MSc Finance Thesis
Hedging H
2
O: A financial approach on the
impact of water scarcity.
Author:
Supervisor:
Wybren Klinker (S2958899)
J.V. Tinang Nzesseu
Abstract
Approaching expected future water scarcity from an investors point of view. Constructing a classification table for the water intensity of the Global Industry Classification Standard sectors, to exclude companies and reweight the Standard & Poor 100 equity index. Excluding aims to minimize exposure to high water intense industries. The reweighting is subjected to maintaining the same expected return and variance of the original index and minimize tracking error to the benchmark. Under an ex-ante tracking error of 1% an 80% reduction in water intensity can be achieved, with the exact same Sharpe-ratio as the Standard & Poor 100.
1. Introduction
Figure 1: Projected global water stress in 2040
Source: Maddocks, A., Young, R.S., Reig, P., World Resource Institute, 2015
This water scarcity has its impact industry wide, since a lot of businesses rely heavily on the supply off clean water. If water becomes more scarce water-intense industries face increasing prices or, in extreme cases, a lack of water supply resulting in negative impact on their corporate performance. If policies are enacted to reduce water usage, this again would result in increasing costs which negatively press on the corporate performance. These negative effects are the reason why I study the feasibility of constructing a portfolio attempting to mitigate these risks to a large extent. The methodological approach is based on the research by Andersson, Bolton and Samama (2016) who constructed a decarbonized index to hedge the risk of an increase in carbon pricing, their study attempts to test if in a similar manner a low-water-intensity index is feasible and yields competitive average returns. Contrary to the approach of Andersson et al. (2016) which focusses on hedging pricing risk, this study will focus on hedging constrained business operations. Carbon pricing can be specified as a financial risk since firms would bear higher costs, instead water scarcity is more an operational risk due to its non-substitutable characteristic. This difference is why the focus lies on intensity of water usage instead of absolute numbers, by dividing it over output in dollars. The water intensity is essentially a ratio which shows the amount of water necessary for a sector to produce $1,000. This low-water-intensity is a tool for investors to hedge against possible negative shocks in global water supply and the inherent risk certain companies are bearing. Constructing and testing this tool is done using the Standard & Poor 100 equity index, hereafter S&P 100. Applying the water intensity scale estimated results in reweighting the index with competitive results compared to its original counterpart. The difference is measured by the tracking error, the deviation of the reweighted portfolios compared to the original S&P 100.
Next to being a hedging tool it also addresses the moral aspect of investing, referred to earlier in the introduction. Investors increasingly want to put their money to good use, as the numbers about impact investing suggest. This view on finance as a science is justly put to words by Robert J. Shiller in his book Finance and the Good Society (2013, p7.):
“Finance is not about ‘making money’ per se. It is a ‘functional’ science in that it exists to support other goals—those of society. The better aligned society’s financial institutions are with its goals and ideals, the stronger and more successful the society will be.”
is structured as follows: First, the literature review on the preceding empirical and theoretical evidence is explained, with a focus on the debate on the link between environmental and financial performance. Second, is the approach to the construction of a water intensity classification and the reweighting of the S&P 100 index. This section is followed by the description of the used model and the results obtained after reweighting the S&P 100 index, ending with a discussion and conclusion on the results.
2. Literature review
In preceding studies, the link between climate change and corporate performance is well established. Both the direct effects of climate change altering the environment firms operate in as the effect of transitional risks caused by policy shifts. Considerable research has been conducted on the effect of environmental policies and regulation on financial performance. Several studies (Little, Muoghalu and Robinson, 1995; Muoghalu, Robison and Glascock 1990; Laplante and Lanoie 1994; Rao, 1996) in which firms with a proactive attitude towards implemented policies and regulation tend to perform better after the new policies are enacted. Cunha et al. (2019) complement these findings with their research on funds adhering to ESG criteria, these funds tend to outperform their benchmark. A probable explanation is that firm adhering to ESG criteria are better resistant against the shocks caused by regulations. In a study by Frederick and Major (1997) they warn about the effect warmer temperatures will have on the hydrological cycle, increasing the possibility of more extreme droughts and floods. Which in turn will have a large impact on the available water supply. This evidence is supported by Ayers et al. (1994) when empirically tested on the Delaware Basin1. More recent findings of the Intergovernmental Panel on Climate Change (2019)
report rising temperatures affecting precipitation patterns, and a decrease in yearly precipitation. As Judith Rosales states in her book Climate Change and Water Resources (2019, p. 23): “It is evident that water resource is a global concern”.
