These approacheswhich also control for a variety of domestic factors as well as global spilloversenable comparisons to be made on which financial markets have been most affected by COVID-19 at the global level, and also enable us to determine whether the virus may have lasting effects on markets. Fourth, we estimate the relative effectiveness of policy responses to COVID-19, including quantitative easing measures that were introduced for the first time in many EMEs.

Overall, we find that emerging markets have been more heavily affected than advanced economies. In particular, emerging economies in Asia and Europe have experienced the sharpest impacts on stock, bond and exchange rates due to COVID-19, as well as abrupt and substantial capital outflows. Quantitative easing and fiscal stimulus measures mainly helped to boost stock prices, notably for advanced overall and emerging economies in Asia. Our findings also highlight the role that global factors and developments in the world's leading financial centers have on financial conditions in EMEs. Importantly, the impact of COVID-19 related quantitative easing measures by central banks in advanced countries extended to EMEs, with significant positive spillovers to EME stock markets in Asia, Europe and Latin America. Going forward, while the ultimate resolution of COVID-19 may be expected to lead to a market correction as uncertainty declines, our impulse response analysis suggests that there may be persistent effects on bond markets in emerging Europe and EME capital flows due to .

The remainder of the paper is organized as follows: Section 2 reviews the related literature on the dynamics of asset prices and capital flows during periods of heightened uncertainty, as well as previous studies on the economic and financial impact of COVID-19. Section 3 presents the data and empirical methodology. Section 4 presents the empirical results. Section 5 concludes.

2 | RELATED LITERATURE find that US stock market movements in the early phase of the pandemic have been more reflective of sentiment than fundamentals. Albulescu (2021) finds that the persistence of the COVID-19 crisis and its related uncertainty amplifies the US financial markets' volatility and has implications for the global financial cycle. Vasileou (2020) finds that US stock markets were not efficient during the pandemic and that behavior and sentiment indicators may be more useful than economic variables in explaining investor decisions. Ramelli and Wagner (2020) discuss the impact of US firms' trade and financial policies on US stock prices during the COVID-19 pandemic. They make the point that investors retreated from the stocks of US firms that were highly exposed to China, in line with the traditional response of markets to increases in uncertainty. As the virus spread to Europe and the US, investors became more concerned about the financial conditions of firms located in these areas, particularly those with high debt and/or low liquidity, with negative repercussions for stock prices. Baker et al. (2020) find that the impact of COVID-19 on US stock market volatility is much greater than that of previous pandemics since 1900, particularly due to the economic ramifications of containment policies. Other papers that focus on the implications of COVID-19 for the US stock markets include Alfaro, Chari, Greenland, and Schott (2020) , Landier and Thesmar (2020) , Mazur, Dang, and Vega (2021) . On the US bond markets, Haddad, Moreira, and Muir (2020) examine disruptions to the US debt market due to COVID-19 and the role played by interventions by the Federal Reserve. They find that while investors initially shifted out of bonds towards more liquid securities to raise cash, Federal Reserve purchases of corporate bonds helped to alleviate the disruption in the bond market. Cristofaro, Gil-Alana, Chen, and Wanke (2020) find that the shock from the COVID-19 pandemic has only temporary impacts on China's Shanghai and Shenzhen Composite Indices, unlike the shock from the 2007-2008 Global Financial Crisis that have permanent impacts on the two indices. Espinosa-Méndez and Arias (2021) find that the COVID-19 pandemic increased herding behavior in European capital markets, where uncertainty drives less informed agents. Takyi and Bentum-Ennin (2020) examine the impacts of COVID-19 on the stock markets of 13 African countries and find that the pandemic had a significant negative effect in most cases.

