Policy Research Working Paper 10557 Russia’s Invasion of Ukraine and Firm Performance in Central Asia The Role of Export Links and Digital Gains Nicolo Dalvit Mariana Iootty Martin Melecky Nithya Srinivasan Finance, Competitiveness and Innovation Global Practice August 2023 Policy Research Working Paper 10557 Abstract This paper studies the effect of Russia’s invasion of Ukraine trade links to Russia suffered greater drops in sales and on the performance of firms in Central Asia. It uses unique employment after the invasion—even though exporters data from the Business Pulse Survey run by the World Bank to Russia may have experienced, on average, higher sales in the Kyrgyz Republic, Tajikistan, and Uzbekistan, which during the studied period. Considering the pre-invasion tracks the sales and employment—along with other main digitization of firms, the findings show that digitization characteristics—of about 1,200 to 1,800 firms in a panel helped firms increase their average employment during the structure. The survey contains two waves before and one studied period. However, the analysis does not find any wave after Russia’s invasion of Ukraine. Using the differ- significant mitigating effect of digitalization associated with ence-in-differences methodology in a regression setup, the the impact of the invasion. analysis finds that Central Asian firms with pre-invasion This paper is a product of the Finance, Competitiveness and Innovation Global Practice. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/prwp. The authors may be contacted at miootty@worldbank.org; mmelecky@worldbank.org; ndalvit@worldbank.org; nsrinivasan5@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Russia’s Invasion of Ukraine and Firm Performance in Central Asia: The Role of Export Links and Digital Gains * Nicolo Dalvit, Mariana Iootty, Martin Melecky, Nithya Srinivasan World Bank Keywords: Ukraine Invasion, Firm Performance, Sales, Employment, Central Asia, Trade link with Russia, Business Pulse Survey, Difference in Differences. JEL Classification: D22, F51, L23, L25, O14. * We thank Arti Grover and Leonardo Iacovone for their helpful comments on the earlier draft of the paper. 1. Introduction Russia’s invasion of Ukraine has had far-reaching consequences through various social, economic, and other channels, creating global ripple effects. The economic impact of the invasion has been felt through multiple channels, including commodity and financial markets, trade and migration links, and business confidence and investor uncertainty. These effects have added to existing concerns about a potential global economic slowdown, increased inflationary pressures, rising levels of debt, and a surge in poverty rates. Russia’s invasion of Ukraine (the invasion) has heightened economic uncertainties and underscored the need for careful analysis and response to mitigate its adverse effects on the global economy and local businesses alike. Central Asian (CA) countries also felt major consequences of the invasion through multiple channels, including the sanctions imposed on Russia and Russia’s recession. Closed trade and payment system links curbed trade, remittances, investment, and tourism, adversely affecting economic growth, inflation, and external and fiscal accounts shortly after the invasion. 1 As a result, the growth of CA economies slowed to 4.1 percent in 2022, from 5.3 percent in 2021. Economic growth is set to remain mediocre in 2023, with the real GDP of CA projected to grow by 4.0 percent (World Bank, 2023a). And businesses in CA countries report that their imports from Russia or Ukraine have decreased or they have stopped sourcing from these countries altogether. They are responding to the resulting growth in production costs by raising their sales prices or coping with reduced profitability (Iootty and Melecky, 2022). A few papers use microeconomic data to examine the effects of Russia’s invasion of Ukraine on business. For example, EBRD (2023) explores the compounding effects of the global pandemic and the invasion on businesses in EBRD countries, focusing on trade and finance transmission channels. The findings suggest that trade disruptions prompted firms to enhance their supply chain resilience by stoking up inputs and diversifying their supplier base. 2 Other studies have examined the invasion's impact on firm financial performance, particularly in relation to rising energy prices and the disruption of energy imports. Ferriani and Gazzni (2023) find evidence of decreased equity returns for European firms listed in the Eurostoxx 600 index, especially for those with high energy intensity and carbon emissions. Bougias et al. (2022) identify higher asset volatility, credit spreads, and default probabilities resulting from the invasion. Hutter and Weber (2022) observed a 1 percent decrease in production for German firms due to the increase in energy prices after the invasion—although turnover increased, reflecting sales from stock, and employment was safeguarded through short-time hiring. Mattera and Soto (2022) found that Spanish energy firms that transitioned to renewable energy sources and implemented ESG strategies and CSR commitments exhibited stronger financial performance and lower dependence on uncertain and weakened markets after the invasion. Our study contributes to the emerging literature on the impacts of Russia’s invasion of Ukraine on business, with a specific focus on Central Asian countries. It examines how firms responded to the early stages after the invasion, depending on their varied trade exposure to Russia and varied levels of 1 More recently, CA countries have been reportedly benefiting from serving as a hub for reexports and arbitrage in financial flows and payments to and from Russia. Specifically, estimations presented in EBRD (2023) show that following the invasion of Ukraine and the trade sanctions imposed on Russia, Kazakhstan was poised to make a small gain (0.4 percent of GDP) as it scales up exports of goods that were previously exported by Russia. 2 The study also highlights the negative spillover effects of evergreening loans to vulnerable firms, which affected the investment, revenue, and employment of healthy firms operating in sectors with a high presence of zombie firms. 2 economic resilience stemming from digitization. In normal economic conditions, firms experience varying growth rates, with some expanding and gaining market share while others contracting. These differential growth patterns are also observed after economic shocks, where firms are affected differently. The asymmetrical impact of large shocks can be attributed to sector-specific effects or variations in firm-level characteristics. 3 As emphasized in World Bank (2023b), certain firm-level characteristics can directly insulate companies from shocks, while others can place them in a better position to adapt swiftly. These characteristics are crucial in determining how firms respond to economic shocks and the speed of their recovery. By analyzing these dynamics, our study aims to enhance the understanding of how firms confront and recover from the economic impacts of the invasion on another country’s economy, providing insights into their vulnerabilities and resilience to better inform policy response options. More broadly, our paper adds to the existing literature on how conflicts in one country can affect the performance of firms in neighboring or third countries through trade, financial, and migration connections. 