1 Introduction

Weighting is essential to convert the various types of environmental impacts into a common indicator using numerical factors based on value choices. Despite the integration of social, political, and ethical values being a controversial topic, weighting is frequently used in Life Cycle Assessment (LCA) studies because of its manageability and comprehensibility. Weighting is defined as one of the steps of LCA (ISO 14040 2006; ISO 14044 2006), and to date, a variety of approaches have been developed in this regard.

The various methods of assessment with the use of weighting factors can be classified as proxy method, midpoint method, and endpoint method (Itsubo et al. 2015). The endpoint method is further classified into two types. In the first type, normalized midpoint scores are weighted among endpoints, and in the second, weighting is performed by multiplying the value per unit of damage to the endpoint and the result of endpoint characterization. Economic assessments are classified as belonging to the latter type. The results of an economic assessment are expressed in monetary values, thus making them easy to understand and communicate as well as allowing their application to cost-effectiveness analyses, among others. On the other hand, methods for converting environmental impacts resulting in health damage and ecosystem decline into economic values are still in the development phase. These results may be strongly biased or incomplete, and their use is often inappropriate depending on how the information is conveyed. It is often difficult to assess such impacts in the first place, given the paucity of information and extensive assumptions.

According to Huppes and Oers (2011), economic assessments are classified into the two approaches: integrated modeling (i.e., External Costs of Energy (ExternE); New Energy Externalities Development for Sustainability (NEEDS, 2006); Weitzman 2007; Tol 2008) or weighting in LCA (i.e., Environmental Priority Strategies (EPS)(Steen, 1999); Life Cycle Impact Assessment Method based on Endpoint Modeling (LIME); Weidema 2009). With respect to the former, while Weitzman (2007) and Tol (2008) evaluated specific domains such as climate change and resource depletion, NEEDS and ExternE offer the advantage of broader domain coverage. Whereas the damage costs of the endpoints are always expressed in monetary values in integrated modeling, such simplification and accumulation is inappropriate, because different types of costs (e.g., market values (repair or replacement) or non-market values (contingent valuation, travel cost, and hedonic pricing) and use or non-use values) are adopted depending on the domains.Footnote 1 For example, NEEDS adopts restoration costs (ecosystem health), damage costs or abatement costs (climate change), restoration costs (building materials), and yield (crops) as willingness-to-pay (WTP).Footnote 2 In addition, the absence of direct weighting across these terms ensures that integrated modeling does not meet the requirements of ISO 14040.

EPS, which is categorized as a weighting approach in LCA, provides monetary evaluation (i.e., WTP) of multi-endpoints related to the four safeguard areas of human health, ecosystem production, biodiversity, and abiotic stock resources. This methodology covers broader domains such as climate change, air pollution, and resource consumption. However, the result of the EPS is not based on the direct weighting procedure among endpoints, but rather, on the WTP that has been estimated by several previous studies conducted in different countries at various points in time. Since it is presumed that WTP varies depending on economic, cultural, and social conditions, such oversimplified accumulation and treatment of these values provides inconsistent results.

Conversely, LIME, another weighting approach in LCA, directly weights the four areas of protection—human health, social assets, biodiversity, and primary production—using choice experiments. In this study, the monetary term (i.e., environmental tax) was included in the questionnaire so that the respondents could directly compare the importance of the four safeguard subjects and the changes in income due to the levying of the environmental tax. Then, WTP for a unit of each safeguard subject can be calculated using a uniform model (Itsubo et al. 2012, 2015). In addition, LIME has broader domain coverage (15 impact categories as midpoint levels) and also considers aspects such as climate change, air pollution, resource consumption, and acidification.

While LIME2 (Itsubo et al. 2012) developed national weighting factors for Japan (which cannot be used universally as they are specific to Japan), LIME3 (Itsubo et al. 2015 and this study) conducts a large-scale simultaneous survey in the G20 countries, using a uniform questionnaire to compare the weighting factors calculated for different countries and to calculate global-scale weighting factors. Part 1 of this paper (Itsubo et al. 2015) solely presented the dimensionless weighting factors, which are normalized to ensure that the total sum of the values for the four safeguard subjects would be 1 focusing on relative awareness among these subjects. Thus, this study is the first to collate evidence on monetary weighting factors (MWFs) on a global scale. The study thus sheds new light on how the MWFs of developing G20 countries differ from those of the developed G20 countries and provides useful information for economic evaluation in future LCA and cost–benefit analysis (CBA) studies considering the global supply chain. Moreover, recent LCA studies use an expanded geographic scope (Hoekstra and Wiedmann, 2014; Hellweg and Canals 2015). For example, Hoekstra and Wiedmann (2014) investigated the global environmental footprint and highlighted several impact categories, such as greenhouse gas (GHG) emissions, water, land, and materials, but they could neither derive the possible relationships among these impact categories nor evaluate the extent of damages in monetary terms. Our results provide the monetary value per unit of damage to the endpoint, and thus, when combined with the findings from previous studies, our results can help the monetary evaluation of an environmental impact at the global scale.