Some studies analyze the global impacts of the pandemic on the volatility of financial markets and cross-border capital flows. Substantial effects on volatility of global stock markets due to COVID-19 have been stressed by Zhang, Min Hu, and Ji (2020) . They find that global financial market risk increased substantially in response to the pandemic. Ly ocsaa, Baumöhl, Výrost, and Moln ar (2020) find that fear of the coronavirusmanifested as excess search volume in Googlerepresents a timely and valuable data source for forecasting stock price variation in the world's 10 largest stock markets. Singh, Dhall, Narang, and Rawat (2020) find evidence for the Overreaction Hypothesis (ORH) in the G20 financial markets during the earlier phase of the pandemic. The financial markets of the G20 economies rebounded when investors began to make rational decisions along with intervention by policy-makers to boost market confidence. Harjoto, Rossi, Lee, and Sergi (2020) examine the impact of COVID-19 on the stock markets of 53 emerging market countries and 23 developed countries. They find that COVID-19 cases and deaths adversely affect stock returns and increase volatility and trading volume in all of these countries, but these effects are heterogenous across emerging and advanced economies. Cases and deaths affected stock returns and volatility in the emerging markets, while only cases of COVID-19 affected stock returns, volatility, and trading volume in the developed markets. Hördahl and Shim (2020) examine the impact of COVID-19 on the relationship between bond portfolio outflows and the exchange rate and long-term interest rates in 19 EMEs. They find that bond portfolio outflows from EMEs are indeed related to currency depreciations and rising long-term interest rates, but with some difficulty in ascertaining the direction of causality. More generally on capital flow dynamics, McKibbin and Sidorenko (2006) show that a pandemic tends to lead to a major shift in capital from the more to the less affected economies. As regards EMEs, Hofmann, Shim, and Shin (2020) suggest that borrowing through local currency bonds has not helped to insulate these economies from financial tensions. Indeed, many EME local currency bond spreads spiked amid sharp currency depreciation and capital outflows. Other work by Souza, Souza, and Silva (2020) find that the COVID-19 pandemic affected the efficiency of capital markets in 44 countries. In the pre-pandemic period, market efficiency was reduced due to individualism and aversion to uncertainty, as well as increased inflation. In the pandemic period, however, market efficiency was increased by individualism and reduced by indulgence. Topcua and Gulal (2020) find that the impact of the COVID-19 outbreak has been the highest in Asian emerging markets whereas emerging markets in Europe have experienced the lowest.

Other previous work has examined the impacts of COVID-19 policy responses by fiscal and monetary authorities on financial markets. Cox et al. (2020) find that the unconventional US monetary policies helped the US financial markets to rebound in late March and April 2020, while they find no evidence that the conventional monetary policy instruments help such rebounds. Caballero and Simsek (2020) discuss the important role of large-scale asset purchases by central banks to cope with downward asset price spirals and severe aggregate demand contractions following a large supply-side shock such as that caused by COVID-19. Adopting an event-study approach, Hartley and Rebucci (2020) find that COVID-19 related quantitative easing measures introduced by advanced and emerging economies had a dampening effect on sovereign bond yields (particularly in emerging economies, many of which had introduced QE measures for the first time). Arslan, Drehmann, and Hofmann (2020) also find that QE announcements by central banks in emerging markets were effective in lowering local currency bond yields and restoring investor confidence. On central bank swap lines introduced in response to COVID-19, Reis (2020a, 2020b) find significant effects in lowering deviations in covered interest parity (CIP). Topcua and Gulal (2020) find that official response time and the size of the stimulus package provided by the governments matter in offsetting the effects of the pandemic. Overall, they found that EMEs with monetary policy frameworks that address the feedback loop between exchange rate depreciation and capital outflows have a greater likelihood of mitigating the detrimental impact of COVID-19.

Our paper contributes to the growing related literature on the impact of COVID-19 on financial markets. The key contribution of our paper relates to its scope across two key dimensions: country coverage with a global sample of 40 economies, and coverage across three asset markets globally and EME capital flows. We also use two comprehensive econometric approaches. As a result, this paper goes far beyond the scope of other papers written in this domain. The prevailing literature tends to focus on the effects of the pandemic on a specific asset market in a small set of countries or even single countries. Thus, an important valueadded in our analysis is that it permits a comparative assessment across different economic regions and asset classes.

