<< Previous Page | Page 6 of 17 | Next Page >>

6. EXTERNAL COST ASSESSMENT

This Chapter seeks to highlight the difficulties faced in seeking to estimate even a range of external costs which might be associated with different waste management options. It is by no means complete (so it is rather similar to studies considered in the previous Chapter in this respect). What we are seeking to do is to open up the discussion about how accurately we can really know what the external costs of different waste management options might be. Given that this is an oft-suggested approach to understanding how to aggregate the effects of different pollutants, we are implicitly raising questions as the suitability of such an approach for choosing between different waste management systems.

Our work suffers from many of the same shortcomings that have affected other studies. We have tried to show quite explicitly why extreme caution would have to be exercised by anyone seeking to make use of these (and not just our) estimates in considering policy options, or in trying to understand, through economic approaches, what the 'best' waste management option might be. Later in this study, we seek to understand the implications of this 'requirement for caution' for both policy makers and local authority decision makers.

As well as discussing environmental repercussions over the whole product cycle as far as possible, the depletion of non-renewable resources and 'ecological rucksack' are each briefly reviewed to capture environmental effects of the different options.

The materials found in MSW that are discussed in the chapter are:

·         steel

·         aluminium

·         paper

·         glass

·         plastic

·         high density polyethylene (HDPE)

·         low density polyethylene (LDPE)

and the processes we have reviewed are

·         transport

·         landfilling

·         incineration; and

·         recycling.

We had hoped to discuss compostables, but we have not done so as we have been unable to source data that we had hoped might be forthcoming in the course of this study (see below).

6.1 Life Cycle Approach

One can compare competing options by taking account of cradle to grave environmental and resource impacts. One such approach is lifecycle assessment (LCA), according to the ISO standard ISO 14040 (International Organisation for Standardisation (ISO), 1997). The Environment Agency (Environment Agency 1997) has issued research reports setting out what constitutes best practice in Life Cycle Analysis for Waste Management. The report suggests LCA should consist of the following four stages.

i) Goal definition and scoping, which defines the system to be studied and the functional unit on which the study is based

ii) Inventory Analysis, which compiles data on resources used and wastes and emissions generated in the form of an inventory table

iii) Impact assessment, which converts the inventory table into an understandable evaluation of the magnitude and significance of the potential environmental impacts; and,

iv) Interpretation, where the inventory and impact assessment results are assessed in line with the goal

and scope of the study. Carrying out a full LCA is time consuming and expensive because of the data collection requirements requiring all resource inputs and environmental discharges for all commodities and economic activities in the product

chain to be assessed. It is also extremely difficult to move from stage ii) through to subsequent stages. Some of the reasons for this will be highlighted in this Chapter.

The list of resources used and emissions generated that could be considered within the analysis is given in Table 23 below. This illustrates the magnitude of the challenge.

Table 23: Environmental Impacts Shown in An LCA

We are not attempting, in this study, to carry out a complete LCA analysis along lines proposed under the ISO standards.¹⁹ However, much of what we are doing applies the essence of the life cycle approach. We seek to use inventory analysis (and we have done no primary work here) to quantify the environmental burdens across the life cycle. We have no doubt that the inventory assumptions used will be questioned. This is the first of many reasons one can give as to why the LCA-based valuation approach will always be open to question. As Hukkinen (1999) puts it in an excellent study:

'The inherent systemic complexities of industrial ecology are compounded by the analytical complexities involved in conducting a life-cycle analysis (LCA), which aims to report the cumulative environmental impact of a product throughout its life-cycle…. The complexities of industrial ecology and the consequent analytical confusion can have a paralysing effect on decision making, when social groups with diverse political and economic agendas use conflicting mental models to understand the system. Scientific uncertainties and complexities frequently open up the platform of both public and corporate environmental politics for yet another LCA expert who can question the 'scientific' validity of all previous LCAs.'

[¹⁹ These include standards in respect of general approach (14040), inventories (14041), and two which are likely to be released soon on impact assessment (14042) and interpretation (14043).]

We are not questioning a specific LCA, but in the spirit of the comment above, we question the claim not so much of LCA, but of any LCA-based valuation, to represent some 'true' value (or even range of values) of the externalities associated with waste management options.

The goal of the LCA is to understand the environmental impacts associated with extracting materials from the waste stream through kerbside collection. The approach would in our view ideally take, as the functional unit, a tonne of municipal waste whose composition would be taken from local compositional studies. It would then use actual data from authorities (such as we have been able to gather) and compare the existing situation, in which some materials are extracted at kerbside, with the situation in which the whole waste stream is landfilled (or incinerated). In essence, we would treat, where possible, the tonne of municipal waste as a set of discrete components which may or not be separated out from the whole.

The boundaries of the analysis are such that we seek to compare, for each of the materials extracted, the environmental impacts associated with materials collection, reprocessing and 'remanufacture' with the alternative route of using one or other linear options (landfill or incineration) and then extracting, processing and manufacturing an amount of raw material required to generate an equivalent amount of the material concerned. Note that this means that we are not considering the environmental impacts associated with the material in use. Issues of functionality lie outside the scope of the study. Waste minimisation is also beyond the study's scope. ²⁰

[²⁰ These omissions are important. There are questions to be asked as to how LCA-based approaches, which seek to assist in some way in the making of waste management decisions, can do so in the spirit of 'sustainability' when to some extent, one takes as given the waste materials which enter the bin. Waste management starts well before the waste is generated. It might be argued that those making decisions concerning what to do with waste are not in a position to influence its generation. This is a matter for debate since some studies suggest that the actions of waste managers (e.g. the provision of wheelie bins - see DETR 1997b) do influence waste generation.]

The 'functional unit' against which impacts will be ultimately be quantified in Chapter 6 shall be a tonne of municipal waste, understood to include fractions of each of the materials listed above (steel, aluminium, paper, glass, HDPE, LDPE) as they arise in non-segregated municipal waste. We are necessarily dependent upon secondary sources of data, which are likely to be of variable quality. Needless to say, we are interested in updating our basic analysis on the basis of what might be claimed to be better information, though it is worth pointing out once more that inventory data is unlikely to be beyond dispute. The Environment Agency should soon be publishing reports that have underpinned its LCA tool, WISARD, containing inventories for different processes. The data in WISARD will also be subject to scrutiny and criticism, the more so since the Agency itself seems very keen to see the tool used by Local Authorities and at the regional level (in Strategic Waste Management Assessments).²¹

[²¹ It should be pointed out that full scrutiny of the tool will be made more difficult by the fact that it is being marketed at quite high cost by a private company. In addition, those who feel the tool can be improved might be reluctant to suggest improvements for the same reason.]

