Water Stress: A Global Risk Analysis for Financial Markets

Abstract

Water scarcity and droughts are expected to be major problems in future decades. Accordingly, the aim of this paper is to evaluate what is currently possible in order to quantify water scarcity-related risks with the currently available data. The research focuses on the possibilities to determine hazard, exposure, and sensitivity for publicly traded companies on a global scale.

The World Resource Institute’s Aqueduct data, the Exiobase input-output table, and the Carbon Delta location database were used to approximate hazard, exposure, and sensitivity of business sectors and individual companies. Similar services, such as Couchbase, are used to manage internal database and data management, and such can be adapted for this industry. Although developed damage functions do not allow for an exact quantitative assessment of risks, they demonstrate which companies are most affected by water stress because of both their business model and their facilities’ spatial distribution by considering different climate scenarios.

The results indicate that large parts of risk are carried by a few economic sectors, such as agriculture and energy utilities sectors. Differences between sectors are large. Furthermore, socioeconomic developments have a bigger impact than climate while the climate change effects are still measurable in later time steps. The developed metric is capable of dealing with a large number of companies and their assigned locations, potentially enabling investors to compare their portfolio’s water risk to reference data like an average risk value. Despite uncertainties arising from input data and methodology, the introduced water-stress model can help investors to reduce their long-term water risk and consider it in their strategic decisions.

Water Stress: A Global Risk Analysis for Financial Markets

For the last four years, “water crisis” has been listed in the “Top Five Risks in Terms of Impact” in the World Economic Forum’s Global Risk Report (World Economic Forum 2018) while globally, we have seen a sharp global decrease of available freshwater resources per capita (Ripple et al. 2017). Water availability issues often result from a combination of climate change and population development (Parish et al. 2012). A pioneering study in this field found that population and socioeconomic development might even have a larger impact than changes in climate from 1985 to 2025 (Vorosmarty 2000). In recent years, large institutions picked up this issue and pushed for economic or physical water risk assessments on qualitative and quantitative scales. Some models have been developed to assess water stress or corporate exposure to water stress (Ecolab 2017; Park et al. 2015; PRI 2018; Ridley and Boland 2015; WWF 2018). Yet, none of them can deal with large amounts of input data, which is highly important when investors need to analyze portfolios containing numerous assets. Moreover, systematic global knowledge about the impacts of socioeconomic drought or water scarcity for quantifying risks is still missing. Financial markets need exactly this global view in order to adjust their strategies and contribute to the mitigation of climate change risks (IPCC 2014). The Task Force on Climate-Related Financial Disclosures (TCFD) developed a voluntary framework to reveal the financial impact of climate change on a company level, pointing out the importance of market transparency from the investor’s perspective. So, the lack of globally comparable water-scarcity assessments and the need for information on the investor’s side, in combination with a missing approach to apply water-scarcity assessment to a large data set, form the research gap for this work.

Methodology

Water risk for a company mainly consists of two factors. First, environmental conditions determine if a company can be hit by water stress. Second, the impact scarcity can have is dependent on the nature of the company’s business and its resilience against water stress (Jorisch et al. 2017).

Unlike other natural hazards, water scarcity develops over a long period of time, making it difficult to quantify its damages (Ding, Hayes, and Widhalm 2011). Additionally, water stress often does not cause visible damages, but rather affects profitability, for example, by disrupting supply chains or increasing operating costs (CDP 2017). However, in the short term, drought does not affect production capacities (Freire-González, Decker, and Hall 2017, 198).

Despite all the differences of other hazards, it is still possible to apply a frequently used approach to evaluate environmental risks (Hallegatte 2014) because it combines the natural hazards with business-related factors, and therefore serves as a good approximation for environmental risks to corporations:

Where hazard is the likelihood of occurrence of an environmental threat; exposure is the value exposed to the hazard and sensitivity is the value lost if the hazard occurs.