A twofold approach is implemented to estimate the impact of COVID-19. First, using a daily data frequency, we use a fixed-effects panel model over the period from January 4, 2010 to August 31, 2020 across 38 advanced and emerging economies to examine the effects of COVID-19 on bond yields, stock prices and exchange rates. For the assessment of the impact on equity and bond flows, data availability over the same time period limits our country sample to 14 EMEs. 2 We consider the following regression in our first stage:

where the dependent variable y i,t is either the 10-year government bond yield, the stock prices, the effective exchange rate, equity flows, or debt flows of country i at time t. Our key explanatory variable COVID19 i,t is defined as daily new cases per one million population. The domestic controls x i,t include the central bank policy rate, a dummy variable for COVID-19 related quantitative easing announcements by central banks, a dummy variable for fiscal stimulus packages announced by national governments in response to COVID-19, a dummy variable for international central bank swap announcements by central banks due to COVID-19, 3 the consumer price index (CPI), the industrial production index, the Citigroup Economic Surprise Index which measures contemporaneous economic surprises in macroeconomic data, and financial market indicators referring to bond yields, stock prices, and effective exchange rates. 4 EPU stands for the US Daily News Index, a measure of global economic policy uncertainty (Baker et al., 2020) . VIX stands for the Chicago Board Options Exchange (CBOE) Volatility Index, a measure of global risk aversion. Controlling for the policy responses to the pandemic also importantly enables us to assess the pure market reaction of COVID-19 on asset prices and capital flows. We also test for spillover effects of quantitative easing measures by advanced economy central banks on emerging markets. δ i represents country fixed effects; and ε i,t is the error term. All of the control variables are lagged by one period to mitigate endogeneity concerns. Table A2 in Appendix A provides details of all variables used, as well as the sources. Second, a structural panel VAR is used to examine the response of financial markets and capital flows to shocks from COVID-19. Crucially, shocks control for a range of domestic and global factors. The panel SVAR is implemented in the same sample used in the first stage. The panel SVAR can be denoted as follows in its general specification, with structural shocks identified by a recursive restriction:

where A L ð Þ is the matrix of the lag polynomial; Y i,t refers to the demeaned value of endogenous variables of country i to accommodate country-specific fixed effects; and μ i,t is a vector of structural disturbances. The ordering of the variables is imposed in a recursive form (Christiano, Eichenbaum, and Evans, 1999) , which results in the following matrix A to fit a just-identified model: 

We place our COVID-19 variable at the top in the ordering, which implies that it will only be affected by contemporaneous shock to itself. Following the COVID-19 variable, we place the global economic policy uncertainty variable second in the ordering, which implies that global factors will be affected by contemporaneous shocks to COVID-19 and itself, but not by contemporaneous shocks to domestic factors or financial market indicators. Importantly, we place the financial market indicators in the last place in the ordering, which is not only based on the assumption that COVID-19 will affect the markets, but also on the consideration of our firststage empirical results that imply the global and domestic factors that are driving the financial markets. Finally, we place our domestic factors in the middle of the ordering. The panel VAR includes three lags, as determined by the Akaike information criterion (AIC).

Prior to examining the results from our panel regressions and panel structural VAR, it is useful to consider the trajectory of global financial markets and capital flows in the aftermath of the COVID-19 outbreak (see Figures A1 to A4 in Appendix A). It can be seen that government bond yields initially declined globally given rising uncertainty amidst a bleak economic outlook, suggesting that investors considered sovereign bonds as safe havens assets at the time. On "Black Monday" (March 9, 2020), financial markets panicked over the worsening of the COVID-19 pandemic and the concomitant oil price war between Saudi Arabia and Russia. Stock markets tanked, while bond yields spiked. Even US Treasuries, usually considered the ultimate safe haven asset, were dropped as investors were desperate for cash (Schrimpf, Shin, & Sushko, 2020; Tooze, 2020a) . Central banks, particularly those in advanced economies, responded quickly with interventions "on an unprecedented scale" and helped to avert "a fullscale meltdown" (Tooze, 2020b) . Large-scale asset purchases of sovereign bonds by the US Federal Reserve, the European Central Bank, the Bank of Japan and other central banks helped to stabilize the situation and led to a significant decline in sovereign bond yields in advanced economies. Following this spike, yields generally trended downwards globally until around June 2020, when yields upticked somewhat, given concerns about a second wave of the virus and the related effects of uncertainty. As well as central bank support packages, substantial COVID-19 related fiscal measures were introduced by national governments, which further reinforced the actions of central banks. Figure A6 in Appendix A demonstrates some cross-country heterogeneity in the magnitudes of the fiscal response to the crisis, with some countries such as Japan and New Zealand bringing in new fiscal measures equating to over 20% of GDP. Our empirical approach controls for the effects of COVID-19 related monetary and fiscal policy measures introduced using announcement date dummy variables. As regards stock prices, these had started to slump already in February 2020, but then dropped sharply at the global level on Black Monday. Stocks recovered somewhat during April, as containment measures imposed by infected countries began to be relaxed, as well as positive spillovers of liquidity measures by central banks to stock markets. On exchange rate and capital flow developments, EMEs as a whole experienced sharp currency depreciations and substantial capital outflows as COVID-19 took hold. This reflected the typical pattern observed in global financial markets during periods of heightened uncertainty. The scale of capital flight, however, was unprecedented: during February and March 2020, EMEs experienced portfolio capital outflows totaling around $100 billion, triple the amount of outflows during the 2008 Global Financial Crisis (Georgieva, 2020) . Indeed, both equity and bond outflows from EMEs were much faster and more pronounced than during previous episodes of EME turmoil, including the 2013 "taper tantrum," the 2015 "China scare," and the 2008 Global Financial Crisis (GFC) ( Figure  A5 in Appendix A). The spike in bond yields after Black Monday was extraordinarily large and steep, but unlike in previous episodes of EME turmoil, bond yields returned to original levels after around a month. Exchange rate devaluation of EMEs was broadly similar to those seen during the GFC, which is also true for stock price changes.