6.2 Background to Our Approach

In the course of conducting a (far from exhaustive) review of relevant work undertaken, we have sought to extract estimates for externality adders associated with different pollutants. These are figures that express the externalities associated with a pollutant or effect in a convenient 'per unit' form, such as £/tonne, or p/vehicle km. We have done this so as to simplify the analysis. All of the studies reviewed above employ this 'externality adder' approach. Purists will point out that this is an unsatisfactory approach and they are probably right. More sophisticated studies, recognising the problems associated with benefits transfer,²² will make use of techniques designed to capture as far as possible the impact at a location under study. For example, in the case of air emissions, modelling will be undertaken to establish the change in pollutant concentration due to those emissions, and to understand how this varies across space (so that changes in the level of exposure can be mapped across the receptors affected). Exposure response functions can then be used to estimate effects, and a final step involves valuing these effects (see e.g. IVM et al 1997; AEA 1997; IIASA et al 1998). It should come as no surprise (and indeed, one can argue that it is the corollary of the fact that benefits transfer is problematic) that these adders vary significantly. For this reason, we have used broad ranges of these adders.

[²² The benefits transfer problem refers essentially to the difficulties inherent in carrying location specific valuations across to different locations.]

There are two reasons for doing this. Both actually cast more fundamental questions about whether this simplified approach is really adequate in the contexts under consideration:

1) The first has to do with the presentation of damage costs in this simple way. Using externality adders implies one is transferring estimates from what are often (not always) location specific studies to other places (benefits transfer). It is well known that even if the underlying dose-response functions are known with certainty (and they are frequently not) and are readily transferable (and they might not be), the environmental effects may not be (and this will be implicit in the function) related linearly to emissions. It makes something of a nonsense of the effort involved in deriving location-specific estimates of the net benefits associated with changes in pollutant concentrations to then imply that a per unit emission factor, derived in local contexts on the basis of the effects of a specific source of emissions on local concentrations (and hence, exposure), can be transferred easily from place to place. This may be a tolerable approach where one is considering similar emissions sources in areas of similar population density and geographical characteristics, and where 'background' levels of the pollutants under investigation are similar. Even then, however, different authors make use of different estimates of the value of statistical life, and this will play a scaling role in quantification of the effects. Furthermore, where threshold effects are believed to be involved (and they may be for dioxins) the assumption breaks down more or less completely. By way of example, it probably makes little sense to make use of adders from studies which have modelled exposure to air emissions resulting from a 100m chimney stack, and then converted the external cost estimates to per tonne values, when the source of the emissions might be a car whose exhaust fumes are much closer to ground level.

2) The second raises more fundamental questions concerning the limitations implicit in the valuation approach. There are problems associated with uncertainties in the underlying science (affecting the reliability of dose-response / exposure-response relationships), the ability to model accurately changes in pollutant concentrations and their distribution across media (introducing errors), and methodological approaches to the valuation of life. Even if, therefore, one was dealing with similar emissions sources in areas of similar population density and geographical characteristics, it would be surprising to find agreement across studies upon the external costs associated with a specific pollutant other than in the statement that the approach is a problematic one (although in fact, this is frequently downplayed). Scientific uncertainty, properly understood, is not something that can be handled through probabilistic analysis. Within reason, one may speculate over the boundaries of that uncertainty, but this can be but speculation. The obvious examples here are the cases of dioxins, where the existence or absence of threshold levels in determining the effects of exposure are subject to debate, and climate change, where the significance of extreme events may yet be enormous - we simply do not know at present. Tinch (1995) refers to the latter as being of 'low probability', but this implies that what is not known - the probability of these events occurring -can be given some quantitative basis. Tinch goes on, however, to cast doubt upon the robustness of the damage estimates associated with global warming. This is in stark contrast to the view expressed in CSERGE et al (1993) where the authors express the view that such estimates are robust to variation on the basis of changes in random variables, even though these are again generated within a probabilistic realm. In a recent DTI-funded study, Dames and Moore adopted an approach used by the free University of Amsterdam where global warming externalities per tonne of CO₂ were assessed using a range £3-£109 per tonne (Ecobalance and Dames and Moore Group 1999).²³

[²³ A problem here is that whilst a study may be methodologically sound in the sense of covering all bases, and explaining key assumptions, at the end of the day, what one is seeking to place a value on is not a set of assumptions, but a real-world effect which may well differ in its manifestations to what was being assumed, and therefore valued.]

In essence, therefore, we are admitting that this approach (which has been adopted in all of the studies discussed in the previous Chapter) is a flawed one. To move beyond this, however, would require location specific modelling work, perhaps involving comprehensive work at 'exemplar sites' designed to facilitate benefits transfer to other sites suitably classified by type. Even this, however, would not overcome the second of the issues discussed above.

Unlike some other studies, we have separated out transport components from the specific options. One reason for doing this is that it allows some understanding of the significance of fuel duty in adding to the costs of different options. To the extent that one accepts this represents an internalisation of the external costs of transport, one can estimate the degree to which the total externality of one or other waste management option is already internalised through fuel duty.

This is a significant change in approach. The work done in 1993 by CSERGE et al which informed the setting of the level of the landfill tax did include transport costs within the different waste management options, but the work was undertaken in the year that the fuel duty escalator was announced. Using the landfill case addressed in that study, the fuel duty per tonne of landfilled waste is close to the mean value of the externality reported by the study for a landfill with energy recovery. In other words, some of the externality associated with landfill with energy recovery in that study is not associated with the landfill per se, but the transport to the landfill. To the extent that a) one believed the landfill tax should be set on the basis of externalities (and we have stated elsewhere reasons why it might not be - see ECOTEC 1997), and b) that the transport element has been internalised by fuel duty, one might suggest that the landfill tax should have been falling as the fuel duty increased. Evidently, similar comments could be applied in the case of incineration, though the transport component assumed in the CSERGE et al (1993) study is less significant than for landfill.

For recycling, to the extent that transport externalities may be significant (and for materials which are collected in lower density forms, as plastics are in kerbside collections, they will be especially so), the fact that transport externalities may be a significant component of the total is interesting. To the extent that the other externalities reported (e.g. those associated with greenhouse gases from materials processing) are not internalised, the current level of internalisation will act to skew choices between waste management options in such a way that the level of recycling is below that which would prevail in the absence of any internalisation at all. This is ironic since the results of most studies (see previous Chapter) seem to suggest that full internalisation would have the opposite effect.

Note here that the climate change levy could have had an effect which reinforced the waste management hierarchy. Yet the detail of its design, notably the outright exemptions (for primary aluminium processing as an electrolytic process) and the levels of exemption proposed for intensive energy users (steel etc.), will reduce the extent to which the price mechanism affects the balance between recycling and primary materials use. This will be further hindered by exemptions for renewable energy, including energy from waste.

We have tried to be reasonably accurate in converting and updating past externality estimates to ensure they are comparable, and are converted accurately into UK currency terms using appropriate deflators and exchange rates. However, the date to which the originals refer is not always absolutely clear. Any inaccuracies will be of limited concern given that:

• Most of them come from relatively recent work so that the impact of exchange rate movements and / or deflators will be relatively small; and

• We are using ranges of values, and the range is typically very large, so that any 'accuracy' lost in the conversion and updating is more or less spurious in the context of the ranges available (it seems rare to find externality adders which are all confined within a range of one order of magnitude).