4 Our analysis aims to shed light on two interconnected dimensions to provide a comprehensive understanding of the effects of Russia’s invasion of Ukraine. Firstly, we examine one exposure factor that conditions the short-term effects of the invasion on firm performance and is expected to influence firms' reactions to such economic shock—we focus on firms’ trade exposure to Russia. We hypothesize that firms with higher trade exposure to Russia have faced distinct challenges (or opportunities) compared with firms with lower trade exposure to Russia. This factor is particularly relevant for Central Asian countries, given their strong economic ties with Russia as a major buyer of commodities and final goods. Therefore, understanding how the level of trade exposure shapes firm-level performance response is crucial for comprehending the differential impact of the invasion shock on private business in Central Asia. Secondly, our study sheds light on possible mitigating factors that can help firms better cope with the impact of the invasion shock and enhance their resilience to similar future shocks. We focus on the role of digital solutions. While firm exposure to trade with Russia can shape the differential impact of the invasion on firm performance even within the same sectors, some factors can help firms better cope with the shock. Previous research has highlighted the benefits of digital technologies for firms in their coping with the effects of the COVID-19 pandemic. Studies conducted by Comin et al. (2022) and Cirera et al. (2022) have shown that firms with higher pre-pandemic adoption of digital and information technologies performed better during the pandemic because of their ability to adapt business strategies in response to the shock. 5 Building on these findings, we aim to explore to what extent the Central Asian firms that adopted digital solutions before the invasion were able to mitigate the adverse impact of the invasion shock on their performance. By examining the adoption and utilization of digital technologies among firms in Central Asia, our study aims to provide insights into the effectiveness of these business strategies in mitigating the negative short-term impacts of the invasion shock and upholding firm performance and resilience. 3 For example, studies by Cirera et al. (2021) and Apehdo et al. (2020) have demonstrated that the burden of the COVID-19 pandemic was disproportionately borne by small and medium-sized firms. 4 For example, Akgündüz et al. (2020) found that the Syrian conflict led to an increase in the number of migrants entering Türkiye, which in turn had a positive impact on the sales performance of existing firms operating in Türkiye, while also expanding the number of new firms being registered in the country. 5 For instance, digital and information technologies can help firms to quickly identify and exploit opportunities to pivot their activity toward less affected markets. 3 We find that firms in the lower 25th percentile of sales and the 75th percentile of employment were more affected in the aftermath of Russia’s invasion of Ukraine than other firms. We try to explain the heterogeneity by examining two candidate structural factors: firms’ export links to Russia and digital gains prior to the invasion. Firms with export links to Russia are estimated to have suffered significantly more in terms of dropping sales and employment. We do not find any significant mitigating effect of digitization after the invasion shock that the existing literature would suggest. However, the results suggest that digitization helped firms increase employment on average during the studied period. While export links to Russia appear to increase sales on average during the studied period, this effect does not survive all the robustness tests we apply. The remainder of the paper is organized as follows. Section 2 describes the employed data and their summary statistics. Section 3 explains the estimation approach and methodology. Section 4 discusses the results. Section 5 concludes and highlights some policy implications. 2. Data We use data from the Business Pulse Survey (BPS) developed by the World Bank to measure the impact of the COVID-19 pandemic on the private sector. The Business Pulse Survey (BPS) is a novel dataset that tracks the potential impact of the pandemic on the private sector along critical dimensions of business performance, such as sales revenue, liquidity, and solvency, labor adjustments, adoption of technology, business expectations and uncertainty, and access to government support programs. In most countries, the survey interviews were conducted over the phone. The data include micro, small, medium, and large businesses across all main sectors (i.e., agriculture, manufacturing, retail, and other services, including construction). The sampling frame excluded micro-firms and/or businesses in agriculture in some countries. In other countries, when micro-businesses were included, the survey instrument offered simplified versions of some questions to facilitate data collection. The sampling frame in most countries where the BPS was not run as a follow-up of the Enterprise Survey was based on censuses from Statistics Agencies, Ministries of Finance or Economy, or business listings from Business Associations and typically only included registered businesses. Our paper uses three waves of data from the Business Pulse Surveys for three Central Asian countries: the Kyrgyz Republic, Tajikistan, and Uzbekistan. The first wave was implemented in August-September 2020, the second wave was implemented in May-July 2021, and the third wave was implemented in March-May 2022. The survey data was collected by interviewing respondents on the phone. In Kyrgyzstan, 1,743 firms were interviewed in the first wave, 1,025 in the second wave, and 1,075 in the third wave. In Tajikistan, 1,690 firms were interviewed in the first wave, 1,032 in the second wave, and 1,031 in the third wave. In Uzbekistan, 1,886 firms were interviewed in the first wave, followed by 1,008 firms and 1,017 firms in the second and third waves, respectively. The sample frame was based on the statistical data from the National Statistical Committees of each country during the first wave in 2020. For the later waves, the list of companies was updated by using lists of business associations and internal lists of entrepreneurs from the survey firm. Data was collected on firms across all four sizes (micro, small, medium, and large) and four sectors (agriculture, manufacturing, retail, and other services). The analysis in our paper is conducted using a balanced panel of firms. To limit the potential noise induced by the variation in our sample of firms across BPS waves, our analysis focuses on the set of firms interviewed in all three consecutive waves of the BPS survey implemented in the three Central Asian 4 countries covered in our analysis. Tables 1 and 2 present the key characteristics of our panel of firms. Given some of the heterogeneity related to the differences in country samples, implementation strategy, and timing of the surveys, we introduce different controls in the analysis. Unless stated otherwise, we include dummies for size (small, medium, and large based on employment), sector (four sector dummies), and country-fixed effects in the analysis. We weigh our results using the stratification weights available at the country level. We also weigh by the initial size of the firm to allow for differential weightage to firms of different sizes in the economy so that we track the aggregate evolution of the sales/employment variable more closely. Table 1 provides further details on panel firms that we use in our analysis. Table 1: Sample Characteristics (Number of Observations) in the Panel Country name Kyrgyzstan Tajikistan Uzbekistan Total Sector Agriculture 141 141 81 363 Manufacturing 165 159 129 453 Retail 183 240 162 585 Other Services 183 195 231 609 Size Micro (0-4) 318 273 243 834 Small (5-19) 246 312 231 789 Medium (20-99) 93 147 129 369 Large (100+) 15 3 0 18 Wave Aug-Sept2020 224 245 201 670 May-July2021 224 245 201 670 March-May2022 224 245 201 670 Table 2: Descriptive Statistics for Sales and Employment Mean p25 Median p75 SD Min Max N Kyrgyzstan Log of sales 