Our panel regression results help to shed more light on developments in global markets and EMEs in particular due to COVID-19. Tables 1 to 4 display the impact of COVID-19 on sovereign bond yields, stock prices, effective exchange rates, and EME capital flows, respectively. Controlling for domestic and global factors, our analysis also sheds light on the impact of monetary and fiscal policy responses to COVID-19 in supporting domestic asset markets and capital flows.

In terms of the magnitudes of the effect of COVID-19 on financial markets, these are notably higher for emerging rather than advanced economies across bond, stock and exchange rates, particularly for European and Asian EMEs. Importantly, these results control for a vast number of domestic and global factors, the coefficient signs of which (where significant) largely are in alignment with priors. 5 On sovereign bond yields, Table 1 shows that over the sample period, COVID-19 has had a significant dampening effect in all advanced economies and emerging economies in Asia. 6 The largest relative effect on yields is apparent for Asian EMEs. It is worth noting that the magnitude of the effect of COVID-19 on yields seems to be much lower across country groups compared to the early stage of the crisis. Beirne et al. (2020) show that, over the period January 4, 2010 to April 30, 2020, the virus had led to a reduction in bond yields in European and Asian EMEs by around 0.24 and 0.14 percentage points respectively. 7 In this updated analysis to the end of August 2020, the sharp impact of COVID-19 has waned somewhat, and this is undoubtedly related to the adjustment of markets to the shock, as well as the longer impact for monetary and fiscal rescue packages to pass through to asset prices. The finding of a negative overall effect of COVID-19 on yields may seem counter-intuitive, as one might expect an increase in COVID-19 cases to worsen financial market turmoil and also increase sovereign bond yields. There are two explanations why the overall effect on bond yields was negative. First, government bonds were perceived as safer assets than corporate bonds, given the corporate sectorwith few exceptions-was very heavily affected by the COVID-19 lockdowns. With many businesses fighting for survival, sovereign bonds were seen as the better alternative, even if the crisis also cast questions on the sustainability of public debt. Secondly, the crisis gave way to an extremely accommodative central bank policy in most places, with slashes in interest rate and new rounds of QE policies in all major advanced economies.