With respect to the last point, mindful of the many caveats which need to be applied, we are aiming at illustrating ranges which are plausible on the basis of the existing literature, and with the understanding that the analysis is a long way from being a complete one. It is better, in our view, to indicate a broad range of possible estimates than either pretending that we can undertake valuations in possession of certain knowledge, or adopting the 'lucky dip' approach to the valuation of externalities (and indeed, this is the advice which is given - it seems rarely to have been followed - by the Better Regulation Unit).²⁴

[²⁴ See www.cabinet-office.gov.uk/regulation/1998/brg/brg_part2_section2.htm. Here, there is guidance on how to treat uncertainty in the context of Regulatory Impact Assessments. Whether the study goes far enough in appreciating the radical nature of uncertainty that can exist is debatable (one is still being asked to estimate the magnitude, or the extent of uncertainty, i.e. to say something about something one might know next to nothing about). There is still pressure for quantification.]

We have grouped 'high' and 'low' valuation factors together. There are two reasons for doing this. One is that several of the factors used relate to pollutants whose values are estimated using estimates of the valuation of life (see below). High values could be assumed to result from high values of life. The other is that the high-value externality adders, to the extent that these are related to air pollution, would perhaps be more relevant for places where population densities are higher (e.g. urban areas). In both cases, there would be reason to believe that where one adder affecting health is 'high', others might be high too (on the basis that the population exposed is significant, or that gas concentrations are low because of high chimney stacks, in all cases).²⁵ There are exceptions to this 'rule,' and in addition, where one is dealing with emissions from different locations, the assumption no longer holds true. We show below what can happen when one changes the adders as they are applied to specific plants. Generally, through creative manipulation and choice of externality estimates, one could make one's ranges broader by choosing high and low contributions where these work in the same direction to increase the net externality (i.e. one would choose high values for positive contributions and low values for negative ones).

[²⁵ There are exceptions to this 'rule' however. Ozone, as derived from VOCs, would be one as it tends to be generated in areas where specific carriers are not present. These carriers exist in the main in areas where NOx is present. Hence, tropospheric ozone may do more damage in areas which are less densely populated (even when its precursors are actually emitted in densely populated urban areas).]

We are not well-placed to know what might be the income elasticity of demand for avoiding the external costs being assessed. Coopers and Lybrand and CSERGE (1996) (and Brisson 1997) work on the basis of an income elasticity of demand of 0.3 (using a figure of 1 for sensitivity). A more elastic demand (as has been hypothesised in the context of some agri-environmental studies) would magnify the effects of increased real incomes over the time after the externality assessment was first made.

Lastly, whilst some studies seek to allocate environmental burdens associated with landfill across the whole life-cycle of the landfill, our principle focus is on marginal changes in the use of one or other type of facility. We have not made any attempt to attribute environmental burdens associated with, for example, landfill engineering, to materials landfilled. The assumption is that externalities associated with landfill engineering are fixed, and only those emissions directly associated with the waste landfilled are taken into account. This may limit the usefulness of this type of approach where the question being raised is one of whether or not to construct one or other facility, although the private cost analysis clearly takes into account the financial side of the equation. Whether the private costs internalise externalities or not, and how well they do this, will depend upon future decisions as to the adequacy of financial provisions for covering the potential for accidents, the requirements (if any) for compensating local residents in the context of decisions to site new facilities, and the ability to enforce operating standards for specific treatment plants.

The marginal / non-marginal distinction is an important point and one that is rarely addressed adequately in the literature. It raises questions as to how, when marginal changes are being considered, to account for disamenity effects where many of them may be poorly (if at all) related to the level of inputs to the facility. For example, where landfills exist, the disamenity associated with the existence of the facility is unlikely to be dealt with best through a pro-rating of that across landfill inputs since the disamenity will be relatively fixed irrespective of inputs (elements related to transport, for example, will not be - see ECOTEC 1998 and EFTEC 1999 for a discussion in the context of aggregates extraction - but most of the analysis undertaken thus far already accounts for some of the transport-related externalities. More local transport disamenity such as litter, dust and dirt, and noise associated with vehicles congregating at the site will not have been included).

6.3 Waste Transport

6.3.1 Residuals

The collection of residual waste for treatment in linear waste management options involves the use of vehicles on a journey which will take them from a depot on to the collection round to pick up waste from bins and bags. Waste will then go either to a transfer station or direct to a landfill site or incinerator. In addition, some bulky waste is collected either for a fee or at zero cost to its producer, from households. There is also the collection of waste delivered to Civic Amenity sites to be considered and the collection of litter from parks, public places, highways etc.

The majority of the waste is collected at the door (although the non-doorstep fraction is a significant fraction of the total). We focus on that element of the waste in this study. As such, we are assuming that the collection approach does not affect the volume put out by the householder. This will not be strictly true. The provision or otherwise of kerbside schemes for dry recyclables will have a bearing on the amount of material taken to bring sites where these are already available. Equally, the provision or otherwise of kerbside systems for organic waste collection will affect the amount of such material taken to civic amenity sites (less will be taken where provision of kerbside collection exists). Transportation of refuse typically takes place in vehicles which may weigh some 24 tonnes, but which have a capacity, typically, of 10-14 tonnes (since they have a dry weight of 10 tonnes or so).

6.3.2 Recyclables and Compostables

The collection of recyclables and compostables can involve a similar collection round to that for residuals, with materials typically being delivered to a depot for sorting (where this has not been done on the round itself). The materials might then be composted at site, or in the case of dry recyclables, transported to reprocessors for their use. Different vehicles may be used for dry recyclables but in our schemes, the vehicles tended to be between 7.5 and 11 tonne vehicles with payloads between 2 and 4 tonnes. Certainly for dry recyclables, vehicles are unlikely to be as fully loaded in weight terms. This may increase the private costs of transport, but it will reduce the associated external costs per load since some externalities associated with transport are related to vehicle weight (e.g. those for road damage and, to some extent, emissions through the relationship to fuel efficiency).

6.3.3 External Cost Analysis

There are a number of impacts associated with transport which should be accounted for in a complete analysis of environmental impacts. These include:

• health effects of vehicle emissions (local);

• effects on global warming through greenhouse gases (GHGs);

• transport related accidents, fatal and non-fatal;

• transport related noise; and

• damage caused to highways.

This is by no means an extensive review of all environmental impacts of transport (see Tinch 1995, Maddison et al 1996). Relatively few studies look at the issue of damage done to the road itself. The system of road tax is now arguably better structured to internalise this, but it is not clear that the nature of the vehicle (maximum load, number of axles etc.) is adequately accounted for in the analysis of external costs. In any case, we are principally interested at this stage in the external costs generated rather than their current level of internalisation (a point to which we return below).

Note that the activity in which those collecting waste are engaged, by virtue of its being carried out on public highways, may expose them to a higher probability of accident than those in other occupations. External costs associated with collection could account for accidents suffered by those engaged in the activity concerned. There is good reason to believe that not only the exposure to traffic, but also the handling of materials in waste, are likely to pose specific hazards. Powell (1992) noted that 'over-3 day injuries' (those which cause workers to be off work for more than three days) are much more common in waste collection than in comparable industries. She noted that both the physical handling of material and the nature of vehicles used could be an issue.

