Short-term effects of COVID-19 lockdown on foreign direct investment: Evidence from Ecuadorian firms.

We exploit the exogenous variation that comes from the COVID-19 and the subsequent lockdown in Ecuador. We estimate a regression discontinuity in time (RDiT) design using official administrative FDI data from January to May 2020. We observe an overall large decrease in FDI inflows. We assess differences across FDI sources and find stronger effects coming from capital increases compared with new firm constitutions. In addition, we find that the negative effects are mostly from inflows coming from North and South American investments. We also assess whether partial reopening of activities positively affects FDI. We do not find any significant effect.

INTRODUCTION

Foreign direct investment (FDI) has grown rapidly since the beginning of the globalization process. Nevertheless, in recent years, there has been a constant decrease due to different exogenous factors (OECD, 2020). In Latin America (LA), this behaviour is similar, considering three stages of growth: the first from the mid‐1990s to the end of the year 2000; the second from 2004 to 2008, which ended with the world economic financial crisis; and the final stage from 2011 to 2013. All of these stages of growth relate to the boom of commodities. However, in the last 5 years, there has been a slowdown in FDI flows in the region.

The decrease in recent years and the fact that FDI is related to the boom of commodities and the end of the super‐cycle of commodities demonstrates that FDI flows are strongly linked to external shocks, which places the countries of the region in a situation of high vulnerability (Carrillo‐Maldonado & Díaz‐Cassou, 2019; Díaz‐Cassou et al., 2020). Specifically, exporters of commodities, with few FDI flows and/or high external financing requirements, are more vulnerable to those external shocks. All these characteristics can influence the degree of impact of a possible crisis caused by the COVID‐19 pandemic. In this line, the OECD (2020) mentioned that the amount of FDI towards developing countries decreased even more because the most developed sectors of these economies, such as agriculture and manufacturing, are the ones mostly affected in a negative way by the pandemic. In the case of Ecuador, there are different pre‐pandemic factors that might have diminished the levels of FDI, such as political instability, bureaucratic procedures, taxes and other limitations that dissuade foreigners from investing in the country. Because of these other variables, we consider an estimation procedure that uses all these factors and allows us to capture directly the effect of COVID‐19.

Moreover, the productive structure of the region has had problems of structural heterogeneity even before the pandemic. These problems have increased, spilling over into economic, social and environmental areas, among others, limiting economic development for several years (ECLAC, 2020). Hence, a delay in the structural changes of LA economies would be expected, not only because of the fall in FDI but also because the pandemic has had negative effects on the labour market, production, productivity and other areas.

Even though FDI has been essential in weathering economic crises (see, for example, Alfaro & Chen, 2012; Desai et al., 2008), unfortunately, the impact that the pandemic has had on different outcomes in the economy has been severe, and FDI is no exception, as it is expected to decrease between 40% and 55% in 2020 in LA (UNCTAD, 2020).

This paper investigates the short‐term impact of the COVID‐19 pandemic on Ecuadorian FDI inflows. In Ecuador, as in other countries, the first cases of COVID‐19 were detected in the main cities. Guayaquil was reported to have cases on 29 February 2020, which led to a lockdown policy applicable to the whole country on 16 March 2020. In addition, mobility and face‐to‐face work hours were suspended. However, the flexibility of mobility and the return to the face‐to‐face workday took place at different periods in each of the cities affected by the pandemic. As such, we exploit the exogenous variation of the COVID‐19 pandemic and its lockdown policy in Ecuador and perform a regression discontinuity in time (RDiT) design as our identification strategy. We further the analysis and assess differences among FDI sources, capital increases and the constitution of new firms. We also explore heterogeneity in terms of the country of origin of the FDI to see which inflows are more affected. Finally, we assess whether partial reopening of activities positively affects FDI. To do this, we focus on the two largest cities in the country. In addition, we conducted three robustness checks on different time bandwidths, functional shapes and a placebo fake time event in order to support our findings. To achieve the objective of the study, we use a rich, firm‐level, real‐time panel dataset that covers the period from 1 January 2020 to 31 May 2020 with all firms that received FDI. Our empirical framework also controls for the time trend and economic factors that affect the whole country like the country risk indicator. We also consider different firm‐specific characteristics such as the economic sector, the type of company and the province in which the firm is located.

Ecuador is particularly interesting in this setting because the largest amount of dollars that comes from FDI is through capital increases (95%), while the rest (5%) is due to the creation of new companies during the period 2013–2017 (Camino‐Mogro, Bermudez‐Barrezueta, & Avilés, 2018). Nevertheless, this participation changed in 2020, where around 57% of FDI was through capital increases and 43% from new companies or new constitutions. In addition, there is a heterogeneous participation of inward FDI from economic activities, firm size and location showing that natural resources and manufacturing sectors, large firms and firms located in Guayaquil, Quito, Manta and Samborondón have the largest FDI. Moreover, Ecuador is highly vulnerable to external shocks because it is highly dependent on commodities prices and, in particular, on oil prices. Additionally, it has high external financing requirements because of its high fiscal deficit and limited monetary policy capacity. Furthermore, the country has very low levels of FDI, so the COVID‐19 pandemic might aggravate the structural problems of this country. Díaz‐Cassou et al. (2020) confirm that the external shock derived from COVID‐19 is leading to a significant recession in Ecuador and that in 2020 and 2021, a contraction of between 7.1% and 11.4% could occur, generating a procyclical fiscal response.