These shocks resulting from this global concern is what a low water intensity portfolio is trying to hedge and can be deconstructed into two parts: the lack of water supply due to a demand exceeding the supply or through an increase in the price of water. Studies on water resource and possible shortages are conducted worldwide, including Canada (He and Horbulyk, 2010), Spain (Roibás, García-Valinas and Wall, 2007) and the US2(Yuhas and Daniels 2006). Each of these studies has a common
outcome, in areas with a limited supply of fresh water the allocation is best regulated by imposing some form of higher pricing. In the case of the US study, they investigate the effect of a large desalination plant in Tampa Bay, producing freshwater out of seawater. This method of producing bears higher costs comparing to accessing a freshwater resource. The authors advocate the use of higher prices and block rates to meet the growing demand. In Canada, the authors study the effect of historically allocated water extraction rights of certain rivers. They propose alternative allocation policies as short-term water trading and volumetric water pricing, their model estimates that these policies have a reducing effect on water usage. Roibás, García-Valinas and Wall (2007) investigate the droughts Sevilla experienced from 1992-1996, where supply cuts were imposed by the government. Their results indicate that instead of
1 A simplified model on the specific dynamics of climate change on freshwater resources can be
found in Robock (1985)
2 For more studies regarding water scarcity in different (US) regions see: Hall (2009), Seo et al. (2008),
imposing cuts a higher pricing strategy is less likely to reduce welfare as drastically when facing water supply cuts. The commonality of these studies is the proposed imposing of higher water pricing, in situations where water supply is either temporarily or structurally restricted. The plausibility of the inherent risk water scarcity has on water intensive industries is theoretically and empirically well established.
3. Design & Methodology
Index funds currently in the market related to water are focussed on companies participating in the water industry, so-called pure-play indexes. There is a difference in approach since the funds are allocated towards one industry only, they all operate in water related businesses. The downside of this strategy is the low diversification, compared to the S&P 100 these indexes are not as well diversified across sectors. In Table 1 an overview of the five largest water focussed exchange traded funds (ETF) is given, with their yield to date and historical monthly volatility compared to the benchmark index.
Table 1: Comparison S&P 100 to largest Water ETF's.
S&P 100 ETF 1 ETF 2 ETF 3 ETF 4 ETF 5 AUM (MM) $7,186.90 $ 1,283.95 $ 807.15 $ 671.97 $ 231.66 $ 23.21 YTD 18.93% 18.85% 13.93% 19.40% 13.02% 14.03% Avg. Monthly Volatility 13% 14% 15% 17% 12% 16% Ratio 1.42 1.31 0.96 1.16 1.08 0.89
* Assets, volatility, and YTD as of 23-12-2020
Assets under management (in millions) and yield to date retrieved from ETF database, AUM of S&P 100 is based on largest ETF: iShares S&P 100. Average monthly volatility retrieved from V-Lab. Ratio displayed divides YTD by monthly volatility. ETF 1 is Invesco Water Resources ETF, ETF 2 is Invesco S&P Global Water Index ETF, ETF 3 is First Trust ISE Water Index Fund, ETF 4 is Invesco Global Water ETF and ETF 5 is Ecofin Global Water ESG Fund.
When comparing the risk return profiles of the listed ETF’s they do not compete with the S&P 100, ETF 1 being the exception, investing in these ETF’s carries more risk. Clearly, investors are willing to carry this risk since these five ETF’s have a summed asset under management of three billion. Reweighting the S&P 100 to carry less risk of water scarcity is not aiming to replace these ETF’s but could offer a welcome addition to an investor’s portfolio holding some of these ETF’s. Both approaches recognize the importance of the commodity water and can complement each other. Since these ETF’s are not as well diversified across sectors as the S&P it would reduce idiosyncratic risk of the portfolio. In the case of shocks to water supply the reweighted S&P is expected to outperform the market and the water tracking ETF’s will also generate higher returns.
exposure to water scarcity risk decreases, when shocks occur the portfolios with the largest reduction in water intensity are expected to perform best.
This theory results in the following set of hypotheses:
1. The S&P 100 can be reweighted so it is less exposed to water scarcity risk. 2. If shocks to the American water supply occur, the reweighted portfolios will outperform the original S&P 100.
The choice of the S&P 100 will add to the applicability of this research if the reweighted portfolios can produce similar expected returns under low tracking error. If this is the case other large market tracking indexes can be reweighted by the same principle. The S&P 100 benchmark used in this study is based on the last three years and summarized in table 2. The new constructed portfolios must conform to the same expected arithmetic mean and standard deviation.
Table 2: Summary Statistics of S&P 100
Standard & Poor 100 (2017-2020) Arithmetic Mean 1.83% Geometric Mean 1.40% Standard Deviation 6.10% Highest Weighted 9.85% Average Weight 0.99% Lowest Weighted 0.06% Number of constituents 101 Changes in constituents 0 Time Periods (Months) 37
Descriptive statistics of Standard & Poor 100 retrieved from Datastream, November 2020. Arithmetic and Geometric mean are monthly returns, standard deviation also in monthly time format. Weights show largest, average, and smallest position of companies in the composition. Time period is from November 2017 – November 2020.