Whether this should properly be accounted for as an external cost depends, arguably, upon whether one believes those facing the hazards involved are well appraised of, and either protected from or compensated for, them. To the extent that one believes that they are, the externality is internalised (such a view is adopted in recent work by Ecobalance and Dames and Moore Group (1999) for the DTI). The belief that such a calculus is being made by employees underpins one approach to the valuation of a statistical life. The hedonic wage approach is founded upon the belief that those undertaking employment consider the remuneration in the context of their exposure to hazards. To the extent that alternative employment opportunities may be limited - and it seems fair to assume that they may be for those employed in waste collection - the remuneration might not reflect this increased exposure to hazards (labour market effects might act to counter the presumption in favour of increased remuneration). Therefore, the health related external costs of waste collection, to the extent that they are based upon average figures, could be understated since the nature of the job places workers at a particular risk, and these might not be reflected in wage rates.

The greater these health related risks are, the more potentially significant it becomes that kerbside collection often (though not always) involves greater time spent in collection activity (in the case of weekly collections of each) than would otherwise be the case. Although the same materials (more or less²⁶) may be being handled, the fact that workers are frequently working in the road itself adds to the dangers associated with the task. This is not explicitly included in the analysis. It is worth pointing out that we have, in speaking to those operating kerbside collection, discussed the issue of injuries. The general response has been that these are fairly rare, and that none were serious in nature. More empirical data on this would be useful (we have not actively sought this, so it may exist). One respondent mentioned the issue of morale in the job, but this was traced to more general concerns (possibly the time of year the research was being carried out - a month or so before Christmas).

[²⁶ The introduction of a kerbside collection for organic material may bring more material into the collected waste stream than in the absence of such a collection.]

We have evaluated transport effects in two ways.

Method 1

In the first approach, emission coefficients for health and noise which come from the Tinch Report (Tinch 1995) and the European Council of Ministers for Transport (ECMT 1998) were used. For the air pollution and noise costs, we have used HGV (heavy goods vehicle) per 000tkm estimates. The higher end of the range in Tinch (1995), taken from the 'urban driving' estimate, may not cover the ECMT figure (depending upon how one updates them). Tinch himself notes that his figures are ' "best estimates" drawn from a survey of the literature. They are intended to show the potential for valuation, and should not be interpreted as "the" value of those [noise and air pollution] effects.' The ECMT (1998) figure has been taken as 5.4p/tkm.

For global warming, the higher value used is from the ExternE estimates as cited in EFTEC (1999) (note the context is similar so the estimates are likely to be transferable). The lower value is the shadow cost estimated in the ECMT (1998) report. The estimates have been updated to account for exchange rates (where, as in the ECMT report, the estimates are in 1991 ECU) and for inflation. These values are shown in Table 24.

Table 24: Valuation Factors Used in Transport Analysis

It is worth pointing out that valuation of transport-related accidents are driven by the separate products of the number of accidents and fatalities, and a measure of the value of life. Whilst some knowledge of the former (at least in the road transport cases) may be gained through statistical analysis, the latter is subject to considerable uncertainty and debate.

Some of the estimates considered in the course of the ExternE study are given in Baranzini (1997) (see Table 25). Other reviews have found variation from ECU 360,000 to ECU 10 million (EFTEC 1996), or elsewhere, from 0.3 to 17.5 MECU (European Commission 1995). The UK Government uses a value of just below £1 million for the purpose of valuing transport deaths and casualties (£902,500 for June 1997 - DETR 1997a), whilst Pearce and Crowards (1995) suggest a value more than double this is more appropriate. The latter is consistent with Metroeconomica's (1996) estimate of ECU 2.8mn, and indeed, most Commission studies, including ExternE, have settled for figures between ECU 2.6-3.0mn - which is close to the Pearce and Crowards (1995) view. Few studies discuss the influence of the nature of the cause of death as a potential factor influencing the value used, which is increasingly seen as an issue in debates on the matter.²⁷

[²⁷ It would appear to be correct to vary one's valuation of life according to the nature of the risk to which people are exposed, at least to the extent that one is evaluating only people's preferences. Sociological studies of risk reveal that lay-people's rationality (and probably that of experts when they are 'acting as' 'lay-people') conflicts with expert assessment of 'risk' (which is not to argue the correctness of one or the other). Factors such as 'dread' and the control which people are able to exert over their exposure to the risk concerned appear to affect their perception of risk, but in ways which are poorly understood at present (for a discussion in the context of waste, see Kasperson et al 1992; Gerrard 1994; more generally, see, e.g., Starr 1976; Slovic 1981; Slovic et al 1994; Horowitz 1994). The approach in work undertaken by NERA (1997) appears to be to pluck multiplicative factors out of the sky to 'account' for these. These seem to have been chosen so as to arrive at results which are 'not too high' and 'not too low', once again downplaying the significance of the uncertainties (perhaps more correctly expressed as ignorance) involved in accounting for such poorly understood impacts upon risk perception. In any case, one suspects (from some detailed consideration of the matter) that were one to find some relationship between risk perception and the nature of hazard, that perception of risks varies in a non-linear manner in relation to the potential consequences.]

Table 25: Valuations of Life Considered in ExternE Project

Source: Baranzini (1997)

Although the closeness of the agreement in some studies and amongst some authors may appear to suggest some form of convergence, in our view, it would be wrong to suppose that such values are necessarily 'correct' by virtue of this agreement. It remains appropriate, in the face of continuing debate, to retain high and low estimates. There appears to be no means of validating these estimates, the only form of validation being that associated with how well a particular study, approached using a specific methodology (which one may or may not accept as valid for the purpose) has been performed. Methodological approaches are still the subject of disagreements, not to mention the question of whether this should be done at all.²⁸

[²⁸ The classic argument for doing so is that policy makers need to allocate resources and therefore make decisions across competing claims. One might ask why, if this is what policy makers do, do they need consultants and economists to do this for them? There is an interesting debate to be had about whether, once aspects of CBA start trying to account for the nature of the hazard to which an individual is exposed, the approach has not finally discovered its own limitations. CBA proponents always claim that individuals, in making decisions, make them on the basis of a cost-benefit analysis. This has always been a questionable assumption (see Sagoff 1989). The fact that a decision has been made cannot lead, ineluctably, to a deduction about the nature of the process by which the decision was arrived at. We live in an extremely complex world, and bombarded by information which carries competing messages, we have, in Scott Lash's words 'no choice but to choose.' In seeking to account for more psychological characteristics, valuation techniques are now trying to capture the spirit of a far more complex rationality through which it hopes to elicit people's preferences. In doing so, it is likely to encounter limits to the degree to which it can be assumed that individuals' preferences correspond to what is assumed to be 'rational' behaviour in the economic context. An exploration of these issues from a different perspective can be found in Sagoff (1994) (see also various chapters in Foster (1997)). Quite apart from the fact that cost-benefit analyses are likely to contain uncertainties and omissions which are not always made completely explicit, such approaches to decision making risk reducing the significance of what some might argue are more fundamental moral and political issues. Some of the responses to contingent valuation questionnaires given by those who are questioned provide a testament to the extent of unease felt by many in going down this route. Interesting examples of the more sceptical attitude to valuation and its use in the field of policy-making are Foster (ed.) (1997) and Vatn and Bromley (1994).]

