<<

Business Intelligence for Sustainable Management

Natural Capital Management

A briefing on multiregional input–output cycle assessment and natural capital valuation techniques

Rick Betita • Director of Research Dr. James Barrett • Vice President of Environmental Economics November 20, 2013

 Modeling GHG, use, and use impacts of the global supply chain

 Determining the natural capital costs of environmental impacts

Today’s increasingly complex and globalized economy is rapidly consuming the ’s natural resources and degrading the services provided by these resources. The depletion of natural capital – the stock of typically undervalued goods and services provided by – is a major threat to a productive global economy. Measuring and managing natural capital is becoming a significant business issue, yet determining the environmental impacts of a deep and sophisticated supply chain remains a challenge. Furthermore, once impacts are calculated, it is difficult to make sense of those impacts and prioritize multiple environmental concerns.

To address these issues, Climate has developed the Natural Capital Management System (NCMS): a scalable, web-based system that enables companies to actively manage natural capital assets. (To download our white paper on managing natural capital assets, visit our NCMS webpage at http://www.climateearth.com/ncms.)

This paper describes the two methodologies underlying our NCMS technology. The first, multiregional input–output life cycle assessment (MRIO-LCA), measures the environmental impacts throughout a global

Business Intelligence for Sustainable Management

supply chain and models the location in the value chain where they occur. The second, natural capital valuation (NCV), allows calculated impacts to be valued, compared, and managed using a consistent unit of measure: the financial value of depleted natural capital.

In this paper, we discuss our current methods for MRIO-LCA and NCV as applied to the NCMS project completed for Webcor Builders in October 2013. This project assessed the natural capital costs incurred during the construction of the San Francisco Public Utilities Commission headquarters building, completed in June 2012. It is part of a larger strategy to enable Webcor to collaborate with future clients in reducing natural capital costs before and during construction, and to provide post-construction results for all major construction projects.

Climate Earth is dedicated to remaining transparent about the application of these emerging technologies. We also strive to partner with leading research organizations and rapidly adopt the most appropriate data and methodologies as they are developed. This strategy of rapid adoption best serves the needs of our clients to go beyond accounting and actively engage in the management of natural capital assets.

Business Intelligence for Sustainable Management

Methodology overview

The calculation of natural capital costs occurs in two steps. First, a multiregional input–output life cycle assessment (MRIO-LCA) model is used to convert company spend data into global supply chain environmental impacts. In the case of the Webcor SFPUC project, we focused on three impact categories: global warming

potential (“GHG”) in kg CO2e, land use impacts (“land”) in acres of forest land, and water use impacts (“water”) in kg water. Second, natural capital valuation (NCV) factors are applied to convert global supply chain impacts into natural capital costs in financial units. Thus, impacts from GHG, land, and water can be compared using a common unit of dollars of natural capital depletion. The figure below illustrates the higher- level process of calculating the natural capital costs from company spend data.

Multiregional input–output life cycle assessment

The multiregional input–output life cycle assessment (MRIO-LCA) model can be further broken down into two subprocesses: MRIO analysis, which models global economic activity, and environmentally-extended input– output LCA (EEIO-LCA), which translates global economic activity into environmental impacts.

MRIO analysis

Traditional input–output analysis methods model the embedded economic activity resulting from direct spend; MRIO analysis extends the results to other countries.

Mapping spend

The first step in MRIO analysis is to each element of spend data to the 430 economic sectors used by the United States Bureau of Economic Analysis (BEA) [1]. Once mapped, the spend data is adjusted to the model year of 2002 using the yearly average Producer Price Index (PPI) values [2], based on the mapped sector and spend year. The SFPUC project spanned a period between 2008 and 2012.

NCMS Methodology White Paper www.climateearth.com 3 Business Intelligence for Sustainable Management

Modeling economic activity

Economic activity is modeled using input–output (I–O) tables from the Comprehensive Environmental Data Archive (CEDA) [3], based on the BEA’s 2002 benchmark input–output accounts. These tables include not only the cost at the final manufacturing facility (the first tier in the supply chain), but also the breakdown of upstream supply chain costs resulting from this first-tier spend. For example, initial demand of architectural metal products involves additional economic activity in upstream sectors – mining metal ores, smelting, making metal parts, and forming subassemblies, as well as generation throughout the supply chain – all of which are calculated from these tables.