3.1 Relative water intensity per sector
To reweight the S&P index into a low water intensity portfolio I construct a measurement system for water intensity (WI). It is important to note once more that water intensity is not equal to water consumption, WI is calculated by cubic litres divided by gross revenue, in $1,000, as stated in equation 1.
(1) 𝑊𝐼 = 𝑚
3
$1,000
could occur, for example the usage of datacentres occurs across most large companies, but this WI is not originating from their main operations. The U.S. Geological Survey defines water consumption as “the part of water withdrawn that is evaporated, transpired, incorporated into products or crops, consumed by humans or livestock, or otherwise removed from the immediate water environment.” (U.S. Geological Survey 2010, p. 49) This definition of water consumption is henceforth used for the calculation of the sectoral water intensity. The constructing of the measurement system relies heavily on several other studies from different types of academic disciplines such as engineering and geography. Since the data on all of the GICS sectors originate from multiple sources, all of the sources will be addressed per sector to create a clear view of how the classification is constructed.
The research of Roson and Damiana (2016) is the largest source of estimating the different water intensity levels of the GICS sectors. Their estimations draw upon multiple studies from different academic disciplines, as well as governmental data collection projects. For each of the sectoral estimations the relevant sources will be discussed. Based on their data most of the GICS sectors can be assigned their appropriate water intensity directly or indirectly. Roson and Damiana (2016) retrieve most of their data from the World Input-Output Database (WIOD) project (Dietzenbacher, Los, Stehrer, Timmer and de Vries, 2013) which provides industrial water consumption and output generated. These results are directly applied to the GICS sector used in this research when applicable. Other sectors can be indirectly estimated from the classification table presented in their research, which is explained per sector. The remaining sectors are estimated via other sources which are addressed per sector. The Consumer Discretionary sector is based on the equally weighted average of the combined WI of textiles, processed foods and light manufacturing retrieved from the estimations of Roson & Damiana (2016). These three represent the largest fraction of the operations within this sector, approximately 85%, based on the sectoral data retrieved from Bloomberg. Consumer Staples takes up the bulk of the relative water intensity, the production of crops is extremely water intense compared to the other sectors. For example, the estimations of different types of agricultural products by Roson and Damiana (2016) puts rice in the most water intense spot with an average water intensity of 31.69 cubic litres per $1,000 output. According to Bloomberg half of the Consumer Staples sector included in the S&P 100 are companies manufacturing food, beverages, and tobacco. A quarter is personal hygiene and the remaining quarter retailing. These exact weightings are used to construct an optimally corresponding WI level. However, due to these large fluctuations between different types of crops there is a strong possibility of negatively impacting certain companies compared to the true ‘poor’ performers. The WI of the Energy sector is exactly the same as the estimated WI of Extraction in the paper by Roson and Damiana (2016). Their estimation is based on research by Mielke, Diaz Anadon and Narayanamurti (2010) on the water usage of the extraction of coal, gas, and oil. Since these resources account for approximately 80% of energy consumption in the USA (EIA, 2019) and the companies included in the Energy sector of the S&P 100 are mainly operating in the fossil fuels, there is no necessity to include the water footprint of renewable or nuclear forms of energy to get a more accurate estimation.
divided by their annual gross revenue of the same year. When there are large differences between the different calculated water intensity estimates the worst performer is given the largest weight to construct the relative WI of the sector. The large deviations are most likely caused by stricter water usage guidelines i.e., ESG reporting with an accountant’s review have higher estimates. By using the strictest guidelines, the estimations used will reflect the real water intensity best. With Telecommunication Services the sectoral estimate is based on the water usage (Verizon, 2019; AT&T 2019) and gross revenue generated by Verizon and AT&T over 2019. Since the difference between the two estimations is merely 0.0021, the average is taken to represent the WI of the Telecommunications Sector. The Financials WI is also based on the reports over 2019 of some representative companies: Bank of America (2019), Wells Fargo (2019) and Goldman Sachs (2019). Based on the annual report of Bank of America the WI is 0.083, which is slightly below that of Wells Fargo (0.096) and above Goldman Sachs (0.033). Assigning 40% weight to Wells Fargo as the ‘poorest performer’ and 30% to the other two results in 0.073 as WI for the Financials sector. The Real Estate WI is coupled with the Financials sector WI, because of their similarities. In 2016 the Real Estate GICS sector was created by removing listed real estate companies and funds from the Financials, due to their growing size and importance. The only constituent of the S&P 100 classified as operating in the Real Estate sector is Simon Property Group, one of the largest real estate investment trust (REIT). These REITs are the main type of companies in the Real Estate sector e.g., in the S&P 1500 98% of these REITS make up the Real Estate sector. The nature of the operations of these firms can be classified as financial product. Which is why these two sectors share the same estimated WI. The ESG reporting from Simon Property Group did not give absolute water consumption data, so their own figures could not be included to represent the Real Estate sector.