In terms of size, the Ecuadorian business environment is made up mostly of micro, small and medium‐sized companies (MSMs) with a participation of 95%, while only 5% are large firms (Superintendencia de Compañias Valores y Seguros, 2018). This business dwarfism and also high levels of informality, poor development of the non‐traditional export sector and lower financial deepening increase the symptoms of low productivity (Ruiz‐Arranz & Deza, 2018), which limits development. In addition, having a higher percentage of small firms in the economy is associated with increases in disparities in Total Factor Productivity (TFP), particularly in LA economies (see, for example, Busso et al., 2013; Camino‐Mogro, Armijos‐Bravo, & Cornejo‐Marcos, 2018; Cole et al., 2005). Finally, it is well known that uncertainty decreases FDI, specifically in countries with less financial development (Choi et al., 2021), political instability and an external debt burden (Lemi & Asefa, 2003) a year before a national election (Julio & Yook, 2016). All of these characteristics are present in the Ecuadorian economic‐political context. On this regard, and considering the environment of social and financial uncertainty created by the COVID‐19 crisis, we expect that FDI flows will decrease during the lockdown because the country has limited financial capacity to cope with it.

Our paper contributes to the growing literature on uncertainty and FDI in the context of a lockdown and partial reopening of economic activities in different cities. It also contributes filling the existing gap of studies on how the COVID‐19 crisis could affect the inflow of foreign investment, focusing on FDI flows in a developing economy that is highly vulnerable to international shocks. Finally, we contribute with an analysis of different geographical areas that invest in companies in the country. We show how the crisis could heterogeneously affect these flows and how the partial reopening of cities in the country could have a heterogeneous effect at the time of receiving FDI.

We structure the remainder of the paper in the following way. In Section 5, we describe how the COVID‐19 crisis affects FDI all over the world and in Ecuador. We also relate the literature to uncertainty theory. Section 3 describes the data and discusses the empirical model. In Section 4, we present our results and robustness checks. Finally, Section 5 concludes with policy recommendations.

COVID‐19 PANDEMIC LOCKDOWN AND FDI

The COVID‐19 pandemic has been considered one of the most disruptive events in history as it is an exogenous shock that has affected not only the world's health systems but also the world economy. The damage caused to the economy is very severe, particularly in the labour market, because several countries declared a lockdown of almost all economic activities and closed their borders to mitigate contagion. This caused the productive apparatus to stop for several months and unemployment began to grow, consequently inequality, especially in countries with limited capacity to overcome the lockdown.

Additionally, the COVID‐19 crisis, as a simultaneous demand and supply shock, affected international trade, where FDI flows began to decline because of the uncertainty surrounding the reaction of markets to the lockdown and the duration of the lockdown in each country. This decline began in developed economies but affected mainly developing countries. A recent report by the ECLAC (2020) mentions that the crisis caused by COVID‐19 has strongly affected FDI and that these effects can be long lasting. These effects were felt more strongly in 2020 because of the different transmission mechanisms (negative).

In this sense, we expect that COVID‐19 would have a negative impact on FDI flows worldwide, particularly in developing economies because of their limited capacity to manage crises properly. ECLAC (2020) and the OECD (2020) show that the crisis could cause a lasting and long‐term recession, as the expected level of global FDI flows in 2021 would represent a 60% decline from 2015.

In Figure 1, we show how the inward FDI of many countries decreased after the first quarter of 2020. We show in panel (a) that there was a steep decrease of inward FDI received by Germany and the United Kingdom. In the same way, in panel (b), we show how the United States had a continuous decrease since the last quarter of 2019. This decrease became even more pronounced after the shock of the global crisis. The recovery of the flows of FDI will be slow, possibly because of the uncertainty of the recovery from the crisis. In addition, there is the possibility of another strict lockdown. All of these uncertainties joined with the high level of debts caused by the crisis could slow the speed of recovery of the economies. Oldekop et al. (2020) mention that the impact of the crisis on public finance is significantly negative for three main reasons: first, the termination of new loans for low and middle‐income countries, which decreases the value of currencies and makes dollar debt repayments more difficult; second, increased government spending on social protection; third, a decrease in public revenue from taxation, which raises debt levels. In the same line, Barbier and Burgess (2020) mention how COVID‐19 makes developing countries more vulnerable. They mention that if progress towards the Sustainable Development Goals was already slow before the crisis, now it is worse. They propose different innovative policies to achieve those goals but assert that it will be difficult.

FIGURE 1

FDI inflows in Europe, USA and Ecuador.

The Ecuadorian case

The COVID‐19 crisis impacted LA and the Caribbean at a time of weak economic and macroeconomic vulnerability (BID, 2020; CEPAL, 2020). In particular, Ecuador is one of the most affected economically and socially, in terms of number of cases and fatalities per million inhabitants (see statistics by Max Roser et al., 2020).

In terms of employment, around 270,000 people lost their jobs during the period between 16 March 2020 (the initial day of the lockdown in Ecuador) and the end of June 2020. This loss happened because Ecuadorian firms were not prepared to face a crisis that prevented economic activity. Carrillo‐Maldonado et al. (2020) mention that 50% of firms were operating with a median of 33 days of resistance without liquidity, about 1 month of operation. Additionally, 25% of firms were highly vulnerable to quarantine or a suspension of economic activities for more than 16 days. Moreover, after 3 months of confinement, Bachas et al. (2020) estimated that only 35% of firms would remain profitable and that almost all firms in the most affected sectors would register losses.