The rest of this paper is organized as follows. Section 2 briefly reviews the survey method and calculation procedures for the MWFs. Section 3 contains details of the estimation results exploring the global and G20 national MWFs. Section 4 extends the analysis to the economic evaluation of environmental impacts on a global scale. Section 5 concludes.

2 Study method

The subjects of the assessment―human health, social assets, biodiversity, and primary production―and research flow have already been described in part 1 of this paper (Itsubo et al. 2015). Therefore, we focus on the calculation procedure for the MWFs. We also briefly review the choice-based experimental design and the discrete choice model used for estimation in this section.

In our survey, the questionnaire provided explanations for the four subjects of assessment and surveyed respondents’ perceptions of these subjects.Footnote 3 After providing these explanations and conducting a question-and-answer session in order to assist the respondents in evaluating what they value, the respondents were allowed to read the questionnaire, which posed several hypothetical choice situations as described below. The respondents received a small remuneration for completing the questionnaire. We considered the attributes of environmental policies using the choice experiment method. Based on the reference values (Itsubo et al. 2015), we determined the alternatives, attributes, and attributes levels, as shown in Table 1.

Table 1 Attributes and levels used in the choice experiment
Full size table

We designed a hypothetical choice set composed of 16 alternatives (8 different choice situations of 2 alternatives) in various combinations of these attribute levels (4 safeguard subjects and environmental tax) using the orthogonal planning method, which is common in choice experimental design and was adopted in a previous study (Itsubo et al. 2012).Footnote 4 The attributes of the choice experiment are (1) human health (loss of health per person), (2) social assets (loss of natural resources per personFootnote 5), (3) biodiversity (loss of species on a global scale), (4) primary production (loss of forests on a global scale), and additional tax (yearly per household). We determined these attribute levels using an LCA.Footnote 6 In the choice experiment, respondents were asked to choose from three policies; policies 1 and 2 were adjusted for hypothetical situations including environmental improvements and the additional payment of environmental tax, and policy 3 referred to the current situation, with no additional environmental tax. Figure 1 shows an example of one of the choice sets provided in the questionnaire. The status quo option (policy 3) was included in all the questionnaires so that the respondents could always compare and contrast the current and hypothetical situations. All respondents were asked the similar eight questions as shown in Fig. 1.Footnote 7

Fig. 1
figure 1

An example of the profiles used in this study. Note: Instead of using quantitative values such as fractional numbers, we adopted visual aids in order to improve respondents’ understanding of possible changes in the four areas of protection as defined by LIME. Before undertaking the choice experiment, we explained to the respondents that the black dots indicate the degree of loss; the higher the number of black dots, the greater the degree of loss. Visual aids are used as risk communication tools and can facilitate the understanding of relatively small risks in the context of health. We selected icon arrays (shown in this figure) from among a variety of tools, such as risk ladders and pie charts. Loomis and duVair (1993) evaluated the effect of visual aids on WTP. For additional details about the application of visual aids, see Lipkus and Holland (1999), Okan et al. (2012), and Garcia and Cokely (2013)

Full size image

The questionnaire was translated into the national language of each surveyed country. The values of the environmental tax and social assets were converted into the local currency of the country being surveyed using the relevant purchasing power parity exchange rate (World Bank 2013).Footnote 8 The Internet survey and face-to-face interview method were primarily adopted for developed countries and emerging countries, to secure as large a number of samples as efficiently as possible and to minimize survey bias, respectively.Footnote 9 Random sampling was implemented for both types of surveys. We obtained response data from 200 to 250 and 500 to 600 households from each of the emerging and developed countries, respectively.Footnote 10

