Policy Research Working Paper 10923 Thirsty Business A Global Analysis of Extreme Weather Shocks on Firms Roberta Gatti Asif M. Islam Casey Maue Esha Zaveri Middle East and North Africa Region & Planet Vice Presidency September 2024 Policy Research Working Paper 10923 Abstract Using global data from the World Bank’s Enterprise Surveys Governance may be an exacerbating factor, with negative that includes the precise geo-location of surveyed firms, this precipitation shocks increasing exposure to corruption. Yet, paper examines how dry spells and precipitation shocks there is also some indication that digitally connected and influence firm performance. The study finds that firms in innovative firms are more resilient to negative precipita- areas that experience dry spells have lower performance tion shocks. Process innovation, website ownership, and in terms of sales. This is particularly true for smaller firms use of technology licensed from foreign firms mediate the and those in developing economies. A higher number of effects of negative precipitation shocks on firm performance. extreme dry days also increases the chances that a firm will However, there is little evidence of adaptation. Negative exit the market. The main channels are largely through labor precipitation shocks have no effect on the presence of green productivity and infrastructure service disruptions such as management practices or green investments for a subset of water and power outages. There is also some evidence of lim- firms for which such data is available. ited access to finance due to negative precipitation shocks. This paper is a product of the Office of the Chief Economist, Middle East and North Africa Region and the Office of the Chief Economist, Planet Vice Presidency. 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 aislam@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 Thirsty Business: A Global Analysis of Extreme Weather Shocks on Firms* Roberta Gatti, Asif M. Islam, Casey Maue, Esha Zaveri JEL Codes: Q1, Q5, H54, O14, D73 Keywords: Precipitation shocks, firm productivity, firm-level analysis, climate change *Roberta Gatti is the chief economist of the Middle East and North Africa Region of the World Bank (email: rgatti@worldbank.org). Asif M. Islam is a senior economist in the Middle East and North Africa Chief Economist Office (aislam@worldbank.org). Esha Zaveri is a Senior Economist in the Office of the Chief Economist of the Planet Vice-Presidency at the World Bank (ezaveri@worldbank.org). Casey Maue is a post-doctoral scholar at the University of Washington School of Environmental and Forest Sciences (cmaue@uw.edu). We would like to thank Hanan Jacoby, Richard Damania, Daniel Lederman, Patrick Behrer and Fan Zhang for comments on an earlier version of the paper. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Thirsty Business: A Global Analysis of Extreme Weather Shocks on Firms* 1. Introduction The variability of rainfall, defined as deviations from its long-term mean, is a growing challenge. Over the past three decades, 1.8 billion people, or approximately 25 percent of humanity, have endured abnormal rainfall episodes each year, whether it was a particularly wet or unusually dry year (Damania et al., 2017). With climate change, deviations from trends are projected to become more pronounced and frequent. Droughts and adverse water supply shocks are a particular concern, with drought frequency and duration rising by nearly a third globally since 2000 (The United Nations Convention to Combat Desertification (UNCCD), 2022) with lasting negative impacts on economic growth in developing economies (Zaveri et al., 2023; Russ, 2020). While the effects of extreme weather events on agriculture and rural areas have received considerable attention, there are also consequences for cities that may have significant implications. The last few years have seen several major cities like Cape Town in South Africa, São Paolo in Brazil, and Chennai in India, face “day zero” type events in which water supplies become threateningly low, with countless more medium-size and small cities experiencing intermittent water supply and water shortages (Zaveri et al., 2021; Singh et al., 2021). Water scarcity can significantly impact households, public services, and critical infrastructure systems, affecting workers and entire communities (Damania et al., 2017; Hyland and Russ, 2019; Islam and Hyland, 2019). The varied effects of extreme weather events on the private sector are not well understood. Firms are a critical engine of economic growth. They generate jobs, provide essential products and services, and encourage innovation. They link cities and towns to global markets. This study explores the effect of 2 droughts (negative precipitation shocks) on a global sample of firms in urban centers. The study finds that negative precipitation shocks hurt firm performance in terms of sales. This is particularly true for smaller firms and those in developing economies. Firms that experience negative precipitation shocks are also more likely to exit the market. An additional extreme dry day leads to a 0.6 percent reduction in sales. The average number of extreme dry days in the sample is 6.7 days such that an increase in extreme dry days of this amount translates to a 3.8 percent reduction in sales. At the sample maximum of 86 days or about 3 months of extreme dry days, the loss in sales can rise to 48.6 percent. Since extreme dry days also lead firms to exit, these estimates may represent an underestimate of the overall impact. The literature has identified several channels through which extreme weather events could affect firms. One channel is through human capital. Hot days could lead to more absenteeism or lower the labor productivity of workers (Somanathan et al., 2021). Another channel is through infrastructure. Negative precipitation shocks (droughts) increase the intensity of water outages that hurt sales (Islam and Hyland, 2019). Droughts may also increase the frequency of power outages (Desbureaux and Rodella, 2019). Access to finance is another channel identified by the literature. Frequent climate shocks might affect the ability of banks to predict outcomes, leading to an increase in interest rates due to additional risk which could, in turn, increase the cost of capital (Kling et al., 2021). Extreme weather events could also lead to balance sheet erosion as firms that experience monetary losses from shocks become more leveraged as they are more likely to get their loan applications rejected and be seen as less creditworthy (Benincasa et al., 2024). Weather shocks can also create liquidity shortages and increase loan defaults, deteriorating credit scores and access to future credit (Aguilar-Gomez et al., 2023). The effects of extreme weather events on investment, however, are not obvious. On one hand, if weather shocks limit access to finance, firms are less likely to invest. On the other hand, firms affected by weather shocks are more likely to invest in fixed assets as they replenish damaged capital (Benincasa et al., 2024). New investments can also lead to vintage effects where replenishment of capital means newer equipment 3 with a lower environmental footprint. Alternatively, firms may become environmentally aware and therefore engage in green investments or green management practices. The Enterprise Surveys allows for the possibility to test some of these channels. The main channels are largely infrastructure service disruptions such as