key: cord-1035857-n1zpez5c
authors: Giofré, Maela
title: COVID-19 stringency measures and foreign investment: An early assessment
date: 2021-08-22
journal: The North American Journal of Economics and Finance
DOI: 10.1016/j.najef.2021.101536
sha: c022fb7a1594691e71361146bcdf22491987ffd7
doc_id: 1035857
cord_uid: n1zpez5c

This paper investigates the evolution of foreign investment in the immediate aftermath of the implementation of COVID-19 government stringency measures. The average stringency index is not correlated with inward investment positions. However, after removing country fixed-effect and controlling for the severity of the outbreak spread, the within-country standard deviation of the stringency index is positively and significantly correlated with inward portfolio investments, at the end of the first quarter of 2020. At the end of the second quarter, the same dispersion measure is instead not associated with a significant change in inward investment. We interpret this evidence as follows. Foreign portfolio investments, typically more volatile and reactive than foreign direct ones, are more responsive, at the end of the first quarter, to governments’ prompt interventions than to gradual ones, thus suggesting that the former policies are perceived as a more serious commitment to stem the spread of COVID-19. In the second quarter, instead, the standard deviation of the index captures the abrupt retreats of the containment measures, together with the timely adoption of policies, thus becoming less informative for foreign investors.

converges on the evidence of a signi…cant impact of COVID con…rmed cases or deaths on …nancial markets' volatility and liquidity. Albulescu (2020) empirically investigates the e¤ect of the o¢ cial announcements regarding the COVID-19 new cases of infection and fatality ratio on the …nancial markets volatility in the United States, and …nds that the coronavirus pandemic is an important source of …nancial volatility. Similarly, Baig et al. (2020) …nd that increases in con…rmed cases and deaths due to coronavirus in the US are associated with a signi…cant increase in market illiquidity and volatility, while declining sentiment and the implementations of restrictions and lockdowns contribute to the deterioration of liquidity and stability of markets. Salisu and Vo (2020) …nd that health-news trends are good predictor of stock returns since the emergence of the pandemic. Li et al. (2021) highlight that both the Covid-19 cases and deaths signi…cantly induce the decline in the Shanghai Stock Exchange index. Ashraf (2020) …nds that stock markets in 64 countries responded negatively and quickly to the growth in COVID-19 con…rmed cases, with a response varying over time and depending on the stage of outbreak.

As far as international investments are concerned, Saurav et al. (2020) highlight that the COVID-19 crisis represents for international enterprises a new and unprecedented source of investor risk that is depressing investor con…dence. OECD (2020a) and OECD (2020d) assess that foreign direct investments of …rms are expected to decline sharply as a consequence of the pandemic and of the stringent public health measures to limit the spread of the COVID-19. Portfolio investments, typically more volatile and reactive than direct investment, react even earlier to the shock that the pandemic in ‡icted on the global economy: emerging market economies have indeed already experienced a massive drop of portfolio investment in ‡ows, because international investors transfer capital back home, or invest in safer assets during periods of uncertainty. Kizys et al. (2021) study the e¤ects of the Oxford COVID-19 Government Response Tracker, whose higher scores are associated with greater stringency, on herding behavior in international stock markets during the coronavirus COVID-19 outbreak. They disclose the presence of herding behavior in the …rst three months of 2020 in 72 countries stock markets' countries, but also highlight that this herding behavior is mitigated by a more stringent government response to the coronavirus crisis, by way of reducing multidimensional uncertainty.

On a purely theoretical level, the relationship between stringency measures and foreign investment is far from obvious. Foreign investors could indeed be averted from investing in a country adopting more radical stringency measures, because it could entail a recession period making less pro…table the assets issued by that country; conversely, foreign investors could be even allured by the assets issued by countries adopting more radical containment policies, because these could be perceived as a severe immediate cost to avoid even higher costs in the near future. The …nal balance of these forces is therefore an empirical matter, and this paper contributes to the existing literature by empirically assessing whether and how foreign investment has reacted in the immediate aftermath of the adoption of COVID-19 government stringency measures.