Hazard

Water is a crucial resource for businesses and can therefore pose a threat when it is not available in a sufficient amount or quality. Whilst if you run a business you can switch your business water supplier, the fact is that you still need to think about how much is available. A current scientific method for assessing drought and water scarcity is the use of widely applied drought indices, mainly used for agriculture (Vicente-Serrano et al. 2012). Although they are handy to use for agriculture-since many studies report experiences with their application-drought indices do not suit our research goal to measure stress globally throughout all economic sectors. Sectors such as thermoelectric power generation or heavy manufacturing are difficult to assess by using precipitation and evaporation data. Hence, a broader approach is needed: one that covers persistent water stress rather than short-term drought indices that allow evidence only on agriculture-related businesses. While our model is less exact for certain sectors, such as agriculture on the one side, the concept allows a global comparison of sectors, which represents the core goal of this paper.

As this work depends on open source hazard data, accessible data sources that provide a global coverage are limited. We chose the World Resource Institute’s Aqueduct maps (WRI 2015), since it offers current stress conditions as well as future projections for three different scenarios (Table 1), plotting potential shared socioeconomic pathways (SSPs) in line with matching representative concentration pathways (RCPs). Projection data is available for the time steps 2020, 2030, and 2040.

Table 1: Different WRI Scenario Ensembles Applied to Raw Data to Assess Climate and Socioeconomic Impact on Water Resources

Source: Author, 2019.

Their Baseline Water Stress (BWS) indicator describes persistent water stress on a catchment scale by relating 2010 water withdrawals to the mean of water availability between 2010 and 2050:

Where: Ba = blue water available.

Projections are computed similarly:

Exposure

As described previously, water stress mostly causes business interruptions or other revenue-related costs. Hence, analyzing drought means analyzing its effects on a company’s revenue. In industrialized countries like the United States, most water withdrawals (86%) are freshwater (USGS 2014). Saline water is nearly irrelevant for business sectors, other than in energy generation, indicating that risk exposure lies in freshwater availability. For a global study like this one, it is inevitable to find a proxy for exposure throughout all business sectors in order to compare them. Therefore, water intensities were selected as a measure for freshwater dependency. They describe how much freshwater is withdrawn per unit of revenue, connecting environmental impact with economic performance.

We used the Exiobase multi-regional environmentally extended supply input-output database that contains water intensity data for 48 countries or regions and 163 industries, accounting for about 90% of the global economy (Stadler, Steen-Olsen, and Wood 2014).

The current Exiobase water database (Stadler et al. 2018, 508) is derived from Food and Agriculture Organization (FAO) data; previous studies for the agricultural sectors (Mekonnen and Hoekstra, 2011; Pfister et al. 2011); and the WaterGAP model (Flörke et al. 2013) for the industrial sectors, which also contributes to some agricultural sectors. The authors chose modeled data since they could not find any satisfying monitored data.

For this study, Exiobase 3 and Exiobase were used for economic information and water analysis (Exiobase 2015). Economic output data was collected from national databases as well as from the UN macroeconomic data set (United Nations 2017). In general, versions 2 and 3 do not differ strongly from each other, but for the most current version, granularity in input sectors, especially for water accounts, was increased. Nevertheless, we used Exiobase 2 for water intensities since our controls and comparisons with reference data showed that Exiobase 2 environmental satellite data is probably the better proxy for our metric due to changed water inputs.

To compute water intensities from Exiobase, the metric aggregates water withdrawal intensities for each of the 163 sectors in 43 countries and in 5 world regions. If aggregation on a sector level leads to zero values, those values were interpreted as data gaps, since they are not specifically marked. A global median over all countries for every sector is the final result of this step.