The water usage footprint of the Health Care sector is retrieved from the paper ‘the environmental footprint of health care: a global assessment’ by Lenzen et al. (2020) Their findings estimate that the US Health Care sector approximately uses 2% of the water resources available. To estimate the WI of the sector the 2% is divided by the sector weight of the S&P 100 of 11.67%3 which proxies for the output generated
by the Health Care sector, which gives an estimated WI of 0.17. The Industrials sector is most difficult to classify due to the substantial number of different types of operations it represents. Roson and Damiana (2016) give an estimation for Heavy Manufacturing of 0.04 m3/$1,000 which results in a relative footprint of 0.3% that cannot be
representative for the Industrials sector. Other sources estimate the water intensity substantially larger; the European Environment Agency (2019) estimate that the European Industrial and Extraction sector combined account for 10.6% of water consumption. Estimations by the United States Geological Survey (2010) estimate that Industrials account for 4% of the total water withdrawal in the United States. Using the same estimation technique as for Health Care the 4% of water withdrawal and 6.35%4
of the representation of the market gives an WI of 0.63. The Information Technology sector is similar to the Financials sector water consumption with one exception. Their datacentres require substantial amounts of water for cooling5. Some of the larger
technology companies’ ESG reporting do include the annual water consumption. Combining the data reported with the annual gross revenue results in deviating
3 Found in table 4 4 Found in table 4.
5 For more in-depth information about the water consumption of datacentres: United States Data Center
estimations of the WI. Google’s data result in a WI of 0.178 (2019) while Facebook has an WI of 0.059 (2018) and Apple 0.049 (2015). Applying the same weights to the highest consumer as with the companies from the Financials sector leads to an aggregate sectoral WI of 0.104. Utilities and Materials sector WI directly originates from the results by Roson and Damiana (2016) which are derived from the WIOD. The Utilities sector is closely related to the Energy sector since most of the large companies in this sector do not only sell and generate their energy but also pump and mine their own fuels. Their respective WI’s do reflect this fact differing only 0.02. For the Materials sector the same caveat is present as for the Consumer Staples, due to the different types of operations included in this sector. Especially the lumber industry is a water intense business, with estimations by Schyns, Booij and Hoekstra (2017) ranging from 300 m3 to 2,600 m3 for a ton of paper product. Some of the firms are thus overestimated
or underestimated due to this generalization.
By normalizing the data from the different sources to water intensity as described in equation 1, the relative water intensity is constructed assigning each of the sectors a fraction summing up to 1. Each of the GICS sectors with their individual WI and relative WI is shown in table 3.
Table 3: Water Intensity per GICS sector, ratio and relative.
Industry Water Intensity (m3/$1,000) Relative Water Intensity Consumer Discretionary 0.40* 2.8% Consumer Staples 10.74* 74.7% Energy 0.70** 4.9% Financials 0.07 0.5% Health Care 0.17 1.2% Industrials 0.63 4.4% Information Technology 0.10 0.7% Materials 0.70** 4.9% Real Estate 0.07 0.5% Telecommunication Services 0.12 0.8% Utilities 0.68** 4.7%
Water intensity for each of the GICS sectors. Water intensity is calculated by dividing the water consumption by gross revenue. Relative water intensity is the fraction of the sum of the average water intensity of the sectors. Sources: ** Directly from Roson and Damiana (2016), * indirectly by combining estimates of Roson and Damiana (2016). Sources of remaining estimates found in text.
Each of the sectors is assigned their relative water intensity, which is derived from finding the sectors average cubic metres of water usage divided by $1,000 of output generated. The relative WI is an indication of the risk exposure of the sector to potential water supply problems. The relative WI figures indicate a severe underweighting of the Consumer Staples sector, since their WI is sizeably higher compared to the rest of the sectors. To take the size of a firm’s operations into account the relative WI is combined with the respective market capitalization rate of each of the companies in the S&P 100 to account for firm size, generated the ranked relative WI shown in equation 2:
By ranking the individual companies on the WI of the sector as their size, restrictive portfolios can be generated removing the largest companies in the most water intense industries. For the sectors which are estimated by some representative companies eq. 2 seems counterintuitive. First, the water consumption is converted to intensity by dividing over gross revenue to be ranked again by their respective market capitalization. Since the companies with the strictest guidelines are assigned a larger weight the use of eq. 2 can give a more accurate estimation of the real exposure to water certain companies face. By generating more accurate estimations the chance of restricting companies without high exposure becomes smaller. Before moving to the section which addresses the construction of these portfolios, some limitations of the above process are addressed.
