Friends of the Earth > Community > Campaigning resources > ISEW > How ISEW terms are calculated
This briefing gives more technical detail for the individual terms in the ISEW. It is taken directly from Chapter 3 of the paper Sustainable Economic Welfare in the UK, 1950-1996 by T. Jackson, N. Marks, J. Ralls and S. Stymne.
The columns below refer to the main table of data, which breaks down the ISEW into its constituent elements for each year.
Column A: Year
Column B: Consumer Expenditure
Column C: Income Inequality
Column D: Adjusted Consumer Expenditure
Column E(+): Services from Domestic Labour
Column F(+): Services from Consumer Durables
Column G(+): Services from Streets and Highways
Column H(+): Public Expenditure on Health and Education
Column I(-): Consumer Durables: difference between
expenditure and value of services
Column J(-): Defensive Private Expenditures on Health
and Education
Column K(-): Costs of Commuting
Column L(-): Costs of Personal Pollution Control
Column M(-): Costs of Automobile Accidents
Column N(-): Costs of Water Pollution
Column O(-): Costs of Air Pollution
Column P(-): Costs of Noise Pollution
Column Q(-): Loss of Natural Habitats
Column R(-): Loss of Farmlands
Column S(-): Depletion of Non-Renewable Resources
Column T(-): Costs of Climate Change
Column U(-): Costs of Ozone Depletion
Column V(+): Net Capital Growth
Column W(+): Net Change in International Position
Column X: Index of Sustainable Economic Welfare
Column Y: Per capita Index of Sustainable Economic
Welfare
Column Z: Gross Domestic Product
Column AA: Per capita Gross Domestic Product
In the following subsections, we describe the individual factors which contribute to the revised UK Index of Sustainable Economic Welfare - these are set out in columns labelled: A, B, C... The methodology used to construct many of these adjustments mostly follows procedures laid down for the US ISEW (Daly and Cobb 1989), and later revisions of the same index (Cobb and Cobb 1994).
Where appropriate we have taken on board methodological amendments suggested by Jackson and Marks (1994). However, we have made three major revisions of the ISEW methodology to account for what we regard as important responses from the critical literature.
These major revisions affect (i) the treatment of income inequality, (ii) the treatment of long-term environmental damage due to global warming, and (iii) costs of ozone depletion. In addition, we have made a number of minor adjustments to other factors to account for the emergence of new studies or better data since the earlier version of the index.
All costs in what follows are real costs (net of inflation) converted to 1990 pounds sterling. In making this conversion we have generally used an implicit GDP inflator calculated from the ratio between published figures on GDP at current prices and GDP in 1990 prices. Where appropriate for methodological reasons we have used different inflators, as indicated in the text.
The letters allocated to the individual factors refer to the columns of the matrix and are as near as possible the same as the column letters in the revised US index (Cobb and Cobb 1994).
The revised UK-ISEW has been prepared for the years between 1950 and 1996 inclusive. The intention was to construct an index directly relevant to current policy decisions, and directly comparable with current economic indicators. It ought to be acknowledged that at this stage, it is still not straightforward to accumulate the appropriate data sets for very recent years.
It is significantly more difficult to provide an up-to-date indicator than it is for conventional measures such as the GDP, where factor-cost economic data is prepared as a matter of course by bodies such as the Treasury, the Department of Trade and Industry and the Office for National Statistics (ONS). However, this is changing rapidly, and new data sets are now available which will make the job of preparing future indicators considerably easier.
For example, the ONS published some environmentally-adjusted satellite accounts for the first time last year (Vaze and Balchin 1996), and estimates of the value of household labour for the first time this year (Murgatroyd and Neuburger 1997).
In spite of these valuable new sources of data and analysis the calculations presented in this paper have relied to some extent on best-guess estimates and extrapolations to extend some parts of the index to 1996.
The revised UK-ISEW takes as its basis the personal consumer expenditure as provided by the UK National Accounts, and published in the Economic Trends Annual Supplements (ET, various years). Personal consumption rose, in fact, faster than GDP during the period in question: growth in GDP was 194%, while growth in personal consumer expenditure was 206%.
This reflects a significant trend in the UK, particularly over the later years of the index, to reduce government fiscal intervention in the economy. It contrasts markedly with policy in certain other countries. For example, in Sweden, the proportion of private consumption has decreased since 1950 while the proportion of public expenditure has risen with respect to GDP (Jackson and Stymne 1996).