12.611 11.002 12.429 13.816 2.011 6.908 23.194 347 Log of employment 2.047 1.099 1.946 2.708 1.148 0 6.908 598 % change in sales during the past 30 days -3.044 -30 0.000 20 47.486 -100 200 377 % change in employment in the last 30 days 94.986 -1.923 0.000 66.667 470.697 -100 6150 372 Tajikistan Log of sales 10.732 9.393 10.463 11.875 1.849 6.908 16.907 468 Log of employment 2.121 1.386 2.079 2.89 1.038 0 5.193 677 % change in sales during the past 30 days -3.769 -20 0.000 10 45.027 -100 200 424 % change in employment in the last 30 days 59.272 0 0.000 37.5 239.036 -92 3500 427 Uzbekistan Log of sales 17.915 16.524 17.728 19.114 1.942 12.722 23.862 388 Log of employment 2.083 1.386 1.946 2.996 1.061 0 5.858 554 % change in sales during the past 30 days 4.181 -30 0.000 20 58.502 -100 200 341 % change in employment in the last 30 days 66.956 -16.667 0.000 39.231 247.435 -100 2900 344 5 Source: authors’ calculations based on Business Pulse Survey data. 3. Methodology To investigate the effect of Russia’s invasion of Ukraine on private businesses in the Central Asia region, this paper exploits the panel dimension of our data. Our empirical analysis is divided into two main parts. First, we provide a descriptive overview of our outcomes of interest, highlighting how the distribution of these firm-level outcomes (sales and employment) changed as Russia’s invasion of Ukraine affected the region. Second, we interact this time variation with the cross-sectional heterogeneity present in our sample of firms. Specifically, we use firm-level exports to Russia as a proxy to measure the exposure of individual businesses to the negative spillovers caused by the invasion. We then analyze the interaction between this proxy and the onset of the invasion to identify the intensity of the invasion shock. Next, we examine the relative performance of firms with higher or lower exposure to the invasion compared to matching peers. This allows us to assess how firms with varying levels of exposure to the invasion were affected by the invasion. Dynamics of firm-level sales and employment First, we analyze the dynamics of outcome variables of interest. The analysis focuses on the differences observed between 2022 when Central Asian economies were affected by the shock triggered by Russia’s invasion of Ukraine, and previous years. We are interested in two outcome variables: firm-level employment and firm-level sales. For every firm j and every year t, the analysis uses the following specification: = + + + + + (1) where is the invasion year dummy, is a vector of dummies for non-invasion BPS waves – one dummy for each of the two pre-2022 waves, is a vector of sector dummies, is a vector of country dummies – one dummy for each of the three countries covered in the analysis, is a set of controls, and is the outcome variable of interest. The analysis uses this specification to describe the evolution of the outcome variable over time. Namely, using OLS with robust standard errors, we estimate specification (1) for the mean of using the quantile regression approach. Analyzing and for the 25 and 75 percentile of th th the entire distribution of an outcome variable of interest, rather than only its mean, is particularly important when considering the effect of an aggregate crisis on individual economic agents. As highlighted earlier, large aggregate shocks can have very heterogenous impacts across individuals and firms, a pattern that can be better understood when looking into the dynamics of higher moments of the (micro)outcome distribution in response to a shock. 6 The varied effect of Russia’s invasion of Ukraine based on a firm’s export exposure to Russia or digitization gains The second specification used in the analysis aims at investigating the invasion’s heterogeneous effects on firms, with special emphasis on Central Asian firms that may have been more exposed to the invasion shock. As already stressed, the analysis first proxies firms’ exposure to the invasion shock by firm-level characteristics (exports to Russia) that are likely to determine a firm’s likely intensity of exposure to the 6 See for example Bloom et al. (2018) and Salgado et al. (2020). 6 invasion shock. It then exploits pre-invasion cross-sectional variation in these proxy variables, interacted with time variation in the shock (invasion vs. non-invasion years) to identify the effect of the invasion on firms that were more or less exposed to the invasion. For every firm j and every year t, the second part of the analysis uses the following specification: , = + + + + + + + + , (2) where is a variable that determines the intensity of the firm’s j ex-ante exposure to the invasion shock. Parameter c thus captures the heterogenous effect of the invasion on firms that were differentially exposed to the invasion. As mentioned earlier, the characteristics used as the term in the analysis – such as trade relations with Russia—captures a firm’s initial exposure to the invasion shock. Section 4 discusses the rationale and justification behind choosing these variables in more detail. An analogous specification can be deployed to examine the empirical evidence about the possible heterogeneous effects of the invasion based on the digitization gains of firms prior to the invasion. The possible mitigating effect of digitization gains on a firm’s export exposure to the invasion shock The final specification used in the analysis aims at assessing whether firm-level digitization gains prior to the invasion shock can act as a mitigating factor in reducing the effect of the invasion on firms with greater exposure to the shock due to their export links to Russia. To this end, the specification uses a triple interaction, interacting the invasion shock, , a proxy of exposure to the shock (initial exports to Russia), , and a measure of a firm-level mitigating factor (digitization gains), , = + + + + + + + + + , (3) + + The coefficient d captures the benefits from digitization gains throughout the estimated sample; coefficient f captures the differential effect of the shock on firms that benefited from the mitigating factor (digitization gains) , and coefficient g captures the additional effect of the mitigating factor (initial digitization gains) on firms that were more exposed to the invasion shock due to export links with Russia. 4. Results The average effect of Russia’s invasion of Ukraine on Central Asian firms This subsection describes the estimated effect of the invasion on two firm-level performance outcomes: sales revenues and labor adjustments, following regression specification (1). For each outcome, we also employ quantile regressions to examine how the invasion's impact differs across different percentiles of firm size (25th, 50th, and 75th percentiles). By analyzing the changes over time in the distribution of sales revenues and labor adjustments, we gain insight into the heterogeneous effects of the crisis on businesses in the region. This approach allows us to go beyond the average effect and understand how the invasion affected firms across the distribution of outcomes (sales and employment). It helps put the unidentified average effect into relative perspective—note that we do not have a counterfactual group to properly identify the average effect because all firms were affected by the invasion to some degree. The estimation results using sales as the dependent variable are reported in Table 3. Overall, we cannot identify any significant average effect of Russia’s invasion of Ukraine on the sales of Central Asian firms (Table 3, column 1). However, we observe important variations across the outcome 7 distributions (sales and employment) when examining the results using quantile regressions. Firms in the 25th and 50th percentiles of the sales distribution experienced a significantly more negative impact on their sales than firms in the 75th percentile (Table 3, columns 2, 3, and 4). The magnitudes of these negative effects for the 25th and 50th percentiles appear comparable. In contrast, larger firms at the 75th percentile of the sales distribution may have faced a much smaller and statistically insignificant sales decline following the invasion. These findings suggest that larger firms may have benefited from absorbing some of the sales declines experienced by firms with small and median sales. In some cases, larger firms may have even gained the entire business previously held by smaller and medium-sized firms. Table 3: Effect of Russia’s Invasion of Ukraine on Firms’ Sales in Central Asia (1) (2) (3) (4) mean 25 percentile th median 75 percentile th Effect of the invasion -0.289 -0.446*** -0.422** -0.268 (0.221) (0.145) (0.206) (0.265) Manufacturing 0.529* 1.172*** 0.843* 0.428 (0.294) (0.358) (0.508) (0.271) Retail 0.604** 0.511 1.308** 0.977** (0.285) (0.324) (0.523) (0.409) Other Services -0.129 0.431 0.475 -0.405* (0.264) (0.322) (0.477) (0.227) Small (5-19) 1.408*** 0.945*** 1.405*** 1.466*** (0.145) (0.143) (0.125) (0.178) Medium (20-99) 2.125*** 1.833*** 2.405*** 2.568*** (0.160) (0.121) (0.178) (0.181) Large (100+) 4.560*** 3.085*** 5.844*** 4.878*** (0.652) (0.479) (0.263) (0.267) Constant 11.784*** 10.657*** 11.396*** 13.020*** (0.275) (0.341) (0.510) (0.240) Observations 1149 1149 1149 1149 R-squared 0.782 0.505 0.571 0.609 Robust standard errors are in parentheses. *** p<.01, ** p<.05, * p<.1 We also observe that manufacturing and retail firms had larger sales than other firms over the studied period. And that small, medium-sized, and large firms had, progressively according to their size, larger sales than micro firms over the studied period. The estimation results for the effect of Russia’s invasion of Ukraine on firms' employment are presented in Table 4. Again, we cannot identify any significant average effect of the invasion on employment at Central Asian firms. However, the results differ from those for sales. We do not observe any significant negative effect on firms in the 25th and 50th percentiles of the employment distribution. Interestingly, we find a significant positive effect of the invasion on firms' employment in the 75th percentile of the employment distribution. This finding provides some support to our earlier hypothesis that larger firms may have absorbed some of the business, both in terms of sales and employment, from smaller and medium-sized firms following the invasion shock. The question is whether there is a structural indicator of exposure to the invasion shock that can better identify this differentiated impact suggested by the 8 estimation results from the quantile regressions for sales and employment. We explore one such measure of the intensity of exposure to the invasion shock next. In addition, the results in Table 4 suggest that manufacturing and other services firms in the 25th percentile of the employment distribution maintained higher employment levels than other firms throughout the study period. As we move up the size categories, smaller, medium-sized, and large firms consistently exhibited higher employment levels than micro firms across the entire employment distribution. This observation suggests that, on average, firms in the manufacturing and other services sectors, as well as firms of larger sizes, were able to maintain relatively higher employment levels compared to micro firms. This finding hints at potentially greater resilience and the ability of these sectors and larger firms to sustain employment during the studied period. Table 4: Effect of Russia’s Invasion of Ukraine on Firms’ Employment in Central Asia (1) (2) (3) (4) mean 25th percentile median 75th percentile Effect of the invasion 0.136 -0.112 0.077 0.425*** (0.154) (0.149) (0.117) (0.094) Manufacturing 0.082 0.320*** 0.077 -0.041 (0.086) (0.072) (0.104) (0.124) Retail -0.098 0.054 -0.146 -0.062 (0.086) (0.065) (0.101) (0.125) Other Services 0.001 0.208** 0.077 -0.041 (0.087) (0.098) (0.092) (0.120) Small (5-19) 1.110*** 1.067*** 1.099*** 1.301*** (0.036) (0.063) (0.031) (0.024) Medium (20-99) 2.306*** 2.245*** 2.303*** 2.505*** (0.048) (0.063) (0.048) (0.036) Large (100+) 4.064*** 3.498*** 4.212*** 4.296*** (0.298) (0.526) (0.055) (1.058) Constant 1.301*** 0.916*** 1.309*** 1.448*** (0.079) (0.108) (0.089) (0.118) Observations 1767 1767 1767 1767 R-squared 0.724 0.490 0.507 0.534 Robust standard errors are in parentheses. *** p<.01, ** p<.05, * p<.1 Varied effect of Russia’s invasion of Ukraine on firms’ performance by firms’ pre-invasion exposure through exports to Russia Table 5 reports the estimation results for regression specification (2) using the exported versus not- exported to Russia (1/0 firm-level) dummy as the exposure measure that we interact with the invasion effect to capture the potentially varied intensity of the invasion’s effect. We expect the invasion’s effect to be more detrimental for firms with export links to Russia before the invasion because of the sanctions imposed on Russia affecting trade, payment settlement, and financial flows. The estimation results indicate that firms with export links to Russia had higher sales on average during the studied period. However, following the invasion, these firms experienced a significant drop in sales, with a decline of approximately 180 log points. This drop corresponds to a reduction of around 86 percent 9 relative to the 17.7 percent decline observed for non-exporting firms on average. The decline in sales among exporting firms was substantial enough to eliminate their sales-volume premium over non- exporting firms. After the invasion, the average sales among firms exporting to Russia were 39 percent lower than the average sales of firms not exporting to Russia—while they were more than three times higher before the crisis. The estimation results show a significant negative impact on employment for firms that exported to Russia compared with firms that did not export to Russia before the invasion. The estimated effect indicates about 120 log points (or around 85 percentage points) larger drop in employment for firms exporting to Russia. This finding could imply that firms not exporting to Russia may have experienced growth in employment during the same period and potentially absorbed a substantial portion of the labor released by firms negatively affected by the decline in the export market to Russia. The firms not exporting to Russia, which serve as a counterfactual in this analysis, could have continued to grow their employment based on the pre-invasion common trend. Additionally, the employment growth may have been supported by hiring immigrants from Russia. Table 5: Effect of Russia’s Invasion of Ukraine on Firms’ Sales and Employment if a Firm Exports to Russia (1) (2) Sales Employment Effect of the invasion -0.195 0.200 (0.229) (0.150) Exported to Russia 1.446*** 0.044 (0.318) (0.170) Invasion x Exported to Russia -1.807*** -1.193** (0.491) (0.534) Manufacturing 0.572** 0.084 (0.267) (0.086) Retail 0.679*** -0.100 (0.259) (0.087) Other Services -0.153 0.005 (0.235) (0.086) Small (5-19) 1.403*** 1.107*** (0.145) (0.036) Medium (20-99) 2.044*** 2.306*** (0.161) (0.049) Large (100+) 4.572*** 4.067*** (0.650) (0.301) Constant 11.707*** 1.297*** (0.263) (0.078) Observations 1149 1767 R-squared 0.789 0.728 Robust standard errors are in parentheses. *** p<.01, ** p<.05, * p<.1 Could digitization drive varying responses of firms’ performance to the invasion shock? The use of digital business solutions can lead to significant benefits for firms, including reduced production costs, upgrading capabilities, and expanded market opportunities. Goldfarb and Tucker (2019) provide a 10 useful review of the economics of digital technologies and find that digitization has changed business economic models by reducing several dimensions of economic costs, such as search, reproduction, transportation, tracking, and verification. 