The European Commission Organisation Environmental Footprint method: comparison with other methods, and rationales for key requirements

As a first observation, it was noted that the reviewed guidance documents for measuring the environmental performance of organisations appear to be considerably less prescriptive than comparable guidance documents for product-level assessments. This may be because product-level standards tend to take the ISO 14044 standard (ISO 2006a) as a starting point, whereas no similar, internationally accepted methodological norm for organisation-level life cycle assessment is available (ISO is currently developing ISO/NP TS 14072 Life cycle assessment—Additional requirements and guidelines for organizations). It may also reflect that life cycle-based organisation environmental footprinting is potentially more challenging than is product-level footprinting—in particular for large organisations producing multiple products and/or operating in multiple sectors.

In general, it appears that most existing methods are intended to support internal use or high-level corporate sustainability reporting rather than to provide a robust and detailed description of environmental performance at an organisational level. Hence, such guidance documents tend to provide recommendations rather than requirements, and often present a range of methodological alternatives instead of fixing a single norm for a given decision point.

To ensure that EC OEF studies are conducted in a consistent and reproducible manner (third core criterion), the EC OEF method must necessarily be highly prescriptive. This sentiment is echoed by Moneva et al. (2006), who observe that “the lack of … restrictive conditions such as a clear definition of the entity boundaries, the development/requirement of integrated indicators or the attachment of an independent verification statement leads to a relaxation of the basic aim, that is, sustainability.”

Few of the reviewed organisation environmental footprint methods require coverage of the entire life cycle with respect to inclusion of supply chain stages associated with organisation activities. Notable exceptions are the Bilan Carbone method and the GHG Protocol Scope 3 Standard. Because the first criterion for a common EC OEF method is that it adopts an approach covering the entire life cycle, such a common method must necessarily diverge from most of the reviewed standards in this respect. However, many of the reviewed methods do provide for optional “Scope 3” life cycle modelling, and hence are complementary to the EC OEF method in this respect when Scope 3-level analyses are undertaken (Table 1).

Of the reviewed methods, only GRI stipulates that a broad suite of relevant environmental impacts must be considered. All other methods refer to single impact categories only. To satisfy criterion 2, the EC OEF method therefore necessarily diverges from the reviewed methods in this respect (Table 2; see Section 3.8 for the required suite of environmental footprint impact assessment models for EC OEF studies).

In product-oriented environmental footprinting analyses, it is standard to define a functional unit and a reference flow. The functional unit, also termed the unit of analysis (as adopted for the EC OEF method), is a quantitative description of the performance of the product system in terms of “what, how much, how well, and for how long” (EC JRC IES 2010a). The reference flow is the amount of product necessary to provide the defined functional unit. Taken together, these serve as the basis for precisely specifying the system to be analysed. Although not employed in any of the reviewed methods (Table 3), these concepts are similarly useful and, indeed, necessary for life cycle-based organisation environmental footprint accountancy. This is because (if appropriately implemented) they firmly ground the analysis in concrete, physical relationships that connect upstream, organisation-level, and downstream processes in a coherent, internally consistent manner. Ensuring such internal consistency is essential for arriving at realistic, physically representative organisation environmental footprint studies. For this reason, the EC OEF method departs from common practice (as per the reviewed guidance documents) by incorporating the concepts of functional unit and reference flow as central elements of organisation environmental footprint studies.

The unit of analysis (functional unit) for an OEF study could, however, potentially be defined in several ways, some of which are incompatible with the defined criteria for the EC OEF method. For example, the unit of analysis might be defined as simply “the organisation within its organisational (site) boundaries”. Here, only a limited subset of “what” considerations are accommodated, while the “how much, how well, and for how long” dimensions are ignored. In this case, the calculated footprint would refer only to the infrastructure of the organisation and would not include site-level activities nor the goods/services provided. Because this approach is not life cycle-based, nor does it allow for a realistic, physically representative OEF, it is insufficient according to our first and fourth criteria.

A second option could be to define the unit of analysis as the organisation, including organisational activities, occurring within the site boundaries during a 1-year time interval. This approach accommodates a subset of both the quantitative and temporal dimensions. Similar to a “Scope 1” analysis, as commonly defined in the context of the reviewed GHG accounting methods, such an approach would effectively limit the assessment to a gate-to-gate analysis instead of a cradle-to-grave analysis, which is likewise unacceptable in light of the first and fourth core criteria.