In spite of the broad welfare theoretic interpretation mentioned in the previous section, there are a number of difficulties in using consumption as the basis for a welfare index. One of the most intractable problems arises from the assumption inherent in this equation that one unit of consumption is much the same as another in terms of delivering welfare.
The fact that this is not generally the case, and that in fact consumption offers a diminishing return in terms of welfare, is one of the reasons for adjusting the aggregate consumption measure for the effects of unequal distribution of incomes in the economy: a pound in the pocket of a rich person is generally worth less than a pound in the pocket of a poor person.
This same effect of diminishing returns also means that each pound in the pocket is worth slightly less than the last pound in the pocket in terms of each individual's welfare. The richer we get the smaller is the welfare improvement associated with a marginal increase in wealth.
Put in another way, consumption tends to be non-linear with respect to the welfare it delivers, whereas measures like the GDP tend to assume linearity. The same problem is reflected in another well-known fact: that welfare is better correlated with relative than it is with absolute levels of income. 16
The problem of non-linearity is the subject of on-going research in economic theory.
Generally speaking, if welfare is convex with respect to consumption, then it is possible to approximate the non-linear relationship with a linear one, at least locally (Dasgupta and Mèler 1991). However, there are some indications that this condition does not always hold.
Indeed, there is plenty of evidence to suggest that increased levels of consumption, particularly the wrong kind of consumption, can impede or even diminish human well-being.
Evolutionary psychologists have evoked what they call the 'mismatch theory' in seeking to explain for example why "rates of depression have been doubling every decade, suicide is the third most common cause of death among young adults in North America, 15% of Americans have had a clinical anxiety disorder" (Wright 1995).
In one sense there is nothing more or less sophisticated here than the time-honoured insight that wealth does not necessarily make people happy. But addressing such an important issue must remain beyond the bounds of the present paper.17
Income inequality has increased considerably in the UK over the period of the study. A time series Gini coefficient18 shows relative stability in the distribution of incomes during the first decade of the study period. During the next decade and a half there was even some improvement in the equality of distribution. But the distribution became considerably more unequal over the last two decades of the study.
It is perhaps worth noting that this is a particular feature of the UK case, by comparison with certain other countries19, not unrelated to the trend in fiscal policy mentioned in the preceding section.It will turn out to have a significant impact on the overall shape of the UK-ISEW. In the original ISEW, the index of inequality was used to provide an adjustment to the welfare measure.
Personal consumer expenditure was 'weighted' by dividing the raw data by the inequality index and multiplying by 100. The weighted values of personal consumption were thus lower the higher the inequality index. There are two specific problems with this approach.Firstly, it provides at best an indication of the relative trend in welfare because the value of the weight applied to personal consumption is calculated relative to 1950.
Secondly, and probably more importantlariness. There is no clear welfare-theoretic interpretation to the difference between actual consumption and weighted consumption for any individual year. Hence there is no way of justifying the particular level of weighted consumption arrived at in terms of welfare.
What the weighted consumption method does essentially is to make a rough and ready relative adjustment to income levels whose direction is determined by the change in income distribution relative to 1950. In the absence of anything better, it is one way in which to indicate relative changes in welfare.
In fact, however, there are other methods available, some at least of which might allow a welfare-theoretic adjustment to incomes on the basis of measured inequalities in income distribution. One such method is the so-called Atkinson index which attempts to measure the equivalent equalised income associated with each unequal distribution of income (Atkinson 1970).
For this study, we have therefore carried out an analysis for the UK based on this method, and used it to adjust personal consumption levels in the UK-ISEW. The Atkinson index falls into a group of inequality indexes based on the social welfare model. Dalton (1920) was the first person to propose measuring social welfare W as the aggregate of the utilities U(yi) associated with each income yi. Thus:
W = Si U(yi)
Dalton is also often referenced as the first to argue that a measure of income inequality could be based on this social welfare model. In practice, of course, what is required to carry out this measurement is a way of relating different incomes to the utility associated with them.
Atkinson suggested that it is possible to derive the total welfare corresponding to a particular distribution of income according to the following formula:
W = Y*[Si (yi/y)(1-e).pi](1/1-e)
where Y is the total income, yi is the mean income of the ith group, y is the mean income of the total income population, pi is the proportion of the total income population in the ith group, and e is a factor which represents the weight attached by society to inequality in the distribution of income.