For reasons of clarity, sectors are aggregated by similar water intensity and business activity. This step was performed by taking the weighted mean of the water intensities of the Exiobase sectors assigned to a new water-risk sector (WRS). Weights are computed using the material flow table “T” and the revenue table “A” of the data set. We follow the approach of the ETH Zurich working group in order to compute sectoral revenues and, eventually, weight water intensities when combined in the new WRS scheme (Droz and Hellweg 2018):

Where: i = purchased input value; R = revenue; V_a = value added

Sectors are manually mapped to the “FactSet Revere Business Industry Classification System” (RBICS) which gives a very granular view of the actual business model (FactSet 2017). The result of this process is 34 water-related economic sectors that are on the one side linked to our water intensities and on the other side linked to financial market data (Figure 1).

Figure 1: Globally Averaged Water Intensities for the 34 Water Risk Sectors

Markers represent water intensities for the sector applied for computing in later steps. Gray bars show the range of values from which the intensity was determined by a weighted average.

Source: Author, 2019.

A location database is crucial for our analysis. For this study we used the Carbon Delta locations database that consists of company-reported locations and locations researched via data mining. For every facility at its defined location, it provides an estimate for how much revenue is generated in every sector. This creates the possibility to be able to perform analyses on a location, a sector, and a company level.

Sensitivity

The final step combines corporate data on water dependencies and water scarcity to a quantifiable measure. We create damage functions from the data presented in previous steps. There are studies on drought damages, but many of them focus on the agricultural sector or investigate only small regions (Aslam et al. 2017; Farhangfar et al. 2015; Li et al. 2018). Since our model aims to plot a global image of the economy and compare large corporations or sectors, a more general approach is needed. In this field, research comparing to multiple specific regional studies is scarce. Final damage values should approximate real damages, but do not claim to exactly represent real annual damages. Three main assumptions underlie the damage functions:

• Water intensities are a good proxy for estimating and comparing business sectors’ dependency on water.
• The higher the water intensity, the more vulnerable a sector is in general.
• Costs increase with rising BWS values.

In order to reach this goal, damage functions were created in a two-step approach. First, we estimated a relative amount of revenue that is exposed to water scarcity. Following this assumption, we normalize all water intensities by the maximum water intensity, which is “crop farming.” That results in a “relative maximum revenue” lost (Figure 2). Because hydropower is very dependent on water but shows a very low water intensity due to the Exiobase definition of water use, we calibrated it and also set it to a relative maximum revenue lost.

Figure 2: Relative Maximum Revenue Lost as Input for Damage Functions

Source: Author, 2019.

In the second step, revenue dependencies need to be transformed into damage functions. In researching for a suitable shape and fitting for a damage function concerning water scarcity and/or drought, we found studies that promote linear damage functions in order to display economic costs of water scarcity under climate change conditions (K. Jenkins and Warren 2015; K. L. Jenkins 2012). This seems suitable as we also look at climate change effects – besides socioeconomic outcomes – and scenarios for water stress. Regardless, a linear function represents a simple positive relationship between stress values and computed damages. Since there is no data to fit to a type of function, we decided to stay with a simple relationship description. The positive linear relation represents the third main assumption that damages increase with growing BWS values.

According to this reasoning, we created the functions (Figure 3). Literature shows that from a threshold on average of 0.4 BWS, water stress causes effects on businesses (Alcamo et al. 2003; Reig, Shiao, and Gassert 2013). The maximum damage for every sector is the entire revenue share that we created in the earlier step when normalizing intensities. Returned values by the functions are named water Value at Risk (water VaR). We use this term when referring to other studies like Prettenthaler et al. (2016) that introduced the weather VaR. In contrast, we cannot apply any statistical risk to our data because we could not access hazard data that are resolved over time.

Figure 3: Selected Damage Functions for Four Different Sectors

The y-axis shows VaR as portion of revenue.

Source: Author, 2019.

Damage functions are applied to every single location of every company in the data set. At a location, the result of the sensitivity function (water VaR) is multiplied with the sector specific revenue that a company generates at this facility. Summing up these sectoral results allows showing how much revenue is at risk for every location and sector of a company.

