key: cord-0835944-bpqgi0r5
authors: Giofré, Maela
title: Foreign investment in times of COVID-19: How strong is the flight to advanced economies?
date: 2022-02-22
journal: nan
DOI: 10.1016/j.mulfin.2022.100735
sha: 338f8793a14569435e67137733539048a76240be
doc_id: 835944
cord_uid: bpqgi0r5

This paper investigates the extent of the flight to advanced economies by foreign investors at the onset of the COVID pandemic. Amid an overall decline of international positions, countries featuring higher GDP per capita, and belonging to the groups of advanced, G7, or Euro-area countries, appear to have been significantly less severely hit by the pandemic than developing countries. In particular, on comparing the growth rates of foreign liabilities at the end of the first quarter of 2020, the wedge between advanced and emerging countries is about 3%, and it is at least twice as large for G7 countries. This wedge is especially significant in the first quarter, and it is paired with the evidence of momentum trading by foreign investors. Our results are robust to the inclusion, as controls, of government stringency measures, alternative indicators of pandemic severity, sample specification and regression methods.

The ongoing COVID-19 pandemic in the world is having a profound impact on the economic and financial system. The growing body of recent literature on the pandemic's effects on financial markets generally reports evidence of a significant impact of confirmed COVID infections or deaths on financial markets' volatility and liquidity (Chebbi et al., 2021; Albulescu, 2021; Baig et al., 2021; Salisu and Vo, 2020; Ashraf, 2020; Li et al., 2021) .

Periods of exceptional uncertainty and crisis typically induce lenders to rebalance their portfolios either in favor of domestic borrowers, the so-called 'flight home' effect (Giannetti and Laeven, 2012) , or in favor of safer assets, the so-called 'flight to quality' effect (Brière et al., 2012) . Papadamou et al. (2021) , among others, identify flight-to-quality episodes when investigating the impact of the COVID-19 pandemic on the time-varying correlation between stock and bond returns in ten countries.

However, the flight-to-quality can occur not only among financial instruments but also among countries, towards economies featuring a higher degree of perceived 'quality'. The retrenchment in international capital flows during crisis periods is indeed a heterogeneous phenomenon among regions, with marked differences between emerging and developed economies (Lane and Milesi-Ferretti, 2017) . Belaid et al. (2021) find an increase in the interdependence between emerging and advanced economies, as a consequence of the COVID-19 spread, which suggests an increase in the transmission of stress and uncertainty among financial markets during the pandemic period. When considering the effects in terms of foreign investments, OECD (2020c) highlights the presence of a large cross-country variation in foreign direct investments and portfolio investments, reproducing the familiar pattern whereby international investors transfer capital back home or invest in safer assets during periods of uncertainty.

The stringent public health measures enacted by governments to limit the spread of the COVID-19 pandemic have induced recession, an erosion of confidence and greater uncertainty (OECD, 2020b) . Moreover, they have caused severe economic disruptions, with a significant impact also on the foreign investment decisions of firms (OECD, 2020a) . (Saurav et al., 2020) highlight that the COVID-19 crisis represents for international enterprises an unprecedented source of investor risk that is depressing investor confidence. This effect may be particularly important for emerging and developing economies, where alternative domestic sources of financing are scarce, so that the overall impact of the pandemic on emerging economies may be particularly severe. Financial markets in emerging economies are typically more vulnerable to global risk sentiment, Therefore, their funding conditions have historically proved to be much more volatile than those of advanced economies. As a predictable consequence, since the onset of the pandemic, developing financial markets experienced a sharp deterioration in investor sentiment, and risk appetite turned into a sudden and adverse reversal of capital flows (OECD, 2021) . Recent reports have indeed documented a flight-to-quality away from emerging countries' liabilities as a consequence of the pandemic (Hevia and Neumeyer, 2020; Levy Yeyati and Valdés, 2020; Bolton et al., 2020) . This paper empirically tests the extent of the flight of international investment to advanced economies, as a first response to the COVID outbreak in the first half of 2020.