3.2 Limitations
The water intensity data of each of the sectors is limited to annual updates. While some companies are actively trying to decrease their water footprint, it can still occur that they are underweighted or restricted from the portfolio. The sectors could also be refined more. Especially within Consumer Staples and Consumer Discretionary there is a wide variation of operations, each with different water footprints not accounted for in this approach. Certain subsectors are punished on account of more water intense subsectors. For the Industrials and Health care sector the estimation techniques are rather crude, due to a lack of firm specific data. In the section describing the construction of low-water-intensity this issue will be addressed more extensively.
3.3 Constructing low WI portfolio.
give an insight into the trade-off between water intensity reduction and the consequential tracking error.
4 Model & Results
To create a ‘free option’ the reweighted benchmark index must be replicated in terms of risk and return. The estimated return and variance are calculated from the last three years of constituents’ stock prices of the S&P 100, which the reweighted portfolios have to match. These constraints on the reweighting guarantee competitiveness in terms of risk return trade-off i.e., the Sharpe ratio has to remain equal to the benchmark’s Sharpe ratio. To account for idiosyncratic risk of specific sectors the original distribution of the S&P 100 is also added in the constrained optimization. The original composition of the S&P 100 is stipulated in table 4. These composition constraints ensure that the newly created portfolios are not over- or underexposed to a particular sector.
Table 4: Industry composition of S&P 100
Industry Fraction of S&P 100 Consumer Discretionary 16.16% Consumer Staples 9.46% Energy 2.16% Financials 9.35% Health Care 11.67% Industrials 6.35% Information Technology 40.27% Materials 0.20% Real Estate 0.12% Telecommunication Services 2.22% Utilities 2.03%
Sectoral distribution of Standard & Poor 100 as of November 2020. Source: Datastream
The reweighting of the portfolios has to follow this distribution as large diversifications could expose investors to sector specific idiosyncratic risk compared to the benchmark. These objective of creating a ‘free option’ on water scarcity give rises to the following optimization problem:
(3) 𝑀𝑖𝑛 𝑇𝐸 = 𝜎2(𝑤 𝑅− 𝑤𝐵)
With σ2 denoting the standard deviation and 𝑤 𝑅 and 𝑤𝐵 respectively the weights of
the reweighted portfolio and the weights of the benchmark. To ensure that the ex-ante tracking error is minimized the benchmark’s characteristics are used as the following constraints in the optimization problem:
(4) ∑ 𝑤𝑖𝑅 = 1 , 𝑤𝑖𝑡ℎ 𝑤𝑖𝑅 > 0
The sum of the remaining constituents in the restricted portfolio must equal 100% not allowing any short positions or weights of zero. Allowing short positions would drastically reduce standard deviation but leaves investors exposed to large fluctuations in the underlying positions of the companies. Maintaining the risk and return characteristics of the benchmark is reflected in the next set of equations:
(5) 𝜎𝑅2 = 𝜎
Equation 5 states that both the standard deviation and the return of the restricted portfolio have to match the characteristics of the benchmark, as summed up in the summary statistics. The expected return matched in the optimization problem is the arithmetic mean return.
The last constraint imposed on the minimization of the TE is in place to preserve the distribution of the GICS sectors. The fraction of the sectors (𝑓𝑖𝑅) in the restricted portfolios are not allowed to deviate more than 1% of the benchmarks distribution (𝑓𝑖𝐵), which results in a lower and upper bound for each of the 11 GICS sectors. Using iterative means the smallest interval, not causing violations of the model, is found. By using a -1% and +1% interval a complete diversification away from a specific sector is not possible, and the model is not violated. There is one instance in which this last constrained is violated, since restricting the 15 highest water intense companies results in removing all of the Consumer Staples constituents. This issue is addressed later in this section. The last constrained is shown in equation 6.
(6) 0.99𝑓𝑖𝐵 ≤ 𝑓𝑖𝑅 ≤ 1.01𝑓𝑖𝐵 𝑓𝑜𝑟 𝑖 = 1, … , 11
With these constraints in place five restricted portfolios can be constructed, incrementally excluding more of the designated ‘poor performers’ in table 3. The optimization problem is solved with the generalized reduced gradient (GRG) method provided by Analytic Solver6 (Frontline Solvers, 2020).
Table 5: Highest water intense corporations.