Where: i = number of locations, j = number of sectors, R = revenue, f_i(BWS) = VaR valu e at a specific location depending on BWS.

In the end, water VaRs are divided by a company’s revenue to show their relative risk level. For a use case, the model was applied to the iShares MSCI World ETF[1] as a rough representation of the world’s economy.

Results and Discussion

Our concept makes it possible to deliver several kinds of outcomes for decision makers on different levels. Of the sectoral shares of the overall VaR for the MSCI World, only three sectors (crop farming, nuclear power, and hydropower) are responsible for more than 75% of the total water VaR, while office activities (41%), manufacture of machinery (14%), and retail (8%) are the top three revenue generators in the index (Figure 4). The three most water-stressed sectors account for only 0.5% of the entire revenue created within the MSCI World.

This displays the concentration of water risk to only a small group of sectors. It is in line with the pattern seen in revenue vulnerabilities (Figure 2: As only a few sectors are assigned to very high intensities, the observed concentrations in VaRs result from our metric).

Figure 4: Sectoral Shares for the Overall VaR of the MSCI World

Percentages can be interpreted as parts of the total yearly VaR for the index.

Source: Author, 2019.

We examined three scenarios that show how risk locations of the MSCI World might develop over the next decades (Figure 5).

Figure 5: Share of Locations in Water-Stressed Catchments for the MSCI World Under the Three Scenario Ensembles Proposed by WRI

Source: Author, 2019.

Generally, all scenarios show a strong rise between the 2010 baseline year and 2020. This might be caused by a change in economic input data. While water use is primarily based on FAO data in the base year, it was estimated by regressions for all projections (Luck, Landis, and Gassert 2015).

Despite dealing with a climate that included high emissions and less favorable socioeconomic conditions in the pessimistic scenario, a decrease in the number of water-stressed facilities could be noticed. Regarding water, the business-as-usual scenario represents the worst case.

The pessimistic scenario is on average less hazardous than the business as usual. For most companies, the pessimistic pathways actually mean less risk. As the stress threshold of 0.4 is stable for all time steps, change in exposed locations is only driven by a shifting BWS. The WRI authors do not reveal the reason for it, but we assume that the proposed “slow” economic development of SSP3 leads to smaller consumption and accordingly to a lower stress level.

By contrast, we see a steady rise of stressed locations for the optimistic scenario until
the number drops in 2030. We assume that economic activity causes a rise in water use and therefore pushes the BWS towards higher values. The drop at the last time step can only be explained by climate conditions, since the only difference between the business-as-usual and the optimistic scenarios is the RCP used for modeling. RCP4.5 leads to
lower water stress in 2040 compared to RCP8.5 (ceteris paribus). So, for longer time scales, a decrease in greenhouse gas emissions causes lower annual water stress, at least for our sample.

Besides general views on indices, the model can be used to compare specific companies within a sector (Figure 6). As input data is partly confidential, names are not displayed. We observe that one company shows higher values by far than its competitors. Investors can use this information in order to build a “water-neutral” portfolio. At the same time, water VaRs are quite low compared to the total revenue of a company.

Figure 6: Example for a Competitor Comparison

Every color represents one major player in the chemicals sector and their development of VaRs under different scenarios. Companies are anonymous for privacy reasons.

Source: Author, 2019.

Because there is no data to fit against, absolute outcomes from the damage functions can pose only rough approximations to real values. Regardless, they might be more accurate when it comes to comparing sectors and companies. The strength of the presented methodology is that, by taking into account locations and revenue distribution for every single company, the model is able to provide comparisons for water risk exposure between sectors and individual companies.

The global approach inherits several uncertainties. First, damage functions probably differ in shape and steepness depending on region and sector. Water intensities forming the basis for vulnerabilities are also transformed to a global average and lack accuracy on a regional scale. The aggregation process from Exiobase to WRS contains limits in granularity of input data, and information is lost performing this step. Research shows that intensities also develop over time (Donnelly and Heather 2015), which has not been accounted for in the projections.