Considering the growth in foreign liabilities as the dependent variable, and after partialling out the growth in the stock market, the severity of the crisis and the stringency of the public containment policies, we confirm the presence of this flight. Amid a generalized decline in foreign investment, we observe that countries featuring higher GDP per capita, belonging to the groups of advanced, G7, or Euro area countries, are significantly less severely hit by the pandemic than emerging countries. In particular, the wedge between advanced and emerging countries is about 3%, and it is at least twice as large for G7 countries, after the first quarter of 2020. Moreover, foreign investors appear to be momentum traders, since a higher (lagged) growth of the equity market index in one country is associated with a higher growth of foreign liabilities. This paper is closely related to the recent work by Giofre' (2021): both of them rely on a similar dataset, and they both investigate the evolution of foreign liabilities at the onset of the COVID virus's spread. However, they differ in their objectives and their contributions to the literature. Giofre' (2021) investigates the different behavior of portfolio and direct investors in order to identify the impact of COVID-induced stringency measures on different categories of foreign investors. She finds that, at the end of the first quarter of 2020, while the stringency measures have no effect, their standard deviation is positively and significantly correlated with inward investments, but only when portfolio investors are considered.

Conversely, this paper considers total foreign liabilities, held by both portfolio and direct investors, because the objective is to detect the presence and magnitude of the flight-to-advanced-economies effect. Moreover, the paper contributes to the literature by testing the hypothesis of momentum trading by foreign investors, during the COVID crisis, thus providing results which can be interpreted in light of the abundant literature connecting investor sentiment with flight-to-quality.

These findings help shed light on the direction of foreign investment in periods of crisis, when cross-border liquidity dries up: in these moments of financial fragility, the emerging countries, which are more reliant on incoming capital are also the ones that suffer the most. Since foreign investors are mostly institutional investors, these results also add to the literature focussing on the behavior of institutional investors in times of crisis: they behave as momentum traders during the COVID pandemic's outbreak, as they have done in previous crises (Baltzer et al., 2019; Bijlsma and Vermeulen, 2016) , and show a significant propensity to leave their positions in emerging economies in favor of advanced ones .

The rest of the paper is structured as follows. In Section 2, we frame our analysis within the literature. In Section 3, we describe the data and provide some descriptive statistics. In Section 4, we sketch the estimable equation. In Section 5, we report the results of the empirical analysis. Section 6 concludes.

Investors typically engage in international investment to improve portfolio performance by reducing the risk of a loss. In times of financial distress, however, the benefits from diversification in different markets decrease because of a stronger interconnectedness among markets. For emerging economies, the lower attractiveness induced by the increased returns correlations with developed countries is paired with a greater institutional fragility and vulnerability to global risk sentiment driven by a spike in investor risk aversion.

This combination of factors makes the borrowing conditions of developing economies historically more volatile than those of advanced economies, so that, in times of crisis, they easily lead to a sudden reversal of capital flows. This paper studies the evolution of international investment at the onset of the COVID pandemic, and therefore relates to two overlapping strands of the empirical finance literature.

The first strand deals with the first factor mentioned above: that is, interconnectedness and financial contagion among markets. Over the past two decades, financial markets have experienced in general a surge of greater international integration (Gamba-Santamaria et al., 2019) .

The correlation between markets is not symmetric, because it is stronger during periods of high volatility than it is during periods of low volatility (Forbes and Rigobon, 2002; Ang and Bekaert, 2015; Kundu and Sarkar, 2016) . According to the definitions in Forbes and Rigobon (2002) , when a high co-movement is present but conditions are stable, we are in the presence of interdependence among markets; instead, contagion occurs if co-movement increases when a crisis arises.