With respect to data on accidents, we have used two estimates. Both are from the DETR. The first (high) set is from DETR (1997a) and relates to casualties from trucks. The second (low) set is from the DETR (1998) and is calculated from figures for HGV traffic volumes and accidents to HGV drivers and passengers. Evidently, this would be expected to be low since it does not include data on pedestrians and the like who may be involved in accidents involving HGVs.

The other aspect where the valuation of life plays an important role (albeit in a non-transparent way in our analysis) is in the health impacts of transport emissions (and air emissions more generally). Here, economic effects are elicited by establishing the effects of emissions on concentrations, usually through atmospheric modelling, and then using dose response functions to estimate the effects on health of these changed concentrations. The modelling of atmospheric concentrations is far from being a precise science partly because the dispersion of pollutants is likely to vary under specific topographical and other local conditions. Dose-response relationships are also the subject of varying degrees of debate (depending on the pollutant). The final step involves valuing the mortality and morbidity effects of the specific pollutants.

Current debates in the valuation literature take the view that mortality effects of air pollution should be treated differently from those associated with, for example, car accidents since in the former case, it is argued that the effect will be to bring forward the deaths of those who would have died soon after anyway.²⁹ Discussions have taken place, therefore, concerning whether, in the case of air pollution, the most appropriate measure for valuing life might be one based upon the value of life years lost (VLYL) or on the value of a statistical life (VOSL) (for a useful discussion, see RPA and Metroeconomica 1999). In the context of air pollution, a recent Department of Health publication decided to use a range of estimates for willingness to pay to reduce the risk of a death brought forward from £2,600 to £1.4 million (DH 1999). Department of Health Ministers subsequently decided that the currently available data 'do not allow the benefits of reducing air pollution to be converted into monetary terms with a sufficient degree of certainty to allow the results to be used in the cost benefit analysis of the NAQS [National Air Quality Strategy]' (DETR 1999b).

[²⁹ The strange thing about this is that, from the perspective of external costs, deaths caused by pollution are less of a concern than those that are treated as accidents on the road. Apparently, because the deaths from pollution are those of vulnerable people, they are attributed less value than if the person were 'less vulnerable.' Moral outrage at murder works in the opposite sense. The more vulnerable the victim the more repugnant the crime. It would be difficult to counter the view that what the pollution is doing is not actually killing the most vulnerable people (this is what the science tells us). It is paradoxical that this is seen as (in relative economic terms) less of a worry than killing younger people. The whole notion of 'bringing death forward' seems to be an attempt to sanitise what is actually a rather unpalatable situation in which we are seeing vulnerable people killed by pollution. Presumably, no self-respecting defence lawyer would state in Court 'sorry, m'lud, my client was merely bringing the deceased's death forward.']

Note that in this analysis, the valuation of mortality only enters into the analysis directly in the context of accidents and injuries related to transport. Indirectly, a valuation of mortality is implicit in externality adders used, however. To the extent that these have been based on studies that made use of mortality estimates based on VOSL (as opposed to VLYL), they will be higher than would be the case had VLYL estimates been used. In this work, we have taken, as high and low estimates for mortality, £6 million and £500,000 respectively.

We have valued congestion using the estimates used by CSERGE in our work for DETR (ECOTEC 1999), these coming from Newbery (1988; 1990).

Method 2

Method 2 is very similar but goes back to first principles in respect of emissions from transport. We have then applied externality adders (used elsewhere in the study - these are shown in Annex 2) to a sub-set of the range of pollutants emitted. Both congestion and casualties are treated in the same way as in Method 1, so that all that changes are the health and global warming estimates which are now derived through estimates of vehicle fuel efficiencies and emissions associated with the relevant fuel type.

6.3.4 Results

We illustrate below, in Tables 26 and 27, the externalities by category for a waste collection system carrying waste in RCVs with average payload 10 tonnes travelling 80km.

Table 26: Method 1, 10 Tonne Payload, 80km roundtrip

NB: Totals are subject to rounding

Table 27: Method 2, 10 Tonne Payload, 80km Roundtrip, Impact Per Tonne Of Waste Transported (£/t)

NB Totals are subject to rounding

The only real differences in these analyses are in the valuation of global warming externalities, and in health effects of pollutants. This illustrates the different results which can be obtained through adopting, on the one hand, the more bottom-up approach in Method 2, and the approach in which one chooses to lump together transport-related impacts on a per kilometre, or per tonne kilometre basis. When one looks at the full analysis, the following points can be made:

• A trivial, though nonetheless important observation is that if we take the low externality adders, the total externality is small. One might suggest that this is tantamount to saying that we are more or less indifferent to what distances waste is moved and in what vehicles when we are not bothered about effects on people's lives, or where we do not think that global warming etc. will have important ramifications. In a sense, if nothing is important, if pollutants do not really cause any harm, and nothing is really changing in ways that need bother us, the issue need not concern us. Whilst this case should not be ruled out as a possibility, policy based on the low externality adder case (given the prevailing uncertainties) is somewhat cynical and potentially leads to significant levels of regret in the future.

• Less trivial is the fact that when aggregating different types of externality, one has to be very careful to account for externalities correctly. The reason for this is that some of the external costs are not directly related to tonnages per se. Some are derived from the number of kilometres a vehicle has travelled. It is inappropriate, in such conditions, to believe that the external costs associated with the transportation of a vehicle carrying ten tonnes of waste will be ten times the externality associated with the movement of one tonne of waste where one has attributed to each tonne a 'distance travelled' equivalent to that moved by the whole vehicle. This poses no great analytical problems, but the problem has not been properly treated in other studies. By way of example, externalities associated with casualties tend to be related to distance travelled. In our earlier work with CSERGE (ECOTEC 1999), taking the tonne of waste as the functional unit, the view adopted was that since each tonne of waste was being transported the same distance (i.e. the whole journey), this distance could be used in calculating the externalities which are related in some way to distance. But clearly, if a ten tonne load is being transported, this approach erroneously attributes some external costs related to each load to each tonne being transported. In this way, externalities associated with casualties owing to movements of waste were over-estimated. Properly treated, the external costs relating to casualties are 'diluted' by the weight of the vehicle's load. As such, one finds that for a 100,000 tonne collection scheme using trucks carrying ten tonnes of waste (they may be 24t RCVs, but they do not carry 24t of waste) the significance of the value attributed to life is small in the calculation of per tonne externalities.³⁰ It increases as one shifts to lower 'payloads'. However, the lower the payload, the more likely fuel consumption is to improve, reducing externalities associated with health and global warming (see below). Even so, at the higher payloads (i.e. especially for residuals), the way in which 'valuation of life' affects our analysis is through the number of casualties associated with transport, and since these numbers are quite low, the effects are relatively small. Note that casualties per tonne of material collected will be far more significant in the case of bring schemes in which journeys are made specifically for the purpose of delivering materials (since the casualty and accident rates are not significantly different for cars, but the number of journeys made to collect one tonne of material may be quite high)³¹.