The response data collected from the survey were statistically analyzed using a random parameter logit model. As already described in part 1 of this paper, the utility function of respondent n choosing alternative i is given by

$$ {U}_{ni}={\beta}_{ns}^{\prime }{x}_{si}+{\beta}_m{m}_i+{\beta}_0\gamma +{\varepsilon}_{ni}\kern02.5em \left( s=\mathrm{HH},\mathrm{SA},\mathrm{BD},\mathrm{PP}\right) $$
(1)

where x si is an attribute vector for a loss of each safeguard subject s—human health (HH), social assets (SA), biodiversity (BD), and primary production (PP)—of an alternative i, and m i is a monetary attribute, which is an additional annual environmental tax in this study, as shown in Table 1. The constant γ equals 1 if alternative i is the status quo (the third alternative in this study; see Fig. 1). ε ni is a random term including all effects due to unobservable information of alternative i and respondent n in our survey. Using the maximum simulated likelihood (MSL) method, we can estimate \( {\beta}_{\mathrm{ns}}^{\prime } \), a random parameter vector; β m, a fixed parameter set as a numeraire; β 0, the preference intensity for a loss of each subject; additional environmental tax; and basic utility of the status quo option.Footnote 11

Because β m indicates the marginal utility of income, marginal willingness-to-pay (MWTP) per household for avoiding a loss of each subject is calculated from the ratio of β s to β m, which is given by the following:

$$ {\mathrm{MWTP}}_{\mathrm{s}}=\frac{\beta_{\mathrm{s}}}{\beta_{\mathrm{m}}}\kern0.5em \left( s=\mathrm{HH},\mathrm{SA},\mathrm{BD},\mathrm{PP}\right) $$
(2)

where MWTPHH, MWTPSA, MWTPBD, and MWTPPP denote MWTP per household for avoiding losses of a day, 1 USD worth resources, 1 species, and 100 million tons, respectively.

The MWFs of each subject are defined as follows:

$$ {\mathrm{MWF}}_{\mathrm{Human}\ \mathrm{health}}={\mathrm{MWTP}}_{\mathrm{HH}}\bullet \frac{F}{D}\bullet 365\kern2.25em $$
(3)
$$ {\mathrm{MWF}}_{\mathrm{Social}\ \mathrm{assets}}={\mathrm{MWTP}}_{\mathrm{SA}}\bullet \frac{F}{D}\kern2.25em $$
(4)
$$ \mathrm{M}{\mathrm{WF}}_{\mathrm{Biodiversity}}={\mathrm{MWTP}}_{\mathrm{BD}}\bullet F\kern2em $$
(5)
$$ {\mathrm{MWF}}_{\mathrm{Primary}\ \mathrm{production}}={\mathrm{MWTP}}_{\mathrm{PP}}\bullet F\kern2.25em $$
(6)

where F denotes the global number of households, and D denotes the global population.Footnote 12 Therefore, the monetary value per unit of damage to each safeguard subject, namely, year, USD, species, and 100 million tons, respectively, is obtained as the corresponding MWF in this study.Footnote 13

3 Results and discussion

3.1 Respondents’ perceptions

Table 2 shows respondents’ perceptions with respect to the importance of each safeguard subject. The respondents in G8 countries tend to view biodiversity (67.2%) and primary production (74.3%), which are the constituents of the ecosystem, as more important, rather than human health (49.7%) and social assets (49.0%), which are the constituents of human society. On the other hand, the majority of respondents from the emerging G20 countries perceive all these losses as very important, with the highest priority being accorded to human health (80.2, 69.6, 74.9, and 77.9%, respectively). In addition, variances in responses between the emerging G20 countries are generally larger than the corresponding values for the G8 countries’ respondents.

Table 2 Respondents’ perceptions of safeguard subjects and environmental problems
Full size table

We also asked the respondents to choose what they perceived to be the most serious problem from among the following 10 alternatives: global warming, unsafe drinking water and lack of sanitary facilities, air pollution, insufficient water supply, diminishing natural resources due to the consumption of minerals and fossil fuels, deforestation, land exploitation or development, unemployment, lack of education, and the rights of women and children. Global warming was considered to be the most serious problem of all for the G8 countries’ respondents, followed by unsafe drinking water and lack of sanitary facilities. A relatively large number of the respondents from the other emerging countries chose air pollution as either the most or the second most serious problem.