The Atkinson index is then defined by:
I = 1-W/Y.
Since welfare falls as the inequality of income distribution rises, the Atkinson index provides an increasing function of inequality in the economy, defined by the difference (normalised with respect to total income) between the total income and the welfare which it delivers.
In a perfectly distributed economy, yi = y for each income group, and so the welfare level is given by:
W = Y*[Si pi](1/1-e) = Y;
and the inequality measure I reduces to 0, as would be expected. The factor e is an important parameter in the measure. It represents society's preference for equality of distribution of incomes20. Since it is possible to conceive of societies which have a positive preference for an unequal distribution of income, it is clear that e can take both negative and positive values21.
When e is zero, society is indifferent to the distribution of income, and welfare again reduces to the total income in the economy:
W = Y*[Si (yi/y).pi] = Y. 22
This parameter therefore allows explicitly for the possibility of attributing different welfare levels according to different attitudes towards inequality in society. In principle the value of e can be determined in a given society by using attitudinal survey data on the level of well-being associated with different income levels.
Schwartz and Winship (1980) suggested that "after reflecting on the different interpretations of e, most sociologists would agree that when using Atkinson's measure to address normative questions, e should be between -0.5 and 2.5".
The income distribution data for the UK has been taken mainly from a report prepared by the Institute for Fiscal Studies providing decile shares of post tax income for the years 1961-1991 (Goodman and Webb 1994, Table 2.3). For the years not covered by this study, we have used an index based on Gini coefficient data to extrapolate the Atkinson index. 23
As far as the parameter e is concerned, we adopted a central value of 0.8 suggested for the UK on studies of consumer behaviour24. However, we are not entirely convinced that it is appropriate to accept a judgement about social welfare inferred from existing patterns of consumption.
The acceptance of (or aversion to) income inequality is an issue which legitimately ought to refer to a social evaluation (for instance by surveys of public attitudes) as well as to market behaviour, and such an evaluation is not yet available for the UK. We have therefore carried out a sensitivity analysis based on the range suggested by Schwartz and Winship.
This sensitivity analysis demonstrates that a slightly higher value of e would considerably depress the index, particularly over the last fifteen years of the study during which income inequality in the UK has reached 'unprecedented' levels (Goodman and Webb 1994, Goodman et al 1997).
Here, values for the Atkinson index for the UK assume an e value of 0.8.
In determining the welfare equivalent of personal consumer expenditures in the UK, we have assumed, for the sake of this study, that it is possible to apply the Atkinson index equally to expenditure as to income25. Thus column D shows personal consumption C multiplied by 1-I, where I is determined by calculating the income-based Atkinson index using a value of 0.8 for the parameter e.
Column D now provides the basis for the rest of the index to which additions and from which subtractions are made to account for the various environmental and social factors detailed below.
It is now widely recognised that household services contribute to economic welfare even though they are not (except for a relatively limited domestic labour sector) traded on the market. The integration of unpaid housework into GDP was recommended by the closing Nairobi Conference of the United Nations Decade for Women.
Agenda 21, the Rio Earth Summit's 'blueprint for sustainability', declares that "unpaid productive work such as domestic work and child care should be included, where appropriate, in satellite national accounts and economic statistics".
The most recent System of National Accounts (SNA 1993) argues against the inclusion of unpaid work within the core accounts on the grounds that "imputed values have a different economic significance than monetary values".
However, it does recommend the use of satellite accounts for production which falls outside the conventional production boundary. One of the advantages of satellite accounts is that they do not necessarily need to be tied to monetary valuations26. In the UK, accounting for housework was the subject of an unsuccessful private member's Bill in 1989.
Recently however, the Office for National Statistics has published a preliminary attempt to provide monetary satellite accounts for the UK for the year 1995.
The ONS study uses time use data from the 1995 Omnibus Survey (Church et al 1996) and a variety of different economic assumptions to compute several different estimates of the value of unpaid household work in the UK, ranging from about 40% to about 120% of GDP (Murgatroyd and Neuburger 1997). The range of economic assumptions in the ONS study relate to two different aspects of the accounting methodology.
The most important of these is the choice of a 'shadow price' for domestic labour which could be set equal to the wage rate of the person doing the work, the wage rate of a domestic worker in the commercial sector, or a general industrial wage rate.