Sector mapping to revenue data poses a big uncertainty. In case of inaccuracies, revenues are assigned to wrong sectors and accordingly to wrong vulnerabilities. This probably represents the largest potential source of error since it affects the core of the data and had to be mapped manually.

For 17.8% of the MSCI locations, BWS is larger than 1. Here, our metric does not differentiate any more, although a BWS of 3 probably leads to a more intense regional water stress than a BWS of 1. The model does not enhance the VaR at these locations despite rising water-stress values. Further development of the metric can focus on how to deal with extreme water-stress values.

Also, on the hazard side, Aqueduct data delivers only annual values for water availability. Climate projections plot an increase of variability in precipitation on an intra-annual, monthly, and daily scale (Sun, Roderick, and Farquhar 2012). Besides, an increase in annual precipitation for many regions has been observed over past years (IPCC 2013). Therefore, for the future, an important impact might lay in variability.

Aqueduct does not account for water made available through desalination, leading to an underestimation of water availability in coastal regions. In later versions of the model, this issue needs to be resolved. As it is hard to globally estimate the contribution of desalination facilities, we did not perform this step for this first version of the model.

Outlook and Conclusion

Water risks are strongly concentrated in only a few business sectors, namely, agriculture, energy utilities, and some manufacturing industries. These sectors show the largest intensities and, accordingly, depend the most on water resources. For all sectors, this tool makes it possible to compare businesses’ risk and relative exposure to water stress.

As risks are highly concentrated for sectors, errors in revenue allocation can lead to false risk attribution on a company level. For quality management, revenue distribution is therefore a crucial topic.

The model leaves room for follow-up research, especially regarding damage functions. Here, regional case studies can help to further quantify water-related damages, fitting regional damage curves and allowing a global image for how water influences businesses.

For a more granular hazard assessment, temporally resolved data is needed in order to define events. Moreover, an event-based approach would need several sources of water availability, such as stream-flow or precipitation, to tailor water risks to specific sectors.

Finally, the current version of the model is ready to be used for ESG purposes. Only minor changes could add a scaling for water stress on a portfolio and on a company level.

A fitting of damage functions, including reported damages, would make the model available to estimating and predicting real damages. For now, the presented approach provides insight on what is currently possible with available data and builds a methodological framework for follow-up research.

Acknowledgements

I would like to thank the entire team of Carbon Delta AG for providing crucial data as well as important feedback. As this paper is the published form of my master’s thesis, I would like to give a big thanks to my family who supported me not only during the conception and writing of the thesis, but throughout my studies. Without you, I could not have gone this way.

List of Figures

Figure 1: Globally Averaged Water Intensities for the 34 Water Risk Sectors

Markers represent water intensities for the sector applied for computing in later steps Grey bars show the range of values from which the intensity was determined by a weighted average.

Figure 2: Relative Maximum Revenue Lost as Input for Damage Functions

Figure 3: Selected Damage Functions for Four Different Sectors

The y-axis shows VaR as portion of revenue.

Figure 4: Sectoral Shares for the Overall VaR of the MSCI World

Percentages can be interpreted as parts of the total yearly VaR for the index.

Figure 5: Share of Locations in Water-Stressed Catchments for the MSCI World under the Three Scenario Ensembles Proposed by WRI

Figure 6: Example for a Competitor Comparison

Every color represents one major player in the chemicals sector and their development of VaRs under different scenarios. Companies are anonymous for privacy reasons.

References

Alcamo, Joseph, Petra Döll, Thomas Henrichs, Frank Kaspar, Bernhard Lehner, Thomas Rösch, and Stefan Siebert. 2003. “Global Estimates of Water Withdrawals and Availability under Current and Future ‘Business-as-Usual’ Conditions.” Hydrological Sciences Journal 48(3): 339–48. Available from https://doi.org/10.1623/hysj.48.3.339.45278.