water and power outages. There is some evidence of effects through labor productivity (sales per worker) and also limited access to finance – negative precipitation shocks decrease the likelihood than firms use banks to financing working capital and have access to overdraft facilities. A new channel uncovered in this study is governance. The intensity of negative precipitation shocks increases exposure to corruption. While there is no effect uncovered regarding weather shocks and investment in machinery and equipment, this could be due to the countervailing effects of limited access to finance and the need to replace damaged capital. However, extreme dry days do increase the probability of investing in land and buildings. Firms that are innovative in terms of process, have website ownership, and use technology licensed from foreign firms experience more muted effects of extreme dry days on firm performance. There is also some evidence that digital technologies and innovation can buffer against climate shocks (Zhao and Parhizgari, 2024; Liu et al., 2023). Finally, for a cross-section of 2019 surveys largely conducted in the Middle East and North Africa and Europe and Central Asia, a green module included in the survey instrument captures green investments and green management practices. This study leverages this new data for a subset of firms where it is available but finds no correlation between precipitation shocks and green management practices or green investments. One explanation could be that firms might adopt such practices only after repeated dry spells over time. However, these survey questions are limited. Adoption of green management practices or investments are captured through a binary variable – where a value of 1 means adoption and 0 implies no adoption - that pertains only to three years prior to the survey. Hence a firm that adopted any green management practices or made green investments 4 years prior may be coded as a zero. Therefore, these findings should be interpreted with care. 4 Several policy implications can be drawn from the findings of the study. First, the results reveal that smaller firms, and those in developing economies are more susceptible to climate shocks. Building resilience among smaller firms and those in developing economies is essential. Second, water and power infrastructure are central to the narrative. As climate change makes precipitation patterns more variable and unpredictable, investments in public water and power infrastructure systems are an important way in which governments can help firms adapt. Second, institutions matter, and governance may play a critical role in how firms fare after extreme weather events. Third, encouraging innovation is one way to build resilience in firms. Finally, extreme weather events may not necessarily lead firms to adapt, and thus other policy interventions would be needed to increase green management practices and green investments. In summary, the study makes the following contributions to the literature. First, the study provides a global analysis of extreme weather events and firm performance. Second, it tests several channels through which the effects of negative precipitation shocks can be identified in the literature while proposing a new channel related to governance. Third, the study explores the types of firms that are more resilient to shocks. And fourth, the study uses a unique dataset to dispel the notion that firms may become environmentally aware after climactic shocks. The rest of the paper is structured as follows. Section 2 describes the data and the empirical approach. Section 3 provides the results with robustness checks. Section 4 concludes. 2. Empirical Approach 2.1 Data 2.1.1 Enterprise Surveys The main data source is cross-sectional firm-level surveys across the world from the World Bank’s Enterprise Surveys (ES). The ES are nationally representative surveys of private formal (registered) firms 5 with 5 or more employees and cover manufacturing and services firms largely collected via face-to-face interviews with business owners or top managers. Our sample for the analysis is restricted to firms that have geo-located information. The final sample consists of about 88,000 firms (depending on the specification) across 174 surveys over 118 economies in the time period 2009-2019. Summary statistics are provided in table A1. The full list of countries and survey years are presented in table A2. In addition, for countries where there were several rounds of surveys, we can track whether the firm exited the market regardless of whether they were re-interviewed in successive waves. We can obtain information on firm exit across 55 countries. The ES methodology includes a consistent definition of the universe of inference, a standard sampling methodology, a standardized survey instrument, and a uniform methodology of implementation. The selection of firms in each country is achieved by stratified random sampling with three levels of stratification: sector, size, and location within the country. Sampling weights are used to correct for unequal probability of selection, ineligibility and non-response. The data are largely collected using Computer- Assisted Personal Interviewing (CAPI) software. The CAPI software collects geo coordinates of the firm’s location that we use to match rainfall and temperature data with the firm-level data. To maintain anonymity of the respondents, the geo-codes are masked around a 2km radius. The surveys are implemented uniformly across countries. Formal training sessions of supervisors and enumerators are undertaken to ensure the best practices are employed consistently. Quality control checks are implemented to guarantee the quality of the data throughout the data collection process. Consistency checks are employed for 10% and 50% batches of the data during the survey to facilitate callbacks to respondents to be undertaken when necessary to verify information. Information on the Enterprise Surveys global methodology and on the sample design and weights computation is available on the website http://www.enterprisesurveys.org. The data have been widely used by several studies analyzing the private sector in developing economies (Besley & Mueller, 2018; Chauvet & Ehrhar, 2018; Hjort & Poulsen, 2019). 6 2.1.2 Precipitation Shocks To measure precipitation shocks, we use reanalysis data produced by the European Centre for Medium- Range Weather Forecasts (ECMWF). More specifically, we use the new land component of the fifth generation of European ReAnalysis (ERA5), hereafter referred to as ERA-5 Land dataset (Muñoz-Sabater et al., 2021). This dataset is produced by the ECMWF as part of the ongoing operations of the Copernicus Climate Change Service (C3S), a subdivision of the Copernicus program, which is the Earth Observation arm of the space program established by the European Commission. ERA5-Land is a global-scale dataset that contains hourly records of more than 50 key meteorological variables (including precipitation) at a 9 km spatial resolution over the period from 1950 to the present. These records are produced by running downscaled meteorological forcings obtained from the ERA5 climate reanalysis 1 through a high-resolution land surface process model developed by ECMWF. For our application, we use the daily aggregated version of the dataset, which is freely provided on the Copernicus Climate Data Store (CDS) and accessible via Google Earth Engine. There are several key features of the ERA5-Land data. First, as detailed below, constructing the precipitation shock measure we favor in our analysis involves normalizing daily rainfall observations against a day-of-year and grid-cell specific climate distribution. Computing the relevant moments of these localized distributions requires a long, consistently measured, time series of daily rainfall data. Reanalysis datasets like ERA5-Land have both these features. Second, with a resolution of 9 km, the ERA5-Land data allow us to measure precipitation shocks precisely in the exact areas where the firms in the Enterprise Surveys are located. Finally, with its complete global coverage, using ERA5-Land means we can construct shock measures for any location in the world, and thus for every single firm in the ES data. By contrast, 1 For an overview of ERA5, see (Hersbach et al., 2020). 