We …nd that the average stringency index is not correlated with inward investment positions.

Also building on the survey by Brodeur et al. (2021) cited above, which emphasizes that the e¤ects may vary not only by the stringency of the social distancing measures, but also by their length of implementation, or degree of compliance, we conjecture that the graduality in the introduction of these policies could matter. In fact, as pointed out by Hale et al. (2020) , as the disease has spread around the world, the governments'restriction policies have di¤ered across countries and over time:

some have rapidly introduced very strict measures in the immediate aftermath of the crisis, such as total lockdown, and then have removed them, as a consequence of a reduction in community transmission; other countries instead reacted with less severe rise and fall of containment measures, as small outbreaks occurred.

We check if this heterogeneity in graduality in the adoption of the stringency measures has signi…cantly a¤ected foreign investors. After partialling out the severity of the outbreak, as captured by new deaths or new cases, and removing country …xed-e¤ects, we observe that, in the …rst quarter of 2020, a higher within-country standard deviation in the stringency measures seems to have made the adopting economy relatively more attractive to foreign portfolio investors. We provide a temptative interpretation of this evidence: the within-country standard deviation could proxy the timeliness of governments'action, in the immediate aftermath of the COVID outbreak. If the prompt adoption of containment measures is interpreted as a serious commitment to restrain the uncontrolled spread of the virus, it could have fostered cross-border inward investment.

The remainder of the paper is structured as follows. In Section 2, we sketch the estimable equation. In Section 3, we describe the data, and provide some descriptive statistics. In Section 4, we report the results of the empirical analysis, with some robustness checks. Section 5 concludes.

Our objective is to assess the evolution of international investments in the aftermath of the adoption of COVID-19 containment measures.

Speci…cally, we empirically test the existence of a relationship between the stringency index in a country and its foreign liabilities or inward investments.

Let's de…ne, …rst, the growth of liabilities as the change from period t to period t + 1, divided by its level t's initial value:

In particular, when considering the …rst quarter (q1) of 2020, the growth of liabilities q1, is the di¤erence between the liabilities at the end of the …rst quarter (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

We consider an alternative de…nition of the dependent variable in the analysis.

To partial out the seasonality of foreign investment allocations, we consider the measure dif f , that is, the di¤erence between the 2020 measure, as de…ned in equation (1) (2)

For instance, dif f q1 is the …rst quarter measure, and is de…ned as the di¤erence between the measure in equation (1a) and its counterpart in 2019, as follows:

We compute this growth in liabilities for the …rst two quarters ( q1 and q2) and for the …rst semester ( s1) of 2020, for di¤erent types of liabilities: total, foreign direct and foreign portfolio inward investments.

We regress the growth in foreign liabilities, di¤erently de…ned, on the average stringency index in the country (SI); and on its within-country standard deviation ( SI), running the following regression:

= + (SI) + ( SI) + controls + "

We are mainly interested in testing the signi…cance, sign and size of the and coe¢ cients. If the adoption of stringent containment measures (SI) deters (or attracts) foreign inward investment, then we should observe a signi…cant negative (or positive) coe¢ cient. As anticipated in the …rst section and better explained in next sections, we also include a measure of the within-country dispersion of the stringency ( SI): a negative would entail an induced reduction in attractiveness, while a positive could reveal an appreciation by foreign investors.

We trade-o¤ a parsimonious speci…cation, due to the low number of observations, with the need to include time-varying regressors, which might contribute to explain the growth in foreign investments, and covariates potentially correlated with our main regressors, whose exclusion could bias the because its change might have a¤ected foreign investment. Second, we control for the number of new COVID-deaths and its within-country standard deviation, as the stringency index is potentially strongly correlated with the health indicators of the epidemic. Finally, we include two binary indicators of economic and …nancial development, to control, for instance, for the presence of any eventual ‡ight to quality propensity by foreign investors.