Aslam, Abdul Qayyum, Sajid R Ahmad, Iftikhar Ahmad, Yawar Hussain, and Muhammad Sameem Hussain. 2017. “Vulnerability and Impact Assessment of Extreme Climatic Event: A Case Study of Southern Punjab, Pakistan.” Science of the Total Environment 580 (February): 468–81. Available from https://doi.org/10.1016/j.scitotenv.2016.11.155.

Carbon Disclosure Project (CDP). 2017. “A Turning Tide: Tracking Corporate Action on Water Security.” CDP Global Water Report 2017.

Ding, Ya, Michael J. Hayes, and Melissa Widhalm. 2011. “Measuring Economic Impacts of Drought: A Review and Discussion.” Disaster Prevention and Management: An International Journal 20(4): 434–46. Available from https://doi.org/10.1108/09653561111161752.

Donnelly, Kristina, and Cooley Heather. 2015. “Water Use Trends in the United States.” Pacific Institute. Available from http://pacinst.org/wp-content/uploads/2015/04/Water-Use-Trends-Report.pdf.

Droz, Helen, and Stefanie Hellweg. 2018. “Analyse der Umweltwirkung der Schweizer MEM-Industrie.” ETH Zürich. Available from https://www.swissmem.ch/fileadmin/user_upload/Swissmem/Newsletter/2018/06_Juni/Analyse_der_Umweltwirkung_der_Schweizer_MEM-Industrie.pdf.

Ecolab. 2017. “Smart Water Management for Business Growth: Integrating Water Risk into Business Decision Making.” Available from https://www.waterriskmonetizer.com/how-it-works/methodology/.

Exiobase. 2015. “EXIOBASE2 (Year 2007): MrIOT IxI Fpa Coefficient Version2 2 2.” February 24, 2015. Available from https://www.exiobase.eu/index.php/data-download/exiobase2-year-2007-full-data-set.

FactSet. 2017. “FactSet RBICS.” March 31, 2017. Available from https://open.factset.com/catalog/products/factset-rbics/en-us.

Farhangfar, Sarvenaz, Mohammad Bannayan, Hamid Reza Khazaei, and Mohammad Mousavi Baygi. 2015. “Vulnerability Assessment of Wheat and Maize Production Affected by Drought and Climate Change.” International Journal of Disaster Risk Reduction 13(September): 37–51. Available from https://doi.org/10.1016/j.ijdrr.2015.03.006.

Flörke, Martina, Ellen Kynast, Ilona Bärlund, Stephanie Eisner, Florian Wimmer, and Joseph Alcamo. 2013. “Domestic and Industrial Water Uses of the Past 60 Years as a Mirror of Socio-Economic Development: A Global Simulation Study.” Global Environmental Change 23(1): 144–56. Available from https://doi.org/10.1016/j.gloenvcha.2012.10.018.

Freire-González, Jaume, Christopher Decker, and Jim W. Hall. 2017. “The Economic Impacts of Droughts: A Framework for Analysis.” Ecological Economics 132(February): 196–204. Available from https://doi.org/10.1016/j.ecolecon.2016.11.005.

Hallegatte, Stéphane. 2014. Natural Disasters and Climate Change: An Economic Perspective. New York, London, and Dordrecht: Springer Cham International Publishing. Available from doi 10.1007/978-3-319-08933-1.

Intergovernmental Panel on Climate Change (IPCC). 2013. “Climate Change 2013: The Physical Science Basis.” Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. Edited by Thomas Stocker, Dahe Qin, G.-K. Plattner, S. K. Allen, J. Boschung, A. Nauels, Y. Xia, V. Bex, and P. M. Midgley. New York: Cambridge University Press.