The models and techniques adopted in the literature to explore interdependence and contagion range from dynamic conditional correlation models (Kenourgios et al., 2011) , through regime-switching models (Baele et al., 2007) and copulas (Rodriguez, 2007) , to wavelet-approaches (Gallegati, 2012; Dimitriou et al., 2020; Amar and Carlotti, 2021) . Yarovaya et al. (2017) explore the concept of asymmetry in return and volatility spillovers across markets. They find that the transmission of negative return shocks and positive volatility shocks dominate, and that the strongest asymmetry occurs in the case of market pairs where the recipient is an emerging market.

The stronger financial integration and interdependence between markets documented in the literature has naturally exacerbated the cross-border effects of the COVID-19 pandemic. Fassas (2020) investigates the connectedness in terms of investors' risk aversion and finds that spillovers exist not only in the returns and volatility of financial assets, but also in market participants' sentiment. In particular, he finds that while the USA was the largest transmitter of sentiment connectedness over the last decade, emerging markets have become powerful transmitter of spillovers, during the COVID pandemic. Akhtaruzzaman et al. (2021) show a significant increase in dynamic conditional correlations in stock returns between China and G7 countries during the COVID period. They also show that optimal hedge ratios increase significantly in most cases, implying higher hedging costs during the period in question.

The second strand of literature is intertwined with the first, but it is mainly related to risk aversion and investor sentiment. The sharp deterioration in investor sentiment and risk appetite drives investors out of emerging financial markets into developed ones, thus fueling one of the many variants of the flight-to-quality effect.

In recent decades, measures of risk aversion and 'sentiment' (Baker and Wurgler, 2007) have received close attention as tools with which to monitor the volatile economic environment (Brière et al., 2012) . The role of investor sentiment in asset pricing has been widely investigated in the literature (Baker and Wurgler, 2007; Da et al., 2014; Stambaugh et al., 2012) . Lee et al. (2002) show that excess returns in the US stock market are contemporaneously positively correlated with shifts in sentiment, and that the magnitude of bullish (bearish) changes in sentiment leads to higher (lower) future excess returns and downward (upward) revisions in volatility. Similar findings have been reported by Bandopadhyaya et al. (2006) , who show that investor sentiment quickly captures relevant news events, and that this sentiment measure accounts for a significant proportion of the changes in the stock market index. Baker et al. (2012) study the international time-series of the cross-section of stock returns. They indicate that these findings are not limited to the USA but extend to the international context, and that sentiment is contagious across markets. On studying differences in investor sentiment between developed and emerging markets, Wang et al. (2021) document that investor sentiment has a more immediate impact in emerging markets, but a more enduring impact in developed ones. Investors in different markets, developed or emerging, may have different distributions of misperceptions, so that the impacts of investor sentiment on stock returns are also likely to be different.

When dramatic unexpected events occur, investor sentiment immediately reacts, and Knightian uncertainty induces investors to shed risky assets in favor of safer claims: that is, it induces a flight-to-quality (Caballero and Krishnamurthy, 2008) . Baele et al. (2019) document, on daily data for 23 countries, specific flights to both quality and liquidity in international equity markets which mainly consist of a flight-to-quality in the US corporate bond market. Bayraci et al. (2018) document a positive co-movement between stock returns and changes in 10-year government bond yields, which signals a flight-to-quality behavior in G7 countries as a result of dramatic changes in investor sentiment and risk aversion at times of market stress. Dimitriou et al. (2020) , by means of a Wavelet Coherence Analysis, find that investors and portfolio managers should search for 'new' safe haven assets during periods of turmoil, such as the global financial crisis and the Eurozone sovereign debt crisis, in order to improve their portfolio diversification strategies.