In order to satisfy these criteria using the concepts of functional unit and reference flow, our analysis points towards the usefulness of understanding the function of the organisation to be primarily the provision of goods and/or services (i.e. products). In this way, the defined unit of analysis (the organisation and, for the purpose of downstream modelling, the functions of its products) provides the necessary basis for linking upstream, site-level, and downstream processes. Indeed, downstream modelling for organisations only becomes intelligible at the product level. Accordingly, the third (required) possible definition for the unit of analysis is the organisation, as product provider (i.e. taking into account the life cycle impacts of the products provided) over a 1-year time interval. On this basis, the reference flow for the analysis (what we term the “Product Portfolio”) is the total amount of products provided by the organisation over the reporting interval. This definition provides space for accommodating all of the necessary components of the functional unit (i.e. what, how much, how well, and for how long), and also satisfies all four of the necessary criteria.

As defined by ISO 14044 (ISO 2006a), the system boundaries for life cycle assessment are “a set of criteria specifying which unit processes are part of a product system and thus determine which processes shall be included within the LCA”. In the context of an OEF study, two levels of system boundary definition are necessary. These are, first, the organisation boundaries and, second, the life cycle (supply chain) boundaries. We refer to the latter as “Organisation Environmental Footprint boundaries”. This section focuses on the definition of the organisation boundaries, while the next (3.6) focuses on the Organisation Environmental Footprint boundaries.

Defining organisation boundaries requires the identification of processes/activities directly attributable to the organisation (as a unit of analysis) versus indirect processes/activities that are linked to organisation activities either upstream or downstream along the supply chain (i.e. indirectly attributable activities). Moneva et al. (2006) observe that “the elaboration of specified guidelines defining the reporting entity boundaries is a complex challenge and has become an absolute necessity”. We agree that providing a consistent basis for defining organisation boundaries is necessary to satisfy the core criteria for a common EC OEF method.

Most of the reviewed guidance documents for organisation environmental footprint studies differentiate between three approaches to defining organisation boundaries (Table 4). First is the equity share approach, where organisation boundaries encompass all activities in which there is an ownership share. Responsibility for associated environmental burdens is subsequently assigned in direct proportion to the ownership share in the activities of concern. Second is the financial control approach, where organisations include within their defined boundaries only those activities over which they have financial control. Third is the operational control approach, where only those activities over which an organisation has operational control are included in the defined organisation boundaries. For further details on these approaches see, for example, WRI and WBCSD (2004). Clearly, in order to achieve increased comparability and reproducibility (criterion 3), the EC OEF method must necessarily deviate from existing guidance by specifying a single basis for defining organisation boundaries.

As EC OEF studies will be, a priori, life cycle-based (Criterion 1), meaning that all activities in the supply chains linked to organisation activities need be accounted for, there is a certain degree of arbitrariness as to how/where organisation boundaries are drawn. The same activities will be included in an EC OEF study as in a study using, for example, the GHG Protocol Scope 3 Standard or similar methods. However, taking into account criteria 3 and 4, we identify several reasons in support of a single, preferred approach.

First, we identify the “control” approach as preferable to the “equity share” approach. As explicitly recognised in existing guidance documents such as ISO WD/TR 14069 and the GHG Protocol Corporate Accounting and Reporting Standard (WRI and WBCSD 2004), the equity share approach is better suited to financial risk management, whereas the control approach is better suited to environmental performance measurement and management. This is because there is greater potential to make management changes in response to insights derived from environmental footprint studies where the organisation has direct influence over the facilities in question—whether via financial or operational control. This is a key intended application of EC OEF studies. More important, however, is that the equity share approach requires modelling all facilities/activities/processes in which the organisation has ownership share, then subsequently attributes burdens in proportion to ownership share. It therefore conflicts with criterion 4 because the attribution of burdens is based on financial considerations rather than on logical, physical cause–effect considerations. In contrast, the control approach defines boundaries at the facility/activity/process level, and allows that burdens be attributed on a physical rather than a financial basis. Hence, the latter allows for constructing an internally consistent life cycle model that quantifies and attributes burdens in direct relation to the material and energy flows linking upstream, organisation-level, and downstream activities, whereas the former does not.