Additionally, FDI in Ecuador decreased by 42% during the crisis. If we compare the same period in 2019, the country lost 107 million USD from FDI. Along these lines, we evidence a 66% decrease in FDI inflows in terms of the number of companies that receive it and 77% regarding capital increases.

In Figure 2, we show how FDI decreased after the lockdown, shown as week 0 in subfigure (a). It is evident that the total amount of inward FDI decreased significantly the first week of the lockdown, then gradually increased but to a lower level (the maximum value is lower than the minimum value before the lockdown). In subfigure (b), we show how inward FDI decreased by country of origin. FDI from China was the first to decrease, then FDI from the Andean Community gradually decreased. FDI from the European Union already shows various ups and downs before week 0, but those decreases lasted for 1 week, whereas after the lockdown, there was a constant decrease for almost 2 months. In subfigures (c) and (d), we show the behaviour of inward FDI received by Guayas and Pichincha, the two main provinces of Ecuador. We show how inward FDI decreased after the lockdown in week 0 but increased after the partial reopening of activities. In the case of Guayaquil, it was on May 20, week 9 after the lockdown, and for Quito, it was on June 3, week 11 after the lockdown.

FIGURE 2

Inward FDI in Ecuador. Note: In Guayaquil and Quito, the reopening took places in different weeks. Guayaquil implement a partial reopening on May 20, and in Quito was on June 3.

The effect of the crisis is notorious on FDI in Ecuador, since it accelerated the problems that already existed at the end of 2019. Moreover, the government's reaction (mechanisms) to attract FDI to the country has been insufficient.

DESIGN AND EMPIRICAL STRATEGY

Data

We use underexplored administrative real‐time data from 2020 obtained from the Superintendencia de Compañías, Valores y Seguros (SCVS). These data include information from investment data sheets, which have information regarding the characteristics of the investment: the name of the shareholders, the country of origin, the type of investment (foreign or domestic), the amount of capital allocation and others. We use firm‐level panel data with weekly information (5152 observations) from 1 January to 31 May 2020.

Real‐time data are information immediately available after its collection from a recent event. In Ecuador, the SCVS receives all of the information of the FDI immediately after the transaction is done. This type of data, instead of putting together information from the past, delivers insights about what is going on in the present, which could help to minimize errors in the analysis (if well collected) and establish action plans (possible changes to policies or actions to promote the investment) more rapidly. Additionally, authors like Malik et al. (2018) mention the importance of real‐time data as more suitable for inference and analysis.

As the aim of this study is to evaluate the short‐term effect of COVID‐19 on the levels of inward FDI, we construct a database that includes the weekly amount of FDI of each firm and include other firm‐specific characteristics. We consider the demographic information (the province where the firm is located) to control for possible heterogeneity effects of inward FDI in provinces with more firms and highly competitive environments. Also, we include corporate information (i.e., the type of firm, whether it is a limited liability company, anonymous society and so on) to control for possible heterogeneity effects of inward FDI in different types of firms that need different control issues for the SCVS. In addition, we take account of industry information (the 2 Digit sector codes from the ISIC Rev 4.0 classification) to control for possible heterogeneity effects of inward FDI in industries that are more specialized and have higher profitability rates. Furthermore, considering that the FDI variable is very volatile, we transform it into logarithms to achieve stability, reduce atypical observations and make the results interpretative.

We include the country risk indicator as a macro variable because this indicator captures factors such as political instability, bureaucratic procedures, taxes and other limitations that dissuade foreigners from investing in the country. We calculate this indicator from the weekly median taken from the daily data published by Ambito. The oil price is extracted from the same source, and we include it in our estimation of the baseline model (without country risk), obtaining similar results to the ones obtained with country risk. We do not include both variables at the same time because they have a high negative correlation, which would cause multicollinearity problems. In Appendix A, we include Table A1 with the Pearson correlation coefficients between the variables used in the analysis. We show that there is a strong negative correlation (more than −0.90) between the two macroeconomic variables, oil prices and country risk.

TABLE A1

Correlation matrix

	FDI (log)	Type of company	Province	Activity	Oil	Risk
FDI (log)	1.00
Type of company	−0.01 (0.71)	1.00
Province	0.01 (0.44)	0.22*** (0.00)	1.00
Activity	−0.01 (0.41)	0.08*** (0.00)	0.10*** (0.00)	1.00
Oil	0.15*** (0.00)	0.00 (1.00)	−0.00 (1.00)	0.00 (1.00)	1.00
Risk	−0.15*** (0.00)	−0.00 (1.00)	0.00 (1.00)	0.00 (1.00)	−0.93*** (0.00)	1.00

Note: Table of Pearson correlations coefficients; p values in parentheses.

p < 0.1.

p < 0.05.

p < 0.01.

Identification and methodology

We take advantage of the completely exogenous effect of the COVID‐19 pandemic that induced a lockdown (and a resulting pause of activities) on 16 March 2020 in Ecuador. Hence, we perform an RDiT design, a quasi‐natural experimental econometric technique that allows us to compare rigorously the impacts of the lockdown on FDI inflows in a time window around the lockdown date because we have high frequent data and we can observe enough data points around the cut‐off (Anderson, 2014; Auffhammer & Kellogg, 2011).