The first of these wage rates assumes that the person values his or her time spent in housework at the same rate as the time spent in paid work, and at the margin the value of an hour of housework is equal to the value of an hour of paid work.
From a production account perspective this would only be valid if time spent in domestic labour constrained the available supply of paid labour, which is in turn only true if the demand for paid labour is unlimited. This perspective also leads to the contentious conclusion that domestic labour carried out by a doctor or a lawyer is worth more than domestic labour carried out by a nurse or a secretary.
Whatever the truth of this assumption from a production account perspective, it is clearly problematic to suggest that the welfare value of household labour is different depending on whether it is a lawyer's family or a nurse's family which benefits from the labour. The main alternative is to value domestic labour according to the cost of employing a domestic worker in the commercial sector.
The assumption implicit here is that at the margin the householder would be indifferent between carrying out the work, and employing a domestic worker to do it for them. For the purposes of this study, we have therefore chosen to value unpaid labour at the wage rate paid to domestic workers in the commercial sector.
This is also the shadow wage rate used by Daly and Cobb (1989) in their original ISEW. The main source of our data on annual hours in relevant household activities was Gershuny and Jones (1987) who studied the changing work/leisure balance in Britain between 1961 and 1984.
Their study uses data from three sample years. The first two samples were conducted in 1961 and 1974/5 by the audience research department of the BBC. The third was a national time-budget sample funded by the Economic and Social Research C and Jones data was divided into 8 broad categories of which domestic labour is one category.
Domestic labour was split into seven subcategories comprising: cooking and washing up, housework, odd jobs, gardening, shopping, child care and travel. Of these categories we decided to exclude gardening on the grounds that it is often a leisure activity rather than 'productive' domestic labour27.
Domestic travel was also excluded because an increase in the time spent in travel (for instance to go shopping or taking children to school) seems to us not to represent any clear increase in welfare. In addition, many shopping activities are recreational in nature rather than 'productive', and since we are excluding leisure from this account, they should also be excluded.
Thus the final account for household labour included cooking and washing up, household repair and maintenance, and child care in the home. The earlier UK-ISEW used the three data points provided by the Gershuny and Jones data set to extrapolate a time series over the period of the study.
Since the latest year of the index was 1990 and the latest year of the time use data was 1985, this was probably a reasonable extrapolation. In extending the index to 1996 however, the extrapolation becomes considerably more tenuous, and we therefore used, as a further data point, the time use data provided by the 1995 Omnibus Study.
Since this data is not categorised in exactly the same way as the earlier data, we have had to adapt some of the categories to fit as closely as possible with the earlier categorisation. In summary, we have included the categories of food preparation, clothing care, care of the home, and home improvement exactly as detailed in the Annex of the ONS study (Murgatroyd and Neuburger 1997).
The category 'care of family/household' from the Omnibus Survey clearly includes additional factors not included in the earlier category 'child care'. In the absence of a detailed breakdown, we have assumed that the time spent in child care was the same in 1995 as it was in 1985 (for both men and women)28.
In order to calculate the total annual hours spent in domestic labour in the UK, we need to multiply the time-use data (expressed in terms of minutes per average day per person by sex) by the total population set for which the time-use data is valid.
The time budget survey by Gershuny and Jones is based on a sample of 25-60 year olds. However, it is obvious that productive work is done at home by those over 60, and by those under 25. We have therefore applied the Gershuny and Jones time-use data to a larger population set, namely from 16 to 6529.
Time series data on hourly wage rates is available back to the 1950s, but in general only for manual, trade-unionised labour. A wage rate for female home and domestic workers was available from 1968 to 1990 (DEmp, various years).
In addition there is a wage rate available from 1968 to 1994 for both male and female workers in a subcategory of miscellaneous services entitled 'other cleaners'30. In order to extend the analysis prior to 1968, we have used annually published data from the Ministry of Labour Gazette (MLG, various years).
Time series data on wages paid in the laundry industry was used to create an index of wage rates over the period 1950 to 1968, and this index was then used to extrapolate shadow wage rates for domestic labour for the early part of the study.
For the later years of the study we have estimated the female domestic wage rate by taking an average of the ratio between the female domestic wage rate and the female 'other cleaners' rate for the 1980s, and applying this ratio to the data on female other cleaners in the 1990s. The final two years' wages data were extrapolated on the basis of GDP growth.