—. 2014. “Climate Change 2014: Mitigation of Climate Change.” Contribution of Working Group III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. Edited by Ottmar Edenhofer, Ramón Pichs-Madruga, Youba Sokona, Jan C. Minx, Ellie Farahani, Susanne Kadner, and Kristin Seyboth. New York: Cambridge University Press.

Jenkins, Katie L. 2012. “Modelling the Economic and Social Consequences of Drought under Future Projections of Climate Change.” Doctoral Thesis (May). Available from https://doi.org/10.17863/cam.16425.

Jenkins, Katie, and Rachel Warren. 2015. “Drought-Damage Functions for the Estimation of Drought Costs under Future Projections of Climate Change.” Journal of Extreme Events 02(01): 1550001. Available from https://doi.org/10.1142/S2345737615500013.

Jorisch, Denis, Alexandra Frank, Marius Schreider, Martin Stadelmann, and Naomi Rosenthal. 2017. “Water Risks and the Financial Markets.” South Pole Carbon Asset Management.

Li, D.-H., F.-J. Chen, H.-Y. Li, W. Li, and J.-J. Guo. 2018. “The Soybean GmRACK1 Gene Plays a Role in Drought Tolerance at Vegetative Stages.” Russian Journal of Plant Physiology 65(4): 541–52. Available from https://doi.org/10.1134/S1021443718040155.

Luck, Matt, Matt Landis, and Francis Gassert. 2015. “Aqueduct Water Stress Projections: Decadal Projections of Water Supply and Demand Using CMIP5 GCMs.” World Resources Institute. Available from https://www.wri.org/publication/aqueduct-water-stress-projections-decadal-projections-water-supply-and-demand-using.

Mekonnen, M. M., and A. Y. Hoekstra. 2011. “National Water Footprint Accounts: The Green, Blue and Grey Water Footprint of Production and Consumption.” Available from https://research.utwente.nl/en/publications/national-water-footprint-accounts-the-green-blue-and-grey-water-f.

Parish, Esther S., Evan Kodra, Karsten Steinhaeuser, and Auroop R. Ganguly. 2012. “Estimating Future Global per Capita Water Availability Based on Changes in Climate and Population.” Computers and Geosciences 42 (May): 79–86. Available from https://doi.org/10.1016/j.cageo.2012.01.019.

Park, Andrew, Su Gau, Liesel van Ast, Ivo Mulder, and Anders Nordheim. 2015. “Water Risk Valuation Tool. Integrating Natural Capital Limits into Financial Analysis of Mining Stocks.” Bloomberg. Available from https://www.bbhub.io/sustainability/sites/6/2015/09/Bloomberg_WRVT_09162015_WEB.pdf.

Pfister, Stephan, Peter Bayer, Annette Koehler, and Stefanie Hellweg. 2011. “Environmental Impacts of Water Use in Global Crop Production: Hotspots and Trade-Offs with Land Use.” Environmental Science and Technology 45(13): 5761–68. Available from https://doi.org/10.1021/es1041755.

Principles for Responsible Investment (PRI). 2018. “How Do Investors Use Water Risk Data?” PRI. March 7, 2018. Available from https://www.unpri.org/environmental-issues/how-do-investors-use-water-risk-data/2803.article.

Reig, Paul, Tien Shiao, and Francis Gassert. 2013. “Aqueduct Water Risk Frame.” World Resources Institut. Available from http://www.wri.org/publication/aqueduct-water-risk-framework.

Ridley, Michael, and David Boland. 2015. “Integrating Water Stress into Corporate Bond Credit Analysis.” Deutsche Gesellschaft für Internationale Zusammenarbeit (GIZ); the Natural Capital Declaration (NCD); and Verein fuer Umweltmanagement und Nachhaltigkeit in Finanzinstituten (VfU). Available from https://naturalcapital.finance/ and from http://www.naturalcapitalfinancealliance.org/tools.