7 rainfall datasets produced by long-running Earth-observing satellites are often lower resolution, have spatial or temporal gaps in data coverage, and exhibit variation in the fidelity and methodology of measurements over time. In our empirical analysis, our preferred measure of precipitation shocks is a variable we call `dry days’, which varies at the annual and secondary sub-national administrative unit (ADM2) level. To construct this variable, we start with the daily total precipitation values observed in all ERA5 grid cells that are contained within the ADM2 units where we observe at least one firm in the ES data. Then, to focus on contemporary climate and match the temporal coverage of the ES data, we restrict the daily data to the period from 1990 to 2021. We then compute the long-run (1990-2021) mean and standard deviation for each day of the year in each grid cell. Using these moments of the local climate distribution, we then classify whether the precipitation observed on each day is more than 1 standard deviation below the long-run average for that particular day of the year. Days that satisfy this condition are considered `dry’ days. Finally, we sum across days within grid-years and average across ERA5 grid cells within ADM2 units to arrive at the final measure of dry days that we use in our primary empirical analysis. In many of our specifications, we also include an analogous measure of `wet’ days, which captures the number of days in a year where rainfall was more than 1 standard deviation above the long-run day-of-year average. Previous studies have often measured rainfall shocks in terms of (normalized) deviations of total cumulative annual or seasonal rainfall from location-specific long-run averages. Standardized drought indices, such as the SPEI or PDSI, have also been widely used. Relative to these standards, our dry (wet) days measure has two key advantages. First, our measure captures extreme dryness (wetness) events even when they occur over a period of just a few days (or even just a single day). This helps us identify nonlinear effects which 8 can be diluted when weather outcomes are more coarsely averaged over time.2 Second, by normalizing daily rainfall values relative to a location- and day-of-year specific climate distribution, our measure effectively summarizes deviations in the timing of rainfall throughout the year. For example, a year where precipitation is exactly equal to the long-run daily average on every day of the year would have zero dry days and zero wet days. But a year with the same total annual precipitation, but where the timing of rainfall throughout the year is significantly shifted, would have many dry and wet days. As a result, when we use dry (wet) days as our measure of precipitation shocks, we can capture effects on firms that result from disruptions to the normal timing of rainfall throughout the year. 2.2 Empirical Approach The empirical approach exploits cross-sectional variation in the ES data combined with precipitation shocks at the ADM2 geographical unit, while accounting for sub-national ADM1 fixed effects. The main analysis estimates the effect of precipitation shocks at the ADM2 level on sales at the firm level while controlling for temperature and other firm-level controls. The following regression is estimated: = 1 ℎ + + + + + (1) Where: i indexes firms, a indexes sub-national ADM1 units, g indexes ADM2 units, s indexes sector (2 digit ISIC level) and t indexes survey-years. Note that ADM2 is more geographically disaggregated unit than ADM1. is the log of sales (in USD). ℎ represents measures of precipitation shocks. The main precipitation shock is number of dry days for a specific ADM2 unit. Average temperature is also accounted for in the estimations. 2 In this way, our dry (wet) days measure is similar in to the `degree-days’ temperature variables used in Schlenker and Roberts’ well-known analysis of U.S. maize yields (Schlenker & Roberts, 2009). 9 We employ a small set of firm level control (X) in the estimation - whether the firm is a multi-firm, age of firm (in logs), size of firm (in logs), exporter status, foreign ownership, and whether the firm has a checking or savings account. Sector (ISIC 2 digit level) and survey year fixed effects are also included in the specification. The main concern of the estimations are omitted variable bias and selection. The key identifying assumption in this analysis is that the experience of extreme dry days in a given year is quasi-random within a given ADM2 unit. A body of empirical climate change literature has exploited random variations in weather to estimate a reduced-form production function-style equation (Dell et al., 2012; Felbermayr et al., 2022; Kotz et al., 2022). Simultaneity bias is less of a concern given that individual firms are unlikely to influence precipitation shocks. To address omitted variable bias concerns, we control for a variety of controls at the firm-level. We also account for time-invariant omitted variables at the ADM1 geographical level through ADM1 fixed effects. Since the number of dry day shocks are exogenous, simultaneity bias is less of a concern. However, we cannot rule out the possibility that firm location selection is endogenous to precipitation shocks. If productive firms move to areas with fewer shocks, then it may appear as though shocks are negatively correlated with firm performance although it is driven to some extent by selection. We try to account for this concern by including controls for determinants of firm productivity. Also, given that our sample only consists of surviving firms, our estimates may understate the true effect if firms are driven out of business due to these negative shocks. We explore potential channels through which precipitation shocks may affect sales. We achieve this by regressing various variables that have been identified in the literature as plausible channels of climatic shocks and other right-hand side variables defined in equation (1). Finally, a subsample of firms in the Middle East and North Africa, and Europe and Central Asia surveyed around the year 2019 were asked specific questions on green investments and adaptation. We exploit this data to evaluate if firms who experience precipitation shocks are more likely to adopt various green measures. 