As can be seen from the regression results, central bank swap line announcements were highly effective in bringing down sovereign bond yields in advanced economies overall, including in Europe and Asia, while interest rate reductions due to COVID-19 also passed through to bond yields across all advanced and emerging economies. There is also some evidence to indicate that fiscal policy has been effective at the global level over the period up to August 31, 2020. In addition, fiscal stimulus packages announced by national governments also seemed to have much stronger effects in this earlier period, both in advanced economies and EMEs in Europe and Asia. Consistent with the earlier analysis in Beirne et al. (2020) , we find that the effects of QE announcements by EME central banks had no significant effect on domestic government bond yields. This is in contrast to the findings of Hartley and Rebucci (2020) central banks in EMEs have had significant dampening effects on bond yields. Unlike that analysis, however, our paper controls for a large number of domestic and global macroeconomic and financial bond yield determinants. It should be also noted that international central bank swaps played no meaningful role for EMEs, which is not surprising: among the beneficiaries of the bilateral currency swaps extended by the Federal Reserve of the United States were only two EMEs, Brazil and Mexico (Gallagher, Gao, Kring, Ocampo, & Volz, 2021) . On the controls, the expected negative relationship between bond yields and both inflation and industrial production holds across the vast majority of regional groups, while global financial market uncertainty as proxied by the VIX is also positively related to bond yields, as expected. It is also notable that, in contrast to Beirne et al. (2020) , QE measures by advanced economy central banks do not spill over to EMEs for the period up to the end of August 2020, with all of these spillover effects taking place during the first phase of the QE response to Turning to the impact on stock markets, Table 2 indicates that while the effect has been marginal relative to new COVID-19 cases confirmed in advanced economies, stock prices have declined most substantially in European EMEs by around 3%, compared to around 1% in EMEs in Asia and Latin America. These are similar to the magnitudes found in Beirne et al. (2020) . In the case of stock markets, we see a strong impact of expansionary monetary policy in advanced economies (through QE and interest rate reductions), which clearly helped to prop up stock markets in advanced economies as well as emerging markets in Asia and Latin America. We also find significant positive spillover effects of advanced economy QE to all EME stock markets. As regards QE announcements by EME central banks, it is notable that these have been effective in Asian EMEs in supporting stock prices. On fiscal stimulus packages, these have been more important for stock markets in emerging rather than advanced economies overall, although stock markets in advanced economies in Asia rallied strongly due to COVID-19 related fiscal measures introduced. The domestic and global controls for the stock market equation are also in accordance with priors, for example, a positive relationship with inflation and industrial production, and a negative relationship with the VIX.

On exchange rates, Table 3 shows that European and Asian EMEs have been most affected, experiencing currency depreciations due to COVID-19, although the magnitude of these effects is not as large overall compared to the effects on stock and bond markets. QE and swap announcements by central banks in advanced economies overall appear to have been effective on exchange rates, while QE in emerging Asian economies has been effective in stabilizing exchange rates. The effect of fiscal stimulus packages on exchange rates is more limited, however, with some evidence of a significant impact at the global level. On asset markets overall, it appears that the initial sharper impact of COVID-19 up to the end of April 2020 (as in Beirne et al., 2020) applies only to the sovereign bond market. Extending the analysis to the end of August yields similar magnitudes of effects of COVID-19 on stock and currency markets.

In regard to EME capital flows, Table 4 indicates that COVID-19 has led to statistically significant outflows of both equities and bonds, reflecting investors' flight to safety. The magnitudes of these effects are much lower in the current estimation period compared to that estimated in Beirne et al. (2020) up to the end of April 2020. This is fully in line with what was observed in international capital markets, where EMEs experienced sharp and substantial outflows at the onset of the crisis. QE measures announced by EMEs' central banks at that time also had significant effects in averting capital outflows, while positive QE spillovers were also evident from advanced economy central banks, that is, advanced countries' asset purchase programs not only helped to lower bond yields and prop up stock markets at home, they also helped to put a stop to capital flight from EMEs. Table 4 shows that over the period up to the end of August 2020, the effect of QE is not statistically significant, in contrast to Beirne et al. (2020) , suggesting that the impact of QE on capital flows has waned over time. Fiscal stimulus packages, however, are supporting EMEs' equity flows. These results have particular significance for Asia, where half of the EME sample used for the capital flow equation is comprised of Asian economies. Across all of the models estimated, we have also examined the role of public health and containment policy introduced in response to the pandemic based on specifications that include the COVID-19 Containment and Health Index by the World Health Organization. We find that these policies are associated with lower sovereign borrowing costs, higher capital inflows to EMEs, and appreciating exchange rates, but negatively related to stock market indices. Moreover, while this alternative specification does not affect the results of our baseline regression, we have opted to maintain as our baseline the model that excludes this variable due to some multicollinearity concerns, particularly in relation to the wider government policy responses to T A B L E 4 COVID-19 impact on EME capital flows the pandemic. Given the importance of our dummy variables for policy responses to COVID-19, we have examined alternative definitions of these dummies as a form of sensitivity analysis. While our benchmark model uses 1-period lags for the policy dummy variables, we also estimated regressions with the policy dummy defined as 1 on and after the announcement date, and the estimations are consistent with the benchmark models. In addition, while the focus of the paper is to examine the impact of COVID-19 (which is exogenous) and the COVID-19 policy responses (which have been robustly examined using different dummy definitions) on asset markets and capital flows, we estimated results using a two-stage least squares approach with a 2-period lag of COVID-19. The estimation results are fully consistent with our benchmark models. Concerns about endogeneity are allayed in this respect. For the macroeconomic and financial controls, using an IV approach with high frequency (i.e., daily) data is fraught with difficulty given the difficulty in identifying suitable IVs. Nonetheless, the results are also robust to a 2SLS approach using 2-period lags of all controls. 9 Turning to the impulse responses from our panel structural VAR models (Figures 1 to 4) , the results indicate some heterogeneity in the responses of asset markets and EME capital flows to COVID-19 shocks, both in terms of statistical significance and duration.