Another variant of the flight-to-quality phenomenon can be observed across countries, rather than across instruments. Cho et al. (2016) show that, in global down markets, capital tends to move from emerging to developed countries. In regard to the COVID pandemic, Papadamou et al. (2021) study the time-varying correlation between stock and bond returns for ten countries during the first quarter of 2020, and they find flight-to-quality episodes. Gupta et al. (2021) focus on the impact of infectious diseases-related uncertainty on the safe-haven characteristic of various assets, and find evidence supporting the hypothesis that the US treasury securities have been used as a safe-haven during the COVID-19 outbreak. Finally, a collection of recent reports provides evidence of flight-to-quality away from Latin America and other emerging economies (Hevia and Neumeyer, 2020; Levy Yeyati and Valdés, 2020; Bolton et al., 2020) . Specifically, Hevia and Neumeyer (2020) document that between 24 February and 30 March 2020, institutional and retail money funds in the US increased their assets by 19%, as a consequence of a fast and huge flight-to-quality and sudden capital outflows from emerging economies.

Our paper reports a systematic and rigorous analysis of the flight to advanced economies during the COVID pandemic, as suggested by the above-reported anecdotal evidence, and provides a quantitative assessment of this effect, within a multivariate framework.

The growth of foreign liabilities, our dependent variable, relies on quarterly data on 53 countries (International Investment Position Statistics, IMF).

The main regressor is a binary variable associated with the status of a country as a member of the 'advanced countries', G7 and Euro groups, following the Economy grouping classification of the Fiscal Monitor database, released by the IMF's Fiscal Affairs Department. An alternative proxy for economic development adopted is the high GDP per capita dummy, a time-invariant binary variable equal to 1 if the GDP per capita is larger than the sample median, and 0 otherwise, based on data collected from the CEIC database. From the same database, we also drew the monthly equity market index data (1-month lagged), which captures the multiple drivers of demand and supply in the recipient stock market.

The NEER (Nominal effective exchange rate, broad index), released by the Bank for International Settlements, is also included in the analysis, since its change might affect foreign investment.

Finally, the source of COVID-related data is a Github ongoing repository of data on coronavirus, the Coronavirus Open Citations Dataset. We drew from this dataset the stringency index ( ), which represents a proxy for the severity of the containment policy measures adopted, and the data on new COVID-deaths and cases per million of inhabitants. These data are originally reported at a daily frequency, but in order to match the quarterly frequency of the dependent variable, we constructed quarterly averages.

In Fig. 1 , we report the distribution and main descriptive statistics of the dependent variable, the growth in foreign liabilities (in %), at the end of the first quarter (panel a) and first semester (panel b): we observe that in panel (a), the average growth measure is about −4%, with a median of −6%, while, in panel (b), the median and mean are about −3%.

In Fig. 2 , we report a similar graphical representation for the growth of the equity market index (in %): in the first quarter (panel a), the mean is -7%, while in the first semester as a whole the drop is almost twice as large (−13% on average).

Our objective is to identify the presence and extent of a flight of foreign investment to advanced economies, at the onset of the COVID-19 outbreak.

We define the growth of liabilities ( ) in the first quarter ( 1) of 2020 as 1: this is the difference between the liabilities at the end of the first quarter of 2020 (March 2020, 03_20 ) and the liabilities at the end of 2019 (December 2019, 12_19 ), scaled by the liabilities at the end of 2019:

(1)

Analogously, for the first semester of 2020, we compute:

To address potential concerns about investment seasonality, we consider the dependent variable . For the first quarter of 2020, 1 is the difference between the 2020 measure, as defined in Eq. (1), and the corresponding measure in 2019:

Analogously, for the first semester of 2020, we compute:

To estimate if and to what extent the growth in foreign liabilities, , differs for advanced economies, we run the following regression:

where ( _ ) is a binary variable taking value 1 if the recipient country considered belongs to the group of advanced economies, and 0 otherwise.

As far as the covariates falling within the residual group of ' ' are concerned, to be noted is that, unfortunately, the low number of observations forced us to state a parsimonious specification. However, luckily, the definition of the dependent variable in difference form allowed us to ignore any country-specific fixed effects, because these were removed by construction.

To estimate the parameters in Eq. (5), we adopted a Robust Least Squares estimation, which is a regression method specifically designed to be robust, or less sensitive, to outliers. 1 In the robustness check section, we subject our results to a standard OLS model and a Quantile regression at the median.