10 3. Results 3.1 Main Results Table 2 provides the main results. An increase in dry days leads to a reduction in sales. The coefficient is statistically significant at the 5% level. In terms of magnitude, an additional extreme dry day leads to a 0.6 percent reduction in sales. The average no. of extreme dry days in the sample is 6.7 days such that an increase in extreme dry days of this amount translates to a 3.8 percent reduction in sales. A one standard deviation increase in extreme dry days ( 12.9 days) results in a 7.3 percent reduction in sales. Note that the average number of extreme dry days includes values of zeroes for areas that did not experience any extreme dry day. At the very extreme, the sample maximum for extreme dry days is 86 days, about 3 months. An increase of extreme dry days of around 3 months results in a 48.6 percent loss in sales. Given that we also find that extreme dry days lead firms to exit, these estimates reflect those of surviving firms and may, thus, underestimate the overall impact. With regards to other covariates in the estimation, wet days are positively correlated with sales, statistically significant at the 1 percent level. This is consistent with Zaveri et al. (2023), who find that positive precipitation shocks can be a boon for the economy. The estimates for firm-level covariates are as expected: the size of the firm, age of the firm, foreign ownership, exporting firms and those that have access to checking accounts have more sales. All coefficients are statistically significant at the 1 percent level of significance. Next, we explore heterogeneities in terms of level of development, region, firm size, and sector. These are reported in table 2. The results show that firms in developing economies are more vulnerable to extreme dry days than high-income economies. The coefficient of extreme dry days is negative and statistically significant for developing economies while for high-income economies the coefficient is statistically 11 insignificant and almost half the size as that of developing economies in absolute terms. Firms in Latin America and the Caribbean are much more vulnerable to extreme dry days than other regions. It is the only region for which the coefficient of extreme dry days is negative and statistically significant, as well as the largest among all the regions in terms of magnitude. The coefficient of extreme dry days is negative for East Asia and the Pacific, Sub-Saharan Africa, and the Middle East and North Africa albeit statistically insignificant. Smaller firms, and firms in service sectors are also more vulnerable to extreme dry days than large firms and manufacturing firms. 3.2 Channels We explore a number of channels highlighted in the literature through which negative precipitation shocks may affect firms. These largely include labor productivity, investment, infrastructure service interruptions (power and water outages), and access to finance. We also consider an additional channel not mentioned in the literature – corruption. The main findings are presented in table 3. Extreme dry days lead to lower labor productivity, defined as sales per worker. This is consistent with Somanathan et al., (2021) that finds hot days lead to absenteeism and therefor lower labor productivity. While we do not see any effect of extreme hot days on full-time employment in the sample, the surveys do not capture absenteeism and health of the workers, and therefore we cannot rule out the possibility of these channels. A prominent channel of the effects of extreme dry days is infrastructure. Extreme dry days lead to higher incidents of water and power outages, as well as larger losses due to electrical outages. This is consistent with the literature that have found that negative rainfall shocks lead to water outages (Hyland and Islam, 2019) and power outages (Desbureaux and Rodella, 2019). Note that extreme dry days do not only lead to power interruptions if hydropower is the source of energy. Water is needed for a variety of energy sources 12 including nuclear energy. Furthermore, extreme dry days may increase the demand for energy, stressing the power infrastructure, and therefore leading to more power outages. Extreme weather events can hurt the ability of firms to access finance. In table 3 we see that extreme dry days are negatively correlated with various access to finance indicators - bank financing of working capital, loan or line of credit, and existence of overdraft facilities. The coefficient of extreme dry days is only statistically significant for bank financing of working capital, and the presence of overdraft facilities. This is consistent with the negative relationship uncovered in the literature between extreme weather events and access to finance. Frequent climate shocks might affect the ability of banks to predict outcomes, and therefore increasing interest rates due to additional risk that results in increasing the cost of capital (Kling et al., 2021; Javadi and Masum, 2021; Brown et al., 2021). Extreme dry days could also lead to balance sheet erosion as firms that experience monetary losses from shocks become more leveraged as they are more likely to get their loan applications rejected and seen as less creditworthy (Benincasa et al., 2024). Weather shocks can also create liquidity shortages, increase loan defaults, deteriorating credit scores and thus access to future credit (Aguilar-Gomez et al., 2023). Similarly, Huang et al., (2018) find that extreme weather events are associated with lower and more volatile earnings and cash flows, resulting in exposed firms holding more cash to generate more financial slack and build resilience. We also explore whether extreme weather days have any effects on investments by firms. The effect of extreme dry days on investment is ambiguous. If weather shocks limit access to finance, firms are less likely to invest. However, firms affected by weather shocks may also be more likely to invest in fixed assets as they replenish damaged capital (Benincasa et al., 2024). In table 3 we find that extreme dry days have no statistically significant relationship with whether a firm invests. However, when we break down investment in physical capital into (a) machine and equipment, and (b) land and buildings, we find different results. Extreme dry days have a positive and statistically significant effect on the likelihood that a firm will invest in land and buildings, consistent with Benincasa et al., (2024). However, there is no statistically significant 13 relationship with whether a firm invests in machinery or equipment. This is a surprising finding. There are some potential explanations. One could be that extreme dry days have adverse effects on concrete, and thus firms may spend on renovating buildings (Wang et al., 2012). This is under the assumption that firms own the buildings they operate in. Another possibility is that supply chain disruptions due to extreme dry days may lead firms to increase storage due to increasing uncertainty. It may also be that exposure to extreme dry days leads firms to become more environmentally aware, and to invest in greener buildings. Unfortunately, the data does not provide information on the purpose of investment – whether for storage or renovations, or for more climate resilient buildings. There is also a possibility that since extreme dry days reduce access to finance, firms that would like to invest in greener fixed assets may prioritize buildings over machinery and equipment. One surprising channel uncovered in the analysis is that firms are more likely to face bribes to get things done in locations exposed to extreme dry days. This may be explained by weakening economic conditions leading to more rent seeking behavior. Alternatively, it may mean that as a resource becomes scarce (in this case water), there may be greater incentives to solicit bribes. 