Defining organisation boundaries is useful from the perspective of describing directly versus indirectly attributable life cycle impacts associated with the activities of an organisation. However, accounting solely for the impacts of activities occurring within the defined organisation boundary does not provide a sufficient representation of an organisation’s environmental impact (Moneva et al. 2006; Schaltegger and Burrit 2010). Instead, life cycle-based environmental accountancy requires an expanded definition of system boundaries that encompasses all relevant supply chain stages (Moneva et al. 2006). Among the methods reviewed, almost all define system boundaries for an analysis in terms of the “3 Scopes” distinction, as developed by the GHG Protocol (WRI and WBCSD 2004; Table 5).

In light of the widespread adoption of the 3 Scopes distinction, there is clearly appeal in similarly employing this concept in the EC OEF method. However, we identify several compelling reasons which suggest an alternative basis for defining system boundaries is preferable for the intended purposes.

First, neither Scope 1 nor Scope 2 analyses are, by definition, life cycle-based, hence they cannot be accommodated in the EC OEF method. Second, the difference between “Scope 2” and “Scope 3” has meaning in the context of non-life cycle-based GHG accounting only, where it may be desirable to include upstream emissions related to the provision of the utilized energy carriers (for example, electricity, steam, heating and cooling energy) in addition to direct “Scope 1” emissions for a selectively expanded gate-to-gate analysis. However, this is not relevant in the context of life cycle-based, multi-criteria environmental accounting studies because “Scope 2” does not provide a useful differentiation for impact categories other than GHG emissions. Finally, although it might be feasible to incorporate a modified conception of the 3 Scopes concept into the EC OEF method, it is likely that such a modified conception would bring more confusion than clarity to both GHG accounting and environmental footprinting. For these reasons, we suggest that it is more useful to employ system boundary terminology that is consistent with life cycle assessment practise rather than (contemporary) GHG accounting practice.

We hence propose adoption of the terms site-level, upstream (i.e. cradle-to-gate) and downstream (gate-to-grave) impacts as a basis for distinguishing between directly and indirectly attributable activities and impacts (Fig. 2). To fulfil the first core criterion (life cycle approach), the organisation environmental footprint boundaries in the EC OEF method should therefore be defined as including all upstream and downstream processes/activities linked to the organisation’s Product Portfolio. This is similar to the GHG Protocol Scope 3 requirement.

Although the EC OEF method is, a priori, life-cycle based, which requires cradle-to-grave environmental accountancy, it should be recognised that, in certain situations, such accountancy may simply not be feasible. For example, where an organisation produces intermediate products of indeterminate fate, it may simply not be possible to arrive at concrete, cradle-to-grave models. Here, the elaboration of sector-specific rules in complement to the EC OEF method will be necessary in order to precisely define use and EoL scenarios as appropriate to the sector and product category, in order to ensure comparability between studies. In the absence of such guidance, cradle-to-gate analyses may, in some circumstances, be defensible.

Following Sangwon et al. (2004), “considering the complex interdependence of processes in modern economies, it would be fair to assume that in general all processes are directly or indirectly connected.” Establishing system boundaries for analyses of OEFs in a systematic manner is therefore challenging. The appropriateness of system boundaries will ultimately be influenced by the completeness of the generic data sets used to represent processes upstream and downstream of the organisation boundary (see Section 3.13 on data quality requirements). Once such boundaries are established, it is also necessary to treat processes within the system boundaries in a similarly systematic manner. In principle, it is desirable that all activities/processes that are attributable to the analysed system within the defined system boundaries are considered in OEF studies. In practice, however, not all of these activities/processes may be environmentally significant, nor may it be possible to acquire the data necessary to include these in the OEF model. For these reasons, cut-off criteria are often established in environmental accountancy models. They provide thresholds for inclusion of environmentally significant processes and flows, in the interest of balancing returns on effort and analytical robustness (Sangwon et al. 2004).

Among the reviewed methods, no common approach to establishing cut-off criteria was identifiable (Table 6). While some methods discourage their use, others specify, at a general level, the basis for determining cut-off criteria. None of the methods presented clearly defined cut-off criteria, nor disallowed their use.