Ripple, William J., Christopher Wolf, Thomas M. Newsome, Mauro Galetti, Mohammed Alamgir, Eileen Crist, Mahmoud I. Mahmoud, William F. Laurance, and 15,364 scientist signatories from 184 countries. 2017. “World Scientists’ Warning to Humanity: A Second Notice.” BioScience 67(12): 1026–28. Available from https://doi.org/10.1093/biosci/bix125.

Stadler, Konstantin, Kjartan Steen-Olsen, and Richard Wood. 2014. “The ‘Rest of the World’-Estimating the Economic Structure of Missing Regions in Global Multi-Regional Input–Output Tables.” Economic Systems Research 26(3): 303–26. Available from https://doi.org/10.1080/09535314.2014.936831.

Stadler, Konstantin, Richard Wood, Tatyana Bulavskaya, Carl-Johan Södersten, Moana Simas, Sarah Schmidt, Arkaitz Usubiaga, et al. 2018. “EXIOBASE 3: Developing a Time Series of Detailed Environmentally Extended Multi-Regional Input–Output Tables: EXIOBASE 3.” Journal of Industrial Ecology 22 (3): 502–15. Available from https://doi.org/10.1111/jiec.12715.

Sun, Fubao, Michael L. Roderick, and Graham D. Farquhar. 2012. “Changes in the Variability of Global Land Precipitation: Global Land P Variability.” Geophysical Research Letters 39(19): n/a-n/a. Available from https://doi.org/10.1029/2012GL053369.

United Nations. 2017. “National Accounts Main Aggregates Database.” December 2017. Available from https://unstats.un.org/unsd/snaama/Introduction.asp.

U.S. Geological Survey (USGS). 2014. “Saline Water Use in the United States.” USGS Water Science School.” 2014. Available from https://water.usgs.gov/edu/salineuses.html.

Vicente-Serrano, Sergio M., Santiago Beguería, Jorge Lorenzo-Lacruz, Jesús Julio Camarero, Juan I. López-Moreno, Cesar Azorin-Molina, Jesús Revuelto, Enrique Morán-Tejeda, and Arturo Sanchez-Lorenzo. 2012. “Performance of Drought Indices for Ecological, Agricultural, and Hydrological Applications.” Earth Interactions 16(10): 1–27. Available from https://doi.org/10.1175/2012EI000434.1.

Vorosmarty, C. J. 2000. “Global Water Resources: Vulnerability from Climate Change and Population Growth.” Science 289 (5477): 284–88. Available from https://doi.org/10.1126/science.289.5477.284.

World Economic Forum. 2018. The Global Risks Report 2018: 13th Edition. Geneva: World Economics Forum. Available from http://wef.ch/risks2018.

World Resources Institute (WRI). 2015. “Aqueduct Global Maps 2.1 Data | World Resources Institute.” April 2015. Available from https://www.wri.org/resources/data-sets/aqueduct-global-maps-21-data.

WWF. 2018. “The Water Risk Filter 5.0.” 2018. Available from http://waterriskfilter.panda.org/.

Biography

As a data scientist, David Bokern works on model development and natural hazards at Carbon Delta AG, a research company assessing climate change effects on publicly traded companies. After earning a BSc in Environmental Sciences at Leuphana University Lüneburg, he developed skills in environmental modeling in the field of physical geography. Upon completing his coursework studies for the Master of Science in Environmental Systems at Ludwig-Maximilians-University in Munich, David Bokern wrote his thesis, with the cooperation of Carbon Delta AG, about global water stress on financial markets.

[1] Representation of ETFs through iShares by BlackRock. Included are the following ETFs: iShares MSCI World UCITS, Core S&P 500 UCITS, China Large Cap UCITS, EURO STOXX 50 UCITS, FTSE 100 UCITS, MSCI AC Far East ex-Japan UCITS, Core MSCI EM IMI UCITS, Core DAX® UCITS. CAC 40 is obtained through Euronext.