3.3 Firm Exit For countries with multiple waves of surveys, the status of firms surveyed in the initial wave is tracked in the succeeding wave. Therefore, we are able to tell if the firm is still in business or have exited the market. For about 23,000 firms that are tracked, there is an exit rate of about 26 percent. The findings in table 4 show that firms that are in locations that experience extreme dry days are more likely to exit the market. This implies that our estimates are largely based on surviving firms and therefore may be on the conservative side. The results in table 4 also show that there are no differential impacts of extreme dry days on firm exit by productivity of firm size. However, manufacturing firms in areas experiencing extreme dry 14 days are less likely to exit while firms with a female top manager are more likely to exit. This provides some indication that women entrepreneurs may be vulnerable to extreme weather shocks. 3.4 Innovation, Digital Connectivity, and Exporters In table 5 we explore whether innovation, digital connectivity and exporting status have any influence in the relationship between extreme dry days and sales. We find that firms that have innovated through improved processes, or have technology licensed from foreign firms tend to debilitate the effects of extreme dry days on sales. This may be explained by more innovative firms being more able to adapt to climate shocks, and therefore are more resilient. Sales of firms that are more likely to own a website are less affected by extreme dry days than firms that do not own a website. While the data does not allow us to explicitly explain this relationship, one could hypothesize that digital connectivity allows firms to access more information that could help them adapt to extreme dry days. Finally, exporters are found to be more resilient to shocks than non-exporters. This suggests that demand from abroad is not affected by localized extreme weather events, and that as long as firms adapt, they may be more likely to survive. This is in contrast to findings in the literature that suggest that exporters are also more likely to be affected than non-exporters (Huppertz, 2023). 3.5 Green Firms The 2019 Enterprise Surveys wave for two regions - Eastern and Central Europe and Middle East and North Africa – has a special green module that captures whether firms adopt climate friendly measure (see table A1 for summary statistics). There is some evidence that firms exposed to extreme weather events are more likely to take steps to adapt by adopting climate-friendly measures (Benincasa et al., 2024). The destruction of physical capital due to extreme weather may result in new investments that lead to vintage effects where replenishment of capital means newer equipment with lower environmental footprint. Alternatively, firms 15 may become environmentally aware and therefore engage in green investments or green management practices. Benincasa (2014) find that firms suffering losses from extreme weather display 12 percentage points greater likelihood of adopting climate-friendly measures. Firms exposed to extreme weather events are more likely to monitor their CO2 emissions. However, low-quality management practices and credit constraints can limit green investments (De Haas et al., 2021). The findings in table 6 show no statistically significant effect between extreme dry days and any of the green measures uncovered in the surveys. However, consistent with the main results of the overall sample, there is a negative effect of extreme dry days on sales and labor productivity – the magnitude is considerably larger than the main sample. Do note that one limitation of the green measures is that they ask whether firms have adopted new measures over the last three years. Thus, firms that may have adopted 4 years ago will be counted as though they have not adopted any measures. 3.6 Robustness – Sensitivity to Country Dominance A concern is that a singly country in the sample may dominate the results. Figure 1 presents the coefficient estimates for extreme dry days as each country is dropped from the sample. Specifically, one country is dropped, the coefficient of extreme dry days is estimated. The country is then returned to the sample as another country is dropped. This process continues until at least every country has been dropped once. Since there are 118 countries in the sample, there are 118 coefficients reported in figure 1, including the 90 percent confidence intervals. As can be seen, the findings are robust to country dominance at the 90 percent level of significance. 4. Conclusion This study uncovered a negative effect of extreme dry days on sales for a global sample of over 88,000 firms across 118 countries. The results reveal that the main channels are through a fall in labor productivity, increase in infrastructure service delivery interruptions, limited access to finance, and a greater exposure to 16 bribery. Firms that are in locations facing extreme dry days are also more likely to exit. At the same time, firms that are innovative, digitally connected, and use foreign technology are more resilient to extreme dry days. However, evidence using a subset of firms for which data on green adoption exists, suggests that firms do not seem to respond to extreme dry days by being more environmentally aware or adopting efficient practices in terms of greening the economy. Several policy recommendations emerge from the study. Climate shocks are an economy-wide concern, especially the intensity of negative precipitation shocks. Thus, labor that reallocates from rural to urban areas due to disruptions in the agriculture sector is still likely to face challenges in urban settings. Furthermore, exposure to shocks does not mean that firms adapt through green investments. Additional interventions may be needed for firms to internalize climate shocks. Finally, innovation and digital connectivity may be key to building resilience against shocks. Access to finance may especially be important, particularly for vulnerable small firms that face credit market failures. The study has several limitations that future research may address. It is unable to unpack the labor related effects such as absenteeism and health. In addition, the universe of firms covered in this study are private formal firms with 5 or more employees. Therefore, the findings may be different for smaller informal firms on one hand, and large state-owned firms on the other end. The study is also unable to identify whether a reduction in sales is due to declining demand, or due to the inability of firms to supply goods. These are important avenues for future research to pursue. 