7 The use of digital solutions also affects firm performance from a capability perspective; it allows for upgrading firm organizational capabilities (agility, digital options, and entrepreneurial alertness) and strategic processes (capability-building, entrepreneurial action, and coevolutionary adaptation) (Sambamuthy et al., 2003). In practice, all these changes across different economic costs and firm capability upgrading mean that digital solutions can reduce production costs; for example, reductions in search costs enable buyers and sellers of products or services to better access the other side of the market, increasing the speed or efficacy of firms finding workers or input suppliers (De Loecker, 2019). Likewise, digital business solutions can also help expand market opportunities, as reductions in search, transaction, or tracking costs allow firms to overcome geographical barriers, penetrate new markets, and increase trade volume (World Bank, 2020). Digital solutions are changing the nature of trade by increasing the scale, scope, and speed of trade, changing how value is created and traded, and giving rise to a new ecosystem for trade. Smaller firms can participate more directly in trade because online platforms significantly reduce the costs of selling across borders (Gonzalez and Ferencz, 2019). The effect of digitalization on market opportunities is particularly relevant for SME exports and internationalization efforts: e-commerce, e-marketing, and e-business are expected to impact SME export practices through better strategic positioning, the adaptation of their offerings, openness, and organization (Dethine et al., 2020).8 Digital solutions are also expected to build firm resilience during economic downturns. Evidence from multiple sectors and countries points to accelerated digital adoption as a potential silver lining from the pandemic. Studies by Comin et al. (2022) and Cirera et al. (2022) have presented evidence for developing economies that firms with higher pre-pandemic technological sophistication performed better during COVID-19, partially because they were better at adopting new technologies in response to the shock. Digital and information technologies can help firms quickly identify and exploit opportunities to pivot their activity toward less affected markets. Abidi et al. (2022) highlight the benefits of digitalization in alleviating scars on firm sales and employment in the Middle East and Central Asia region. At a global level, Copestake et al. (2022) found that digitally-enabled firms faced a lower decline in sales during the pandemic compared with digitally-constrained firms—suggesting that digitalization acted as an additional coping mechanism during the pandemic. Nose and Honda (2023) found evidence of heterogeneous resilience to uncertainty shocks between less and more digitalized firms. More digitalized firms weather shocks better, with smaller drops in sales and profits, while less-digitalized firms are worse off, with long-lasting scars. 7 Specifically, digital solutions can reduce search costs, enlarging the potential scope and quality of search. In digital environments, goods can be replicated at zero cost, making them often non-rival. The role of geographic distance changes as the cost of transportation for digital goods and information is approximately zero. Digital technologies make it easy to track any one individual's behavior. Finally, digital verification can make it easier to certify the reputation and trustworthiness of any individual, firm, or organization in the digital economy. 8 At the aggregate level, digital solutions can help structural transformation by fostering the servicification of manufacturing while increasing opportunities for developing countries to diversify in traded goods and services. However, it is unclear if digitalization facilitates better positioning of developing countries in global markets or if it narrows the scope for their participation and upgrading opportunities in global value chains due to relatively larger benefits for developed countries (Matthess and Kunkel, 2020). 11 The estimation results from regression specification (3), which examines the potential heterogeneous effect of the invasion based on firm-level digitization, are presented in Table 6. The findings reconfirm that firms engaged in exports to Russia before the invasion experienced higher sales throughout the studied period. They also reconfirm that the invasion had a more pronounced negative effect on the sales of exporting firms to Russia compared to other firms. In addition, the estimated coefficients for the impact of digitalization on sales and employment are positive and statistically significant. This suggests that firms that had adopted digital solutions before the invasion exhibited higher average sales and employment during the studied period, with approximately 41 log points higher sales and 24 log points higher employment. The estimated coefficient for the mitigating effect of digitization concerning the invasion impact is insignificant. This could be due to the overall lack of significance in the invasion effect across different types of firms (the average effect in Table 3) or the insignificant or possibly slightly positive effect of the invasion on firms that did not export to Russia before the invasion. Table 6: Can Digitization Gains be Associated with Varying Effects of Russia’s Invasion of Ukraine on Firms? (1) (2) Sales Employment Affected by the invasion -0.115 0.083 (0.329) (0.106) Export to Russia 1.370*** 0.001 (0.339) (0.167) Use of digital solutions pre-invasion 0.389* 0.251*** (0.201) (0.074) Invasion x Export to Russia -1.800*** -1.241** (0.547) (0.540) Invasion x Digital -0.110 0.158 (0.415) (0.222) Manufacturing 0.621** 0.053 (0.283) (0.082) Retail 0.675** -0.157* (0.269) (0.086) Other Services -0.143 -0.036 (0.248) (0.085) Small (5-19) 1.371*** 1.075*** (0.145) (0.037) Medium (20-99) 1.970*** 2.244*** (0.160) (0.051) Large (100+) 4.515*** 4.004*** (0.659) (0.291) Constant 11.406*** 1.157*** (0.319) (0.085) Observations 1145 1761 R-squared 0.791 0.734 Robust standard errors are in parentheses. *** p<.01, ** p<.05, * p<.1 12 Can digitization mitigate the effect of Russia’s invasion of Ukraine on exposed firms’ performance? Exploring the triple interaction of invasion, export exposure, and digitization. Further investigation was carried out using a triple interaction with invasion, export exposure, and digitization by running regression specification (3). However, the results were not consistently significant across sales and employment outcomes. This inconsistency likely arises because the triple interaction was estimated on less than 1% of the sample observations, which is too little to achieve stable estimates and reliable inferences. Even though we have tried two alternative specifications: with digitization and non- digitization dummies, the lacking positive observations on the triple interaction could not have been alleviated. The exploratory results are reported in Tables A.3 and A.4 in the annex. Namely, the triple interaction of invasionXexportsXdigitalization (or invasionXexportsXnon-digitalization) dropped out from the regressions for sales because of its low variation. In the regression for employment, the triple interaction with digitization was estimated as significantly negative (positive for non-digitization). Again, this result is preliminary and based on weakly identified triple interaction that accounts for less than 1% of the observations. If such a finding were to be taken as a possible hypothesis for future research, it would ask: why digitized firms may have experienced greater repercussions from the invasion—even though the expectation based on the literature would be the opposite? One plausible conjecture