To estimate the parameters in equation (3) 

We consider inward investment in 53 countries, upon data availability. Data on foreign liabilities are drawn from the International Investment Position Statistics, released by the IMF, which provides information on foreign assets and liabilities, classi…ed in several categories and instruments, at a quarterly frequency. In our analysis, total liabilities (total inward investments) are also split into foreign direct and foreign portfolio investments.

The source of COVID-related data is a Github ongoing repository of data on coronavirus, the Coronavirus Open Citations Dataset.

We draw from this dataset our main regressor, the stringency index (SI), which represents a proxy for the severity of the containment policy measures adopted, and the data about 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 construct quarterly averages and within-country quarterly standard deviations.

We include in our speci…cation other three controls. First, the NEER (Nominal e¤ective exchange rate, broad index), released by the Bank for International Settlements. Then, we include two binary indicators of economic and …nancial development, i.e., the GDP per capita and the market capitalization per GDP, drawn from CEIC data.

In Figure 1 and 2, we report the distribution of the dependent variable. Figure 1 relies on the measure de…ned in equation (1), for both the …rst quarter ( q1, panels #a) and the second quarter ( q2, panels #b): Panels (1a) and (1b) refer to total inward investments, panels (2a) and

(2b) to foreign direct investment, while panels (3a) and (3b) refer to portfolio investments. We can observe, …rst, that the measure is more negatively skewed in the …rst quarter, than in the second quarter. Second, the distribution of portfolio inward investment is more negatively skewed than the distribution of direct investment. Figure 2 is similar to Figure 1 , but relies on the dif f measure de…ned in equation (2): a comparison across quarters and types of investments (direct or portfolio) reveals the same pattern observed in Figure 1 . By comparing Figure 1 and 2, we can notice that, in the …rst quarter, the distribution of the measure controlling for seasonality, de…ned in equation (2a), is more negatively skewed than the one relying on equation (1a), especially for portfolio investment.

In Figure 3 , we report the distribution and main descriptive statistics of the quarterly stringency index and its within-country standard deviation. Panels (#a) refer to the …rst quarter, while panels (#b) refer to the second quarter. Panels (1a) and (1b) focus on the average quarterly stringency index (SI), while panels (2a) and (2b) refer to the within-country standard deviation ( SI).

The average stringency index, whose original values range 0-100, in the …rst quarter is about 19, while in the second quarter is about 71, thus disclosing the dramatic tightening of the anti-COVID 19 containment measures.

However, panels (2a) and (2b) provide an additional piece of information on the stringency index:

by comparing the within-country standard deviation in the two quarters, we observe that its average 8 J o u r n a l P r e -p r o o f is signi…cantly larger in the …rst than in the second quarter (26 versus 10). From Figure 3 , we learn that, though the average stringency index SI in the …rst quarter is about one-fourth of its level in the second quarter, the average within-country standard deviation is about 2.5 times larger: the adoption of containment measures has been more abrupt in the immediate aftermath of the COVID spread, in order to face the challenge of the unprecedented event and its severe consequences. 2 4 Regression analysis

In Table 1 , we report the main …ndings of our regression analysis for the …rst quarter, under a Robust Least Squares estimation. Columns (1a) and (1b) report results relative to total foreign inward investment, columns (2a) and (2b) refer to foreign direct investments, while columns (3a) and (3b) refer to portfolio investments. Columns (#a) rely on the liabilities growth measure , whose structure is de…ned in equation (1), while columns (#b) rely on the dif f measure, whose structure is de…ned in equation (2).

As anticipated in Section 2, we are forced to keep a parsimonious speci…cation, because we can rely on a quite limited country sample.

It is worth stressing that the dependent variable is de…ned in di¤erence form, which allows us to ignore any problem related to country-speci…c …xed e¤ects, removed by construction.

We are however concerned, on the one hand, about time-varying regressors which might concur to explain the growth in foreign investments, and, on the other hand, about covariates potentially correlated with our main regressors, whose exclusion could bias the estimated coe¢ cients.