Consequently, our empirical strategy is to leverage the sharp discontinuities of economic activities at the start of the lockdown and employ a (sharp) RDiT design. Cattaneo et al. (2020) argue that the motivation of the RDD approach is that within a relatively narrow window of time around an event (in this case the lockdown), the unobserved factors that influence the dependent variable (FDI inflows) are likely similar. Therefore, observations before the event provide a counterfactual group that can be compared with observations after the same event. To perform the RDiT, we formally estimate the treatment effect as the FDI inflows variations in Ecuadorian firms around the lockdown date: where is the number of weeks before and after the official lockdown date. We subsequently employ an approach based on Imbens and Lemieux (2008). We estimate the model when adopting an RDiT approach, using a cut‐off date of 16 March 2020, when the president declared the lockdown, remote work and restricted mobility. where are the FDI inflows received in firm at week , which is our main outcome. The parameter of interest is , the (local average treatment) impacts of the lockdown on FDI, for which we obtain an unbiased estimate under the RD assumption that does not change discontinuously upon the policy introduction (Anderson, 2014). Our treatment variable is , which is a dummy variable that is equal to one for all weeks after the lockdown on 16 March and zero for all preceding weeks. We use data from 1 January to 31 May 2020 for our main specification, which allows 10 weeks on each side of the cut‐off. Our running variable is that represents the number of weeks before and after the official lockdown date. To provide robust analysis and make comparisons, we allow the function to have different functional forms to flexibly control for variations in FDI that would have occurred in the absence of the lockdown. This function includes a linear model and a quadratic model of . We use a bandwidth of 11 weeks before and after the lockdown date as our preferred time bandwidth, but we also present results for different bandwidths to investigate the duration of the lockdown impacts. Additionally, is a country risk variable (measured in logs) that could influence country FDI inflows. Finally, is a time fixed effect, is a province fixed effect, is a type of company fixed effect, is an industry fixed effect and is the error term. We report standard errors robust to heteroskedasticity.

We also use Equation 2 to look at several sources of heterogeneity in FDI flows such as FDI by type, like new constitutions and capital increases, and by region of origin of the inward FDI. This classification allows us to estimate the impact of decreases in FDI because of the COVID‐19 crisis and determine the largest negative impact by type and by demographic origin.

Finally, besides the effect of the lockdown imposed throughout the country, we also assess whether the partial return to activities in the two most important cities (Guayaquil and Quito) has an effect on FDI inflows. Giofré (2021) suggests that, when allocating their investment abroad, foreign portfolio investors value speed in the adoption of government containment policies more than their average stringency, as a commitment to stability, lower uncertainty and then higher adjusted risk returns in the near future. Therefore, we estimate the following equation: where our new parameter of interest is , the (local average treatment) impact of the return to activities on FDI. Our treatment variable is , which is a dummy variable that is equal to one for all weeks after the partial return to activities for each city and zero for all preceding weeks. It is important to mention that we perform this analysis only for Guayaquil and Quito, which concentrate around the 75% of total FDI in Ecuador in the last 10 years. More important is that these two cities have different dates of partial return to activities, so, we use different time windows (and separate regressions) according to each city. We expect the partial return to activities to have a positive impact on the FDI in each city only if the global uncertainty about COVID‐19 decreases. In other words, we expect to find that the flexibility of the containment measures has a positive effect on FDI because the expectations of foreign investors might improve if investors trust the measures previously taken by the government to contain the spread of COVID‐19.

MAIN RESULTS

Before showing the results of the estimates, we present graphical evidence of the effect of the COVID‐19 lockdown on FDI inflows in Ecuadorian firms during the period from January 2020 to May 2020. In this regard, and similar to Barnes et al. (2020) and Dang and Trinh (2021), we run an RDiT regression of FDI inflows (in logs) against the lockdown cut‐off date. In Figure 3, we present the RDiT graph, which shows that FDI dramatically decreases a week after the lockdown date in Ecuador. In this sense, we find significant discontinuity and a decrease at the cut‐off date, but we also observe that the reduction of FDI inflows decreases slightly at higher bandwidths. However, the recovery of FDI several weeks after the lockdown does not reach the levels of FDI before the lockdown.

FIGURE 3

Regression discontinuity plot for FDI inflows analysis.

Once we show that there is discontinuity in the FDI inflows of Ecuadorian firms, we estimate the effect of the lockdown on FDI inflows using Equation 2. In Table 1, we present our main results using a bandwidth of ±11 weeks around the lockdown date. This bandwidth allows us to analyse a window from January to May 2020. In the same table, we show the results when we consider all FDI inflows and then divide into capital increases and new constitution to get detailed information about which type of FDI is more/less affected by the lockdown. In addition, we present our results comparing an estimation with and without control variables to show the robustness of our model specification. In all the cases, our preferred models are those that include control variables, as in Equation 2 (models 2, 4 and 6).

TABLE 1

COVID‐19 lockdown and FDI inflow: Bandwidth ±11 weeks

	Total FDI		Capital increase		New constitution
	(1)	(2)	(3)	(4)	(5)	(6)
Panel A: Linear model
PostLockdown_it=1	−0.540*** (0.087)	−0.406*** (0.109)	−0.186*** (0.065)	−0.146* (0.079)	−0.095 (0.793)	−0.558 (1.027)
risk_t		−0.251*** (0.069)		−0.067 (0.048)		−0.101 (0.664)
Panel B: Quadratic model
PostLockdown_it=1	−0.608*** (0.062)	−0.410*** (0.110)	−0.197*** (0.045)	−0.135* (0.077)	−0.260 (1.164)	−1.285 (1.210)
risk_t		−0.0192*** (0.069)		−0.057 (0.050)		−0.213 (0.482)
Mean before lockdown	0.640	0.640	0.227	0.227	6.851	6.851
Province FE	No	Yes	No	Yes	No	Yes
Type FE	No	Yes	No	Yes	No	Yes
Industry FE	No	Yes	No	Yes	No	Yes
Time FE	Yes	Yes	Yes	Yes	Yes	Yes
Observations	5,152	5,152	4,975	4,975	177	177

Note: Robust standard errors in parentheses. The running variable is the number of weeks from the lockdown date. All regressions include week dummies. The control variable of uncertainty (risk) is reported. Panel A uses the running variable in linear form. Panel B includes the quadratic term of the running variable.

p < 0.1.

p < 0.01.