The I–O tables in CEDA, representing economic activity in the United States, are also used to model foreign economies under the “import assumption” [4] (also known as “autonomous ” [5] and “mirrored economy” [6]). For the SFPUC project, the first tier of spend was assumed to occur entirely within the United States. In subsequent tiers, US spend was split between domestic and imported spend using country recipes.

Creating country recipes

Country recipes indicate the source of production for the average commodity purchased in a particular country. For the SFPUC project, country recipes were limited to modeling the US supply of commodities under the “uni-directional trade” assumption, which reduces data requirements without introducing large errors [4].

For a particular commodity (sector), the fraction of US purchases originating from other countries is calculated from the import matrix, a supplementary table accompanying the BEA’s 2002 benchmark I–O accounts [7]. Global trade flow data from the Commodity Trade Statistics database (UN Comtrade) [8] is used to further break down imports into individual countries of origin.

For example, according to the BEA’s import matrix, 18.5% of “flat glass manufacturing” originated outside of the US. Of those imports, 22% came from China and 17% came from Germany. Thus, the first three country recipe values for “flat glass manufacturing” in the US are:

 US: 81.5% (= 100% − 18.5%)  China: 4.1% (= 18.5% × 22%)  Germany: 3.2% (= 18.5% × 17%)

For a particular sector and country (e.g. “flat glass manufacturing” in the US), country recipe values over all source countries sum to unity.

Each country recipe contains a “long tail” of countries that contribute very small percentages to the US supply of commodities. To reduce the number of calculations performed, ten countries were selected based on where foreign impacts were likely to occur. As a result, the original country recipes were collapsed into values for the US, the ten selected countries, and the “rest of world” (ROW) containing the sum of values for all other countries.

Modeling the international supply chain

The basis of traditional I–O models is the ( ) matrix, also known as the Leontief inverse after the Nobel Prize-winning work of Wassily Leontief [9]. It allows for the calculation of economic activity resulting from

NCMS Methodology White Paper www.climateearth.com 4 Business Intelligence for Sustainable Management

spend in an economy made up of many interdependent sectors, and represents the sum of first-tier purchases, second-tier purchases resulting from first-tier purchases, and so forth.

In the MRIO model developed by Climate Earth, this convergent infinite series is broken down into its individual parts, allowing for allocation of domestic spend across countries (representing commodity imports) between each tier of the calculation. These iterative calculations are repeated until total allocated spend reaches a threshold of 99% of the total embedded spend (calculated by multiplying the Leontief inverse by company spend, as in traditional I–O analysis), at which point closed economies are assumed and the remaining economic activity is calculated internally for each individual country using the Leontief inverse.

The result of these calculations is the modeled embedded economic activity in each country-sector (e.g. glass manufacturing in China, wood product manufacturing in ) resulting from company spend.

EEIO-LCA

Environmentally-extended input–output life cycle assessment (EEIO-LCA) augments traditional input–output analysis with environmental accounts, which give average impacts per dollar of economic activity in a given sector. These are calculated by dividing the total direct impacts per sector over a given year (in impact units, e.g. the fruit farming sector’s direct water withdrawals in kg of water) by the total output of that sector in that year (in units of model year dollars, e.g. the fruit farming sector’s output in USD2002).

The results of MRIO analysis provide embedded economic activity by sector and country. In EEIO-LCA, these values are multiplied by impact factors for GHG, land use, and water use from CEDA; though these numbers are based on US data, they are used to represent direct sector emissions globally due to limited data availability from other countries. Calculated impacts are then summed across sectors to give impacts by country, which provides the input for natural capital valuation.