Elsewhere in environmental accountancy, cut-off criteria are sometimes defined in terms of a proportion of energy or mass flows at the inventory level, or as a specified maximum contribution of the cut-off flows to the impact categories considered (for example, see EC JRC IES 2010a; ISO 2006a; and Sangwon et al. 2004). Because the EC OEF method does not provide recommended methods for energy or mass flow analysis, neither of these options are available as potential cut-off criteria (see subsequent section on impact assessment methods).

For the purpose of the EC OEF method, if cut-offs were to be entertained then they must necessarily be identified on a systematic basis, so as to ensure reproducibility/consistency among studies (criterion 3), and on the basis of environmental impacts in order to ensure that no environmentally relevant flows are cut off (criterion 4). The option of defining cut-off criteria based on a predetermined level of contribution to impacts appears, at first instance, feasible in the context of the EC OEF method because required impact assessment methods are specified. However, we foresee that some non-trivial, practical problems would arise due to the multi-criteria nature of the EC OEF method, which render this option unattractive. While it would seem straightforward to determine flows to be cut off on the basis of impact contributions for a single criteria assessment (for example, if only GHG emissions are considered), this will not be the case where multiple impact categories are considered. Flows will contribute differently to each impact category, meaning that different flows will need to be cut off. This would effectively require determining cut-offs for each impact category relative to the threshold, with parallel final impact assessments and interpretations in each case.

Moreover, in order to determine what may be cut off, the whole system must be first modelled in some way (be it based on a kind of screening); hence, potentially cut-off flows might as well anyway be included in some form in the analysis. In light of these considerations, it was decided that the EC OEF method will disallow the use of cut-offs. It should be noted, however, that the recommended data quality provisions effectively allow for the use of “proxy data” in place of cut-offs (see Section 3.13).

Impact categories refer to specific categories of environmental impacts considered in an environmental accountancy study. These are generally related to resource use (for example, fossil fuels and mineral ores) or emissions of environmentally problematic substances, such as GHGs or toxic chemicals. Environmental footprint impact assessment methods use models for quantifying the causal relationships between the material/energy inputs and emissions associated with organisation activities (inventoried in the Resource Use and Emissions Profile) and each environmental footprint impact category considered. Each impact category hence has an associated, stand-alone environmental footprint impact assessment model and impact indicator.

The purpose of environmental footprint impact assessment is to group and aggregate the collected inventory data (Resource Use and Emissions Profile) according to the respective contributions to each impact category. This subsequently provides the necessary basis for interpretation of the environmental footprint results relative to the goals of the footprint study (for example, identification of supply chain “hot spots” and options for improvement). The selection of environmental footprint impact categories must therefore be comprehensive in the sense that they cover all relevant environmental issues related to the activities of the organisation of interest.

All reviewed methods other than the Global Reporting Initiative provide for single criterion environmental assessments—most commonly, organisation GHG accounting but also including water footprinting (Table 7). This single-criterion approach is clearly in conflict with both the first and second core criteria for the EC OEF method. For many of the GHG accounting methods, which require accounting for the six “Kyoto” GHGs only, the fourth criterion is also compromised. GRI, in contrast, stipulates that all relevant impacts (social, economic, and environmental) should be considered. However, supporting models are provided for only a limited subset of these impacts (contradicting criterion 3).

A broad suite of impact assessment categories and supporting models are available for potential use in the EC OEF method. In order to meet criteria 2 (multi-criteria assessment), 3 (reproducibility/consistency), and 4 (maximise physical representativeness), the EC OEF method must specify: (1) a required suite of relevant environmental impact categories and (2) the best available impact assessment models to employ. The International Reference Life Cycle Data System (ILCD) includes a suitable, systematic review of available methods and supporting models as a basis for its recommendations (EC JRC IES 2011a). It was therefore decided that the EC OEF method will adopt the suite of recommended mid-point methods (because of its lower level of uncertainty than endpoint methods) from the ILCD Handbook (Table 7). These must necessarily be updated over time to reflect evolving best practices in this field. For details regarding the requirement methods, models, category indicators and sources, see EC JRC IES 2011a.

It should be noted that in many contexts, data availability may currently be insufficient to support application of all 14 of the specified impact categories. This applies, in particular, to modelling abiotic resource depletion—water, land use, and toxicity impacts.