The results on bond yields indicate that there is a statistically significant response of bond yields to COVID-19 shocks (defined as one-unit shocks) at periods over a longer time horizon (40-50 days) for Emerging Europe and Latin America. Moreover, for Emerging Europe, the response remains persistent. For Emerging Asia and advanced economies overall and particularly in Europe, we find that bond yields respond to COVID-19 shocks significantly at periods earlier in the time horizon (1-10 days). For other regions, we do not find a significant result. For stock markets, we find that the responses to COVID-19 shocks are statistically significant and negative across the sample of all countries, but that this is largely driven by advanced economies in Asia and Europe. Moreover, the statistical significance of the effect wanes after around 10 days. As regards exchange rates, we find that the impulse responses are mostly not significant except, however, in the case of Emerging Asia where COVID-19 shocks led to net depreciating effects on exchange rates up to a horizon of just after 10 days, after which point the responses to shock become insignificant. This effect of shocks at the early horizon stage is in line with expectations and also in line with what was observed at the outbreak of COVID-19, that is, depreciating exchange rates in EMEs linked to large capital outflows. Indeed, we also F I G U R E 4 Response of capital flows to COVID shock. Notes: Impulse responses with 95% confidence bands in dashed lines are reported. The vertical axis is the response of markets in percentage points to a one unit rise in COVID cases per one million of the population, and the unit of the horizon axis refers to 1 day [Color figure can be viewed at wileyonlinelibrary.com] find that the IRFs are highly significant in the case of capital flow reactions to COVID-19. For equity flows, the response is negative and highly significant across the entire time horizon. For bond flows, we also find a negative and highly significant effect up to around 10 days, after which point the effect becomes statistically insignificant. Importantly, the direction of the impulse responses across all asset markets and capital flows are fully consistent with the signs of the COVID-19 coefficients estimated in the earlier panel regression. 10 Moreover, these impulse responses also control for the same full set of domestic and global factors. Finally, and also in line with the panel analysis, permanent effects of the shocks are most pronounced in magnitude for emerging economies in Europe and Asia.

This paper provides an empirical analysis of the impact of the COVID-19 pandemic on global financial markets and EME capital flow dynamics, as well as the effectiveness of policy responses. Against the backdrop of globally interconnected financial markets, we examine the impact across sovereign bond and equity markets, as well as exchange rates and capital flows. Our analysis enables a comparative assessment to be made across advanced and emerging economies. Controlling for a large number of domestic and global financial and macroeconomic factors, our results suggest that COVID-19 has had the most substantial effects on financial markets in European and Asian EMEs. Moreover, EME equity and bond outflows appear to be directly linked to COVID-19, given investor risk aversion and flight to safety. Sovereign bond markets in EMEs appear to have been most affected by COVID-19, compared to the magnitude of the effects on stock prices and exchange rates. In terms of the magnitude of the effect on markets, comparing the current results to Beirne et al. (2020) indicates that COVID-19 had a much more substantial impact on sovereign bond yields and capital flows in the first phase of the crisis, up to the end of April 2020. In addition, while COVID-19 will ultimately subside, our results suggest that markets may experience some persistent effects, particularly for bond markets in emerging Europe and EME capital flows.