that could be explored in the future if more data becomes available is that certain digitization of firms may leave firms still vulnerable to the impact of certain large shocks, such as shifts in monetary policy. For example, the digitization of central Asian firms may have been concentrated in production technologies and led to higher leveraging in the pursuit of greater financial efficiency before the invasion shock. This greater leverage made them more vulnerable to the subsequent interest rate hikes implemented by monetary authorities in response to rising inflation and inflation expectations. To support this conjecture, Figure 1A plots the probability of falling into arrears in the six months following the invasion shock for digitized versus non-digitized firms by firm size in Figure 1. We can observe that the probability of falling into arrears is higher for digitized firms than non-digitized firms across all firm sizes but micro firms. This stylized fact of our data helps support our conjecture for the unexpected result of digitalization and invasion-trade exposure that links digitization to possible financial overleveraging as digitized firms pursue further efficiencies at the cost of decreasing their resilience to possible interest rate hikes. This conjectural evidence could motivate future research into the possibly more context-dependent effects of digitization on the efficiency and resilience of firms. Robustness checks We conduct several robustness checks by re-estimating equation (2) in alternative specifications. First, we re-estimate equation (2), adding the clustering of standard errors at the level of countries to account for possible covariance of the shocks across firms within the same country—thus controlling for possible unobservable common characteristics such as country-level policies. These estimation results are reported in Table A5. Table A5 shows similar results to our baseline estimation without clustering, but we lose the significance of the effect of pre-invasion exports and digital gains on sales. That is, the positive effect of digitization on employment is robust to clustering, and so is the negative effect of the double 13 interaction of invasion with exports on both sales and employment. The double interaction of invasion and digitization remains insignificant. Second, we re-estimate equation (2) clustering at the industry level to account for industry-specific spillovers and common factors such as industry-specific disruptions in commodity markets or supply chains. The result reported in Table A6 suggests that clustering at the industry level generally produces more significant results and further reinforces the baseline results. Namely, the effect of pre-invasion exports on sales remains significant but only at the 10% level, and we lose the significance of pre-invasion digitization on sales. The effect of pre-invasion digitization of employment and the double interaction of invasion and exports on sales and employment remain significant at 10% and 5%, respectively. Third, we try to maximize the number of available observations by estimating equation (2) on a sample of repeated cross-sections rather than a time-series panel in our baseline. The sample summary statistics are in Table A7. The cross-sectional data contains 5,319, 3,065, and 3,123 observations in waves 1 (2020), 2 (2021), and 3 (2022) across the three countries—compared with 670 observations in the panel dataset in each of the waves. The largest share of firms belongs to the retail sector and services sector (about 32 and 29 percent), and the firms are most frequently micro and small firms (about 41 and 40 percent of observations) in the repeated cross-section data. In the panel dataset, the largest share of firms belongs to the services and retail sector (30 and 29 percent) and are micro and small firms (about 41 and 40 percent). The quantile regression results in Tables A8a and A8b suggest that the sales and employment in the 25th quantiles (firms with lower sales) were more negatively affected than those in other quantiles. This possible heterogeneity differs from that estimated for the panel, where we observed a relatively more negative effect of the invasion on sales at the 25th and 50th percentiles for sales and the 75th percentile for employment. Table A9 reports the results of re-estimating equation (2) on a sample of repeated cross-sections with (in our case) stricter clustering at the country level. Note that, unlike the panel dataset, we cannot track all firms before and after the invasion. Moreover, we faced issues with increased survey attrition after the invasion. Therefore, while we can improve the pre-invasion trend estimates, we do not improve as much the post-invasion estimates while not exactly matching the characteristics of the additional firm-level observations before and after the invasion. The results reported in Table A9 confirm that the effect of the double interaction invasionXexports on employment remains robust even to changing the sample. The effect of invasionXexports on sales remains negative but with a larger estimated standard error and does not reach the conventional significance levels. The same applies to pre-invasion exports and digital gains. Overall, we interpret the results of the applied robustness test as indicating the robustness of the negative effect of the invasion through export links with Russia on firms’ performance (employment and sales). The positive effect of pre-invasion digitization gains on employment also appears generally robust. The positive effect of pre-invasion export links with Russia is less robust—especially to country-level clustering of errors (spillovers). 14 5. Conclusion This paper examined the impact of Russia’s invasion of Ukraine on firms in Central Asian countries, focusing on firms with pre-invasion trade links with Russia as a measure of exposure to the invasion shock and firms with pre-invasion digitization gains as a potential mitigant of the shock to explain possible heterogeneity in the Ukraine invasion effects. The study found that Central Asian firms that had prior export ties to Russia experienced significant negative effects on their sales and employment after the invasion shock. The analysis also considered firms' pre-invasion digitization as a potential mechanism for coping with the invasion's repercussions. However, no significant mitigation effect of digitization on firm performance is found, possibly due to mixed effects on firms that did not export to Russia. The paper's findings have important policy implications, highlighting the potential trade-off between the efficiency gains associated with concentrated export links and the importance of geographic export diversification for resiliency to geopolitical shocks. Based on unstable results for digitization as a mitigant of risks from concentrated exports given possible geopolitical shocks, we encourage future research into the possible trade-off that firms' digitization in particular areas can deliver between firms’ efficiency in normal times and coping with unexpected large economic shocks. More recently, countries in Central Asia boosted their trade thanks to the intermediary (transit) trade in goods from Western countries to Russia (EBRD, 2023 and ADB 2023). 9,10 After sanctions were introduced, the trade turnover of the European Union with Russia’s neighbors to the East increased many times, while their Russia trade also increased. In 2022, the intra-EAEU trade was up 40 percent. ADB (2023) reports that, for Central Asian countries, reexports accounted for close to 50 percent of the rise in exports to Russia. Moreover, Central Asian countries have replenished their own budgets through an influx of labor, capital, and remittances. In addition, the countries increased the sale of their own oil and gas, the demand for which has hiked after the large drop in EU gas supplies from Russia. With the relocation of Russian business to Central Asian countries, the arguably temporary inflow of skilled labor from Russia, and a short-lived windfall reexport margin (ADB 2023), the total effects of Russia’s invasion of Ukraine on Central Asian business—and its dynamics over time—needs to be further studied using updated firm-level data and rigorous empirical methods. 