Rank Company name Relative WI Cumulative Relative WI
1 WALMART 19.0% 19.0%
2 PROCTER & GAMBLE 15.4% 34.4%
3 COCA-COLA 10.1% 44.5%
4 PEPSICO 8.9% 53.4%
5 COSTCO WHOLESALE 7.5% 60.9%
6 PHILIP MORRIS INTL. 5.3% 66.2%
7 CVS HEALTH 3.9% 70.1% 8 MONDELEZ INTERNATIONAL CL. A 3.7% 73.8% 9 ALTRIA GROUP 3.3% 77.1% 10 COLGATE-PALM. 3.3% 80.4% 11 AMAZON.COM 2.6% 82.9% 12 DUPONT DE NEMOURS 2.1% 85.0% 13 KRAFT HEINZ 1.8% 86.8% 14 WALGREENS BOOTS ALLIANCE 1.5% 88.2% 15 APPLE 0.8% 89.0%
Ranking 15 highest water intense companies. Relative water intensity to the total of summed WI of S&P 100. Cumulative relative WI shows the summed WI of the largest 15 companies in the index.
6 Analytic Solver extends the capabilities of the built-in Solver extension for Excel. Using Analytic Solver gave
The companies listed in table 3 are ranked according to equation 2, accounting for both water intensity of their sector as their size, measured by market capitalization rate. The relative WI is the fraction which the company contributes to the total S&P 100 water intensity. Removing all 15 companies with high water intensity would reduce the overall exposure to water intense industries by approximately 89%. Based on this ranking five portfolios are constructed, each restricting three extra companies from the benchmark’s constituents list. In the remainder of this thesis the newly constructed portfolios will be addressed as low water intensity (LWI) and the remaining number of constituents e.g., LWI-98 and so on.
Finding the new weights by optimizing under the constraints given by eq. 4-6 have to be transformed into the actual ex-ante TE. The TE is calculated in the following way:
(7) 𝑇𝐸 = √(𝑤𝑅− 𝑤𝐵)′𝐶𝑜𝑣𝑀𝑎𝑡(𝑤𝑅− 𝑤𝐵)
Figure 2: Efficient Frontier of restricted S&P 100
Figure 2 shows the trade-off between the reduction of WI by restricting more companies listed in table 5 and the tracking error to the original S&P 100. The outlying square represents the fully restricted index which is not efficient due to fully excluding the consumer staples sector.
The figure clearly shows that as the reduction in water intensity becomes larger the tracking error increases. The TE from the benchmark compared to LWI-98 of 0.69% is not substantially large, for a hypothetical investor willing to invest under the threshold of 1% TE the LWI-92 is still feasible. The applicability of the reweighting based on WI ends with the LWI-89 portfolio since this restriction cut-off still leaves room for companies in the Consumer Staples sector. Due to the limited constituent list of the S&P 100 this problem is less likely to occur in larger equity indexes.
The fact that the agricultural sector is exposed to droughts and other forms of water supply fluctuations is nothing new, but from a financial point of view this sector does carry this risk more compared to other sectors. The consensus among investors is that Consumer Staples is often a safe haven in times of crises. Due to the necessity of the goods it manufactures demand will not fluctuate very strong. A recent example of the performance of the sector in the Covid-19 crisis was researched by Smales (2020). The sectors performing best under the circumstances of the Covid-19 crisis are Consumer Staples, Healthcare and Information Technology. Stressing the importance of recognizing this risk is a valuable addition to the perception of this particular sector. As the literature suggests water scarcity is a growing problem in the coming decades and higher pricing of the liquid could induce substantial costs on food producing companies. Investors and investment managers need to be aware of the higher exposure to this specific type of risk for the Consumer Staples sector.
5. Discussion & Conclusion
Reweighting the Standard & Poor 100 equity index using the water intensity of the different GICS sectors is a feasible strategy to partially hedge against the risk of water scarcity. By classifying and ranking the companies most exposed to the risk of water scarcity the S&P 100 constituent list can restrict some of the ‘poor performers’. Removing certain companies can be done while maintaining a well-diversified allocation of components weighting across the largest sectors. Under the threshold of a 1% ex-ante tracking error compared to the original S&P 100 investors maintain the
Benchmark LWI-98 LWI-95 LWI-92 LWI-89 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 0.01% 0.69% 0.82% 0.91% 1.10% R e d u cti o n o f Wate r In te n si ty Tracking Error
Efficient Frontier restricted S&P 100
same expected return and standard deviation whilst reducing their exposure to high water intense companies by 80%. Restricting all 15 companies ranked highest in WI caused the model to break constraints, making this portfolio not feasible in the approach described. The violation of these constraints is caused by effectively removing all companies classified as operating in the Consumer Staples section. Since the model aimed to maintain the sectoral distribution of the benchmark this portfolio does not fall on the efficient frontier. The complete removal of the Consumer Staples sector is due to its high relative WI of 74.7%, which exposes this sector to the risk of water scarcity. Reflecting on the two hypotheses, the first hypothesis cannot be rejected. By actively excluding companies exposed to water scarcity risk the reweighting of the S&P 100 into the LWI portfolios creates less exposed portfolios. The second hypothesis cannot be accepted nor rejected since the constructing of the portfolios remains purely theoretical. Next, to being theoretical portfolios large shocks in the water supply have not yet occurred making it impossible to verify the second hypothesis.