17 References Aguilar-Gomez, Sandra, Emilio Gutierrez, David Heres, David Jaume, Martin Tobal (2023) “Thermal stress and financial distress: Extreme temperatures and firms’ loan defaults in Mexico.” Journal of Development Economics (forthcoming) Benincasa, Emanuela, Frank Betz, and Luca Gattini (2024) “How do Firms Cope with Losses from Extreme Weather Events?” Journal of Corporate Finance 84 102508 Besley, Timothy and Hannes Mueller (2018) “Predation, Protection, and Productivity: A Firm-Level Perspective.” American Economic Journal: Macroeconomics 10(2): 184-221. Brown, James R., Matthew T. Gustafson, and Ivan T. Ivanov (2021) “Weathering Cash Flow shocks.” The Journal of Finance 76 (4): 1731-1772 Chauvet, Lisa and Helene Ehrhart (2018) “Aid and Growth: Evidence from Firm-Level Data.” Journal of Development Economics 135: 461-477. Damania, Richard, Sebastien Desbureaux, Marie Hyland, Asif Islam, Aude-Sophie Rodella, Jason Russ, and Esha Zaveri (2017), Uncharted waters: The New Economics of Water Scarcity and Variability. World Bank Publications. De Haas, Ralph, Ralf Martin, Mirabelle Muûls, and Helena Schweiger (2021). “Managerial and Financial Barriers to the Net-Zero Transition.” CEPR Discussion Paper No. DP15886. Dell, Melissa, Benjamin F. Jones, and Benjamin A. Olken (2012) “Temperature shocks and economic growth: Evidence from the last half century.” American Economic Journal: Macroeconomics 4(3): 66- 95. Desbureaux, Sebastien and Aude-Sophie Rodella (2019). “Drought in the City: The Economic Impact of Water Scarcity in Latin American Metropolitan Areas.” World Development 114:13-27 Felbermayr, Gabriel, Jasmin Gröschl, Mark Sanders, Vincent Schippers, Thomas Steinwachs (2022) “The Economic Impact of Weather Anomalies.” World Development 1515: 105745 Hersbach, Hans, Bill Bell, Paul Berrisford, Shoji Hirahara, Andras Horányi, Joaquin Muñoz-Sabater, Julien Nicolas, Carole Peubey, Raluca Radu, Dinand Schepers, Adrian Simmons, Cornel Soci, Saleh Abdalla, Xavier Abellan, Gianpaolo Balsamo, Peter Bechtold, Gionata Biavati, Jean Bidlot, Massimo Bonavita, Giovanna De Chiara, Per Dahlgren, Dick Dee, Michail Diamantakis, Rossana Dragani, Johannes Flemming, Richard Forbes, Manuel Fuentes, Alan Geer, Leo Haimberger, Sean Healy, Robin J. Hogan, Elías Hólm, Marta Janisková, Sarah Keeley, Patrick Laloyaux, Philippe Lopez, Cristina Lupu, Gabor Radnoti, Patricia de Rosnay, Iryna Rozum, Freja Vamborg, Sebastien Villaume, Jean- Noël Thépaut (2020). The ERA5 global reanalysis. Quarterly Journal of the Royal Meteorological Society, 146(730), 1999–2049. https://doi.org/10.1002/qj.3803 Hjort, Jonas, and Jonas Poulsen (2019) “The Arrival of Fast Internet and Employment in Africa.” American Economic Review, 109 (3): 1032-79. 18 Huppertz, Maximilian (2024) “Sacking the Sales Staff: Firm Reactions to Extreme Weather and Implications for Policy Design.” mimeo https://maxhuppertz.github.io/files/max_huppertz_jmp.pdf Hyland, Marie and Jason Russ (2019), “Water as destiny–the long-term impacts of drought in sub- saharan africa.” World Development, 115, 30–45. Islam, Asif and Marie Hyland (2019), “The drivers and impacts of water infrastructure reliability–a global analysis of manufacturing firms.” Ecological Economics, 163, 143–157. Javadi, Siamak and Abdullah-Al Masum (2021). “The Impact of Climate Change on the Cost of Bank Loans.” Journal of Corporate Finance 69 102019 Kling, Gerhard, Ulrich Volz, Victor Murinde, and Sibel Ayas (2021). “The Impact of Climate Vulnerability on Firms' Cost of Capital and Access to Finance.” World Development 137 105131 Kotz, Maximilian, Anders Levermann, and Leonie Wenz (2022) "The Effect of Rainfall Changes on Economic Production." Nature 601:223–227. https://doi.org/10.1038/s41586-021-04283-8 Liu, Xueqing, Kunpeng Zhang, and Yayun Ren (2023) "Does Climate Warming Affect Labour Productivity in Emerging Economies? - Evidence from Chinese-listed Firms." Applied Economics 55(24): 2801-2814. Muñoz-Sabater, Joaquin, Emanuel Dutra, Anna Agustí-Panareda, Clement Albergel, Gabriele Arduini, Gianpaolo Balsamo, Souhail Boussetta, Margarita Choulga, Shaun Harrigan, Hans Hersbach, Brecht Martens, Diego G. Miralles, Maria Piles, Nemesio J. Rodríguez-Fernández, Ervin Zsoter, Carlo Buontempo, Jean-Noel Thépaut (2021). ERA5-Land: A state-of-the-art global reanalysis dataset for land applications. Earth System Science Data, 13(9), 4349–4383. https://doi.org/10.5194/essd-13-4349- 2021 Russ, Jason (2020), “Water runoff and economic activity: The Impact of Water Supply Shocks on Growth.” Journal of Environmental Economics and Management, 101, 102322 Schlenker, Wolfram and Michael J. Roberts (2009). “Nonlinear temperature effects indicate severe damages to U.S. crop yields under climate change.” Proceedings of the National Academy of Sciences, 106(37), 15594–15598. https://doi.org/10.1073/pnas.0906865106 Singh, Chandni, Garima Jain, Vibhas Sukhwani, and Rajib Shaw (2021), “Losses and Damages Associated with Slow-onset Events: Urban Drought and Water Insecurity in Asia.” Current Opinion in Environmental Sustainability, 50, 72–86. Somanathan, E., Rohini Somanathan, Anant Sudarshan, and Meenu Tewari (2021). “The Impact of Temperature on Productivity and Labor Supply: Evidence from Indian Manufacturing.” Journal of Political Economy 129(6): 1797-1827. The United Nations Convention to Combat Desertification (UNCCD) (2022), Drought in Numbers. UNCCD Publication. Wang, Xiaoming, Mark G. Stewart, and Minh Nguyen (2012). “Impact of Climate Change on Corrosion and Damage to Concrete Infrastructure in Australia.” Climatic Change 110, 941–957 (2012). https://doi.org/10.1007/s10584-011-0124-7 19 Zaveri, Esha, Jason Russ, Amjad Khan, Richard Damania, Edoardo Borgomeo, and Anders Jagerskog (2021), Ebb and flow: Volume 1. Water, Migration, and Development. World Bank Publications. Zaveri, Esha, Richard Damania, and Nathan Engle (2023), “Droughts and Deficits: The Global Impact of Drought on Economic Growth.” Policy Research Working Paper 10453. Zhao, Le and A. M. Parhizgari (2024) “Climate Change, Technological Innovation, and Firm Performance.” International Review of Economics and Finance 93: 189-203 20 Table 1: Summary Statistics Variable Obs Mean Std. dev. Min Max Log of Sales (USD) 88,060 12.771 2.093 4.489 22.732 Log of Sales per Worker (USD) 87,541 10.066 1.699 0.797 18.978 Extreme Dry Days (Admin 2) 88,060 6.677 12.931 0 86 Extreme Water Days (Admin 2) 88,060 43.444 9.459 12 71 Average Temperature 88,060 18.357 7.141 -8.759 31.847 Average Temperature Squared 88,060 387.981 260.081 0.000003 1014.204 Firm is part of a larger firm Y/N 88,060 0.167 0.373 0 1 Log of age of firm 88,060 2.596 0.773 0 5.394 Log of size 88,060 2.775 1.097 -1.099 9.741 Direct exports 10% or more of sales Y/N 88,060 0.126 0.331 0 1 Foreign ownership Y/N 88,060 0.097 0.295 0 1 Establishment has checking or savings account Y/N 88,060 0.878 0.327 0 1 Firm purchased fixed assets Y/N 87,554 0.423 0.494 0 1 Firm Purchased Machinery and Equipment Y/N 85,120 0.392 0.488 0 1 Firm Purchased Land and Buildings Y/N 82,107 0.096 0.295 0 1 Website Y/N 87,904 0.512 0.500 0 1 Average No of Incidents of Water Shortages per Month (0 if no shortage, manf firms only) 44,887 0.891 6.414 0 365 Average No. of Power Outages per Month (0 if no shortage) 85,178 5.313 22.939 0 14000 Losses due to Electrical Outages (% of Annual Sales) 87,666 2.202 7.196 0 100 Establishment has a Line of Credit or Loan Y/N 86,690 0.354 0.478 0 1 Firm Use Banks to Finance Working Capital Y/N 86,077 0.299 0.458 0 1 Proportion of Working Capital Financed by Banks (%) 86,345 11.333 22.367 0 100 Establishment has Overdraft Facility Y/N 85,348 0.397 0.489 0 1 Firm Exit Y/N (conservative) 23,257 0.096 0.295 0 1 Firm Exit Y/N (extended) 23,257 0.259 0.438 0 1 Senior Management Time Spent in Dealing with Requirements of Government Regulations (%) 80,221 9.573 15.492 0 100 Firm Expected to Make a Payment to Get Things Done Y/N 80,282 0.173 0.378 0 1 Female Top Manager Y/N 87,910 0.177 0.382 0 1 Firm has a Female Owner Y/N 87,142 0.339 0.473 0 1 Manufacturing Sector Y/N 88,042 0.295 0.456 0 1 ISO Certification Ownership Y/N 86,093 0.164 0.370 0 1 Tech licensed from foreign firms Y/N 48,961 0.142 0.349 0 1 New/Significantly Improved Process Introduced over last 3 years 82,786 0.290 0.454 0 1 New Products/Services Introduced Over Last 3 