We include the (one-month lagged) growth in the Nominal E¤ective Exchange Rate (NEER), a measure of the appreciation of the economy's currency against a broad basket of currencies, because its change might have a¤ected foreign investment. 3 Second, we control for the number of new COVID-deaths and its within-country standard deviation: the stringency index is likely correlated with this speci…c indicator of the epidemic, as it represents the government reaction to contain new cases, deaths, and intensive-care treatments. 4

Finally, we include two binary indicators of economic and …nancial development, to account for a potential di¤erent ‡ow of foreign investments towards high versus low developed countries:

for instance, according to the ‡ight to quality rationale, in the presence of a global shock, foreign investors would deviate their investments to more stable and developed economies. 5

First of all, we observe a signi…cant negative coe¢ cient of the constant term. The constant's coe¢ cient represents the mean of the dependent variable, if all regressors are set to zero. We observe that the dynamics of the growth in foreign liabilities for countries with low economic and …nancial development (since also the indicators of development are set to zero) is strongly negative, from -10% to -33%, depending on the growth de…nition and on the subset of liabilities considered, which is consistent with the substantial average decrease in foreign investment after the COVID outbreak, already observed in Figure 1 (panels 1a, 2a and 3a).

As far as our main regressors are concerned, we …rst observe that the average stringency index (SI) does not a¤ect the growth in inward liabilities in the …rst quarter of 2020. However, the coe¢ cients of the within-country standard deviation ( SI) are positive and highly signi…cant for foreign portfolio investment (columns (3a) and (3b)), for both measures considered: a one-unit increase of the SI pushes inward portfolio investment from 0.64 to 0.66%. If we recall that, in the …rst quarter, the mean of SI is 26 (the median is 27), we point to an economically sizeable e¤ect.

The coe¢ cient of SI for total liabilities, driven by the strong signi…cance of its portfolio investment component, is about 0.3%, while the coe¢ cient for direct inward investment is signi…cant only in column (2a).

As far as the other regressors are concerned, the coe¢ cients related to the COVID-new deaths per million are never signi…cant, while the appreciation of the currency is (marginally) signi…cant J o u r n a l P r e -p r o o f Journal Pre-proof only for foreign direct investment, and only in the second speci…cation of the growth measure. The binary regressors capturing the economic and …nancial development of the receiving economy seem to deliver contrasting, and non systematic results: on the one hand, countries with higher than median GDP per capita seem to witness a dampened drop in the growth of foreign liabilities; on the other hand, countries with higher than median market capitalization per GDP, seem to su¤er even more the contraction in foreign inward investment. Though the evidence about the relationship between liabilities'growth and development is worth-investigating, we must underline that, also in this case, the signi…cance of the coe¢ cients is usually quite weak and far from systematic across instruments and dependent variable's speci…cations.

To allow an immediate comparison over time, across various types of liabilities, and di¤erent de…nitions of the dependent variable and regressors, we report in a singe (1a) and (1b) report results relative to the …rst quarter of 2020, already displayed in Table 1 , columns (2a) and (2b) refer to the second quarter, while columns (3a) and (3b) refer to the …rst semester of 2020. 6 We observe that, after removing country …xed-e¤ects and controlling for the severity of the epidemic spread across economies, as captured by the number of new COVID-death/new COVIDcases per million, the only signi…cant coe¢ cients refer to the e¤ect of the within-country standard deviation ( SI) for portfolio investors, and only at the end of the …rst quarter of 2020, as already found in Table 1 . The coe¢ cients are never signi…cant in the last four columns, which refer to the 2020 second quarter and …rst semester.

From this table, we derive three main considerations. First, portfolio investors reveal to be the J o u r n a l P r e -p r o o f Journal Pre-proof ones more sensitive to the stringency index. Second, the allocation of foreign portfolio investors seems to have been a¤ected by the within-country standard deviation of the stringency index, rather than by its average level. Third, the e¤ect is signi…cant only at the end of the …rst quarter.

Let's try to provide an interpretative key to these pieces of evidence.

First of all, we are not surprised to observe that portfolio investments are more a¤ected by the policy measures adopted, because they are typically more volatile and reactive than foreign direct investments, and then were expected to react earlier also to the pandemic shock.