[³⁰ Obviously, the 'high' and 'low' range in respect of impacts on health (as opposed to casualties) are in some way related to differing valuations placed upon life so this does enter the analysis indirectly through other routes.

³¹ Obviously, at higher densities of bring sites, journey distance for those doing the 'bringing' will fall, and indeed, use of a vehicle may become completely unnecessary.]

• The greatest contributions to the total externality associated with moving waste from one place to another come, potentially, through congestion, global warming, and the effects on noise and health. Clearly, congestion effects will vary depending upon the route the vehicle takes/has to take, as well as the time at which the journey is made (some studies attempt to account for variation owing to the these by differentiating by level of urbanisation and on and off peak periods. This is not easy especially since these change, and they can be different in different areas). The same things can be said for effects on noise and health (since these will be related to population exposure). Noise and health effects will, however, also be amenable to influence through adequate vehicle maintenance and use of modern vehicles with suitable emissions abatement equipment. The great unknown is global warming, and the effects of transport upon this are invariant with respect to location of emissions. They do, of course, vary with distances travelled, and also with the vehicle load. In the case of kerbside schemes, these externalities will fall (per tonne of material collected) as participation rates increase. At the same time, depending upon the relative rates of growth of residuals and the kerbside scheme, and depending upon whether the collection takes place at the same time as the residuals, the external costs per tonne of residual collection will rise.

Note that in an earlier study (ECOTEC 1999), we did attempt to understand the extent to which transport related externalities were already internalised through fuel duty. This was not done in the study by CSERGE et al (1993) since the escalator was only introduced in 1993. Effectively, we can calculate an implied level of internalisation per tonne of waste transported in a given phase (on the basis of the fuel consumed per tonne of waste transported and the existing level of fuel duty). We estimate this to be some £1.2 per tonne of waste (assuming a round trip for a 10 tonne load of 80km).

Since transport externalities were used in the CSERGE et al (1993) assessment of the external costs of landfilling and incineration, then to the extent that the assessment of external costs was used in support of the tax level, there would be reason to believe that the internalisation of some of these 'landfill' externalities would suggest lower levels of landfill tax. The current level of fuel duty, applied to the rural landfill scenario in the CSERGE et al (1993) report (return journey of 80km, so total of 160km), would effectively internalise around £2.31 per tonne of waste assuming 16 tonne trucks with a 10 tonne load returning empty from the landfill (i.e. 16km per tonne landfilled). This assumes a fuel consumption of 0.32 km/l of fuel (from White et al 1995). This is interesting since mean values of the externalities from landfill as measured by CSERGE et al (1993) are lower than this for urban landfills with energy recovery and only marginally greater for all other types of landfill examined. Increased fuel efficiency and lower transport distances would, of course, lower the duty per tonne of waste.

6.4 Landfill

Emissions associated with landfill are a subject of some debate. Estimates of landfill gas generation have been given in Aumonier and Warren Spring Laboratory (both reviewed by CSERGE et al 1993), Powell (1992), USEPA (1998) and Entec (1999a) among others. Relatively little information exists concerning the external costs of landfill on the environment, a somewhat surprising statement since it lies at the bottom of the waste management hierarchy, and is therefore arguably deserving of attention. Indeed, if there is uncertainty about its impacts, one might reasonably question the logic behind its position at the base of the hierarchy. Reinforcing its position at the base, however, is the view that however well engineered they may be, landfill liners (natural or otherwise) will not contain waste indefinitely. Quite apart from the issues associated with (temporary) land-take, therefore, there is a perception that at a more fundamental level, the practice of landfilling simply passes on a problem created in one generation to another in (possibly very many) years to come. This has been debated more seriously in connection with hazardous and radioactive waste landfills and depositories (see Gerrard 1994 for an account of the US experience).³²

[³² In the case of radioactive wastes (where the 'future generations' issue is most pertinent, Gerrard (1994) reports that the US Department for the Environment 'spent several million dollars designing a "keep out" sign for WIPP [the Waste Isolation Pilot Plant] that would be effective for 10,000 years and recognisable by any future earthling.']

The work that was undertaken by CSERGE et al (1993) prior to the landfill tax concentrated primarily upon GHG emissions and upon transport to the landfill site (see above). That study did not distinguish between the CO₂ emissions that arise from biogenic sources and those that do not. The argument given was that this would not alter the analysis significantly. Yet this assumes that the estimates of damage associated with GHGs are fairly well understood (and implicitly, that they are believed to be small). Furthermore, it is a statement that has to be made relative in the context of an analysis which focuses only on a subset of the total external costs, and where sensitivities in respect of landfill gas collection and combustion are ignored. Collection and combustion of landfill gas has the net effect of converting CH₄ to CO₂, making the question of how one accounts for proportion of the GHG emissions which emerge as CO₂ rather more important (since more CO₂ is produced, but much of this may be from biogenic sources).

The CSERGE et al (1993) report offers the view that estimates of the valuation of damages associated with global warming as have been made are relatively robust, yet it alludes to studies which seek to deal with 'uncertainty' through use of random variables with a triangular probability distribution.³³ This view is at odds with that of ECMT (1998) (and in spirit, that of Tinch 1995 - see above) who, in taking what one might call a precautionary approach to the issue of uncertainty, used a value of 50ECU/kg rather than the $20 per tonne used in the CSERGE et al (1993) study. These are three orders of magnitude apart. It seems that if one believes that one must attribute values to phenomena whose outcome is uncertain, the use of wide variations is likely to be if not the appropriate way, then the only way to deal with that uncertainty (rather than to pretend that one has certain knowledge of something about which one has admitted one does not) if indeed one believes one can within this sort of analysis.³⁴

[³³ If the so-called uncertainty is being approached through probabilistic analysis, it loses the characteristic of being uncertain. Uncertainty as defined here is qualitatively different to inaccuracy, or error in measurement. One might be able to ascribe boundaries or probabilistic assessments to the latter, but not to the former.

³⁴ A study for the World Bank by Hagler Bailly et al (1997) made use of shadow price values of $5, $20 and $40, but even this choice was arbitrarily made.]

Elsewhere, it has been usual in valuation of the effects of biodegradation under landfill conditions to ignore the releases of CO₂ on grounds that these are emissions which would have occurred anyway and that they are part of the carbon cycle. The argument is that these sources of CO₂ are not the consequence of anthropogenic releases into the atmosphere per se, but are releases that would have occurred anyway (USEPA 1998). The methane component, on the other hand, can be considered anthropogenic in character. It would be consistent with this view not only to ignore the CO₂ emissions from landfill (on the basis that all are biogenic), but also to subtract from any valuation of the emissions of methane from landfill the value of the equivalent emissions of CO₂ which would have occurred had the material been biodegrading outside landfill. As far as we can see, this has not been done in any external cost study thus far.