In Panel A of Table 1, we present the results of a linear model and the results using a quadratic model (in Panel B of the same table). Our results are robust to the use of a quadratic model. Overall, we find that the COVID‐19 lockdown has a strong, statistically significant, negative impact on FDI inflows at the 1% level and at the 10% level for FDI inflows received from capital increases. This evidence remains qualitatively similar regardless of the inclusion of control variables and different functional forms of the running variable. More specifically, we find that the lockdown leads to a 63% reduction of total FDI inflows in Ecuadorian firms and a 64% reduction in FDI capital increases. Nevertheless, the impact of the lockdown on new constitutions from FDI is around 8%. This effect is not statistically significant at standard levels. When we use a different functional form of the running variable (quadratic form in panel B), the results are similar. For example, the lockdown led to a 64% reduction of total FDI inflows and a 59% reduction in FDI capital increases.

Using the same bandwidth (±11 weeks) in Table 2, we show our results by country of origin of FDI inflow. We present the results for North America, China, the European Union and South America, because they represent around 85% of total FDI inflows in 2020. Using Equation 2, we present the results in the same manner as in Table 1 (linear and quadratic model, with and without control variables). Our results suggest that the lockdown has strong, statistically significant, negative impacts on FDI inflow at the 1% level for investments from South American and North American countries (significant at the 10% level) and that no impact is found for FDI inflows in China and European Union countries. In particular, we find that the lockdown led to 84% reduction of total FDI inflows that come from South American countries and a 74% reduction from North American countries.

TABLE 2

COVID‐19 lockdown and FDI inflow by region: Bandwidth ±11 weeks

	North America		China		European Union		South America
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
Panel A: Linear model
PostLockdown_it=1	−0.674*** (0.215)	−0.444* (0.241)	−0.145 (0.228)	0.127 (0.278)	−0.457* (0.265)	−0.370 (0.365)	−0.617*** (0.116)	−0.505*** (0.142)
risk_t		−0.430** (0.175)		−0.509** (0.200)		−0.163 (0.220)		−0.209** (0.091)
Panel B: Quadratic model
PostLockdown_it=1	−0.726*** (0.170)	−0.449* (0.237)	−0.570*** (0.172)	−00058 (0.266)	−0.614*** (0.169)	−0.406 (0.379)	−0.599*** (0.083)	−0.473*** (0.144)
risk_t		−0.268* (0.155)		−0.496** (0.202)		−0.202 (0.235)		−0.122 (0.089)
Mean before lockdown	0.602	0.602	0.556	0.556	0.753	0.753	0.601	0.601
Province FE	No	Yes	No	Yes	No	Yes	No	Yes
Type FE	No	Yes	No	Yes	No	Yes	No	Yes
Industry FE	No	Yes	No	Yes	No	Yes	No	Yes
Time FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Observations	736	736	483	483	897	897	2,392	2,392

Note: Robust standard errors in parentheses. The running variable is the number of weeks from the lockdown date. All regressions include week dummies. The control variable of uncertainty (risk) is reported. Panel A uses the running variable in linear form. Panel B includes the quadratic term of the running variable.

p < 0.1.

p < 0.05.

p < 0.01.

This evidence is in line with the amount of FDI by country of origin that Ecuadorian firms receive. The United States (included in the North American region) and the Andean Community of Nations (includes South American countries) are the main countries that invest in Ecuadorian firms but are also the most affected by the COVID‐19 crisis, excluding China.

Overall, we find a large, statistically significant, negative impact of the COVID‐19 lockdown on FDI inflows in Ecuadorian firms. In addition, our results show that the country risk indicator ( ) is particularly important in our model and shows that an increase of country risk leads to a decrease in FDI inflow. Our results cannot be compared with other research since there are no studies that analyse the lockdown and FDI inflows.

Robustness checks: Different bandwidths and placebos of fake lockdown dates

In this subsection, we present several robustness checks for our main results. First, we change the main bandwidth (±11 weeks) to ±5 and ±8 weeks. Second, we change the date of the lockdown, simulating a fake lockdown one to 3 weeks before the real lockdown week in Ecuador.

In Table 3, we show our results with a bandwidth of ±5 weeks from the COVID‐19 lockdown by using Equation 2. Again, our results suggest that the COVID‐19 lockdown has strong, statistically significant impacts on decreasing FDI inflow at the 5% level. Nevertheless, when we divide the FDI inflows by capital increases and new constitutions, none of these types has a significant result because of the lockdown. We find that the lockdown leads to a 37% reduction of total FDI inflows when we use this new bandwidth. We also show that our results are robust to the inclusion of control variables and different functional forms of the running variable (see panel B). In addition, in Table 4, we present the results of FDI inflows by country of origin. We show that the impact is large and statistically significant at the 10% level for China. This evidence suggests that the lockdown has a −157% impact on FDI inflows received from China. For the other regions, the impact of the lockdown is not statistically significant at standard levels.