The average of the male and female rates was then multiplied (for each year) by the total hours spent in productive domestic labour in order to arrive at a monetary value to be included in the index31. The historical wage rate we have used as a shadow wage for domestic labour increased in real terms by a factor of just over 2.6 during the period of the study.
Since the hours spent in domestic labour decline slightly, and the population between 16 and 65 increases slightly, this leads to an increase in the value of household services in the UK between 1950 and 1996 by a factor of 2.8.
In view of the relative magnitude of household services in the index as a whole, this increase exerts a considerable (positive) influence over the shape of the index.
In the US index, the value of household labour grew by a factor of only 1.59 between 1950 and 1986, exactly keeping pace with population growth (Cobb and Cobb, 1994). Daly and Cobb (1989) seem to have argued for a constant shadow cost for domestic labour.
"Even if the time of those performing the work were now valued more highly in the market-place (a problematic assumption given the long-term stagnation in real wage rates), that should not affect the value of housework". Given these differences between the UK index and the US index, we looked at two further ways of accounting for the value of household labour.
Firstly, we looked at the rate of growth of average real wage rates (rather than specific wage rates) in the UK over the period. Although this average real wage rate grows slightly less fast than the wage rate for domestic labour, it would still imply a factor of two increase in the value of household services over the period.
Secondly, we looked at the impact of choosing a constant shadow wage rate for domestic labour over the time period, on the grounds that the real value of such labour remains unchanged. Choosing the 1990 wage rate for domestic labour (for example) we find that the value of household services remains more or less constant over the years 1950 to 1996.
It seems clear that there are methodological issues here which are complex and difficult to resolve. Is time more productive or less than it was in 1950? Is leisure more valuable or less? And if one is to use a constant shadow wage rate over the period, should this shadow wage rate be closer to wage rates paid today or to wage rates paid in 1950?
Given the impact which these methodological decisions have on the shape of the index, there is clearly considerable scope for further work here.
For the purposes of this exercise, we decided to retain the shadow wage rate based on the historical (varying) wage rates applicable to domestic labour.
In the light of the overall index, this decision is conservative with respect to the shape of the index over the study period, even though it has a considerable impact (in the early years) on the absolute magnitude of the difference between the ISEW and the GDP curve.
Had we chosen a constant 1990 wage as a shadow wage rate, the UK-ISEW would have started much closer to the GDP curve, but the shape of the ISEW curve would have differed even more dramatically from the shape of the GDP curve32.
A part of what is denoted as consumer expenditure in the national accounts relates to expenditure on consumer durables.
However these expenditures do not, strictly speaking, represent consumption within the accounting period since durable goods may last up to ten or twelve years, contributing to welfare over a period of time to come. Equally, some of the 'consumption' during the accounting period will relate to expenditures on durable goods from previous accounting periods.
It has therefore been argued by Daly and Cobb that the appropriate way to treat expenditures on consumer durables is to account for the flow of services arising during the accounting period from the net stock of consumer durables in that period.
The Daly and Cobb ISEW therefore adds the value of the service provided by the stock of consumer durables under column F and later (Column I) subtracts the expenditure on durables.
For the US-ISEW, Daly and Cobb took year on year data on the net stock of consumer durables, and multiplied this value by a percentage designed to reflect the ratio between services and stock over the study period.
In the initial version of the index, this ratio was taken as 10% roughly in line with the ratio between housing services and net housing stock. In the revision to the index (Cobb and Cobb 1994), the 10% figure was revised upwards to 22.5% on the grounds that this "would account for both the combination of imputed value (interest) and depreciation".
For the UK, we have based our adjustments for this column on estimates by Patterson (1992) of the flow of services from consumer expenditure. Patterson's results are presented in terms of total consumer expenditure, rather than expenditure on consumer durables as a separate category of expenditure.
In the UK index, therefore, instead of adding in the services from consumer durables (in column F) and subtracting the expenditure on consumer durables (column I) we have only one column (which we place under column I) which subtracts the differences calculated in the Patterson work between consumer expenditure and the value of the services flowing from that expenditure.
Further discussion of this adjustment is presented below.