9 https://www.russia-briefing.com/news/ebrd-report-anti-russian-sanctions-are-boosting-central-asian- economies.html/ 10 https://www.adb.org/sites/default/files/publication/863591/ado-april-2023-special-topic.pdf 15 References Abidi, N., Sakha, S., & Herradi, M. E. (2022). Digitalization and Resilience: firm-level evidence during the COVID-19 Pandemic. IMF Working Paper, 2022(034). https://doi.org/10.5089/9798400201073.001 Akgündüz, Y. E., Bağır, Y. K., Cılasun, S. M., & Kirdar, M. G. (2023). Consequences of a massive refugee influx on firm performance and market structure. Journal of Development Economics, 162, 103081. https://doi.org/10.1016/j.jdeveco.2023.103081 Bachmann, R., Baqaee,D., Bayer,C., Kuhn,M., Löschel,A., Moll,B., Peichl,A., Pittel,K., & Schularick, P. (2022). What if? The economic effects for Germany of a stop of energy imports from Russia. CESInfo Econ Pol Policy Report (36) Baqaee, D., Moll, B., Landais, C., & Martin, P. (2022) The economic Consequences of a Stop of Energy Imports from Russia. French Council of Economic Advisors, Focus 2022 (084). Bloom, N., Floetotto, M., Jaimovich, N., Saporta-Eksten, I., & Terry, S. (2018). Really uncertain business cycles. Econometrica, 86(3), 1031–1065. https://doi.org/10.3982/ecta10927 Bougias, A., Episcopos, A., & Leledakis, G. N. (2022). Valuation of European firms during the Russia– Ukraine war. Economics Letters, 218, 110750. https://doi.org/10.1016/j.econlet.2022.110750 Cirera, X., Vargas Da Cruz, M., Grover, A., Iacovone, L., Medvedev, D., Pereira Lopez, M. M., & Reyes, S. (2021) Firm Recovery during COVID-19: Six Stylized Facts. Policy Research Working Paper Series, 2021 (9810). Comin, D. A., Cruz, M., Cirera, X., Lee, K. M., & Torres, J. (2022). Technology and resilience. NBER Working Paper, 2021 (19644). Copestake, A., Estefania-Flores, J., & Furceri, D. (2022). Digitalization and Resilience. IMF Working Paper, 2022(210). https://www.imf.org/en/Publications/WP/Issues/2022/10/28/Digitalization-and- Resilience-525207. Dethine, B., Enjolras, M., & Monticolo, D. (2020). Digitalization and SMEs’ export Management: Impacts on resources and capabilities. Technology Innovation Management Review, 10(4), 18–34. https://doi.org/10.22215/timreview/1344 European Bank for Reconstruction and Development (2022). Transition Report 2022−23: Business Unusual, London 2022. Ferriani, F., & Gazzani, A. (2022). The impact of the war in Ukraine on energy prices: Consequences for firms’ financial performance. Social Science Research Network. https://doi.org/10.2139/ssrn.4216406 Goldfarb, A., & Tucker, C. (2019). Digital Economics. Journal of Economic Literature, 57(1), 3–43. https://doi.org/10.1257/jel.20171452 16 Hutter, C., & Weber, E. (2022). Russia-Ukraine war: short-run production and labour market effects of the energy crisis. IAB-Discussion Paper, 2022(10). Iootty, M. & Melecky, M. (2022) Making the pulse of business in Central Asia following the Russian invasion of Ukraine. Blog published on Eurasian Perspectives. Available at: https://blogs.worldbank.org/europeandcentralasia/taking-pulse-business-central-asia-following- russian-invasion-ukraine Li, H., Wu, Y., Cao, D., & Wang, Y. (2021). Organizational mindfulness towards digital transformation as a prerequisite of information processing capability to achieve market agility. Journal of Business Research, 122, 700–712. https://doi.org/10.1016/j.jbusres.2019.10.036 López González, J. & Ferencz, J. (2018). Digital Trade and Market Openness. OECD Trade Policy Papers, No. 217, OECD Publishing, Paris. http://dx.doi.org/10.1787/1bd89c9a-en Mansell, R. (2001). Digital opportunities and the missing link for developing countries. Oxford Review of Economic Policy, 17(2), 282–295. https://doi.org/10.1093/oxrep/17.2.282 Mattera, M., & Soto, F. G. (2022). Dodging the bullet: overcoming the financial impact of Ukraine armed conflict with sustainable business strategies and environmental approaches. The Journal of Risk Finance, 24(1), 122–142. https://doi.org/10.1108/jrf-04-2022-0092 Matthess, M., & Kunkel, S. (2020). Structural change and digitalization in developing countries: Conceptually linking the two transformations. Technology in Society, 63, 101428. https://doi.org/10.1016/j.techsoc.2020.101428 Nose, M., & Honda, J. (2023). Firm-level Digitalization and Resilience to Shocks: Role of Fiscal Policy. IMF Working Paper, 2023(034). Salgado, S., Guvenen, F., & Bloom, N. (2019). Skewed business cycles. https://doi.org/10.3386/w26565 World Bank (2020). World Development Report 2020: Trading for development in the age of global value chains. In Washington, DC: World Bank eBooks. https://doi.org/10.1596/978-1-4648-1457-0 World Bank (2023a). Europe and Central Asia Economic Update, Spring 2023: Weak Growth, High Inflation, and a Cost-of-Living Crisis. © Washington, DC : World Bank. http://hdl.handle.net/10986/39617 World Bank (2023b). Private Sector Resilience: What have we learnt from the COVID-19 pandemic? Forthcoming. 17 Annex Table A1: Effect of Russia’s Invasion of Ukraine on Sales if Firm Exports to Russia (1) (2) (3) (4) mean 25th percentile median 75th percentile Affected by the invasion -0.195 -0.342** -0.215 -0.094 (0.229) (0.172) (0.248) (0.288) Export to Russia 1.446*** 1.595** 1.275*** 0.693 (0.318) (0.633) (0.416) (1.311) Invasion x Export to Russia -1.807*** -1.659 -1.914 -2.209 (0.491) (1.409) (1.300) (2.496) Manufacturing 0.572** 1.040*** 0.819* 0.317 (0.267) (0.306) (0.483) (0.373) Retail 0.679*** 0.509* 1.333** 0.822* (0.259) (0.303) (0.522) (0.474) Other Services -0.153 0.300 0.278 -0.599 (0.235) (0.354) (0.469) (0.403) Small (5-19) 1.403*** 0.929*** 1.437*** 1.510*** (0.145) (0.128) (0.174) (0.170) Medium (20-99) 2.044*** 1.833*** 2.259*** 2.520*** (0.161) (0.080) (0.211) (0.256) Large (100+) 4.572*** 3.172*** 5.734*** 4.921*** (0.650) (0.625) (0.336) (0.402) Constant 11.707*** 10.702*** 11.207*** 13.087*** (0.263) (0.386) (0.486) (0.358) Observations 1149 1149 1149 1149 R-squared 0.789 0.516 0.581 0.611 Robust standard errors are in parentheses. *** p<.01, ** p<.05, * p<.1 18 Table A2: Effect of Russia’s Invasion of Ukraine on Employment if Firm Exports to Russia (1) (2) (3) (4) mean 25th_percentile median 75th_percentile Affected by the invasion 0.200 0.000 0.125 0.468*** (0.150) (0.134) (0.123) (0.077) Export to Russia 0.044 0.357** 0.223 0.095 (0.170) (0.170) (0.353) (0.154) Invasion x Export to Russia -1.193** -1.070 -1.309** -1.362 (0.534) (2.174) (0.628) (2.139) Manufacturing 0.084 0.315*** 0.065 -0.022 (0.086) (0.119) (0.102) (0.138) Retail -0.100 0.092 -0.159 -0.062 (0.087) (0.104) (0.098) (0.139) Other Services 0.005 0.224** 0.065 -0.040 (0.086) (0.113) (0.089) (0.137) Small (5-19) 1.107*** 1.071*** 1.139*** 1.300*** (0.036) (0.064) (0.031) (0.032) Medium (20-99) 2.306*** 2.212*** 2.323*** 2.504*** (0.049) (0.071) (0.047) (0.047) Large (100+) 4.067*** 3.465*** 4.212*** 4.296*** (0.301) (0.249) (0.056) (1.040) Constant 1.297*** 0.874*** 1.301*** 1.430*** (0.078) (0.124) (0.087) (0.132) Observations 1767 1767 1767 1767 R-squared 0.728 0.498 0.511 0.536 Robust standard errors are in parentheses. *** p<.01, ** p<.05, * p<.1 19 Table A3: The Effect of Triple Interaction of Invasion, Exports, and Digitization on Sales and Employment (1) (2) Sales Employment Affected by the invasion -0.115 0.076 (0.329) (0.106) Export to Russia 1.370*** 0.001 (0.339) (0.167) Use of digital solutions pre-invasion 0.389* 0.250*** (0.201) (0.074) Invasion x Export to