3.2 Estimation results

3.2.1 Preference intensities

Table 3 displays the estimation results using utility function (1) for the G20, G8, G20 excluding G8, and each of the G20 countries. The number of respondents in the G20, G8, and G20 excluding G8 countries is 6183, 4146, and 2037, respectively. The corresponding log likelihood ratios are 0.47, 0.45, and 0.51, all of which are sufficiently high for the discrete choice model.Footnote 14 , Footnote 15 We assumed that the parameters are distributed normally.The mean and standard deviation values are reported, except for an environmental tax that is set as a numeraire. “Yes” is displayed with its direction in the column “Status quo” if β 0 is statistically significant, and “No” is displayed otherwise.

Table 3 Preference intensities for the four subjects and environmental tax
Full size table

First, the estimated preference intensities of all four subjects are statistically significant and negative at the 1% level, since utility decreases as a result of increased damage. The parameters of environmental tax are also negative and statistically significant at the 1% level in every area except Mexico (listed in italics). All the estimates for the status quo parameter are negative, thus indicating that, on average, the respondents prefer an improvement over the current status in these countries.

3.2.2 Monetary weighing factors

Table 4 shows the MWFs calculated using the results in Table 3 and Eqs. (2) to (6). According to the estimates obtained using the pooled data of all G20 countries, the MWFs are 10,000 USD for human health (per year), 0.77 USD for social assets (per USD of resources), 6.6 billion USD for biodiversity (per species), and 3.1 billion USD for primary production (per 100 million tons). Similarly, 7300 USD for human health, 0.47 USD for social assets, 6.3 billion USD for biodiversity, and 2.8 billion USD for primary production are obtained as the MWFs for the developed G8 countries. All the results are lower than the corresponding values of the emerging countries: 20,000 USD for human health (2.7 times), 1.7 USD for social assets (3.6 times), 7 billion USD for biodiversity (1.1 times), and 3.7 billion USD for primary production (1.3 times). These gaps are particularly prominent in the MWFs of human health and social assets, while a small difference exists between the MWFs of biodiversity and primary production. The results based on the pooled data indicate that the loss of health and social assets are perceived as more urgent and crucial in the emerging countries relative to the developed G8 countries.

Tabl 4 Weighting factors expressed in monetary values
Full size table

Table 4 also shows country-specific MWFs.Footnote 16 For the developed countries, the MWFs of human health range from 2900 USD (USA) to 16,000 USD (Italy), and the corresponding values for the emerging countries are 6700 USD (Russia) to 84,000 USD (Argentina). This result is consistent with the dimensionless weighting factors calculated in part 1 of this paper. The health of residents constitutes one obvious difference between the developed and emerging countries. Respondents from emerging countries tend to suffer poor health because of environmental problems, such as air pollution, as described in Table 2. This constitutes an immediate threat, which, conversely, is already restricted to some extent in developed countries.

The MWFs of social assets range from 0.08 (USA) to 0.95 (Italy) in the developed countries, and from 0.71 (Brazil) to 11 (Argentina) in the emerging countries. This result indicates that the respondents in developed countries evaluate social assets with some discount, while those in emerging countries place more value on social assets than their private income. This viewpoint may arise because, unlike the respondents in emerging countries, those from developed countries perceive social assets to be substitutable by other technologies and international trade in exchange for money.

The MWFs of biodiversity range from 1.7 billion USD (USA) to 11 billion USD (Italy) in developed countries, and from 3.4 billion USD (Brazil) to 19 billion USD (Argentina) in emerging countries. The MWFs of primary production range from 810 million USD (USA) to 4.8 billion USD (Italy) in developed countries, and from 1.9 billion USD (Brazil) to 11 billion USD (Argentina) in emerging countries. In contrast to the dimensionless weighting factors calculated in part 1 of this paper, two subjects related to the ecosystem are negatively, although weakly, correlated with individual income levels.Footnote 17

3.2.3 Cross-country heterogeneity

Regarding the standard deviations, there is considerably greater variation in the emerging countries for all subjects. These variations are approximately 6 times (human health), 12 times (social assets), and more than double (biodiversity and primary production) the corresponding values in developed countries. For two subjects related to human society, the variations in the MWFs are quite small within the developed countries but considerably large among the emerging countries. This finding indicates that while the MWFs related to human society, namely, human health and social assets, are internationally transferable within developed countries, caution must be exercised while considering such transfers among emerging countries or between developed and emerging countries. Moreover, the variations in the MWFs of biodiversity and primary production, which are related to the ecosystem, are considerably large in both developed and emerging countries. Thus, these MWFs must be used judiciously when considering international transfers even within developed countries.