Second, we observe that, at the end of the …rst quarter of 2020, portfolio investors have not been driven by the average stringency of the policy adopted, but rather by the sharpness of the government reaction as opposed to the graduality of the interventions.

Third, the di¤erence across quarters is systematic. This evidence is consistent with the work of Ashraf (2020) , which suggests that stock markets have quickly responded to COVID-19 pandemic, and that this response has varied over time, depending on the stage of outbreak, with a stronger negative market reaction during early days of con…rmed cases. The prompt adoption of containment measures in the immediate aftermath of the COVID crisis, compared to more gradual ones, might have been interpreted by foreign investors as a serious commitment to face the negative consequences of an uncontrolled spread of the virus. In the second quarter, instead, the standard deviation of the index within a country also seizes the retreat of the containment measures: when this index gets blurred and only loosely correlated with the timeliness in the implementation of rigorous containment measures, it also becomes loosely correlated with cross-border investment.

It is worth stressing that the aim of this research is to establish the existence of a connection between the COVID restrictive measures and foreign investors'allocation choices. It does not aim to assess either the quality of the containment measures imposed by di¤erent countries or the appropriateness of their timing, as it would require a throughout investigation on the implementation of di¤erent policies in di¤erent economies, which is far beyond the scope of this paper. Indeed, both the degree and the speed of adoption of containment measures in di¤erent countries can be strictly related to the severity of the e¤ects of the COVID spread, which has shown a notable degree of J o u r n a l P r e -p r o o f Journal Pre-proof cross-country heterogeneity. The reasons behind this heterogeneity are quite obscure at the moment, and scienti…c research will hopefully make them clearer in the near future. Accordingly, it can be argued that governments cannot be blamed or praised for the measures adopted, as these have been country-speci…c reactions to country-speci…c conditions, in terms of severity of cases, deaths, and pre-existing e¢ ciency of the national health system. By controlling for the new COVID deaths (or the new COVID cases) per million of inhabitants and by removing country …xed-e¤ects, we try to partial out this cross-country heterogeneity: the eventual e¤ects of the level (SI) and the withincountry dispersion of stringency measures ( SI) are therefore computed on top of the severity of the spread, as accounted for by these epidemic indicators. 7

In the remainder of the paper, we undergo our …ndings to a bunch of robustness checks, to infer the strengths and limits of the analysis.

In the following tables, we check the sensitivity of our …ndings to alternative controls (Table 3) , speci…cations of the country sample (Table 4) , and estimation strategies (Table 5 ).

In Table 3 , we include the variable "new COVID-cases per mn of inhabitants" (and its withincountry standard deviation), as an alternative to "new COVID-deaths per mn of inhabitants" (and its within-country standard deviation). Ashraf (2020) …nd that stock markets reacted more proactively to the growth in number of con…rmed cases as compared to the growth in number of COVID deaths.

We therefore check whether our …ndings are a¤ected by the introduction of this alternative covariate.

We observe a pattern qualitatively very similar to Table 2 , in which the only signi…cant coe¢ cients are those related to SI in the …rst quarter for foreign portfolio investment: quantitatively, the e¤ect is reduced but still important, since a one-unit increase in the dispersion index leads to 0.4% larger inward portfolio investments.

In Table 4 , we test whether our …ndings survive to the exclusion of speci…c countries from the 7 The correlation coe¢ cient between the average stringency index (SI) and the "new COVID-deaths per mn" average is equal to 0.22 (signi…cant at 5% level) in the …rst quarter, and 0.04 (non signi…cant) in second quarter. The correlation coe¢ cient between the corresponding within-country standard deviation ( SI) is not statistically signi…cant (equal to 0.17 in the …rst quarter, and -0.04 in the second quarter).

J o u r n a l P r e -p r o o f Journal Pre-proof sample. In columns (1a) and (1b) of Table 4 , we exclude China from the sample. China has been the …rst country to be struck by the COVID spread, several weeks before other countries. The e¤ect of the stringency index and its dispersion measure could therefore have been distorted by China's asynchronic timing of lockdown and loosening measures, in the …rst and second quarter.