To investigate the presence and magnitude of the flight of international investment to advanced economies, we regress the growth of foreign liabilities on the binary variables capturing the 'advanced' dummy, and on a bunch of control covariates, which are progressively added to the econometric specification: we aim to test the flight-to-quality hypothesis, according to which foreign investors, as a consequence of the COVID outbreak, would deviate their investments to more stable and developed economies.

We first consider countries with a high GDP per capita, i.e., a GDP per capita larger than the median; then the 'advanced economies' as defined by the Economy grouping classification of the IMF, and, finally, two sub-groups, G7, and Euro area, to detect possible differential effects for the top advanced economies (G7), and for one specific developed area (Euro area countries).

We perform a regression analysis under a Robust Least Squares estimation, relative to the end of the first quarter (columns (#a)), and first semester of 2020 (columns (#b)).

The basic specification, in Table 1 , includes two control variables besides the 'advanced' dummy. The first is the growth in the Nominal Effective Exchange Rate (NEER), a measure of the appreciation of the economy's currency against a broad basket of currencies, because its change may affect foreign investment. 2 The second control variable is the number of new COVID-deaths per million of inhabitants, a direct health indicator of the epidemic. Indeed, the growing number of recent studies on the impact of the COVID event on financial markets generally finds evidence of a significant impact of confirmed COVID cases or deaths on financial markets' volatility and liquidity (Albulescu, 2021; Baig et al., 2021; Salisu and Vo, 2020; Ashraf, 2020) .

We observe a significant negative coefficient of the constant term. When the advanced dummy is included, the constant's coefficient represents the average dynamic of the dependent variable for the excluded 'non-advanced' group. The size and significance of the constant's coefficient is then consistent with the average decrease in foreign investment after the COVID outbreak, as already observed in Fig. 1 .

Depending on the grouping dummy considered, this general decrease ranges from −6% to −10% in the first quarter, and from −4% to −7%, in the semester as a whole, thus showing a slow recovery in the second quarter of 2020. This evidence is also consistent with the literature documenting the general decline in international investment during crisis periods (Milesi-Ferretti and 1 Among Robust Least Squares, we adopt the M-estimation developed by Huber (1973) , with Huber Type III standard errors (Huber (1981) , pp. 173) Our results are robust to other alternative Robust Least Squares methods, such as the S-estimation and the MM-estimation. 2 We include its one-month lagged value, to avoid endogeneity issues. See Appendix A, for a definition of this covariate. (#b), 1 refers to the end of the first semester of 2020, as defined in Eq. (4). ***, **, and * indicate significance at the 1, 5, and 10% levels, respectively.

Control for growth in equity market index (1-month lagged).

This table follows the same structure as Table 1 , with an additional covariate: the growth in the equity market index (diff equity market index). Tille, 2011; Cetorelli and Goldberg, 2011; Raddatz and Schmukler, 2012; Haas and Horen, 2013) . The coefficients of the developed economies' dummy are all positive and statistically significant in the first quarter (columns (#a)): recipient countries with a GDP per capita higher than the median experience a 4.4% larger growth in foreign liabilities, while for countries classified as 'advanced' the percentage is 3.4. When the two sub-groups, G7 and Euro area, are considered the percentages are 5.2 and 3.8.

The other control covariates do not display any statistically significant impact on the dependent variable. In the first semester (columns (#b)), the results are instead non systematic, with a significant coefficient only for the Euro area dummy (2.8%) and for the appreciation of the exchange rate. This table follows the same structure as Table 2 , with two additional regressors: the Stringency Index ( ) and the standard deviation of the stringency index ( ).