3.2.4 Population-weighted MWFs

As described above, the MWFs estimated using the pooled data are useful for an overview of the data; nevertheless, they are highly influenced by the composition of the sample. Table 5 shows the population-weighted averages of the estimated factors of each G20 country after adjustment according to the current population statistics (World Bank 2013). For MWFs based on environmental tax, each population-weighted average of the MWFs (23,000 USD for human health, 2.5 USD for social assets, 11 billion USD for biodiversity, and 5.6 billion USD for primary production) is larger than the results obtained using the simply pooled data by giving more weightage to the preferences of respondents from the emerging countries. In addition, the differences between the values of the developed and emerging countries become larger (5200 USD, 0.43 USD, 4.4 billion USD, and 2.1 billion USD in developed countries, and 27,000 USD, 2.9 USD, 13 billion USD, and 6.3 billion USD in emerging countries, respectively) when considering weighted MWFs. Figure 2a, b shows the representative values of this study, which are global-scale and country-specific MWFs related to human society and the ecosystem, respectively.

Table 5 Summary of the population-weighted monetary weighting factors obtained in this study
Full size table
Fig. 2
figure 2

a Global-scale and country-specific MWFs related to human society. b Global-scale and country-specific MWFs related to the ecosystem

Full size image

3.3 Comparison with previous studies and validity of our results

3.3.1 Results of Japanese respondents in LIME2

In order to compare this study’s results with those of LIME2 (Itsubo et al. 2012), Table 6 shows the reference values and MWFs of the two studies. First, note that the questionnaire used for the current study provided the respondents with the reference values of damages on a global scale, while the national reference values for Japan were provided in the LIME2 questionnaire. The global reference values reflect damages that have occurred at locations relatively far away from the respondents, while the national reference values reflect the damages (half of which originated in Japan in LIME2) that have occurred closer to home (i.e., Japan). The reference values corresponding to each damage event are 4 days per person annually (double that in LIME2), 60 USD per person annually (less than one-third of 200 USD in LIME2), 100 species in the world annually (100 times the corresponding value in LIME2), and 20 billion tons in the world annually (100 times the corresponding value in LIME2).

Table 6 Comparison between LIME2 and LIME3—the results of Japanese respondents
Full size table

Next, because the parameter of social assets—another monetary parameter—was set as the numeraire in LIME2 and the basic statistics (i.e., population) used are at the national scale, we recalculated the global-scale MWFs defined above based on the results of LIME2 so that we may compare the MWFs estimated using the data of the Japanese respondents in our study. The values are generally less than those of LIME2, except for social assets. For MWFs based on environmental tax, the MWFs are 3300 USD a day per person (three-quarters of 4400 USD), 0.49 USD for 1 USD worth per person (differs by one order of magnitude from 0.03 USD), 4.1 billion USD for one species (differs by two orders of magnitude from 240 billion USD), and 2.2 billion USD for 100 million tons (differs by one order of magnitude from 77 billion USD).

For the loss of human health, the world and national averages were presented to the respondents in our questionnaire and for LIME2, respectively. Our pre-survey and interview data obtained before the main large-scale survey indicated that Japanese respondents perceived their health risks to be lower than the world average. This fact led to lower MWFs in our study even though they evaluated the same subjects. Regarding the loss of social assets, the MWFs of LIME2 are significantly small (0.03 USD). Note that the subjects evaluated are inherently different, namely, the loss of resources in the world in the current study and the loss of resources in Japan alone in LIME2. This fact indicates that Japanese respondents probably place more value on the global-scale loss of mineral and fossil resources. This interpretation is consistent with the fact that Japan lacks such resources and relies heavily on resources imported from other countries. Another possible reason of the noticeable difference between the MWFs of social assets is the growing interest in mineral and fossil resources in Japan from 2005 to 2013.Footnote 18

Regarding the loss of biodiversity and primary production, note that the respondents of the two studies were provided different information pertaining to these damages. The Japanese respondents evaluated the global-scale environmental damage in the current study and domestic environmental damage in LIME2. The trend that the respondents are willing to pay more for local environmental conservation than for conservation in foreign countries has also been observed in a previous study on environmental economic valuation (Lindhjem et al. 2011, Tsurumi et al. 2015). For the respondents, WTP for biodiversity differs according to the location of the problem. In addition, the MWFs at the national scale, which are the representative values of LIME2, appear in the last four rows for reference.Footnote 19

3.3.2 Previous economic valuation studies

The results of previous studies on the economic evaluation of the four subjects except for social assets, which have no comparative values, are listed in Table 7. For consistency, all the figures have been converted to show the values for 1 year’s loss of life for human health, one species for biodiversity, and 100 million tons for primary production.