TABLE 3

COVID‐19 lockdown and FDI inflow: Bandwidth ±5 weeks

	Total FDI		Capital increase		New constitution
	(1)	(2)	(3)	(4)	(5)	(6)
Panel A: Linear model
PostLockdown_it=1	−0.575** (0.293)	−0.581** (0.291)	−0.163 (0.239)	−0.147 (0.238)	−1.816*** (0.661)	−0.745 (1.662)
risk_t		−0.230 (0.326)		−0.220 (0.253)		0.171 (3.038)
Panel B: Quadratic model
PostLockdown_it=1	−0.823*** (0.191)	−0.665*** (0.266)	−0.323** (0.150)	−0.194 (0.215)	−2.065** (0.917)	−0.309 (5.313)
risk_t		−0.477 (0.297)		−0.309 (0.246)		0.908 (7.350)
Mean before lockdown	1.566	1.566	0.637	0.637	6.789	6.789
Province FE	No	Yes	No	Yes	No	Yes
Type FE	No	Yes	No	Yes	No	Yes
Industry FE	No	Yes	No	Yes	No	Yes
Time FE	Yes	Yes	Yes	Yes	Yes	Yes
Observations	990	990	921	921	69	69

Note: Robust standard errors in parentheses. The running variable is the number of weeks from the lockdown date. All regressions include week dummies. The control variable of uncertainty (risk) is reported. Panel A uses the running variable in linear form. Panel B includes the quadratic term of the running variable.

p < 0.05.

p < 0.01.

TABLE 4

COVID‐19 lockdown and FDI inflow by region: Bandwidth ±5 weeks

	North America		China		European Union		South America
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
Panel A: Linear model
PostLockdown_it=1	−0.679 (0.581)	−0.715 (0.582)	−2.110* (1.118)	−2.099* (1.141)	−1.220 (1.331)	−1.190 (1.338)	−0.393 (0.339)	−0.403 (0.342)
risk_t		−1.299 (0.796)		0.405 (1.106)		1.084 (1.317)		−0.341 (0.419)
Panel B: Quadratic model
PostLockdown_it=1	−0.805** (0.368)	−0.498 (0.486)	−1.446* (0.750)	−1.737* (1.002)	−1.640* (0.890)	−1.679 (1.273)	−0.643*** (0.223)	−0.458 (0.310)
risk_t		−0.927 (0.661)		0.877 (0.859)		0.116 (1.253)		−0.556 (0.369)
Mean before lockdown	1.652	1.652	1.331	1.331	1.898	1.898	1.456	1.456
Province FE	No	Yes	No	Yes	No	Yes	No	Yes
Type FE	No	Yes	No	Yes	No	Yes	No	Yes
Industry FE	No	Yes	No	Yes	No	Yes	No	Yes
Time FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Observations	143	143	44	44	132	132	517	517

Note: Robust standard errors in parentheses. The running variable is the number of weeks from the lockdown date. All regressions include week dummies. The control variable of uncertainty (risk) is reported. Panel A uses the running variable in linear form. Panel B includes the quadratic term of the running variable.

p < 0.1.

p < 0.05.

p < 0.01.

In Table 5, we present the results with a bandwidth of ±8 weeks from the COVID‐19 lockdown. Our evidence is similar to our main results. The COVID‐19 lockdown has a strong, statistically significant, negative impact on FDI inflows at the 1% level, on FDI inflows of capital increases at the 10% level. We find that the lockdown leads to a 90% reduction of total FDI inflows in Ecuadorian firms and a 70% reduction in FDI capital increases. This evidence is robust to other functional forms. Moreover, in Table 6, we show the impact of the lockdown on FDI inflows, by country of origin with a bandwidth of ±8 weeks from the cut‐off. The results show that there is a negative impact of the lockdown of −112% on FDI flows received from the North American region and a 107% decrease of FDI inflows from South American countries. Again, our results show that FDI inflows that come from countries in North and South America have the largest impact on inward FDI.

TABLE 5

COVID‐19 lockdown and FDI inflow: Bandwidth ±8 weeks

	Total FDI		Capital increase		New constitution
	(1)	(2)	(3)	(4)	(5)	(6)
Panel A: Linear model
PostLockdown_it=1	−0.877*** (0.153)	−0.859*** (0.242)	−0.274** (0.117)	−0.245* (0.141)	−0.337 (0.926)	−0.232 (1.296)
risk_t		0.029 (0.184)		−0.029 (0.132)		−0.035 (1.006)
Panel B: Quadratic model
PostLockdown_it=1	−0.902*** (0.106)	−0.902*** (0.243)	−0.277** (0.084)	−0.230* (0.132)	−1.106 (1.143)	−0.661 (1.004)
risk_t		0.001 (0.179)		−0.039 (0.130)		−0.182 (0.706)
Mean before lockdown	0.958	0.958	0.351	0.351	6.783	6.783
Province FE	No	Yes	No	Yes	No	Yes
Type FE	No	Yes	No	Yes	No	Yes
Industry FE	No	Yes	No	Yes	No	Yes
Time FE	Yes	Yes	Yes	Yes	Yes	Yes
Observations	2,635	2,635	2,154	2,154	121	121

Note: Robust standard errors in parentheses. The running variable is the number of weeks from the lockdown date. All regressions include week dummies. The control variable of uncertainty (risk) is reported. Panel A uses the running variable in linear form. Panel B includes the quadratic term of the running variable.

p < 0.1.

p < 0.05.

p < 0.01.