Our analysis has tried to shed light on an important question to consider as the composition of waste being sent to different options changes over time. Indeed, since consideration of the matter might shed light upon the desirability of sending different wastes to different disposal options, we have sought to model the externalities from landfill in such a way as the model can incorporate changes in waste composition sent to landfill.³⁵ In doing this we have relied on estimates of methane emissions which come from only one source (Barlaz 1998), which is recognised as a problem by the USEPA in its work (from where these estimates are taken - see Annex 3). It is interesting to note that some materials are treated as net sequesters of carbon in this model since their carbon is deemed of biogenic origin and is assumed to degrade incompletely in landfills.

[³⁵ It is also worth questioning at a more fundamental level the presumption that the emissions of CO₂ which occur outside landfill conditions are truly not of anthropogenic origin. The activities which constitute the cycling of carbon are, it could be argued, being artificially speeded up. The rate at which photosynthetic product is extracted from the land (and the way in which its production may be speeded up, for example through the use of synthetic fertilisers, production of which uses natural gas to fix nitrogen) has consequences for the cycling of carbon and for the fluxes of GHGs. However, for the purposes of this study, we do not investigate the matter further.]

The model also allows for varying estimates of the rate at which methane is oxidised through the landfill cap, though estimates used in USEPA are 10%. We have also allowed for flexibility in terms of performance in respect of gas recovery from the landfill, and hence, in the case where the collected gas is used for energy recovery rather than flaring, efficiency of energy conversion. This allows us to model the situation for three types of landfill:

1. one where no gas collection occurs (so there are net emissions of CH₄, with any CO₂ emissions assumed to be biogenic).

2. one where gas collection occurs and all the collected gas is flared (converting CH₄ to CO₂, hence reducing the costs associated with collected gas since there is oxidation to CO₂ of biogenic origin).

3. one where collected gas is used for energy recovery (so the same oxidation effect occurs, with the added benefit of displacing energy).

Displaced energy is treated in a separate module, which uses high and low emissions factors for air pollutants to arrive at high and low estimates of avoided costs per MJ of energy generated. This is done for three cases - that where one assumes the marginal energy source displaced is coal-fired, that where one assumes the marginal source displaced is from the UK average mix (avoided emissions from these were based on ETSU 1997, see Annex 4), and that where no displacement is assumed. Hence, there are four non-zero values for the avoided external costs associated with energy generation (each of the two displacement cases with high and low adders, respectively, applied).

Note that in this system expansion, the avoided externalities associated with gas collection and flaring / gas collection and energy recovery depend upon a number of factors:

• the volume of gas generated;

• the composition of the gas (in particular, its calorific value, dependent principally on the proportion that is methane);

• the gas collection efficiency;

• in the case of energy recovery, the efficiency of that recovery process; and

• the assumption made about which source of energy, if any, is being 'displaced'; and

• the emissions data pertaining to that source.

We noted in the previous Chapter that these figures and assumptions are crucial in arriving at figures for the net externality attributable to specific waste treatment options. Methane generation is discussed in more detail below. The assumption concerning avoided external costs deserves further comment, however, because of its critical influence on the analysis.

6.4.1 A Note on Avoided Externalities Associated with Energy Recovery from Waste Treatment Facilities

It has been customary in analyses of this nature to treat energy recovery as having beneficial impacts through the displacement of other sources of energy. There are a number of issues that one would need to account for in dealing with the issue. The first involves whether one is really displacing anything, and whether it might not be the case that, because energy use is expanding, nothing is being displaced as such. One is simply sourcing energy from different (new) sources. It makes sense to recover energy from incineration plants since the process itself implies generation of energy (if not necessarily its recovery). This approach would hold that no energy source is being displaced per se.

If energy use is expanding, or more generally, if one looks at the longer-term, the question of which if any source is really being displaced, or replaced (even in a shrinking scenario, plant is replaced), might be reduced to one of 'which source is not being introduced that would otherwise have been introduced?' Two approaches might be relevant here. The first would be to make the observation that the principal new source of energy is gas. There might, therefore, be an argument that one should consider gas-fired power as the source being displaced. The second might look to the longer term. Which energy sources are we seeking to develop in the future? In this case, the answer might be 'renewables of one or other type', especially if the Government is keen to meet its target of 10% for the proportion of energy supplied by renewables in the future. The displaced source could, therefore, be an alternative source of renewable energy. Even here, however, it could be argued that government renewables targets are set on the understanding that energy from waste will be a contributor.

More commonly, in studies of this nature, it has been common to focus on marginal changes. The incremental increase in energy from waste capacity would displace the marginal source of electricity. It is this perspective that has led those carrying out this type of analysis to treat recovered energy as though it were displacing coal, or the average source of energy supplied.

We understand that this view was also adopted in the Environment Agency's WISARD mode though for different reasons. The argument that seems to have been employed is that energy from waste displaces coal since it replaces energy sources which are not base load. Some argue that this is a difficult argument to sustain since incinerators are generating more or less continuously.

We have used both the standard assumptions, as well as the assumption that no displacement effect is occurring. We do this since the effect of the 'replacing coal' and 'replacing average energy mix' assumptions are controversial. Using the no displacement scenario not only allows one to see the effects of these assumptions on the results but also reflects our belief that the standard assumptions are controversial and likely to generate disagreements. It may, in any case, not be appropriate to apply the usual assumptions where one is considering non-marginal changes in the supply of energy from waste treatment plants.

6.4.2 A Note on Methane Emissions and Energy Generation from Landfill Gas

Methane emissions from landfills are not incredibly well understood. A range of estimates could be generated from different studies in the public domain. CSERGE et al (1993) looked at estimates from Aumonier and from Warren Spring Laboratory (WSL), and found ranges for best estimates of methane generation of between 53-81 m³ per tonne of MSW. The full range, from the low estimate assuming 20% methane oxidation, to the high estimate from Aumonier, was from 25-117 m³ per tonne. Powell's (1992) mini-survey estimated recoverable quantities of the order 100 m³ per tonne (in which case, the actual quantities would presumably be much higher). Entec (1999a) on the other hand, used much higher figures of the order 400-500 m³ landfill gas per tonne of MSW of which 50% was assumed to be methane (i.e. 200-250 m³methane per tonne MSW).³⁶ Using the composition figures we have taken, the USEPA (1998) methane generation figures give 50 m³ at 5% oxidation rates, and only 42 m³ at 20% oxidation rates. It should be noted, therefore, that these are relatively low estimates of methane generation. Methane emissions in our analysis come from USEPA figures, not because we feel these are 'correct', but principally because they allow us to link methane generation to specific components of the waste stream.³⁷ No methane emissions factors are given for emissions from screenings, textiles and miscellaneous combustibles, which together comprise 18% of MSW in our compositional data. Hence, methane emissions per tonne of MSW are sensitive in our model (and obviously in practice too) to the waste composition. In particular, looking at the USEPA data by material type (see Annex 3), methane generation is sensitive to the distribution of 'paper' across paper and board types, as well as to the distribution across putrescible components, especially the relative proportion of food scraps.