Generally speaking, US government expenditures on public services are excluded from Daly and Cobb's ISEW on the grounds that they are 'largely defensive in nature'. However, Daly and Cobb regarded some government expenditures in the US as non-defensive, and therefore included them as positive contributions to welfare in the index.
In particular, expenditures on streets and highways were deemed to be non-defensive. It was therefore appropriate to include some measure of the services flowing from government expenditure on streets and highways in the index.
For the US case, the annual value of services flowing from streets and highways was estimated by takingreets and highways, estimating the net stock, and then subtracting a proportion to account for non-welfare services (commuting). In the UK as in the US, some local government services such as water, sewerage, and public transit, are provided for a fee, and are therefore already implicitly included within personal consumption.
In contrast to the US however, UK roads can be considered to be financed through a taxation system on vehicle usage, and through tax on fuel.
Since these payments are already included in personal consumption it is inappropriate to add an additional expenditure here. We have therefore omitted this column from the UK index, for the moment. We acknowledge that this is an incomplete solution to the problem.
In accordance with the basic methodology of the index (see section 2 above) we should ideally subtract the vehicle tax component of personal consumption (expenditure), and add in the services flowing from the expenditure.
However, we have not been able to find sufficient data to carry out this ideal calculation.
In an expenditure-based measure of GDP, it is customary to include public expenditures on health and education as a contribution to public consumption.
The question of whether or not to include such expenditures in an index of welfare is complex.
It is clearly the case that some at least of the money which governments spend on health and education contributes to the physical well-being of the nation, the building up of its skill base, and the enjoyment of its citizens.
On the other hand, we have already noted above that the absence of equilibrium markets for public goods and services makes it difficult to use consumption in this sector as a proxy for welfare provided. It is problematic to measure output-based concepts of health and education on the basis of input-based expenditure assumptions, and to impute increasing welfare levels on the basis of increasing expenditures.
Furthermore some health expenditures may be purely defensive against activities counted elsewhere in the economy as consumption, and it would seem inappropriate to include both sets of expenditures as additive contributions to welfare.
For example, it has been estimated that treatment of smoking-related illness in UK costs the National Health Service in excess of œ600 million per year (HEA, 1993).
Since consumer expenditure on smoking is counted under personal consumption as a contribution to welfare, it is clearly inappropriate to add the costs of treating the diseases caused as a direct result of that expenditure into the welfare measure.
Indeed it may even be appropriate to subtract such costs from welfare.There is a further problem in relating expenditures in the public sector to increases in human capital: greater knowledge base and skills or better health.Daly and Cobb cite, for example, statistics illustrating an increase in 'restricted activity days' in the US coincident with an increase in expenditures on health.
As for educational expenditures, they cite Thurow's (1975) 'job competition model' of the relationship between education and income, in which "workers are hired on the basis of their 'relative position in the labor queue', which is determined more by their academic degrees than by their actual job-related skills" (Daly and Cobb, 1989 p403).
The point here is that academic performance is used more to create or reinforce relative differences in earning ability than to generate overall increases in the productivity and economic welfare of the nation. Daly and Cobb do not deny that education and health expenditure contribute to human capital. They simply argue that the relationship between expenditure and the services flowing from human capital formation is difficult to establish.
They therefore choose to exclude investment in human capital from their calculations "even though we recognise its theoretical importance in sustainable economic welfare" (op cit, p402).
As a result of this choice, Daly and Cobb included only that fraction of public expenditures on health and education which they believed to represent non-defensive consumption expenditures.
On this basis, the revised ISEW methodology (Cobb and Cobb 1994) includes one half of all medical expenditures, the excluded half being assumed to be defensive expenditures, and one half of all higher education expenditures, which are assumed to represent pure consumption.
We have followed this revised procedure in the UK. Data on public expenditures in health and education have been taken from the Annual Abstract of Statistics (AAS, various years33), converted to constant 1990 prices using the implicit GDP inflator.
This column uses an analysis due to Patterson (1992) to compute the difference between expenditure on consumer goods, and the service flowing from those goods during the accounting period (see column F).
Although this procedure is not generally carried out in the national accounts, it is not without precedent in the economic literature. For example, when Friedman proposed his permanent income hypothesis he used the term consumption "to designate the value of the services that it is planned to consume.
The term is generally used to designate the actual expenditures on goods and services. It therefore differs from the value of the services it is planned to consume because of additions to or subtractions from the stock of consumer goods 34."