The recent literature has emphasized, during crisis periods, a distinctive momentum trading behavior of financial institutions and foreign investors, with a substantial increase in sales of past loser stocks during the market decline (Bijlsma and Vermeulen, 2016; Baltzer et al., 2019) . In particular, Baltzer et al. (2019) document that foreign investors, which are predominantly institutional investors, increased their momentum trading in correspondence to the volatile periods connected with the evolution of the financial crisis 2007-2008. Bijlsma and Vermeulen (2016) more specifically analyze the behavior of insurers, which are among the largest global investors, and which typically have relatively large foreign asset positions. They find that, whereas insurers are generally contrarian or neutral traders, they may turn into momentum traders during a flight-to-quality related to a crisis, such as the European sovereign bond one.

Since we focus precisely on the flight-to-quality of foreign investment during the COVID crisis, we need to account for momentum trading.

In Table 2 , we include the 1-month lagged growth of the stock market index in order to check for the presence of this momentum trading behavior. 3 To be noted is that, on the one hand, this index is related only to the stock market, while our dependent variable includes all types of financial liabilities; on the other hand, this is a country index, while the dependent variable refers only to foreign holdings, thus excluding domestic ones. However, because of the spillover among financial instruments and across borders, this index can reflect the global investor sentiment towards the country, which is driven by both fundamental and behavioral factors. The inclusion of this covariate, therefore, besides making it possible to control for momentum trading, also partially alleviates the problem of omitted variables connected with our parsimonious specification, because it captures a synthesis of the drivers of investors' decisions, also those of the foreign investors. 4 8 M. Giofré Table 4 Sensitivity analysis -Additional regressor: 'new COVID-cases per mn'. This table replicates Table 3 , with an additional covariate: 'new cases of COVID per mn' (and its standard deviation).

The results indeed provide evidence of the momentum trading behavior of foreign investors: the coefficient of the growth of the equity index is positive and statistically significant, at the end of the first quarter, in all specifications; the effect is also economically significant: a 1% increase in the growth of the equity index induces an increase in the growth of total liabilities ranging from 16.2% to 19.3%. After partialling out the growth of the stock index, the effect of the 'developed countries' dummy is still positive and significant, with a reduction of the coefficient of the 'advanced' dummy (from 3.4% to 2.4%) and of the Euro area dummy (from 3.8% to 3.2%). As far as the first semester results are concerned, we observe that the Euro area dummy slightly increases (from 2.8% to 3.1%) and the 'advanced' dummy becomes significant (4.6%).

As a consequence of global COVID pandemic, governments have been forced to introduce restriction policies, differing in terms of severity and timing. On the one hand, some countries imposed very harsh measures immediately after the outbreak of the crisis, and then removed them after the evidence of a reduction in contagion; on the other hand, other countries chose to react to the epidemic's spread with moderate increases and reductions of containment measures (Hale et al., 2020) . Ahmed et al. (2020) emphasize that the access of emerging economies to international capital markets is strongly influenced by the health of global financial markets, but also by the extent of the spread of the COVID virus, and by the restrictive measures that it has prompted.

We therefore include the Stringency Index ( ), a measure of the severity of the containment policy measures adopted by different governments to react to the COVID virus's spread. We also add its standard deviation, since Giofre' (2021) finds that a higher within-country standard deviation in the stringency index positively affected foreign portfolio investors, at the end of the first quarter of 2020. 5 9 M. Giofré This table replicates Table 3 , but the country sample excludes China.

In Table 3 , we add the stringency index ( ) and its standard deviation ( ) to the econometric specification; we also add the standard deviation of the number of new COVID deaths per mn, for consistency and to allow comparability with Giofre' (2021).

We observe that the stringency index and its standard deviation do not affect inward investment, while both the growth of equity market index and the grouping dummies show positive, statistically significant, and economically relevant coefficients.

In Tables 4 to 8, we subject our findings to several sensitivity analyses. In Table 4 , we add the pandemic indicator 'number of new COVID cases per mn' to the one referred to COVID deaths. Tables 5 and 6 consider different country sample specifications, excluding, respectively, China and offshore countries. The last two tables consider alternative estimation methods, such as OLS (Table 7) and Quantile regression (Table 8) .