years 83,978 0.348 0.476 0 1 Management Practices Quality 14,947 0.506 0.211 0 1 Experienced at least one bribe payment Y/N 58,019 0.178 0.382 0 1 Small and Medium Enterprise Y/N 88,060 0.925 0.264 0 1 21 Table 2: Precipitation Shocks and Sales Model OLS with Admin 1 Fixed Effects Dependent Variable Log of Sales (USD) Developi Europe Latin Middle Small and Services High Sub- Full ng East Asia and America East and South Medium Large Manufactur and Sample Income Saharan Sample Economi and Pacific Central and North Asia Size Enterprises ing Other Economies Africa es Asia Caribbean Africa Enterprises Sectors coef/se coef/se coef/se coef/se coef/se coef/se coef/se coef/se coef/se coef/se coef/se coef/se coef/se Extreme Dry Days -0.006** -0.007** 0.004 -0.004 0.013 -0.012*** -0.002 0.007 -0.005 -0.007** 0.003 -0.003 -0.006* (Admin 2) (0.003) (0.003) (0.010) (0.009) (0.008) (0.004) (0.015) (0.011) (0.006) (0.003) (0.005) (0.003) (0.003) Extreme Water Days 0.008*** 0.009*** 0.006 -0.002 0.008*** 0.007 0.007 -0.010 0.003 0.009*** -0.006 0.003 0.011*** (Admin 2) (0.002) (0.003) (0.004) (0.007) (0.003) (0.005) (0.006) (0.007) (0.007) (0.002) (0.005) (0.003) (0.003) Average Temperature 0.080** 0.112** 0.072 0.096 0.028 0.145 0.105 -0.377 -0.146 0.082** 0.094 0.059 0.097** (0.037) (0.044) (0.076) (0.072) (0.062) (0.233) (0.365) (0.286) (0.250) (0.039) (0.103) (0.048) (0.049) Average Temperature -0.002 -0.002* -0.003 -0.002 0.001 -0.003 -0.002 0.009 0.003 -0.002 -0.004 -0.002 -0.002 Squared (0.001) (0.001) (0.003) (0.002) (0.003) (0.006) (0.009) (0.006) (0.006) (0.001) (0.004) (0.001) (0.002) 12.298** Constant 8.070*** 7.306*** 10.106*** 8.259*** 9.290*** 7.397*** 7.738** 9.763*** 7.897*** 9.932*** 8.635*** 7.825*** * (0.315) (0.407) (0.557) (1.245) (0.414) (2.401) (3.425) (3.458) (2.408) (0.328) (0.839) (0.438) (0.406) Controls YES YES YES YES YES YES YES YES YES YES YES YES YES Admin 1 Fixed Effects YES YES YES YES YES YES YES YES YES YES YES YES YES Sector (2 DIGIT ISIC) YES YES YES YES YES YES YES YES YES YES YES YES YES Fixed Effects Year Fixed Effects YES YES YES YES YES YES YES YES YES YES YES YES YES Number of observations 88,060 73,909 14,133 9,425 31,055 12,485 11,040 10,954 13,093 70,537 17,365 50,138 37,769 Adjusted R2 0.664 0.625 0.672 0.665 0.647 0.685 0.671 0.676 0.577 0.618 0.549 0.736 0.639 note: *** p<0.01, ** p<0.05, * p<0.1. Robust Standard Errors Clustered at the Size-Sector-Location strata. Main control variables include Extreme Water Days (Admin 2), Average Temperature, Firm is part of a larger firm Y/N, Log of age of firm, Log of size, Direct exports 10% or more of sales Y/N, Foreign ownership Y/N, Establishment has checking or savings account Y/N 22 Table 3: Channels Model OLS with Admin 1 Fixed Effects Average No of Losses Firm Firm Firm Incidents of Average No. Firm Firm due to Use Establishm Expected Purchased Log of Sales Water of Power Purchased Purchased Electrical Banks to ent has a Establishment to Make a Machinery Dependent Variable per Worker Shortages per Outages per Fixed Land and Outages Finance Line of has Overdraft Payment and (USD) Month (0 if no Month (0 if no Assets Buildings (% of Working Credit or Facility Y/N to Get Equipment shortage, manf shortage) Y/N Y/N Annual Capital Loan Y/N Things Y/N firms only) Sales) Y/N Done Y/N coef/se coef/se coef/se coef/se coef/se coef/se coef/se coef/se coef/se coef/se coef/se Extreme Dry Days -0.005** 0.034*** 0.076** 0.0003 -0.0004 0.003*** 0.101*** -0.002* -0.001 -0.003*** 0.002*** (Admin 2) (0.003) (0.013) (0.030) (0.001) (0.001) (0.000) (0.016) (0.001) (0.001) (0.001) (0.001) Extreme Water Days 0.008*** 0.021* -0.006 -0.0004 0.0004 -0.001*** -0.007 0.001 0.001 0.002** 0.000 (Admin 2) (0.002) (0.011) (0.029) (0.001) (0.001) (0.000) (0.010) (0.001) (0.001) (0.001) (0.001) Average Temperature 0.072* -0.218 -0.177 0.003 0.007 0.001 -0.138 0.005 -0.014 -0.015 0.016 (0.037) (0.158) (0.346) (0.013) (0.014) (0.008) (0.200) (0.012) (0.013) (0.013) (0.011) Average Temperature -0.002 0.012* 0.011 -0.0001 -0.0002 -0.00002 0.013 0.00001 0.0004 0.0003 -0.001 Squared (0.001) (0.007) (0.014) (0.000) (0.000) (0.000) (0.009) (0.000) (0.000) (0.000) (0.000) Constant 8.163*** -1.213 3.353 0.171 0.077 0.004 -0.401 -0.096 0.102 0.064 0.006 (0.315) (0.755) (2.414) (0.130) (0.132) (0.073) (1.055) (0.114) (0.124) (0.115) (0.096) Controls YES YES YES YES YES YES YES YES YES YES YES Admin 1 Fixed YES YES YES YES YES YES YES YES YES YES YES Effects Sector (2 DIGIT YES YES YES YES YES YES YES YES YES YES YES ISIC) Fixed Effects Year Fixed Effects YES YES YES YES YES YES YES YES YES YES YES Number of 87,541 44,808 85,171 87,553 85,114 82,104 87,666 86,074 86,688 85,347 80,274 observations Adjusted R2 0.490 0.149 0.299 0.186 0.189 0.122 0.291 0.172 0.219 0.324 0.259 note: *** p<0.01, ** p<0.05, * p<0.1. Robust Standard Errors Clustered at the Size-Sector-Location strata. Main control variables include Extreme Water Days (Admin 2), Average Temperature, Firm is part of a larger firm Y/N, Log of age of firm, Log of size, Direct exports 10% or more of sales Y/N, Foreign ownership Y/N, Establishment has checking or savings account Y/N 23 Table 4: Firm Exit Model OLS with Admin 1 Fixed Effects Dependent Variable Firm Exit Y/N coef/se coef/se coef/se coef/se coef/se Extreme Dry Days (Admin 2) 0.005** 0.005* 0.005** 0.005** 0.005** (0.003) (0.003) (0.003) (0.003) (0.003) Extreme Dry Days (Admin 2) X 0.000005 Labor Productivity (0.000) Labor Productivity (deflated, USD) -0.011 (0.020) Extreme Dry Days (Admin 2) X Log -0.0002 of Size (0.000) Extreme Dry Days (Admin 2) X -0.002* Manufacturing (0.001) Extreme Dry Days (Admin 2) X 0.003* Female Top Manager (0.002) Female top manager Y/N -0.108* (0.062) Extreme Water Days (Admin 2) -0.002 -0.002 -0.001 -0.001 -0.003 (0.003) (0.005) (0.003) (0.002) (0.002) Average Temperature -0.016 -0.014 -0.016 -0.016 -0.015 (0.021) (0.021) (0.021) (0.021) (0.021) Average Temperature Squared 0.0005 0.0004 0.0005 0.0005 0.0005 (0.001) (0.001) (0.001) (0.001) (0.001) Constant 0.592*** 0.687*** 0.568*** 0.589*** 0.606*** (0.176) (0.248) (0.180) (0.175) (0.175) Controls YES YES YES YES YES Admin 1 Fixed Effects YES YES YES YES YES Sector (2 DIGIT ISIC) Fixed Effects YES YES YES YES YES Year Fixed Effects YES YES YES YES YES Number of observations 23,212 23,070 23,212 23,212 23,178 Adjusted R2 0.133 0.133 0.133 0.133 0.133 note: *** p<0.01, ** p<0.05, * p<0.1. Robust Standard Errors Clustered at the Size-Sector-Location strata. Main control variables include Extreme Water Days (Admin 2), Average Temperature, Firm is part of a larger firm Y/N, Log of age of firm, Log of size, Direct exports 10% or more of sales Y/N, Foreign ownership Y/N, Establishment has checking or savings account Y/N 24 Table 5: Innovation, Digital Connectivity, and Exporters Model OLS with Admin 1 Fixed Effects Dependent Variable Log of Sales (USD) coef/se coef/se coef/se coef/se Extreme Dry Days (Admin 2) X Website 0.004** (0.002) Website Y/N 0.301*** (0.022) Extreme Dry Days (Admin 2) X Tech Licensed from Foreign 0.008** Firms (0.004) Tech licensed from foreign firms Y/N 0.226*** (0.036) Extreme Dry Days (Admin 2) X Process Innovation 0.005*** (0.002) New/Significantly Improved Process Introduced over last 3 0.147*** years (0.023) Extreme Dry Days (Admin 2) X Exporters 0.006** (0.003) Extreme Dry Days (Admin 2) -0.009*** -0.004 -0.003 -0.006** (0.003) (0.003) (0.003) (0.003) Extreme Water Days (Admin 2) 0.008*** 0.002 0.006*** 0.008*** (0.002) (0.002) (0.002) (0.002) Average Temperature 0.081** 