Patterson calculated the stock in each year for each of a set of durable goods, categorised according to the UN classification.
He then calculated a service value flowing from each class of goods by applying a user cost "which can be thought of as having three components: the interest foregone in holding a consumer durable; the depreciation charge; and the capital gains or losses over the holding period".
He used lifetime assumptions for each category based on the 'medium-life' estimates provided by the former Central Statistical Office (CSO).
The value of the services flowing from the stock of consumer durables was then added to the expenditure on non-durable goods to obtain the value of services flowing from consumer expenditure.Patterson's calculations covered the period from 1964 to 1989. For the earlier and later years of the study we used a regression analysis to extend the calculation.
For the purposes of presentation in this study, we calculated the difference between the time series service value extrapolated from Patterson's work and the total consumer expenditure (as given in Economic Trends) at constant 1990 prices. This difference is then subtracted from the index in column I.
In theory, the separate accounting of expenditures on durables and the value of services received from the stock of durables could allow us to pick up changes in the obsolescence of durable goods in the economy.
For example, short-term obsolescence of durable goods tends to inflate consumer expenditure without adding to the service flowing from the stock, whereas improvements in the durability of goods would increase the service value associated with those goods without increasing personal consumption.
In practice, the task of computing the flow of services from a stock of durables is dependent on a complex set of information about lifetimes, opportunity costs, and depreciation charges as well as some means of evaluation of the services flowing.
It is worth pointing out that, in Patterson's work, assumptions about lifetime and depreciation rate are taken as constant over the period of the study.
This means that we cannot identify here any potential reductions in the service value of consumer expenditure on durables as a result of increased technical, economic or fashion-driven obsolescence.
When lifetimes are assumed shorter, the depreciation rates are higher. A study by the former CSO indicates that the value of the stock of consumer durables may be as much as twenty percent lower (in 1986) using short-life assumptions, than using the medium-life assumptions35.
This implies that any tendencies towards increased obsolescence over the study period would increase the difference between expenditure on consumer durables and the value of the services flowing from them even more than is witnessed in the present analysis.
The current result should therefore be regarded as conservative with respect to increasing obsolescence36.There is clearly scope for more detailed work on the impacts of obsolescence on welfare.
In column H, selected public expenditures on health and education were included as non-defensive contributions to welfare. If the accounting for expenditure on health and education is to be consistent, we need to subtract from total consumption those expenditures in the private sector which are deemed to be defensive.
As in the case of public expenditure, one half of the expenditure on health and one half of the expenditure on further education are taken to be non-defensive consumption expenditures. The rest of private expenditure on health and education are taken as defensive expenditure, and subtracted from the total index.
In the absence of data at the level of national year on year private expenditures on health and education, we have relied on data from the Family Expenditure Surveys37 in order to calculate the percentage of total weekly family expenditure on health and education, and the relevant portion of that expenditure deemed defensive under the criteria discussed above.
We have then applied these percentages to consumer expenditure at constant 1990 prices, to derive estimates of the total national defensive private expenditure on health and education.
The cost of commuting to work is regarded as lar patterns of urbanisation and settlement, and is therefore subtracted from personal consumption. Our estimation of these costs for the UK is made by multiplying the total expenditure on each mode of passenger transport (rail, bus and coach, cars) by the percentage of total mileage attributable to commuting.
The underlying assumption here is that the costs of each mode of transport are proportional to the mileage. Whilst this is clearly not true at the individual level, it is not an unreasonable assumption in the aggregate.
Public transport expenditure tends to have higher costs per mile for shorter commuting-type journeys, and the costs of shorter car trips are higher per mile than those for longer recreational trips.
On the other hand, some of the costs of motoring are fixed capital costs, and it is not clear in what proportion (if at all) these costs should be allocated to commuting journeys. Total expenditures on car, rail, underground, bus and coach transport were obtained from two sources.
For the years 1952 to 1972 transport expenditure statistics were published by the Ministry of Transport in Passenger Transport in Great Britain (MoT, various years). For the years 1964 to the present day, data on public transport expenditures were published by the Department of Transport in Transport Statistics GB (DoT, various years)38.
Since there is relatively close agreement between the different sets of statistics in the overlapping years, we used the DoT data for the years 1964 to the present, and the MoT data for the years 1952 to 1964.