In Table 4 , we add another pandemic indicator -the variable 'new COVID-cases per mn of inhabitants' (and, for consistency, its standard deviation) -besides the one relative to the number of new COVID deaths per mn. Ashraf (2020) finds that stock markets reacted more proactively to the growth in the number of confirmed cases as compared to the growth in the number of COVID deaths. The two indicators are, predictably, quite correlated ( = 0.57 in the first quarter and = 0.30 in the first semester). The coefficients of the grouping variables and the momentum trading are qualitatively unchanged for the first quarter, whilst in the first semester, only the coefficient of the Euro area is confirmed as significant. 6 6 In Table 10 in Appendix B, we report the results of the regression in which the 'new COVID-cases per mn of inhabitants' is included as an alternative to 'new COVID-deaths per mn of inhabitants'. The coefficients of the variables of interest remain substantially unchanged. This table replicates Table 3 , but the country sample excludes offshore countries (according to the classification in Damgaard et al. (2018) ).

In Tables 5 and 6 , we test whether our findings are robust to the exclusion from the sample of some countries that may have driven the results.

In Table 5 , we exclude China from the sample of countries. China was the first country to be hit by COVID, several weeks before other countries. Our findings may therefore be distorted by China's asynchronous timing of lockdown and loosening measures in the first and second quarter of 2020.

In Table 6 , adopting the classification proposed by Damgaard et al. (2018) , we exclude from the sample potential offshore financial centers, to ensure that our results are not driven by economies distorting investors' decisions for reasons that are hard to control in our analysis. From our original sample, Hong Kong, Ireland, Luxembourg, the Netherlands and Singapore are excluded.

The exclusion of China and offshore centers confirms our findings, both qualitatively and quantitatively. 7 Interestingly, under these specifications the coefficients of the standard deviation of the stringency index become statistically significant. This finding is consistent with the results of Giofre' (2021), where, in the restricted samples, the coefficient of the standard deviation of the stringency index was positive and significant, not only for the growth in portfolio liabilities, but also for the growth in total liabilities, our dependent variable.

In Tables 7 and 8 , we run the same regression specification of Table 3 , but using alternative estimation techniques.

The results under the OLS specification in Table 7 are similar to the benchmark, with the exception of the coefficient of the 'advanced' economies in column (2a), which loses statistical significance. Table 8 reports the results of a Quantile regression. The quantile regression estimates the conditional median, rather than the conditional mean of the response variable as in the method of least squares, and therefore, similarly to the Robust Least Squares 7 Alternative offshore classifications are adopted in Tables 11 and 12 of Appendix B, and the associated regressions deliver similar results. See Appendix A, for details on the countries excluded in various sample specifications. This table replicates Table 3 , but the model is estimated through standard OLS techniques. method adopted in the main specification, it is more robust against outliers in the response measurements. The results are similar to the standard ones, with the exception of the coefficient of the 'high GDP' binary variable, which loses statistical significance.

This paper tests the existence and extent of the flight to advanced economies by foreign investors in the aftermath of the COVID outbreak. Amid a generalized decline in international investment typical of distress periods, we observe that advanced countries with a higher GDP per capita, belonging to the G7 group, or to the Euro area have been significantly less severely hit by the pandemic than emerging countries. In particular, comparing the growth in foreign liabilities at the end of the first quarter of 2020, the wedge between advanced and emerging countries is about 3%, and it is at least twice as large for G7 countries. This wedge is statistically significant in the first quarter horizon, and it is paired with systematic momentum trading by foreign investors. Our results are robust to the inclusion, as controls, of government stringency measures, alternative measures of pandemic severity, sample specification and regression methods.

Our findings also provide information to regulators and market participants on how foreign investors behave during periods of financial turmoil: understanding the drivers and the directions of investors' sentiment is not only of interest for asset pricing, but it has important implications for portfolio diversification and investment strategies. The flight-to-quality occurs in periods of high volatility, and this phenomenon is strictly monitored both by portfolio managers for trading strategies and by investors for its effects on the overall stability of the financial system.