On policy responses, our findings show that the impact of conventional monetary policy reactions to COVID-19 dominate the effect on sovereign bond markets compared to other policy responses. The magnitude of the effect is about twice as large for advanced compared to emerging economies. Unlike the earlier analysis in Beirne et al. (2020) , the impact of QE on bond yields is not statistically significant over the sample period up to the end of August 2020. As regards stock markets, interest rate reductions were more effective in advanced compared to emerging economies overall by a factor of around two. In addition, QE measures in advanced economies, as well as spillovers to EMEs, helped to boost domestic stock prices by around 2% and 13%, respectively. QE spillovers from advanced economies to emerging Asia boosted stock prices by around 8%, while QE by central banks in emerging Asia helped to increase stock prices by around 14%. The impact of fiscal policy on stock markets was confined to Asian advanced economies and EMEs overall. At the global level, while the magnitude of the effect of policy responses on exchange rates was much lower than in other markets, COVID-19 fiscal stimulus packages helped to boost equity flows by around 6%.

Heightened uncertainty due to the COVID-19 pandemic has clearly affected the financial markets of EMEs more detrimentally than advanced economies. However, it appears that EMEs have performed strongly in their policy responses to the pandemic. While fiscal stimulus packages have contributed to restoring confidence in local markets, many EME central banks have embarked successfully on quantitative easing for the first time. Our results would suggest that these monetary policy measures have been particularly effective in the case of Asian EMEs, supporting stock prices. While Beirne et al. (2020) show that these QE measures also helped to stabilize capital flows up to end April 2020, our results indicate an important capital flow stabilizing role for fiscal policy and interest rate policy responses over the period up to end August 2020. Moreover, given the scale of bond and equity capital outflows from EMEs, particularly at the onset of the crisis, our results highlight the importance of strengthening the domestic investor base to be less reliant on international portfolio investment, corroborating findings by Hofmann et al. (2020) . Going forward, the COVID-19 crisis illustrates the need for concerted efforts at bolstering domestic financial resource mobilization in EMEs, and for reducing exposure to international portfolio capital and financial contagion. The extent of capital outflows also strengthens the case for reviving discussions around the management of capital flows and on the development of the global financial safety net. Finally, our results highlight the importance for EMEs to develop further their overall policy toolkits to respond to spikes in financial market volatility and crisis episodes, notably with the use of QE measures. With conventional monetary policy having easing limits and fiscal policy space constrained by excessive public debt, using QE policies can be a potent stimulator in domestic markets, particularly where inflation expectations are contained and exchange rates are flexible.

ENDNOTES 1 An earlier working paper version of this work is based on a sample period from January 4, 2010 to April 30, 2020 (see Beirne, Renzhi, Sugandi, & Volz, 2020 ). In the current paper, we draw some comparisons with this earlier analysis as regards the COVID-19 market impact and policy response effectiveness. 2 See Table A1 in Appendix A for the full list of countries by regional sub-group. 3 We considered including dummies for the announcement of a program with the International Monetary Fund (IMF). However, only two countries in our sample received IMF support during the COVID-19 crisis: Pakistan (Rapid Financing Instrument, April 16) and Chile (Flexible Credit Line, May 29, 2020). At the time of writing, South Africa is in negotiations with the IMF. Please refer to Table B1 in Appendix B for details of the announcement dates of various key policy responses to COVID-19. 4 We drop the asset market indicator from x i,t if it is used as the dependent variable in the left-hand side of the regression. 5 As can be seen from Figures A7 to A10 in Appendix A, the panel models estimated appear to reflect well the trajectory of actual global asset market prices and EME capital flows. 6 Given that our COVID-19 variable is defined as one new confirmed case per one million of the population, our results therefore imply a substantial accumulated effect of COVID-19 on financial markets and capital flows. 7 See Table A3 in Appendix A for an alternative specification in relation to the effect of COVID-19 on EME sovereign bond yields, the results of which are fully in line with our baseline. 8 It is also notable that the results of the effects on bond yields are positive and significant for the overall sample and the sample of all EMEs, while mostly negative and significant elsewhere. There are a number of possible valid explanations for this. First, the full panel is a much more heterogeneous set of countries with a different level of daily cases of COVID-19. The use of fixed effects in such a set-up may not always be sufficient to fully capture that heterogeneity. Second, the subsample of all EMEs contains 15 economies, and do not reflect the sum of Emerging Europe, Latin America and Emerging Asia (we have not shown the results for "Other EMEs," containing Tunisia and South Africa, given the difficulty in ascribing an interpretation to this subpanel). Finally, the subsample analysis of Emerging Asia has less policy controls (e.g., CB swaps) compared to other advanced economy panels. For all of these reasons, there can be some differences in the signs of the effects found across the panels of different country groupings. It follows that making comparisons of the results across different country groups are subject to caveats. 9 Given the extremely large volume of these additional results, they are not provided in the paper, but are available from the authors upon request. 10 We have also computed the full set of IRFs across all variables used in the analysis. The huge volume of results prevents us from including them in the paper, but they are all available from the authors upon request. Verikios, G., Sullivan, M., Stojanovski, P., Giesecke, J., & Woo, G. (2011) The global economic effects of pandemic influenza (General Paper G-224, Centre of Policy Studies). Melbourne: Monash University. Ramelli, S., & Wagner, A. F. (2020) . Feverish stock price reactions to 