By comparison with Table 2 , we observe that the exclusion of China reduces the impact of SI on foreign portfolio investors from 0.64 to 0.56%, when considering the measure of column (1a), and from 0.66 to 0.65%, when considering the dif f measure of column (1b): the size of the coe¢ cient is a¤ected, but the e¤ect remains still sizeable and signi…cant.

In columns (2a) to (4b) of Table 4 , we exclude from the sample potential o¤shore …nancial centres, to make sure our results are not driven by economies distorting investors'decisions for reasons hard to control in our analysis. We consider three di¤erent classi…cations proposed by the literature: columns We con…rm that, at the end of the …rst quarter of 2020, foreign investors'allocation is a¤ected by the within-country dispersion of the stringency index, rather than by the stringency index itself.

Di¤erently from previous …ndings, however, we observe, that SI systematically a¤ects all types of foreign investors, direct and portfolio, under both speci…cations of the dependent variable. The impact on foreign direct investors, negligible in the full sample-case, becomes not only statistically signi…cant, but also economically sizeable, ranging from 0.43 to 0.62%. However, consistently with our previous …ndings, the impact remains larger for foreign portfolio investors than for direct investors.

Also the size of the e¤ect is signi…cantly boosted after the exclusion of o¤shore countries, ranging from 0.71 to 0.80%, depending on the speci…cation. Interestingly, the exclusion of o¤shore centres makes our …ndings more general and systematically valid. 8 Finally, Table 5 reports results for the …rst quarter of 2020, under di¤erent econometric model 8 Tables 4a and 4b in Appendix B report the corresponding tables for the second quarter and the …rst semester. The coe¢ cients are non signi…cant, with the exception of one (marginally) signi…cant negative coe¢ cient for the average SI, in column (1a) of Table 4a , relative to foreign direct investment, but only relative to one of the two measures of the dependent variable. 14 J o u r n a l P r e -p r o o f speci…cations.

The results under the OLS speci…cation (columns (1a) and (1b)) are qualitatively similar to the ones under the Robust Least Squares approach of Table 2 : the coe¢ cient in column (1a) is reduced (from 0.64 to 0.56%), while the coe¢ cient in column (1b) is marginally increased (from 0.66 to 0.68%).

In columns (2a) to (4b), we report the results of a Quantile regression. The quantile regression estimates, relative to the least squares regression, are more robust against outliers in the response measurements: whereas the method of least squares estimates the conditional mean of the response variable, quantile regression estimates its conditional median (or other quantiles). In columns (2a) and (2b), we report the conditional 25th percentile, in columns (3a) and (3b) the median, and in columns (4a) and (4b) the 75th percentile of the response variable.

We observe, …rst, that the coe¢ cient of the SI factor is again systematically signi…cant only for foreign portfolio investors: there is only one (marginally) signi…cant coe¢ cient in the total investment panel, and one in the direct investment panel, both at the 25th percentile, but this signi…cance is not robust across both measures of the dependent variable. Focusing on panel III, referred to foreign portfolio investment, we observe that the coe¢ cients vary over percentiles and de…nitions of the measure. The coe¢ cient of the measure of dispersion SI is (marginally) signi…cant at the 25th percentile of the response variable (0.55, column (2b)), but only when de…ned as in equation (2), while it systematically a¤ects the median and 75th percentile of the response variable, under both de…nitions of the measure. Interestingly, there is no precise ranking in the e¤ect of SI over these two percentiles: according to the measure de…ned in equation (1), the e¤ect on the median of the response variable is larger than its e¤ect on the 75th percentile, and vice versa when considering the alternative dif f measure de…ned in equation (2).

Overall, the results of Table 5 indicate that our …ndings are not driven by the choice of the econometric model. 9 9 Tables 5a and 5b in Appendix B, referred, respectively, to the second quarter and the …rst semester of 2020, con…rm the absence of signi…cant e¤ects in the second quarter of 2020. The only exception is relative to column (4b) of panel II (foreign direct investment) of Table 5b : there are two (marginally) signi…cant negative coe¢ cient for the average SI and its standard deviation SI, but only for one of the two measures of the dependent variable.