Conclusion

Climate Earth’s MRIO-LCA model includes a few assumptions about non-US economies and their associated emissions. However, it uses publicly available data and well-established methodologies to provide a directionally accurate view of a very complex picture. Furthermore, the model is built to allow for modifications as new economic or environmental data is found or made available.

NCMS Methodology White Paper www.climateearth.com 5 Business Intelligence for Sustainable Management

Natural capital valuation

Environmental impacts are translated into financial costs in the natural capital valuation (NCV) step. Since the process of NCV will vary somewhat by environmental impact, we will illustrate the calculation techniques applied to the SFPUC project. For land use and water consumption, environmental costs are dependent on where the activity takes place; emissions are indifferent to where the emissions take place, so a single global number is sufficient to account for those costs.

Note that as part of our process, these assumptions and interim model results will differ for each project. Our practice is to discuss interim conclusions with each client and to fully document interim and final results. Since final results are applied as factors in the NCMS database, factors can be adjusted to run “what if” scenarios and may be tuned over time as customers uncover improved primary data about upstream impacts. The process of managing natural capital assets is a dynamic one, and the NCMS system is designed to support the dynamic nature of improving methods of analysis and measurement.

Discount rates and

The discount rate is a way of measuring the degree to which people prefer present benefits to future benefits and discount future costs relative to current costs. Steep discount rates place a much higher value on current costs compared to future ones: a 10% discount rate devalues a cost by 10% for every year in the future it occurs. This might be the case for someone with a short life expectancy, for example. Young children sometimes exhibit an extreme version of this behavior when they display a strong preference for dessert before dinner rather than after. Conversely, a lower discount rate of 3% values future costs much more closely to their current value. A discount rate of 0% implies that future costs are valued exactly the same as present ones, no how far in the future they might occur.

The choice of a discount rate in economic analyses of this type is extremely important because environmental costs do not typically occur all at once, but are spread out over time. The most severe impacts of , for example, may not be felt for 50 or 100 years in the future. A discount rate of 10% applied over a 100 year time horizon reduces the present valuation of those costs by a factor of 99.997%, making them essentially meaningless. At a societal discount rate of 10%, the value of avoiding even catastrophic climate change 100 years in the future is essentially zero.

For short to medium term economic costs, a discount rate in the range of 6% is common in economic analyses. For long-term costs, such as those associated with climate change, much lower discount rates are typically applied. In its effort to frame the long-term costs of climate change, the British Government’s Stern Review on the Economics of Climate Change [10] applied a discount rate of 1.4% to future costs of climate change in developing its estimate of the societal costs of current carbon emissions. This discount rate is lower than most economists apply in their analyses, and the report and its primary author, Sir Nicholas Stern, received a fair amount of criticism from the economics profession for that choice. Stern and others have defended that choice by noting that discounting has traditionally been used for much shorter time horizons than those implied by climate change and that the types of costs that climate change threatens (such as the loss of entire

NCMS Methodology White Paper www.climateearth.com 6 Business Intelligence for Sustainable Management

countries to rising ) requires a different approach to discounting. Specifically, the fact that many of the potential costs generated by a changing climate are very unlikely but carry almost incalculably high costs makes traditional discounting theories inappropriate. In a somewhat critical review of the Stern analysis, prominent Harvard economist Martin Weitzman agreed with this essential point [11].

This is relevant to our analysis for two reasons. The first is that in choosing a discount rate to apply to future environmental costs, we agree with Stern’s central point that the long-term nature of the costs involved requires a higher discount rate than traditionally applied. However, like Weitzman, we stopped short of fully accepting Stern’s 1.4% for land use change and water consumption impacts, in large part because both imply environmental costs that are much more immediate than climate change. As a compromise, we chose a slightly higher rate for land use change impacts for the SFPUC study. (Specific rates selected for the SFPUC project are available on request for academic or research review.)

At the same time, the Stern Review is perhaps the most comprehensive attempt to assess the economic costs of climate change, and importantly for us, of the present costs of greenhouse gas emissions. For that reason, we chose Stern’s value of the societal costs of emissions for our valuation. That value of $110 per metric ton of carbon and carbon equivalents is higher than some analyses, notably those in the $25 range by Tol [12], and much lower than others, such as those in the $900 per ton range by Ackerman and Stanton [13].