This study has several limitations which are discussed in the following part. The reweighting of portfolios could be more refined by splitting the currently used 11 sectors into their respective subsectors. Some of the companies are being penalized by being put into the same classification as higher water intensity companies. For example, the forestry and paper industry are highly water intensive increasing the relative WI of the Materials sector. Which is also the case for Consumer Discretionary, Industrials and Consumer Staples. The limitation to these 11 sectors does create over- and underweighting, specifying the classification table to industry groups or even industries can give a better approach to reweighting. Creating a more accurate classification system justifies the use of other reweighting approaches. As discussed earlier underweighting the companies with large exposures is viable if the estimations become more accurate. Another limitation of the research is the static nature of this classification table, companies are becoming more efficient in their water consumption and the recycling of the water used in their production process. Water intensity is therefore expected to decline over time which is not captured into this model, by actively reclassifying the sectors on annual or semi-annual basis it becomes a more precise tool to hedge against water intensive operations. Both these limitation stem from the same cause: lack of transparent data. For the construction of this relatively crude categorization of the sectors a substantial amount of different academic estimations had to be included, these estimations are relatively generalized due to the fact that actual water usage is data not willingly disclosed by corporations. While environmental groups pressured politics to ensure disclosing of carbon emissions this is not the case for water consumption. Making it mandatory to disclose these types of environmental performance indicators into annual reporting will give a more transparent view of the risk’s companies are exposed to. Furthermore, it could induce companies to take on greater action to reduce their water consumption, making their operations less water intensive and less vulnerable to shocks to their supply.
References.
Andersson, M., Bolton, P., and Samama, F. (2016). Hedging climate risk. Financial Analysts Journal, 72(3), 13-32.
Apple. (2016). Apple Environmental Responsibility Report 2016. Retrieved from:
https://images.apple.com/environment/pdf/Apple_Environmental_Responsibility_Rep ort_2016.pdf
AT&T. (2019). 2019-2020 Summary.
Retrieved from: https://about.att.com/ecms/dam/csr/2019/library/corporate-responsibility/2019-2020-Summary.pdf
Ayers, M.A., Wolock, D.M., McCabe, G.J., Hay, L.E., and Tasker, G.D. (1994). Sensitivity of water resources in the Delaware River Basin to climate variability and change. US Government Printing Office.
Blackhurst, B.M., Hendrickson, C., and Vidal, J.S.I. (2010). Direct and indirect water withdrawals for US industrial sectors. Environmental science & technology, 44(6), 2126-2130.
Cunha, F.A.F.D.S., de Oliveira, E.M., Orsato, R.J., Klotzle, M.C., Cyrino Oliveira, F.L., and Caiado, R.G.G. (2020). Can sustainable investments outperform traditional benchmarks? Evidence from global stock markets. Business Strategy and the
Environment, 29(2), 682-697.
Department of Energy. (2016). United States data centre energy usage report (No. LBNL-1005775). Retrieved from: https://eta.lbl.gov/publications/united-states-data-center-energy
Dietzenbacher E., Los, B., Stehrer, R., Timmer, M. and de Vries G. (2013). The Construction of World Input-Output Tables in the WIOD Project. Economic Systems Research, 25, 71–98.
European Environment Agency. (2019). Use of freshwater resources in Europe. Retrieved from: https://www.eea.europa.eu/data-and-maps/indicators/use-of-freshwater-resources-3/assessment-4.
Facebook. (2019). Sustainability Report 2019.
Retrieved from: https://sustainability.fb.com/asset/2019-sustainability-report/ FTSE Russell. (2020). An introduction to Sustainable Investment.
Retrieved from: www.ftserussell.com/research/introduction-sustainable-investment. Frederick, K.D., and Major, D.C. (1997). Climate change and water resources. Climatic change, 37(1), 7-23.
Goldman Sachs. (2019). Sustainable Finance: the imperative and the opportunity. Retrieved from: https://www.goldmansachs.com/s/sustainability-report/index.html Google. (2018). Google Environmental Report 2019. Retrieved from:
https://www.gstatic.com/gumdrop/sustainability/google-2019-environmental-report.pdf Hall, C. (2009). Politically feasible, revenue sufficient, and economically efficient municipal water rates. Contemporary Economic Policy, 27(4), 539–554.