Specific data (also referred to as primary data) refer to directly measured or collected data representative of activities at a specific facility or set of facilities. These should be used for as many processes and activities as possible so as to best reflect the physical reality of organisation activities (core criterion 4). Requirements among the guidance documents reviewed regarding specific data differed widely (Table 8). ISO 14064 and ISO WD/TR 14069, Bilan Carbone, and DEFRA require specific data for activities with in the defined organisation boundary, whereas the GHG Protocol Corporate and Scope 3 Standards suggest that the choice of specific versus generic data should reflect the goals of the study. CDP Water Disclosure and GDI provide no specifications at all in this regard.

To best reflect physical reality, the collection of specific data is mandatory in the EC OEF method for all processes/activities over which the organisation has financial and/or ownership control (and, hence, access to the necessary data) and for background processes/activities where appropriate. In some exceptional cases, generic (secondary) data may be more representative for a certain activity/process within or outside of the organisation boundary. In these cases, the EC OEF method requires that any such generic data be used—for example, where supply chains are not fixed because certain inputs are sourced from multiple suppliers in response to changing market prices.

Generic data (also referred to as secondary data) refer to data that are not directly collected, measured, or estimated, but rather sourced from a third-party life cycle inventory database or other source. Most often, specific data will not be available for upstream and/or downstream processes, hence generic data will be used. Requirements among the guidance documents reviewed regarding generic data differed widely (Table 9).

The EC OEF method necessarily permits use of generic data for processes/activities outside of the organisation boundary, as well as in exceptional instances where specific data are unavailable for processes/activities within the organisation boundary. In all cases, the most representative data must be sourced (core criterion 4). For example, sector-specific generic data are preferred over multi-sector generic data. To further improve consistency (core criterion 3), preferred sources for generic data are provided in the EC OEF method guidance document. All such data must fulfil the minimum stipulated data quality requirements (see Section 3.13).

Allocation problems arise wherever input flows and resultant emissions need to be partitioned among co-products or co-processes. Such problems are characteristic of several distinct situations that may potentially arise in OEF studies. If a process or facility that is under the financial and/or operational control of the reporting organisation provides several products, and only a subset of these are attributable to the organisations’ defined Product Portfolio, then allocation will need to be performed between these goods/services. This includes both (a) between products produced independently of each other and (b) between co-produced products. In these situations, all inputs and emissions linked to the process must be partitioned between the product of interest and the other co-products in a principled manner in order to ensure both reproducibility/consistency (criterion 3) and physical representativeness (criterion 4). However, a variety of competing bases may be potentially applied to resolve allocation problems.

The ISO 14044 standard (ISO 2006a) provides the most widely recognised allocation hierarchy. However, there is no convergence between the reviewed methods as to a preferred allocation hierarchy (Table 10), and several depart from ISO 14044. For example, economic allocation is disallowed in the Bilan Carbone standard in order to increase physical representativeness (ADEME 2007), while system expansion is excluded from the GHG Protocol Scope 3 Standard (WRI and WBCSD 2011). Nor does the ISO hierarchy provide a sufficiently detailed basis for identifying appropriate allocation solutions. In particular, the wording of the ISO hierarchy is vague, hence allowing for alternative interpretations. A review of relevant LCA literature suggests that competing interpretations are the norm (Ekvall and Finnveden 2001; Guinee et al. 2002; Brander and Wylie 2011; Pelletier and Tyedmers 2011; Wardenaar et al. 2012). It was therefore deemed necessary that the EC OEF method articulate more precisely what is meant by each step in the adopted multi-functionality hierarchy (criterion 3), and also that the preferred allocation solutions be those that result in physically representative modelling outcomes (criterion 4). However, it is recognised that further prescription (i.e. fixed allocation scenarios and factors) at the level of sectorial guidance documents will be necessary in the future to further increase the extent to which both of these criteria are achieved.

Taking ISO 14044 as a starting point, the EC OEF method requires application of the allocation hierarchy presented in Table 10. Similar to ISO 14044, sub-division or system expansion represent preferred solutions to (avoiding) allocation problems. We clarify, however, that expanding the system to include the functions of the co-products refers here to the calculation and reporting of impacts at the level of all co-products (i.e. at the system level). Subsequent steps of modelling alternative co-product supply chains and subtracting the inventories in order to isolate the inventory attributable to the co-product of interest (i.e. substitution) is explicitly excluded from this step of the hierarchy (instead, substitution is defined as two specific applications of allocation—see below). This is clearly preferable according to criterion 4 because the potential for the use of arbitrary or poorly representative substitution scenarios as preferred solutions is reduced.