COVID-19 and the United States financial markets' volatility

Aggregate and firm-level stock returns during pandemics

Central Bank bond purchases in emerging market economies

An updated assessment of the economic impact of COVID-19 (ADB Briefs No

Central Bank swap lines during the COVID-19 pandemic

Central Bank swap lines: Evidence on the effects of the lender of last resort

The unprecedented stock market impact of COVID-19

The coronavirus and the great influenza pandemic: Lessons from the "Spanish flu" for the Coronavirus's potential effects on mortality and economic activity

2020. Financial market and capital flow dynamics during the COVID-19 pandemic (ADBI Working Paper 1158)

Interdependence and contagion in global asset markets

Basel: Bank for International Settlements

A model of asset Price spirals and aggregate demand amplification of a COVID-19 shock (NBER Working Paper No. 27044)

Monetary Policy Shocks: What Have we Learned and To What End?

Emil Verner 2020. Pandemics Depress the Economy, Public Health Interventions Do Not: Evidence from the

What explains the COVID-19 stock market?

Modelling stock market data in China: Crisis and coronavirus

Effects of COVID-19 pandemic on international capital markets

COVID-19 effect on herding behaviour in European capital markets

Safety first: Expanding the global financial safety net in response to COVID-19

Economic effects of the 1918 influenza pandemic: Implications for a modern-day pandemic

Confronting the crisis: Priorities for the global economy

When selling becomes viral: Disruptions in debt markets in the COVID-19 crisis and the Fed's response (NBER Working Paper No. 27168)

How do equity markets react to COVID-19? Evidence from emerging and developed countries

An event study of COVID-19 central Bank quantitative easing in advanced and emerging economies (NBER Working Paper No. 27339)

Emerging market economy exchange rates and local currency bond markets amid the COVID-19 pandemic

International Monetary Fund (IMF). (2020, April). The great lockdown. World Economic Outlook

Longer-run economic consequences of pandemics (Federal Reserve Bank of San Francisco Working Paper No. 2020-09)

Earnings expectations in the COVID crisis

Fear of the coronavirus and the stock markets

The potential impact of COVID-19 on GDP and trade: A preliminary assessment

COVID-19 and the March 2020 stock market crash: Evidence from S&P1500

Global macroeconomic consequences of pandemic influenza (CAMA Miscellaneous Publications). Canberra: Centre for Applied Macroeconomic Analysis

The global macroeconomic impacts of COVID-19: Seven scenarios (CAMA Working Paper No. 19/2020). Canberra: Centre for Applied Macroeconomic Analysis

Leverage and margin spirals in fixed income markets during the COVID-19 crisis

The outbreak of COVID-19 and stock market responses: An event study and panel data analysis for G-20 countries

The impact of COVID-19 on stock market performance in Africa: A Bayesian structural time series approach

Is the coronavirus crash worse than the 2008 financial crisis? Foreign Policy

How coronavirus almost brought down the global financial system

The impact of COVID-19 on emerging stock markets

Behavioral finance and market efficiency in the time of the COVID-19 pandemic: Does fear drive the market?

Taiwan Pakistan