Induced land use change

Induced land use change is the conversion of to different states primarily though . For the SFPUC project, Climate Earth used the MRIO-LCA method discussed in the previous section to calculate the number of hectares of forest land conversion needed to support the project and identified the countries in which those impacts occurred. As mentioned above, the environmental impacts of the project were spread across a number of countries. The Climate Earth model estimated that 98.8% of the land use impacts occurred in 16 different countries. The distribution of impacts was highly concentrated, however, with 2 countries (the U.S. and Indonesia) experiencing over 90% of the impacts. Because of the thin tail of the distribution in the case of the SFPUC project, we focused on the five countries that would be most impacted, which together accounted for just under 96% of the total land use impacts of the project.

For these five countries – the United States, Indonesia, Thailand, Canada, and – we searched the environmental and ecological economics literature for natural capital accounting studies that had valued the costs of deforestation in these countries in economic terms. The theory of natural capital accounting begins with the premise that various (such as tropical , coral reefs, tidal marshes and others) provide ecosystem services that have some value but which the beneficiaries typically pay little or nothing for. As a result, these while these services are clearly valuable, they have no market value. For example, tropical forests provide flood prevention, water purification, carbon sequestration and other services that, in their absence, would either have to be obtained through other means (buying bottled water) or foregone (faster concentration of atmospheric carbon) in their absence. Natural capital accounting studies attempt to quantify these services and determine the value associated with them using a variety of methods such as replacement cost, hedonic pricing, and damage avoidance costs [14].

In calculating the value of forest services, we were able to identify at least one study per country that valued the services provided by a hectare of forest land. Unless the original research indicated otherwise, we assumed that the lost services were replenished over 35 years, so that in the year of conversion, each

NCMS Methodology White Paper www.climateearth.com 7 Business Intelligence for Sustainable Management

converted hectare produced 1% of its usual services, and regrew at a constant percentage, reaching full service provision in the 35th year.

Some studies reported their results having already calculated a net present value. Those that did all used a discount rate other than the one used in our own calculations. For these studies, we deconstructed the NPV at their discount rate to produce a stream of undiscounted benefits, applied the regrowth algorithm described above, and re-discounted them to produce an NPV that is consistent across the different countries.

In each case, we performed all of the calculations in the currency used in the original research, which for most was a local currency. We then converted those values to US dollars using Purchasing Power Parity data from the World Bank [15] and adjusted them for inflation to bring them to constant US 2012 dollars using the implicit GDP deflator [16].

Purchasing Power Parity (PPP) is similar to an exchange rate in that it is a way of converting currency of one type to another. However, exchange rates are based on the financial value of currencies on global markets, which are subject to international financial and in some cases to intentional exchange rate interventions by national governments. In contrast PPP is based on the cost of buying a fixed bundle of goods in each country. The cost of that bundle in a country’s native currency is compared to the cost of buying that bundle in the United States. Particularly for developing countries and countries with a strong interventionist policy in exchange rate management, PPP provides a better comparison of the value of goods and services between countries than official exchange rates do.

Water withdrawal

As with land use changes, water withdrawals associated with the SFPUC project are spread out over a number of different countries. However, the concentration of costs is much less pronounced than with land use. In this case, 21 countries account for just under 97% of water withdrawals (measured by volume). For the purposes of this study, we focused on the 9 countries that together account for 92.8% of withdrawals. Extending further down the distribution meets with rapidly diminishing returns, so that to reach the 96% threshold as we did in the land use calculations would require that we include an additional 9 countries. To make the question more tractable, we included only those countries that accounted for more than one half of one percent of the total water withdrawals associated with the project.