0.055 0.073* 0.079** (0.037) (0.048) (0.038) (0.037) Average Temperature Squared -0.002 -0.002 -0.001 -0.002 (0.001) (0.001) (0.001) (0.001) Direct exports 10% or more of sales Y/N 0.212*** 0.250*** 0.227*** 0.212*** (0.026) (0.031) (0.026) (0.027) Constant 8.061*** 8.749*** 8.072*** 8.076*** (0.312) (0.428) (0.313) (0.315) Controls YES YES YES YES Admin 1 Fixed Effects YES YES YES YES Sector (2 DIGIT ISIC) Fixed Effects YES YES YES YES Year Fixed Effects YES YES YES YES Number of observations 87,903 48,889 82,782 88,060 Adjusted R2 0.668 0.745 0.664 0.664 note: *** p<0.01, ** p<0.05, * p<0.1. Robust Standard Errors Clustered at the Size-Sector-Location strata. Main control variables include Extreme Water Days (Admin 2), Average Temperature, Firm is part of a larger firm Y/N, Log of age of firm, Log of size, Foreign ownership Y/N, Establishment has checking or savings account Y/N 25 Table 6: Green Practices and Precipitation Shocks (MENA ECA 2019 sample) Model OLS with Admin 1 Fixed Effects More Machinery Adopt any Climate- Heating and Log of Log of Sales Water and Energy Energy Friendly Cooling Dependent Variable Sales per Worker Management Equipment Management Efficiency Energy Improveme (USD) (USD) Y/N Upgrades Y/N Measures Generation nts Y/N Y/N Y/N On site Y/N coef/se coef/se coef/se coef/se coef/se coef/se coef/se coef/se Extreme Dry Days -0.043** -0.044** -0.003 -0.004 -0.006 0.0004 0.005 0.004 (Admin 2) (0.021) (0.021) (0.004) (0.004) (0.006) (0.009) (0.009) (0.006) Extreme Water Days 0.005 0.004 0.002 0.003* -0.001 0.002 0.005** 0.0001 (Admin 2) (0.005) (0.005) (0.002) (0.001) (0.002) (0.002) (0.002) (0.002) Average Temperature 0.084 0.102 0.024 0.009 0.030 -0.024 -0.001 -0.035 (0.073) (0.074) (0.029) (0.023) (0.032) (0.029) (0.031) (0.029) Average Temperature -0.002 -0.003 -0.001 -0.001 -0.001 0.001 -0.0002 0.001 Squared (0.003) (0.003) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) Constant 8.835*** 8.877*** -0.264 -0.156 0.107 0.083 -0.113 0.303 (0.561) (0.564) (0.200) (0.174) (0.247) (0.221) (0.245) (0.224) Controls YES YES YES YES YES YES YES YES Admin 1 Fixed Effects YES YES YES YES YES YES YES YES Sector (2 DIGIT ISIC) YES YES YES YES YES YES YES YES Fixed Effects Year Fixed Effects YES YES YES YES YES YES YES YES Number of 22,310 22,208 22,509 22,234 23,194 22,900 22,860 23,469 observations Adjusted R2 0.661 0.475 0.205 0.182 0.225 0.207 0.199 0.207 note: *** p<0.01, ** p<0.05, * p<0.1. Robust Standard Errors Clustered at the Size-Sector-Location strata. Main control variables include Extreme Water Days (Admin 2), Average Temperature, Firm is part of a larger firm Y/N, Log of age of firm, Log of size, Direct exports 10% or more of sales Y/N, Foreign ownership Y/N, Establishment has checking or savings account Y/N 26 Table A1: Summary Statistics MENA ECA Sample Variable Obs Mean Std. dev. Min Max Log of Sales (USD) 22,310 13.013 1.783 5.439 20.456 Log of Sales per Worker (USD) 22,208 10.378 1.436 0.797 17.494 Extreme Dry Days (Admin 2) 22,310 0.294 1.311 0 18 Extreme Water Days (Admin 2) 22,310 44.530 7.133 15 66 Average Temperature 22,310 13.546 4.682 -1.394 26.869 Average Temperature Squared 22,310 205.405 123.655 0.000 721.934 Firm is part of a larger firm Y/N 22,310 0.090 0.286 0 1 Log of age of firm 22,310 2.681 0.718 0 5.268 Log of size 22,310 2.693 1.019 0 9.210 Direct exports 10% or more of sales Y/N 22,310 0.156 0.363 0 1 Foreign ownership Y/N 22,310 0.066 0.249 0 1 Establishment has checking or savings account Y/N 22,310 0.945 0.227 0 1 Water Management Y/N 20,494 0.163 0.369 0 1 More Climate-Friendly Energy Generation On site Y/N 20,232 0.107 0.309 0 1 Machinery and Equipment Upgrades Y/N 21,094 0.419 0.493 0 1 Energy Management Y/N 20,828 0.246 0.431 0 1 Heating and Cooling Improvements Y/N 20,770 0.342 0.474 0 1 Adopt any Energy Efficiency Measures Y/N 21,262 0.274 0.446 0 1 27 Table A2: Countries and Survey Years Economy Year N Economy Year N Economy Year N Economy Year N Afghanistan 2014 111 Eswatini 2016 85 Malta 2019 171 Slovenia 2019 386 Albania 2013 279 Ethiopia 2011 421 Mauritania 2014 79 South Africa 2020 65 Albania 2019 317 Ethiopia 2015 726 Mexico 2010 1,293 South Sudan 2014 648 Argentina 2010 912 Finland 2020 644 Moldova 2013 299 Sudan 2014 235 Gambia, Argentina 2017 799 The 2018 97 Moldova 2019 346 Suriname 2018 148 Armenia 2013 236 Georgia 2013 273 Mongolia 2013 316 Sweden 2020 445 Armenia 2020 426 Georgia 2019 470 Mongolia 2019 351 Tajikistan 2013 243 Azerbaijan 2013 232 Ghana 2013 502 Montenegro 2013 86 Tajikistan 2019 245 Azerbaijan 2019 125 Greece 2018 528 Montenegro 2019 117 Tanzania 2013 371 Bangladesh 2013 1,350 Guatemala 2010 352 Morocco 2013 179 Thailand 2016 760 Belarus 2013 281 Guatemala 2017 238 Morocco 2019 345 Timor-Leste 2015 69 Belarus 2018 561 Guinea 2016 45 Mozambique 2018 500 Togo 2016 115 Belgium 2020 565 Honduras 2010 249 Myanmar 2014 500 Tunisia 2013 485 Benin 2016 102 Honduras 2016 237 Myanmar 2016 559 Tunisia 2020 459 Bolivia 2010 39 Hungary 2013 180 Namibia 2014 268 Türkiye 2013 709 Bolivia 2017 96 Hungary 2019 772 Nepal 2013 469 Türkiye 2019 1,436 Bosnia and Herzegovina 2013 292 India 2014 8,685 Netherlands 2020 750 Uganda 2013 443 Bosnia and Herzegovina 2019 306 Indonesia 2015 1,177 Nicaragua 2010 261 Ukraine 2013 695 Bulgaria 2013 260 Iraq 2011 735 Nicaragua 2016 285 Ukraine 2019 1,095 Bulgaria 2019 617 Ireland 2020 269 Niger 2017 102 Uruguay 2010 381 Burundi 2014 143 Israel 2013 399 Nigeria 2014 1,681 Uruguay 2017 217 North Cambodia 2016 294 Italy 2019 656 Macedonia 2013 325 Uzbekistan 2013 361 North Cameroon 2016 162 Jordan 2013 518 Macedonia 2019 295 Uzbekistan 2019 1,063 Venezuela, Central African Republic 2011 131 Jordan 2019 312 Pakistan 2013 340 RB 2010 185 Chad 2018 141 Kazakhstan 2013 414 Panama 2010 12 Viet Nam 2015 862 Papua New West Bank Chile 2010 816 Kazakhstan 2019 1,098 Guinea 2015 46 and Gaza 2013 377 West Bank China 2012 2,159 Kenya 2013 600 Paraguay 2010 307 and Gaza 2019 305 Colombia 2010 887 Kenya 2018 827 Paraguay 2017 308 Yemen, Rep. 2010 267 Colombia 2017 892 Kosovo 2013 174 Peru 2010 785 Yemen, Rep. 2013 244 Costa Rica 2010 412 Kosovo 2019 162 Peru 2017 678 Zambia 2013 606 Kyrgyz Croatia 2013 282 Republic 2013 209 Philippines 2015 1,004 Zambia 2019 548 Kyrgyz Croatia 2019 368 Republic 2019 313 Poland 2013 370 Zimbabwe 2011 558 Cyprus 2019 172 Lao PDR 2016 344 Poland 2019 688 Zimbabwe 2016 565 Czechia 2013 207 Lao PDR 2018 283 Portugal 2019 787 Czechia 2019 482 Latvia 2013 243 Romania 2013 457 Djibouti 2013 106 Latvia 2019 291 Romania 2019 780 Russian Dominican Republic 2016 198 Lebanon 2013 275 Federation 2012 2,822 Russian Congo, Dem. Rep. 2013 422 Lebanon 2019 352 Federation 2019 1,111 Ecuador 2010 314 Lesotho 2016 126 Rwanda 2011 189 Ecuador 2017 333 Liberia 2017 126 Rwanda 2019 355 Egypt, Arab Rep. 2013 1,345 Lithuania 2013 203 Senegal 2014 305 Egypt, Arab Rep. 2016 1,374 Lithuania 2019 342 Serbia 2013 327 Egypt, Arab Rep. 2020 2,795 Luxembourg 2020 158 Serbia 2019 292 El Salvador 2010 281 Madagascar 2013 227 Sierra Leone 2017 115 Slovak El Salvador 2016 570 Malawi 2014 310 Republic 2013 168 28 Slovak Estonia 2013 128 Malaysia 2015 702 Republic 2019 417 Estonia 2019 177 Mali 2016 153 Slovenia 2013 230 29 Extreme Dry Days -0.016 -0.014 -0.012 -0.008 -0.006 -0.004 -0.002 0 -0.01 C1 C4 C7 C10 C13 C16 C19 C22 C25 C28 C31 C34 coef C37 C40 C43 C46 C49 C52 Figure 1: Sensitivity Country Dominance C55 CI -90 C58 C61 C64 C67 C70 Country Dropped C73 C76 C79 CI +90 C82 C85 C88 C91 C94 C97 C100 C103 C106 C109 C112 C115 C118 30