This paper investigates the evolution in foreign investment in the immediate aftermath of the adoption of government stringency measures to restrain the spread of COVID-19. Foreign investors could be averted from investing in a country adopting more radical stringency measures, because it could entail a recession period making less pro…table the assets issued by that country. Conversely, foreign investors could be allured by the assets issued by countries adopting more radical containment policies, because these could be perceived as a severe immediate cost to avoid even higher costs in the near future.

We observe that the quarterly average stringency index in each country does not a¤ect inward investment. However, the within-country standard deviation of the stringency index does. In particular, we observe that, after controlling for the severity of the COVID-contagion and removing country …xed-e¤ect, a higher within-country standard deviation in the stringency index makes the adopting countries relatively more attractive for foreign portfolio investors, but only at the end of the …rst quarter of 2020. An increase of one unit of the within-country standard deviation of the stringency index pushes inward portfolio investments from 0.4% up to 0.8%, depending on the speci-…cation adopted. Being the average within-country standard deviation of the stringency index equal to 26 in the …rst quarter, we point to a sizeable average impact on foreign portfolio investments.

This evidence can be interpreted as follows. At the end of the …rst quarter of 2020, the growth in foreign portfolio investments, typically more volatile and reactive than foreign direct ones, responds to governments'prompt and severe reactions more than to gradual ones, since the former can represent for foreign investors a more serious commitment to stem the spread of COVID-19. At the end of the second quarter, instead, the standard deviation of the index within a country also captures the retreat of the containment measures: when this index gets blurred and only loosely correlated with the timeliness in the implementation of rigorous containment measures, it also becomes loosely statistically correlated with cross-border investment.

This early evidence seems to suggest that foreign portfolio investors, when allocating their investment abroad, value, more than the average stringency of the government containment policies, the (1a), in columns #a, or as in equation (2a), in columns #b. Columns (1a) and (1b) refer to foreign total, columns (2a) and (2b) to foreign direct, and columns (3a) and (3b) to foreign portfolio inward investment. ***, **, and * indicate signi…cance at the 1, 5, and 10% levels, respectively. q1 (mar2020-dec2019)/dec2019 (3), for the …rst quarter (columns (1a) and (1b)), the second quarter (columns (2a) and (2b)), and the …rst semester (columns (3a) and (3b)) of 2020. Panel I, II and III refer, respectively, to total, direct and portfolio inward investment. The econometric speci…cation, as in Table 1 , also includes the controls reported at the bottom of the table (the number of the new COVID deaths per mn, its standard deviation, the (one-month lagged) quarterly appreciation in the nominal e¤ective exchange rate, and binary indicators of economic development and …nancial development). ***, **, and * indicate signi…cance at the 1, 5, and 10% levels, respectively. This table is the same as Table 2 , but the covariate "new number of COVID deaths per mn" is replaced by the covariate "new cases of COVID per mn", with its corresponding within-country standard deviation. This table is the same as Table 2 , but the sample excludes China (columns (1a) and (1b)), or o¤shore countries (from columns (2a) to (4b)), according to three alternative o¤shore de…nitions: columns (2a) and (2b) follow the classi…cation in Damgaard et al. (2018) , columns (3a) and (3b) follow Zoromé (2007), columns (4a) and (4b) follow Lane and Milesi-Ferretti (2017) (see Appendix A.1 for details). This table is the same as Table 2 , but under alternative econometric speci…cations: OLS in columns (1a) and (1b), and Quantile regressions in columns (2a) to (4b). In our analysis, we consider and report as regressors both the quarterly overall mean of the daily stringency index (SI j ) and its quarterly standard deviation ( SI j ), computed within each country over the corresponding quarter.