Russia -1.800*** 0.459** (0.547) (0.188) Invasion x Digital -0.110 0.168 (0.415) (0.222) Invasion*Digital*Export to Russia -1.741*** (0.532) Manufacturing 0.621** 0.052 (0.283) (0.082) Retail 0.675** -0.158* (0.269) (0.086) Other Services -0.143 -0.037 (0.248) (0.085) Small (5-19) 1.371*** 1.078*** (0.145) (0.037) Medium (20-99) 1.970*** 2.247*** (0.160) (0.051) Large (100+) 4.515*** 4.007*** (0.659) (0.291) Constant 11.406*** 1.155*** (0.319) (0.085) Observations 1145 1761 R-squared 0.791 0.734 Robust standard errors are in parentheses. *** p<.01, ** p<.05, * p<.1 20 Table A4: The effect of Triple Interaction of Invasion, Exports, and Lacking Digitization on Sales and Employment (1) (2) Sales Employment Affected by the invasion -0.225 0.244 (0.285) (0.195) Export to Russia 1.370*** 0.001 (0.339) (0.167) No digitization -0.389* -0.250*** (0.201) (0.074) Invasion x Export to Russia -1.800*** -1.281** (0.547) (0.545) Invasion x No-digital 0.110 -0.168 (0.415) (0.222) Invasion x No-digital x Export to Russia 1.741*** (0.532) Manufacturing 0.621** 0.052 (0.283) (0.082) Retail 0.675** -0.158* (0.269) (0.086) Other Services -0.143 -0.037 (0.248) (0.085) Small (5-19) 1.371*** 1.078*** (0.145) (0.037) Medium (20-99) 1.970*** 2.247*** (0.160) (0.051) Large (100+) 4.515*** 4.007*** (0.659) (0.291) Constant 11.795*** 1.405*** (0.276) (0.084) Observations 1145 1761 R-squared 0.791 0.734 Robust standard errors are in parentheses. *** p<.01, ** p<.05, * p<.1 21 Table A5: Robustness Check of the Invasion’s Effect on Sales and Employment Using Country-Level Clustering (1) (2) Sales Employment Affected by the invasion -0.115 0.083** (0.131) (0.018) Export to Russia 1.370 0.001 (0.536) (0.105) Use of digital solutions pre-invasion 0.389 0.251* (0.311) (0.060) Invasion x Export to Russia -1.800** -1.241** (0.301) (0.236) Invasion x Digital -0.110 0.158 (0.241) (0.211) Manufacturing 0.621 0.053 (0.677) (0.136) Retail 0.675 -0.157** (0.619) (0.027) Other Services -0.143 -0.036 (0.929) (0.127) Small (5-19) 1.371* 1.075*** (0.378) (0.012) Medium (20-99) 1.970** 2.244*** (0.275) (0.050) Large (100+) 4.515*** 4.004*** (0.311) (0.167) Constant 11.406*** 1.157** (0.630) (0.175) Observations 1145 1761 R-squared 0.791 0.734 Standard errors clustered at the country level are in parentheses *** p<.01, ** p<.05, * p<.1 22 Table A6: Robustness Check of the Invasion’s Effect on Sales and Employment Using Industry-level Clustering (1) (2) Sales Employment Affected by the invasion -0.115 0.083 (0.205) (0.072) Export to Russia 1.370* 0.001 (0.441) (0.175) Use of digital solutions pre-invasion 0.389 0.251*** (0.224) (0.042) Invasion x Export to Russia -1.800* -1.241** (0.580) (0.355) Invasion x Digital -0.110 0.158 (0.384) (0.110) Manufacturing 0.621*** 0.053* (0.063) (0.022) Retail 0.675*** -0.157*** (0.079) (0.022) Other Services -0.143* -0.036** (0.046) (0.008) Small (5-19) 1.371*** 1.075*** (0.208) (0.033) Medium (20-99) 1.970*** 2.244*** (0.059) (0.019) Large (100+) 4.515*** 4.004*** (0.173) (0.082) Constant 11.406*** 1.157*** (0.323) (0.063) Observations 1145 1761 R-squared 0.791 0.734 Standard errors clustered at the industry level are in parentheses *** p<.01, ** p<.05, * p<.1 23 Table A7: Sample Summary Statistics for Repeated Cross-Sections Panel A: Number of Observations Country name Kyrgyzstan Tajikistan Uzbekistan Total Sector Agriculture 718 733 523 1974 Manufacturing 1026 617 825 2468 Retail 1120 1323 1122 3565 Other Services 939 978 1374 3291 Size Micro (0-4) 1727 1433 1430 4590 Small (5-19) 1386 1594 1462 4442 Medium (20-99) 542 631 882 2055 Large (100+) 66 6 63 135 Wave Aug-Sept2020 1743 1690 1886 5319 May-July2021 1025 1032 1008 3065 March-May2022 1075 1031 1017 3123 Panel B: Descriptive Statistics for Sales and Employment Mean p25 Median p75 SD Min Max N Kyrgyzstan Log of sales 12.428 11.002 12.206 13.592 1.85 6.908 23.194 1573 Log of employment 1.896 1.099 1.609 2.639 1.114 0 8.006 3475 % change in sales during the past 30 days 3.022 -20 0.000 20 50.269 -100 400 1715 % change in employment in the last 30 days 84.909 0 0.000 33.333 972.888 -100 29900 1053 Tajikistan Log of sales 10.621 9.21 10.373 11.807 1.892 5.298 17.622 1768 Log of employment 2.048 1.386 1.946 2.773 1.021 0 5.394 3498 % change in sales during the past 30 days -4.487 -25 0.000 10 45.158 -100 350 1721 % change in employment in the last 30 days 61.901 0 0.000 33.333 274.76 -95 4900 1124 Uzbekistan Log of sales 18.076 16.811 17.910 19.337 2.101 11.513 28.73 1759 Log of employment 2.146 1.386 2.079 2.996 1.146 0 8.039 3594 % change in sales during the past 30 days 10.443 -20 0.000 30 60.33 -100 500 1680 % change in employment in the last 30 days 45.807 0 0.000 33.333 185.883 -100 2900 989 Source: authors’ calculations based on Business Pulse Survey data. 24 Table A8a: Effect of Russia’s Invasion of Ukraine on Firms’ Sales Using Repeated Cross-Sections (1) (2) (3) (4) mean 25th percentile median 75th percentile Affected by the invasion -0.215 -0.312** -0.216 -0.102 (0.363) (0.128) (0.253) (0.229) Manufacturing 0.735 0.625*** 1.058*** 0.483** (0.429) (0.192) (0.231) (0.240) Retail 1.270* 1.119*** 1.598*** 1.072*** (0.434) (0.235) (0.191) (0.226) Other Services 0.402 0.438*** 0.477** 0.163 (0.614) (0.120) (0.188) (0.214) Small (5-19) 1.274** 1.204*** 1.344*** 1.463*** (0.191) (0.067) (0.095) (0.081) Medium (20-99) 2.429*** 2.226*** 2.653*** 2.763*** (0.207) (0.122) (0.134) (0.103) Large (100+) 3.868** 3.908*** 3.871*** 4.033*** (0.605) (0.199) (0.390) (0.527) 11.246*** 10.382*** 11.043*** 12.189*** (0.591) (0.154) (0.192) (0.214) Observations 3703 3703 3703 3703 R2 0.7502 0.502 0.555 0.547 Robust standard errors are in parentheses *** p<.01, ** p<.05, * p<.1 Table A8b: Effect of Russia’s Invasion of Ukraine on Firms’ Employment Using Repeated Cross-Sections (1) (2) (3) (4) mean 25th percentile median 75th percentile Affected by the invasion -0.046 -0.201* -0.054 0.204 (0.072) (0.107) (0.097) (0.160) Manufacturing -0.012 0.062 0.000 -0.030 (0.069) (0.063) (0.043) (0.066) Retail -0.102 -0.143*** -0.121** -0.063 (0.043) (0.052) (0.048) (0.062) Other Services 0.040 -0.028 0.078 0.000 (0.192) (0.065) (0.067) (0.086) Small (5-19) 1.196*** 1.045*** 1.204*** 1.354*** (0.052) (0.023) (0.017) (0.021) Medium (20-99) 2.472*** 2.305*** 2.473*** 2.708*** (0.061) (0.030) (0.025) (0.029) Large (100+) 4.063*** 3.769*** 3.791*** 4.348*** (0.285) (0.084) (0.103) (0.569) Constant 1.297** 1.034*** 1.220*** 1.386*** (0.136) (0.052) (0.037) (0.061) Observations 8307 8307 8307 8307 R2 0.7286 0.517 0.501 0.506 Robust standard errors are in parentheses *** p<.01, ** p<.05, * p<.1 25 Table A9: Robustness Check of Invasion’s Effect on Sales and Employment Estimated on the Sample of Repeated Cross-Sections Using Country-Level Clustering (1) (2) Sales Employment Affected by the invasion 0.051 -0.139 (0.330) (0.120) Export to Russia 0.280 0.114 (0.513) (0.056) Use of digital solutions pre-invasion 0.506 0.080 (0.302) (0.034) Invasion x Export to Russia -0.831 -1.346** (0.567) (0.301) Invasion x Digital -0.362 0.306 (0.257) (0.264) Manufacturing 0.702 -0.022 (0.480) (0.103) Retail 1.224* -0.085 (0.417) (0.080) Other Services 0.333 0.052 (0.661) (0.226) Small (5-19) 1.225** 1.152*** (0.161) (0.050) Medium (20-99) 2.309*** 2.416*** (0.186) (0.101) Large (100+) 3.774** 3.997*** (0.550) (0.328) Constant 10.915*** 1.307** (0.481) (0.179) Observations 3536 6008 R-squared 0.752 0.705 Standard errors clustered at the country level are in parentheses *** p<.01, ** p<.05, * p<.1 26 Figure A.1: The probability of falling into arrears for digitized versus non-digitized firms by firm size Probability of falling into arrears .25 Average probability of falling into arrears .2 .15 .1 .05 0 Micro (0-4) Small (5-19) Medium (20-99) Large (100+) Does not expect to fall into arrears It is/expects to fall into arrears Source: Author’s calculations based on the Business Pulse survey data 27