Table 7 Economic values related to losses in the four subjects
Full size table

Value of statistical life (VSL) estimates are widely used by many US and European agencies, such as US Department for Transportation (DOT), the US Environmental Protection Agency (EPA), UK Department for Transportation (DFT), and European Commission, in policy assessments pertaining to mortality risk reduction. For comparison, these values are converted into Value of Statistical Life Year (VSLY), given byFootnote 20

$$ \mathrm{VSLY}=\frac{\mathrm{VSL}}{\left(1-{\left(1+ r\right)}^{- L}\right)/ r} $$
(7)

where L is the remaining lifetime (assumed to be 40 in Table 7), and r is the rate at which the utility is discounted (assumed to be 0 in Table 7). Though it is inappropriate to make a simple comparison with the results of our study (because the values of VSL employed by these official agencies are mainly based on wage risk and road safety studies using private market data), we observe that these values are larger than the results of our study.

VSL estimates from various studies show a wide range, from 7292 USD to 86.3 million USD, with median 2.4 million USD, depending on the risk valuation context such as cause of death (e.g., traffic accident, wage risk, cancer, other illness, and environment-related), immediate or future risk, change in public or private risk, and small or large risk reduction (Lindhjem et al. 2011).Footnote 21 The information regarding loss of health in our questionnaire can be specified as environment-related, latent, public risk, and small change in risk (4 days annually), and all these contexts cause smaller mortality values (Arberini and Scasny 2011; Lindhjem et al. 2011). For other LCA methodologies, 100,000 USD per year of lost life (YOLL), adopted in EPS derived from previous VSL estimates, is also reasonably larger than the result of our study.Footnote 22 NEEDS, which is proximate to the subject evaluated in our study in terms of methodology and the definition of mortality, employs 48,000 USD as the mean value of the increase in life expectancy.Footnote 23 In the context of public programs using interview data, our study and NEEDS share the same idea of life expectancy evaluation, but the scales of such impacts differ (national in NEEDS and global in LIME), leading to smaller values in our study. This interpretation is consistent with our result showing smaller values for developed countries, where the respondents have a relatively lower baseline risk perception than those in emerging countries.

WTP for avoiding extinction of a species has been estimated by several studies. Table 7 shows the comparable values calculated based on the results of previous studies and world statistics. While these studies commonly evaluated accessible local species such as water fringe, cheirotonus jambar, and the iriomote wild cat (Mitani and Kuriyama 2005; Yoshida 2013; MOEJ 2014, respectively), we evaluate abstract species present in locations relatively far away from the respondents and on the global scale. This difference in approach leads to somewhat smaller values for our study.

For primary production, TEEB (2009) estimates the potential benefits of several biomes, including forests, based on the results of 104 case studies (507 values). Assuming average yearly plant growth per hectare as 10 ton, the comparable values can be calculated as shown in Table 7. As is the case with biodiversity, the respondents in the studies commonly evaluated biomes close to their respective habitations, thus explaining the relatively smaller values in our study.

The MWFs of our study are generally smaller as compared to the previous studies. This is because our questionnaire informed the respondents about the global-scale damages. Although the gaps are probably underestimated in each subject (sometimes an order of a magnitude smaller), these gaps are still regarded to be within the reasonable range. We obtained reasonable and conservative values as MWFs for the G20 countries.

4 Further discussions

Since the weighting factors derived here represent the economic value of one unit of damage, the economic value of the environmental impact related to each safeguard subject can be derived by multiplying its relevant reference value and its MWF. Therefore, the world total economic value (WTEV) of the annual environmental damage caused by human activity can be obtained by Eq. (8):

$$ \begin{array}{cc}\hfill {\mathrm{WTEV}}_{\mathrm{Impact},\kern0.75em \mathrm{annual}}=\sum_s{\mathrm{RV}}_s\cdot {\mathrm{MWF}}_s\hfill & \hfill \left( s=\mathrm{HH},\mathrm{SA},\mathrm{BD},\mathrm{PP}\right)\hfill \end{array} $$
(8)

where RV s is the reference value of each safeguard subject (Itsubo et al. 2015). The results are presented in Table 8. The WTEV of annual impacts was estimated to be approximately 5.1 trillion USD, which is equal to 6.7% of world’s total GDP in 2013. In addition, using the MWFSAs, we arrive at a result of approximately 2.5 trillion USD, which is equal to 3.3% of the world’s total GDP.