[³⁶ EIRU (1992) report similarly large ranges in a review of theoretical studies.

³⁷ It has been suggested that the USEPA (1998) figures are low partly for political reasons (since this reduces the US contribution to global warming from landfill gas).]

Discrepancies are magnified when one looks at the assumed energy delivery from MSW landfilled. Calorific values for 1 m³ of landfill gas of the order 19MJ/m³ landfill gas have been quoted (Entec 1999a). Almost equivalently (if one assumes 50% of the gas is methane) a calorific value for methane of 39.75MJ/m³ was used by Manley 1990 (in Powell 1992). Entec (1999a) then estimate energy content of landfill gas per tonne of MSW by:

1. estimating gas collection efficiencies; and

2. the percentage of landfill gas utilisation over the lifetime of the landfill,

and then multiplying these factors together, along with the calorific value mentioned above, to arrive at a calorific value of the gas collected. This is then further reduced by a factor representing the efficiency of the engine used to generate electrical output (Entec (1999a) use a figure of 40% for the engine efficiency).

Using the approach taken by the US EPA, for every Metric Tonne Carbon Equivalent (MTCE) of CH₄ collected one generates 646 kWh of electrical energy, which translates to approximately 115-150 kWh per tonne MSW (depending upon assumptions concerning oxidation rates). This can be compared with an estimate of only 79 kWh in CSERGE et al (1993) and 298-475 kWh (worst and best case scenarios) in Entec (1999a). In our study, we have used US EPA (1998) methane data and then followed an approach similar to that used by Entec (1999a) to derive energy recovery figures. This has meant converting from MTCE CH₄ from the US EPA report to m³ of CH₄ for the purposes of understanding energy generation (and we have assumed a conversion factor of 238 m³ per MTCE CH₄). This means that at 35% engine efficiency rates and 30% landfill gas collection efficiency, the energy generated is 59 kWh, whilst at 60% collection efficiency, the energy generated is 118kWh (in our model, this is independent of oxidation rates).

Note that this approach does still introduce the question about whether one is interested in marginal changes or those over the lifetime of the landfill. Arguably, if one is landfilling in the period after gas collection has begun, but well before closure, the distinction is irrelevant. Early in the lifetime, and late in the lifetime, the issues are more important since then, the composition of landfilled waste affects the efficiency of gas capture. Different materials decay at different rates, and degrade more or less completely under landfill conditions. EIRU (1992) note that factors affecting methane production include particle size, size of lysimeters, refuse composition (including nutrient content, i.e., carbon, nitrogen, phosphorus; and pH), density, temperature, moisture, size and depth of landfill, site geology, nature of intermediate cover, nature of lining or capping, local climate, and the presence or otherwise of sludge or other methanogenic inoculum, or biomethanation inhibitors.

Materials such as paper are known to degrade less well than putrescibles under landfill conditions (see Table 28). What our model does not account for in any way is the rate of landfill gas generation. This is important in considering the economic feasibility of landfill gas collection, and consequently, the net environmental impact of landfilling. Removal of relatively inert materials from the waste stream can increase the rate of methane production, bringing forward the onset of economically feasible levels of gas generation and reducing the time period of landfill stabilisation. This means that over the life of the landfill, a greater proportion of the gas generated would be collected, and more would be available for electricity generation and conversion of CH₄ to CO₂. Work by EIRU (1992) suggested that degradability of waste changed significantly after the introduction of a recycling scheme in Stockbridge. This is shown in Table 29. The key changes are that the readily degradable fraction of residual waste landfilled increases whilst the inert fraction falls.

Table 28: Biodegradability of Waste Components

Material Biodegradability (%) ¹ Degradability Category (%)

Sources: ERL (1990), EIRU (1992) and Mosey and Mistry (1991)

Table 29: Changes In the Degradability of Household Waste Collected in Stockbridge

Source: EIRU (1992) calculated using data from WSL and Poll (1991)

In the ideal world, one would model gas generation with more dynamic profiles. The nature of waste landfilled influences the completeness of gas collection, though this is also influenced (for a specific waste fraction) by the period at which one landfills the material relative to closure. This is illustrated graphically in Figure 3 below, in which it is assumed that landfill gas collection becomes 'cost ineffective' below a certain rate (note the curves are drawn for illustrative purposes only and are not intended to be perfect representations of the post-closure situation). The volume of emitted gas is equivalent to the integral under the decay curve once the rate of generation has fallen below the cost-effectiveness cut-off.

Figure 3: Effect of Rate of Gas Generation Post-closure on Uncollected Gas Volumes

In environmental terms, the smaller is the area ABC, then other things being equal, the better will be the performance of the landfill. On the other hand, removing paper would, under the USEPA assumptions, remove a net sequester of carbon. Arguably, it then becomes important to know what alternative use is being made of the paper, but as we shall see, a clear-cut decision as to what is likely to be the 'best' option is likely to be elusive.

6.4.3 Results

The DETR estimate regarding waste composition are the subject of considerable disagreement amongst those who believe that the significance of the putrescible fraction in particular has been understated in that compositional analysis. This appears to be a common view among those who have conducted analysis of actual waste streams as opposed to conducting 'waste analyses' on the basis of what may be outdated, or simply incorrect, linkages between Acorn social groupings - themselves outdated - and waste generation.

Some information on composition can be found for London in Ecologika (1998). Like that study, work by Network Recycling in South Gloucestershire also suggests a putrescible fraction of the order 40%. We have used a composition as shown in Annex 5 which we suspect is a reasonable approximation to the composition of municipal waste. It should be pointed out. However, that there is no obvious set of statistics to use in this area. The actual typical tonne will vary across authorities.

Questions can be asked as to which values for the key variables should be used. The USEPA (1998) reports sources and commentators as suggesting that oxidation at the cap could range from 5-40%, whilst gas collection efficiency might range from 60-95%. We have accepted the former range, but the figures for the latter are relatively high. Willumsen (1997) suggests that only about 25 to 50% of the gas produced in landfills is recoverable. ETSU (1996) also suggests that collection efficiency was unlikely to be greater than 50%. We show

cases with 40% and 70% gas collection efficiency. On the efficiency of the engine, Entec (1999a) use 40% as the 'best case'. We have used a range from 25% to 40%. The total externalities are shown for differing combinations of the energy recovery and landfill gas modules (we have combined the high externality gas estimates with the high externality estimates for displaced energy, under different assumptions concerning the energy source displaced, and vice versa).

Tables 30-33 show our results for the composition of MSW we have used. What we have done is to start with the low oxidation, low gas collection efficiency and low engine efficiency scenarios and we have changed each of these, in turn, to the higher figure (generating the 4 tables). The effect of using wide ranges of external cost estimates produces results that are, unsurprisingly, rather different to those in which relatively no ranges in per unit externalities were attributed to the emissions of specific pollutants. In particular, the high externality adders highlight the significance of undertaking measures to collect gas, and to recover energy from it since this places a higher premium on the replacement of other energy sources.

Table 30: Estimates of Some of the External Costs of Landfilling a Tonne of MSW (£ per tonne MSW)