TABLE 6

COVID‐19 lockdown and FDI inflow by region: Bandwidth ±8 weeks

	North America		China		European Union		South America
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
Panel A: Linear model
PostLockdown_it=1	−1.160*** (0.408)	−1.120* (0.589)	−0.571 (0.549)	−0.655 (0.919)	−0.483 (0.480)	−0.461 (0.769)	−0.941*** (0.194)	−0.974*** (0.315)
risk_t		−0.062 (0.411)		0.133 (0.634)		−0.036 (0.585)		0.052 (0.243)
Panel B: Quadratic model
PostLockdown_it=1	−1.045*** (0.259)	−1.082* (0.573)	−0.895*** (0.323)	−0.774 (0.888)	−0.712* (0.364)	−0.490 (0.780)	−0.905*** (0.128)	−0.968*** (0.314)
risk_t		0.038 (0.417)		−0.122 (0.709)		−0.226 (0.561)		0.063 (0.234)
Mean before lockdown	0.994	0.994	0.976	0.976	1.038	1.038	0.905	0.905
Province FE	No	Yes	No	Yes	No	Yes	No	Yes
Type FE	No	Yes	No	Yes	No	Yes	No	Yes
Industry FE	No	Yes	No	Yes	No	Yes	No	Yes
Time FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Observations	340	340	153	153	459	459	1,309	1,309

Note: Robust standard errors in parentheses. The running variable is the number of weeks from the lockdown date. All regressions include week dummies. The control variable of uncertainty (risk) is reported. Panel A uses the running variable in linear form. Panel B includes the quadratic term of the running variable.

p < 0.1.

p < 0.01.

Additionally, we show that the negative impact of the COVID‐19 lockdown increases after the first weeks of confinement, and it begins to decrease more slowly from week 11 after the lockdown. However, the impact is still quite large, which shows that its effect is persistent over time. The effect might persist because of the uncertainty generated by the crisis and by the policies that might be implemented to cope with a possible outbreak that could generate a new lockdown.

Finally, we offer a set of placebo tests to strengthen the causal interpretation we give to our results. In this regard, we change the date of the official lockdown, simulating that it was 1, 2 or 3 weeks before the real date. We expect not to find a negative and significant coefficient in any of these cases or a positive and significant coefficient. In Table 7, columns (2), (4) and (6) present the results of estimates of Equation 2 with the same fixed effects and control variables as the baseline equation. Our results of estimates with full fixed effects support our main identification strategy.

TABLE 7

COVID‐19 lockdown and FDI inflow: Placebo of fake week of lockdown

	−1 week		−2 weeks		−3 weeks
	(1)	(2)	(3)	(4)	(5)	(6)
Panel A: Linear model
PostLockdown_it=1	−0.396*** (0.112)	−0.319 (0.225)	−0.101 (0.127)	−0.372* (0.221)	−0.149 (0.140)	0.014 (0.153)
risk_t		−0.080 (0.146)		−0.624*** (0.132)		−0.454*** (1.006)
Panel B: Quadratic model
PostLockdown_it=1	−0.374*** (0.087)	−0.258 (0.233)	−0.076 (0.107)	0.297 (0.217)	−0.051 (0.116)	−0.025 (0.154)
risk_t		−0.095 (0.147)		−0.396*** (0.146)		−0.052 (0.139)
Mean before lockdown	0.633	0.633	0.626	0.626	0.659	0.659
Province FE	No	Yes	No	Yes	No	Yes
Type FE	No	Yes	No	Yes	No	Yes
Industry FE	No	Yes	No	Yes	No	Yes
Time FE	Yes	Yes	Yes	Yes	Yes	Yes
Observations	4,704	4,704	4,256	4,256	3,808	3,808

Note: Robust standard errors in parentheses. The running variable is the number of weeks from the lockdown date. All regressions include week dummies. The control variable of uncertainty (risk) is reported. Panel A uses the running variable in linear form. Panel B includes the quadratic term of the running variable.

p < 0.1.

p < 0.01.

Reopening activities in Guayaquil and Quito

In this subsection, we analyse how partial reopening activities (20 May and 3 June 2020) might have impacted FDI inflows in the two most important cities in Ecuador (Guayaquil and Quito). We start our analysis by showing graphical evidence of the effect of the partial return to activities (reopening) on Ecuadorian FDI inflows in Guayaquil and Quito. We plot the results for each city in Figure 4, which shows that FDI inflows increase after a week of the partial return to activities in each city (as expected). Moreover, we show that the effect is larger in Guayaquil than in Quito. Nevertheless, the increase of FDI inflows in each city seems to be weak after several weeks of the reopening. This suggests that the impact of reopening policies in each city does not generate an effect strong enough to recover losses in terms of FDI. This might be because of the uncertainty created by the crisis and negative expectations from investors regarding Ecuadorian market.

FIGURE 4

Regression discontinuity plot for FDI inflows analysis of partial reopening activities in Guayaquil and Quito.