Finally, the reported evidence suggests that policymakers should continue to provide liquidity to developing economies. This is because in moments of financial fragility, when cross-border liquidity dries up, the emerging countries -which are more reliant on incoming capital -are the ones that suffer the harshest reversal of capital flows.

The availability of a longer time span may stimulate further future research on the topic addressed by this paper, and provide a thorough overview on the evolution of international investment during and after the COVID pandemic. This table replicates Table 3 , but the model is estimated through a Quantile regression (at the median).

The impact of COVID-19 on emerging markets economies' financial conditions

Albulescu, C., 2021. COVID-19 and the United States financial markets' volatility

Who drives the dance? Further insights from a time-frequency wavelet analysis of the interrelationship between stock markets and uncertainty

International asset allocation with regime shifts

Stock markets reaction to COVID-19: Cases or fatalities?

The determinants of stock and bond return comovements

Flights to safety

Deaths, panic, lockdowns and US equity markets: The case of COVID-19 pandemic

Global, local, and contagious investor sentiment

Who trades on momentum?

Measuring investor sentiment in equity markets

Stock-bond co-movements and flight-to-quality in G7 countries: a time-frequency analysis

Emerging and advanced economies markets behaviour during the COVID-19 crisis era

Insurance companiesš trading behaviour during the European sovereign debt crisis: Flight home or flight to quality?

COVID-19 in latin america: How is it different than in advanced economies

No contagion, only globalization and flight to quality

Collective risk management in a flight to quality episode

Global banks and international shock transmission: Evidence from the crisis

The COVID-19 pandemic and stock liquidity: Evidence from S&P 500

Flight-to-quality and correlation between currency and stock returns

The sum of all FEARS investor sentiment and asset prices

Piercing the veil

Are there any other safe haven assets? Evidence for şexoticť and alternative assets

Risk aversion connectedness in developed and emerging equity markets before and after the COVID-19 pandemic

A wavelet-based approach to test for financial market contagion

Volatility spillovers among global stock markets: measuring total and directional effects

The flight home effect: Evidence from the syndicated loan market during financial crises

Infectious disease-related uncertainty and the safe-haven characteristic of US treasury securities

Running for the exit? International bank lending during a financial crisis

Risk of openness index: When do government responses need to be increased or maintained?

A perfect storm: COVID-19 in emerging economies

Robust regression: Asymptotics, conjectures and Monte Carlo

Robust Statistics

Financial crises and stock market contagion in a multivariate time-varying asymmetric framework

Return and volatility interdependences in up and down markets across developed and emerging countries

International Financial Integration in the Aftermath of the Global Financial Crisis

Stock market volatility, excess returns, and the role of investor sentiment

COVID-19 in Latin America: How is it different than in advanced economies

COVID-19 and stock market nexus: evidence from shanghai stock exchange

The great retrenchment: international capital flows during the global financial crisis

Foreign direct investment flows in the time of COVID-19. Tackling coronavirus (COVID-19): Contributing to A global effort

Global financial markets policy responses to COVID-19. WP, OECD directorate for financial and enterprise affairs

OECD investment policy responses to COVID-19. Tackling Coronavirus (COVID-19): Contributing To A Global Effort

Flight-to-quality between global stock and bond markets in the COVID era

On the international transmission of shocks: Micro-evidence from mutual fund portfolios

Measuring financial contagion: A copula approach

Predicting stock returns in the presence of COVID-19 pandemic: The role of health news

The impact of COVID-19 on foreign investors: early evidence from a global pulse survey. Global Investment Climate

The short of it: Investor sentiment and anomalies

Investor sentiment and stock returns: Global evidence

Asymmetry in spillover effects: Evidence for international stock index futures markets

Supplementary material related to this article can be found online at https://doi.org/10.1016/j.mulfin.2022.100735.