Also in line with ISO 14044, the second tier of the adopted allocation hierarchy requires the identification of an underlying physical relationship as the basis for partitioning input/output flows. Here also, we provide further specification/clarification to that provided by ISO 14044 in two ways. First, we identify substitution based on a direct, empirically demonstrable relationship as a possible form of physical allocation. An example of such a substitution is when manure nitrogen is applied to agricultural land, directly substituting for an equivalent amount of the specific fertilizer nitrogen that the farmer would otherwise have applied. Here, the system is expanded to include production of the specific nitrogen fertilizer, and the animal husbandry system from which the manure is derived is credited for the substituted fertilizer production. This is distinct from assumed substitutions “at the margin” based on hypothetical market models, as is typical of much consequential LCA modelling. Secondly, we add the term “relevant” as an important and necessary descriptor for an underlying physical relationship. In this context, the relationship should be relevant either in the sense that it reflects how input flows determine the proportions of output flows, or in terms of how specific characteristics of the input flows relate to the functions provided by the co-products.

The third tier of our adopted allocation hierarchy, again similar to ISO 14044, refers to allocation based on some other relevant relationship. As the last tier of the hierarchy, these “last resort” solutions should only be applied when it is not possible to identify a tier one or tier two solution. We include here allocation based on hypothetical substitution scenarios. Another example frequently invoked in relation to this tier is economic allocation, although such a solution is in clear contradiction with criterion 4 as it typically results in model outcomes that reflect market relationships rather than physical relationships (Pelletier and Tyedmers 2011, 2012).

Dealing with multi-functionality of products is particularly challenging when recycling or energy recovery of one (or more) of these products is involved. Several approaches exist and are currently applied for solving multi-functionality problems at EoL. Only a few of the reviewed methods provide recommendations in this respect, and these are varied (Table 11).

A detailed description of and comparison with the approaches analysed for the development of the EoL formula for the EC OEF method is provided in Allacker et al. (2013). Briefly summarised, three common approaches can be distinguished:

For the EC OEF method, the choice of approach was based on the following (non-exhaustive) considerations, in particular to ensure alignment with criterion 1 (coverage of all life cycle stages) and criterion 4 (maximise physical representativeness):

Although most of the methods evaluated do not specify minimum data quality requirements, some provide recommendations and guidance (Table 12). For example, ISO14064-3 stipulates a process for validation and verification of reported GHG emissions data, depending on the level of assurance required by the user of the data/intended application. Assessing the completeness, consistency, accuracy, and transparency of the provided information is required, with the objective to determine whether errors, omissions, and misinterpretations could affect the user’s decisions.

The issue of data quality is particularly important with respect to criteria 3 (reproducibility/comparability) and 4 (physical representativeness). As a result, the EC OEF method necessarily goes further than the reviewed methodologies in order to ensure that data quality objectives are met.

We identified three related data quality issues that must be considered, and for which we deemed methodological specification to be necessary:

To meet the criterion of reproducibility and comparability (criterion 3), it is insufficient to simply derive and report data quality indicators via a data quality assessment. This is because, despite that transparency with respect to data quality is increased, deriving conclusions on the basis of poor and/or uncertain data remains possible. Instead, these steps must be complemented with minimum data quality requirements. Towards this end, the EC OEF method establishes a mandatory, semi-quantitative data quality assessment scheme, with defined quality criteria with respect to: technological representativeness; geographical representativeness; temporal representativeness; completeness; parameter uncertainty; and methodological appropriateness and consistency. Additional documentation, nomenclature, and review requirements are also specified (European Commission 2010b).

While the goal is total reproducibility/comparability between EC OEF studies, which requires parity in data quality, in reality achieving perfect data representativeness will rarely be achievable. In order to balance feasibility and robustness, the EC OEF method takes a stepwise approach to data quality requirements. Specifically, “good quality” data (as determined via a semi-quantitative data quality assessment following the defined criteria) are required for data representing at least 70 % of the environmental impacts caused by the organisation (in each environmental impact category considered). Lower quality requirements (as assessed based on expert judgement) apply for the data representing activities responsible for the remaining 30 % of environmental impacts (of which 10 % may be satisfied using “proxy data” in cases of data gaps). A detailed assessment scheme is provided.