We then estimate the effects of partial reopening activities on FDI inflows in Guayaquil and Quito using the sharp RDD model presented in Equation 3. We use a bandwidth of ±8 weeks for each city. We also control for country risk and type of company, industry and time fixed effects. In addition, we present our results using a linear and quadratic model of the running variable. In Table 8, we show the results of the impact of partial reopening activities on FDI flows in Guayaquil and Quito. We prefer the linear model with all controls (columns 2 and 4). Our evidence suggests that the partial reopening activities in Guayaquil have a positive impact on FDI flows, but this impact is not statistically significant at standard levels. On the other hand, the partial reopening activities in Quito have a negative impact on FDI flows, but it is not statistically significant at standard levels.

TABLE 8

Partial reopening activities and FDI inflow: Guayaquil and Quito

	Guayaquil		Quito
	(1)	(2)	(3)	(4)
Panel A: Linear model
PostLockdown_it=1	0.474* (0.287)	0.451 (0.307)	0.118 (0.346)	−0.122 (0.346)
risk_t		−0.214 (0.797)		−1.795 (1.304)
Panel B: Quadratic model
PostLockdown_it=1	0.582** (0.281)	0.501 (0.333)	0.102 (0.347)	−0.119 (0.342)
risk_t		−0.610 (1.098)		−1.053 (0.958)
Mean before lockdown	0.065	0.065	0.123	0.123
Type FE	No	Yes	No	Yes
Industry FE	No	Yes	No	Yes
Time FE	Yes	Yes	Yes	Yes
Observations	527	527	646	646

Note: Robust standard errors in parentheses. The running variable is the number of weeks from the lockdown date. All regressions include week dummies. The control variable of uncertainty (risk) is reported. Panel A uses the running variable in linear form. Panel B includes the quadratic term of the running variable.

p < 0.1.

p < 0.05.

Overall, we expected a significant impact of reopening policies on FDI inflows, but we do not find this result in the two largest city of Ecuador. This evidence supports the idea that the recession caused by the crisis might be difficult to reverse. Despite the reopening of the economy, these policies did not have a positive impact on FDI, perhaps because of the uncertainty of another longer and more severe lockdown or even a pronounced recession in Ecuador.

CONCLUSIONS

In this paper, we present for the first time the short‐term impact of the COVID‐19 lockdown on FDI inflows received by firms. For this purpose, we use an RDiT design based on real‐time data of FDI registered by Ecuadorian firms in the supervisory and regulatory institution of companies: the Superintendencia de Compañías, Valores y Seguros (SCVS). In addition, we analyse the short‐term impact of the lockdown on two different types of FDI and by country of origin. Additionally, we study the short‐term impact of partial reopening activities on FDI inflows in the two most important cities in the country.

By examining the unexpected lockdown imposed on the whole country on 16 March, our empirical analysis reveals that the COVID‐19 lockdown significantly reduces FDI inflows in Ecuador in the short term. Our results show a large decrease in FDI inflows (−63%), including FDI capital increases (−64%). We also find evidence that the lockdown effects are greater for FDI inflows from North and South American countries, where the impact is a large decrease of 84% and 74% of FDI, respectively. Our results are robust to a set of different tests such as a linear or a quadratic model of the running variable, changes in the bandwidth and a placebo lockdown date.

Furthermore, our results suggest that the impact of the COVID‐19 lockdown on FDI inflows in Ecuadorian firms seems to be a short‐term effect. With a short bandwidth, the negative impact is −37%. Then, when we amplify the bandwidth to ±8 weeks, the impact on FDI inflows is −90%, and finally, with a bandwidth of ±11 weeks, the impact is −63%. This evidence suggests that FDI inflows into the country are not recovering quickly, which might be because of the uncertainty generated by the crisis regarding the markets. Fu et al. (2021) argue that investors were unwilling to invest in a country where COVID‐19 was spreading rapidly or invest in foreign markets when the home market was experiencing a higher death rate. In this line, Adam et al. (2020) mention that other lockdown policies might increase the economic shock impact when countries seek to exit the present one, and this could negatively affect expectations of restarting economies. This could delay the structural change, particularly in developing countries.

Finally, we assess the impact of partial reopening activities on FDI inflows in Guayaquil and Quito. Our findings suggest that the partial reopening activities in these two cities have no impact on the recovery of FDI inflows. Again, uncertainty may play an important role in the slowness of the recovery of the FDI in the two largest cities of Ecuador. Specifically, the management of the crisis and the chaos generated in each city might have affected the recovery of FDI inflows in each city.

This research is the first empirical work that assesses the impact of a lockdown and a partial reopening of activities on FDI inflows. Our findings contribute to various strands of the literature and to policy debate. We contribute with evidence of the negative impact that lockdowns have on FDI, so closing borders and restricting mobility might be negative signs for foreign investors, especially in the context of a small and developing economy. Furthermore, our results contribute to a scenario of global uncertainty. A possible mismanagement of an economic and social crisis can also be a bad sign for international investors, and this can delay/impede the arrival of fresh capital.

Moreover, our results show insights for policy makers in two dimensions. First, reopening the economy quickly without having properly managed the crisis does not ensure that FDI flows arrive faster, so public policymakers should focus their efforts on adequately managing the crisis, providing reliable information and reopening priority economic activities. Second, policymakers should look for mechanisms to improve the reception of FDI (faster) after this economic and social crisis. In this sense, tax exemptions to new company constitutions and reinvestment of multinational profits should be a priority to create jobs and lessen the shock in the labour market.

We believe that this paper could be the beginning of future works that analyse how the impact of isolation policies in different settings may affect FDI and other outcomes of international trade. In addition, future research could be conducted with a longer time horizon when the data are available, so that the long‐term effects and the impact on other outcomes, such as company exports and imports, can be assessed.