Source: https://github.com/OxCGRT/covid-policy-tracker

New COVID death per mn (and its within-country standard deviation) This is a daily variable, reported by the countries'authorities. In our analysis, we consider both the quarterly average of new COVID-19 deaths and its standard deviation, computed within each country over the corresponding quarter. This covariate closely follows the stringency index: it is always included in the regression speci…cation in the form of the stringency index. To avoid eventual zeros at the denominator (when considering the growth rate of "new COVID death per mn" in the second quarter), we add 0.0001 to the corresponding values.(it is negligible relative to both the average and the standard deviation).

Source: https://github.com New COVID cases per mn (and its within-country standard deviation) This is a daily variable, reported by the countries'authorities. In our analysis, we consider both the average quarterly number of new COVID-19 cases and its standard deviation, computed within each country over the corresponding quarter. This covariate closely follows the stringency index: it is always included in the regression speci…cation in the form of the stringency index.

Source: https://github.com Nominal E¤ective Exchange Rate BIS e¤ective exchange rate Nominal, Broad Indices Monthly averages; 2010=100. The NEER regressor is included with the same structure as the dependent variable. For instance, if the dependent variable is q1 as de…ned in equation (1a), then the regressor included is (N EER 03_20 N EER 12_19 )=N EER 12_19 Source: Bank for International Settlements Economic Development GDP per capita (year: 2019, or latest available data). The regressor included is a binary variable equal to 1 if the GDP per capita is larger than the sample mean, and 0 otherwise. Source: CEIC data Financial development Market capitalization to GDP (year: 2019, or latest available data). The regressor included is a binary variable equal to 1 if the market capitalization per GDP is larger than the sample mean, and 0 otherwise. Source: CEIC data Table 4a . Sensitivity analysis: sample speci…cation (second quarter 2020)

This table is the same as Table 4 , but refers to the end of the second quarter of 2020, rather than to the …rst quarter. This table is the same as Table 4 , but refers to the end of the …rst semester of 2020 (relative to the end of 2019), rather than to the …rst quarter. This table is the same as Table 5 , but refers to the end of the second quarter of 2020, rather than to the …rst quarter. This table is the same as Table 5 , but refers to the end of the …rst semester of 2020 (relative to the end of 2019), rather than to the …rst quarter. 

Inequality in the impact of the coronavirus shock: Evidence from real time surveys

Covid-19 and the united states …nancial markets'volatility. Finance Research Letters , forthcoming

Stock markets reaction to covid-19: Cases or fatalities? Research in International Business and Finance 54

Deaths, panic, lockdowns and us equity markets: The case of covid-19 pandemic

Covid-induced economic uncertainty

Global Economic Prospects

Covid-19 crisis fuels hostility against foreigners. IZA Discussion Paper 13250

Global supply chains in the pandemic

Measuring labor supply and demand shocks during covid-19

A literature review of the economics of covid-19

The short-term economic consequences of covid-19: Occupation tasks and mental health in canada

O¤shore countries In Tables 5, 5a and 5b, we restrict the sample to exclude potential o¤shore countries. Columns (2) and (2a) refer to the o¤shore classi…cation speci…ed in Damgaard et al. (2018). From our original sample, Hong Kong, Ireland, Luxembourg, the Netherlands and Singapore are excluded. Columns (3) and (3a) refer to the o¤shore classi…cation speci…ed in Zoromé (2007). From our original sample Cyprus, Hong Kong, Ireland, Latvia, Luxembourg, Malta, Singapore, Switzerland and United Kingdom are excluded. Columns (4) and (4a) refer to the o¤shore classi…cation speci…ed in Lane and Milesi-Ferretti (2017). From our original sample Belgium, Cyprus, Hong Kong, Ireland, Luxembourg, Malta, the Netherlands, Singapore, Switzerland and the United Kingdom are excluded.

The Stringency Index is a daily aggregate measure of the overall stringency of containment and closure policies. It is calculated by taking the ordinal value and adding a weighted constant if the policy is general rather than targeted, if applicable, which are then re-scaled by their maximum value to create a score between 0 and 100. More information can be found at Oxford's Government Response Tracker, https://www.bsg.ox.ac.uk/research/research-projects/coronavirusgovernment-response-tracker

The growth in liabilities ; at quarterly or semi-annual level (end of period), follows equation (1):