Table 8 Annual global damage cost
Full size table

Compared with the results of the Stern Review (Stern 2007), which estimates the economic damage due to climate change (2.5 to 10 trillion USD), and the findings of TEEB (2009) (several trillion USD), which estimates the economic damage accruing from the loss of an ecosystem service, our result is reasonable although it appears to be an underestimate for the following reasons. First, the reference value of social assets (4.49E−11) does not contain of the impacts of global warming because of its undeveloped functional model and lack of data. We should include the loss of social assets (such as damages arising from agricultural, forestry, and fishery production, and energy consumption) and land loss due to floods (all of which are considered in assessment reports such as that of the IPCC) in future estimations. In addition, currently, the loss of human health due to toxic releases and the loss of primary production due to acidification are also excluded from our reference values. The reference value regarding the loss of biodiversity is also underestimated in our study because we exclude damages other than that caused to vascular plants and the impact of soil erosion due to mining in surrounding areas.Footnote 24

5 Conclusions

This study calculated the MWFs for the G20 countries by conducting a large-scale survey using a uniform questionnaire. Random sampling was implemented, and a total of 6400 responses were obtained. In all the areas surveyed, the results were statistically significant at the 1% level, with the exception of monetary attributes for Mexico. Using the population-weighted values of each G20 community provided the following estimates of the MWFs: 23,000 USD for human health (per year), 2.5 USD for social assets (per USD of resources), 11 billion USD for biodiversity (per species), and 5.6 billion USD for primary production (per 100 million tons). The differences between the results of the developed and emerging G20 countries are considerably large, with the values generally being smaller in the former in purchasing power parity (USD) terms. The significant variations in these estimated results indicate that caution should be exercised while transferring the MWFs of biodiversity and primary production (which are related to the ecosystem) internationally or even within developed countries. However, the MWFs of the subjects related to human society (human health and social assets), which show minor variations among the G8 countries, may be transferred within developed countries.

Our results indicate that the WTEV of annual impacts is approximately 5.1 trillion USD, which is equal to 6.7% of the world’s total GDP and comparable to the findings of the Stern Review (Stern, 2007) and TEEB (2009), which calculated global economic damage due to climate change and loss of ecosystem service, respectively. While the main goal of this study is not to examine absolute impacts, rather impacts relative to some defined unit, we obtained reasonable MWFs on a global scale compared with previous studies and provide useful evidence of the difference in MWFs between developed and emerging countries.

The reference values in this study reflect the damages that have occurred on a global scale, and the evaluators are people in various different regions (e.g., 19 countries from the G20). Therefore, the MWFs of our study refer to the willingness to pay for a marginal reduction in the global environmental impact for different geographical scales (e.g., G20, G8, G20 excluding the G8 states, and each country). From this aspect, country-specific MWFs could be different from other national weighting factors, reflecting the damages that have occurred closer to home. This is a topic for further research and could be of significant help in conducting a cost–benefit analysis using LCA. For an LCA practitioner, these MWFs can be chosen depending on the aims of the application. In general, the reliability of country-specific weighting factors is higher than those at the global scale. If LCA users wish to share the calculated results internationally (e.g., among the member countries of the G8), it would be useful to know the weighting factors of the group countries (e.g., the MWF of the G8). If users share the results domestically, the country-specific MWFs would be helpful; however, it must be noted that the values reflect damages incurred on a global scale.

One of the biggest advantages of direct weighting in the LIME method is that unlike other integrated models, the data-generating process can be fully controlled across different countries/regions if the questionnaire is well designed. This means that with the LIME methods, we can directly compare individual behaviors in different countries and investigate how country-specific attributes (such as living standards or regulations) affect their WTP for each safeguard subject. Establishing cross-country heterogeneity allows us to formally link local estimates with global ones. Additionally, cross-country studies help extrapolate future value developments from the current estimates, especially when country-specific attributes are expected to change over time.