Unlike land use conversion, water is not a that provides ecosystem services. Rather, it is itself a service (more appropriately a consumable good) provided by biomes. As such, water withdrawal has not been subject to the same type of valuation research as biomes have. As a result, the best intellectually consistent measure we found for water valuation across all of the countries we examined is the price of water in these countries. This is dissatisfying in a number of dimensions, mostly centered around the fact that water prices do not reflect scarcity well, only the price that residents and businesses have to pay for water use. In most cases, water is provided by a local or national government agency that often intentionally or otherwise does not present its customers with the full price of providing that water. In light of this, these numbers must be seen as a lower bound of the value of the water withdrawals associated with the project.

Of the countries we included in the study – the United States, Indonesia, China, Thailand, Canada, Mexico, Japan, and Germany – six are members of the OECD and we obtained national averages of retail water prices from a single source that compared water prices among OECD countries. We converted the values to 2012 US dollars using Purchasing Power Parity and the GDP deflator as described above. For the 3 non-OECD countries

NCMS Methodology White Paper www.climateearth.com 8 Business Intelligence for Sustainable Management

– Indonesia, China, and Thailand – we found World Bank citations for local water prices. In cases where we found prices for multiple states, regions or other areas, we took a simple unweighted average of those values. In the case of Indonesia, the source material noted that the water was provided at a rate that only covered 80% of full cost recovery, so we pro-rated the price by that factor. Since water consumption is concurrent with the project, there was no need to discount the numbers.

Conclusion

The main goal of each Climate Earth project is to provide a reasonable assessment of the natural capital costs most relevant to the company or industry the company represents. For the construction industry and the SFPUC project discussed in this paper, we chose to examine the costs of greenhouse gas emissions, land use conversion, and water consumption associated with construction and its resulting supply chain.

In developing estimates, we attempt to be as comprehensive and consistent as possible and to use conservatively high assumptions wherever we have to make a decision based on our judgment. While we believe the underlying science is sufficiently advanced to achieve our goal of providing a reasonable assessment of likely environmental costs, we also recognize that certain basic inconsistencies will arise.

One inconsistency involves the use of different discount rates. For example, we used the value of $110 per metric ton of carbon equivalent (in 2012 US dollars) from the Stern Review on the societal costs of greenhouse gas emissions. As noted above, this estimate is based on a discount rate of 1.4%, while our cost estimates for induced land use change are discounted at a slightly higher rate. This implies that a dollar’s worth damages associated with climate change 30 years from now are worth more than a dollar’s worth of damages from deforestation. This inconsistency is fundamentally unresolvable because using a higher discount rate for climate change essentially reduces the long term costs of climate change to near zero, while using a lower discount rate for deforestation in our view would put an inappropriately high value on the medium-term costs associated with induced land use change.

This numerical inconsistency is logically resolvable, however at least in part. The societal discount rate is a function of pure time preference, long-term consumption growth rate projections, and risk aversion. The potential costs of climate change include so-called “fat tail” outcomes: low-probability but extremely high-cost outcomes at the tail end of the probability distribution. The uncertainty over the downside risk associated with increased greenhouse gas emissions is much higher than the risk associated with the long-term impacts of deforestation which are relatively much more well-known. For a given level of risk aversion, greenhouse gas emissions must therefore be associated with a higher willingness to pay to avoid the more risky costs associated with those emissions. The increased uncertainty around potential outcomes demands a higher valuation of potential future costs and thus a lower discount rate.

A second and less resolvable issue is one of embedded costs in the existing research. In the case of the SFPUC project, the cost of greenhouse gas emissions were embedded in the deforestation studies we relied on. In each study, the authors included in their estimation of ecosystem services the value of the carbon sequestered by the forests in question. These values were based on an avoided cost of carbon emissions, and each used a societal cost of carbon that is different from the value in the Stern Review and they were not consistent with each other. The differences in values used in each study reflect the state of the research at the time each study was written. Because the studies typically presented their results in terms of the value of a hectare of the biome as a whole, it was impossible to decompose the valuations into their constituent parts and replace the carbon sequestration term with the Stern value of $110 per ton of carbon.