He, L., and Horbulyk, T.M. (2010). Market-based policy instruments, irrigation water demand, and crop diversification in the bow river basin of southern Alberta. Canadian Journal of Agricultural Economics/Revue canadienne d'agroeconomie, 58(2), 191-213.
Intergovernmental Panel on Climate Change. (2019). Climate Change and Land. Retrieved from: https://www.ipcc.ch/srccl/
Jain, P.C. (1987). The effect on stock price of inclusion in or exclusion from the S&P 500. Financial Analysts Journal, 43(1), 58–65.
Klassen, R.D., and McLaughlin, C.P. (1996). The impact of environmental management on firm performance. Management Science, 42(8), 1199-1214. Laplante, B., and Lanoie, P. (1994). The Market Response to Environmental Incidents in Canada: A Theoretical and Empirical Analysis. Southern Economic Journal, 60(3), 657-672.
Lenzen, M., Malik, A., Li, M., Fry, J., Weisz, H., Pichler, P.P., and Pencheon, D. (2020). The environmental footprint of health care: a global assessment. The Lancet Planetary Health, 4(7), e271-e279.
Little, P., Muoghalu, M.I., and Robison, H.D. (1995). Hazardous Waste Lawsuits, Financial Disclosure, and Investors’ Interests. Journal of Accounting, Auditing & Finance, 10(2), 383-398.
Maddocks, A., Young, R.S., and Reig, P. (2015). Ranking the world’s most water-stressed countries in 2040. World Resources Institute.
Retrieved from: https://www.wri.org/blog/2015/08/ranking-world-s-most-water-stressed-countries-2040
Mielke, E., Anadon, L.D., and Narayanamurti, V. (2010). Water consumption of energy resource extraction, processing, and conversion. Belfer Center for Science and International Affairs.
Muoghalu, M.I., Robison, H.D., and Glascock, J.L. (1990). Hazardous waste lawsuits, stockholder returns, and deterrence. Southern Economic Journal, 357-370.
Principles for Responsible Investment. (2020). Annual Report 2020. Retrieved from: https://www.unpri.org/annual-report-2020/delivering-our-blueprint-for-responsible-investment/responsible-investors/support-investors-incorporating-esg-issues Rao, S.M. (1996). The Effect of Published Reports of Environmental Pollution on Stock Prices. Journal of Financial and Strategic Decisions, 9(1), 25-32.
Robock, A. (1985). An updated climate feedback diagram. Bulletin of the American Meteorological Society, 66(7), 786–787.
Roibás, D., García-Valiñas, M.Á., and Wall, A. (2007). Measuring welfare losses from interruption and pricing as responses to water shortages: an application to the case of Seville. Environmental and resource economics, 38(2), 231-243.
Rosales, J. (2019). Climate change and water resources. Burlington, Vermont: Delve Publishing.
Roson, R., and Damania, R. (2016). Simulating the macroeconomic impact of future water scarcity: An assessment of alternative scenarios. University Ca'Foscari of Venice, Dept. of Economics Research Paper Series No, 7.
Rossi, G., and Cancelliere, A. (2013). Managing drought risk in water supply systems in Europe: a review. International Journal of Water Resources Development, 29(2), 272–289.
Schyns, J.F., Booij, M.J., and Hoekstra, A.Y. (2017). The water footprint of wood for lumber, pulp, paper, fuel, and firewood. Advances in Water Resources, 107, 490– 501.
Seo, S., Segarra, E., Mitchell, P.D., and Leatham, D.J. (2008). Irrigation technology adoption and its implication for water conservation in the Texas high plains: a real options approach. Agricultural Economics, 38(1), 47–55.
Shiller, R.J. (2013). Finance and the good society. Princeton, New Jersey: Princeton University Press.
Smales, L.A. (2020). Investor Attention and the Response of US Stock Sectors to the COVID-19 Crisis. Review of Behavioral Finance.
Social Investment Forum Foundation (2010). Report on Socially Responsible Trends in the United States.
Retrieved from:
https://www.ussif.org/files/Publications/10_Trends_Exec_Summary.pdf
Thomsen, R. (1993). Future droughts, water shortages in parts of western Europe. Eos, Transactions American Geophysical Union, 74(14), 161–165.
Verizon. (2019). Verizon sustainability 2019 independent accountants review report. Retrieved from:
https://www.verizon.com/about/sites/default/files/Verizon-sustainability-2019-Independent-Accountants-Review-Report.pdf Wells Fargo. (2019). Climate Change Issue Brief. Retrieved from: https://www.wellsfargo.com/about/corporate-responsibility/environment/
World Bank. (2017). Turbulent Waters: Pursuing Water Security in Fragile Contexts. Retrieved from: https://openknowledge.worldbank.org/handle/10986/26207