NCMS Methodology White Paper www.climateearth.com 9 Business Intelligence for Sustainable Management

In the absence of a common methodology for valuing natural capital, either through these efforts or through new primary research done expressly for commercial application, these types of inconsistencies will likely persist. However, multiple efforts to create a uniform framework for valuing the natural capital of nations are underway that might avoid these issues in the future. As noted in the introduction, Climate Earth’s strategy is to work closely with the research and NGO communities to rapidly adopt best practices and to contribute our practical knowledge to the research when applicable. Notable research projects underway include the System of Environmental Economic Accounting (SEEA) being developed by the United Nations [17] and the WAVES Partnership [18].

Regardless of the as yet unresolved issues, we are confident that the current science, if used in an open and transparent manner, is more than adequately advanced to form a solid representation of the environmental costs associated with business activity. We also believe that the costs of delaying will be much higher to business and the than any costs associated with business beginning to manage natural capital assets in a systematic way.

NCMS Methodology White Paper www.climateearth.com 10 Business Intelligence for Sustainable Management

References

[1] U.S. Bureau of Economic Analysis (BEA), “U.S. Benchmark Input-Output Accounts, 2002,” 2007.

[2] U.S. Bureau of Labor Statistics, “Producer Price Index (PPI).” [Online]. Available: http://www.bls.gov/ppi/.

[3] S. Suh, “CEDA 4.0 User’s Guide.” 2010.

[4] G. P. Peters and E. G. Hertwich, “The application of multi-regional input-output analysis to industrial ,” in Handbook of input-output economics in industrial ecology, Springer, 2009, pp. 847–863.

[5] M. Lenzen, L.-L. Pade, and J. Munksgaard, “CO2 multipliers in multi- input-output models,” Econ. Syst. Res., vol. 16, no. 4, pp. 391–412, Dec. 2004.

[6] A. H. Strømman and A. Gauteplass, “Domestic fractions of emissions in linked economies,” Industrial Ecology Programme, Norwegian Unviersity of Science and Technology, Trondheim, Norway, Working Paper no. 4/2004, 2004.

[7] U.S. Bureau of Economic Analysis (BEA), “Import Matrix from the 2002 Benchmark Input-Output Accounts.” 2008.

[8] United Nations Statistics Division (UNSD), “Commodity Trade Statistics Database.” 2012.

[9] Wikipedia, “Input-output model.” [Online]. Available: http://en.wikipedia.org/wiki/Input- output_model.

[10] N. Stern, The Economics of Climate Change: The Stern Review. 2007.

[11] M. L. Weitzman, “A Review of the Stern Review on the Economics of Climate Change,” J. Econ. Lit., vol. XLV, pp. 703–724, 2007.

[12] R. S. J. Tol, “The Social Cost of Carbon: Trends, Outliers and Catastrophes,” Economics: The Open- Access, Open-Assessment E-Journal. 2008.

[13] E. A. Stanton and F. Ackerman, “Climate Risks and Carbon Prices: Revising the Social Cost of Carbon,” Economics: The Open-Access, Open-Assessment E-Journal. 2012.

[14] R. Costanza, R. D’Arge, R. de Groot, S. Farber, M. Grasso, B. Hannon, K. KLimburg, S. Naeem, R. V O’Neill, J. Paruelo, R. G. Raskin, P. Sutton, and M. van den Belt, “The value of the world’s ecosystem services and natural capital,” Nature, vol. 387, pp. 253–260, 1997.

[15] The World Bank, “PPP conversion factor (GDP) to market exchange rate ratio.” [Online]. Available: http://data.worldbank.org/indicator/PA.NUS.PPPC.RF.

[16] U.S. Bureau of Economic Analysis (BEA), “BEA National Economic Accounts: Gross Domestic Product (GDP).” [Online]. Available: http://www.bea.gov/national/index.htm#gdp.

[17] United Nations Statistics Division (UNSD), “System of Environmental-Economic Accounting (SEEA).” [Online]. Available: http://unstats.un.org/unsd/envaccounting/seea.asp.

[18] WAVES, “Wealth Accounting and the Valuation of Ecosystem Services.” [Online]. Available: http://www.wavespartnership.org/waves/.

NCMS Methodology White Paper www.climateearth.com 11