Estimates for 1950 and 1951 were extrapolated linearly. The current costs were converted to constant 1990 prices using the implicit consumer expenditure inflator. Data on the breakdown of motoring costs between businesses and consumers were available only until 1985. For the later years of the study we used a ten year average to calculate the proportion of total motoring costs spent by consumers.
The National Travel Surveys provide data on the relevant proportion of passenger miles attributable to commuting for the years 1965, 1973, 1976, 1979, 1986, 1991, 1993-95. For the intervening years we have used a linear interpolation.
In the revised version of the US-ISEW Cobb and Cobb introduced a new column to account for personal expenditure on pollution abatement and control, such as the purchase of air filters or on-line water filters.
They argued that these costs should be subtracted from personal consumption on the grounds that they are defensive in nature.
The time series data on personal pollution control used in the revised US-ISEW was supplied by Carson and Young in their contribution to Cobb and Cobb (1994).
For the UK, we have assumed that household expenditure on environmental protection represented 5% of the total estimated environmental expenditure of œ40 billion in 1990, in line with estimates in Economic Trends (ET, 1992).
In the absence of time series data for the UK, we have indexed this figure according to the series data in the revised US index.
This assumption is defensible only to the extent that patterns of environmental awareness and consumer spending are similar between the UK and the USA.
For the years after 1990 we have assumed an annual growth rate in environmental expenditure equal to the average annual growth rate between 1988 and 1990.
This column estimates the costs of automobile accidents, and subtracts them from personal consumption on the grounds that they are defensive expenditures, and should not be counted as additional contributions to welfare. Since defensive hospital and medical costs are assumed already to have been counted for under the adjustments made in columns H and J, however, this column accounts only for the non-injury related costs of road accidents.
Taylor (1990) estimated the average (non-injury) costs of road accidents involving injury at œ2,201 in February 1988, and the costs of accidents not involving injury at œ739. 39 In the same work the author estimated the ratio of accidents not involving injury to the number involving injury as 8 to 1.
Time series data on the total number of accidents reported to the police was taken from Department of Transport data (DoT, 1993) and for later years from the Annual Abstract of Statistics. 40
In the absence of more extensive data on changes both in the costs of accidents and the ratio involving injury, we have used the 1988 figures for each of these parameters and applied them to the time series data on total numbers of accidents to get estimated costs for road accidents over the period 1950-1996.
To estimate the changes in the impacts of water pollution over the period from 1950 to 1996, we have used the national river quality surveys of 1958, 1970, 1975, 1980, 1985-1994. In the absence of data relevant to the years prior to 1958, we have assumed a rate of decline in water quality equal to the rate of improvement in water quality evident in the years immediately after 1958.
Implicitly we are assuming that the commencement of water quality surveys coincided with a specific effort to improve water quality, prior to which there had been some deterioration in quality as a result of the accumulation of industrial activity. River quality for the years 1995 and 1996 is assumed to be the same as for 1994.
It should be pointed out that the question of water pollution is not entirely answered by reference to the quality of inland rivers and canals. Water pollution also affects tidal rivers and estuaries, and - perhaps more importantly - groundwater supplies. The adoption of an index based on river quality assumes that the quality of other waters more or less follows the same patterns.
For tidal rivers and estuaries, the data indicates that this is indeed the case. The question of trends in groundwater pollution is considerably more complex. Groundwaters are less susceptible to point-source water pollution than are the rivers into which industrial pollution is directly discharged. On the other hand, they are susceptible to non-point source discharges such as atmospheric deposition, run-off from agricultural chemicals, and toxic leachates from landfill.
Because many of these burdens are cumulative, it is likely that groundwater quality has deteriorated, despite improvements in river water quality. However, in the absence of data we have not been able to reflect this possibility.
Creating an index of water quality from the river quality surveys was complicated by two specific factors. Firstly, the classification itself changed in 1980. The river surveys up to and including 1980 used a classification which divided the total length of inland rivers and canals into four classes: unpolluted, doubtful, poor and grossly polluted.
Later surveys are divided into five categories: good quality (two classes 1A and 1B), fair quality, poor quality and bad quality. Although the two classifications are not completely analogous with one another, it is possible to make some comparison on the basis of the year 1980, for which data exists in both series.
The second complication in creating an index is in choosing what to reflect in the index. One index of pollution was created by using the reciprocal of the percentages of the total length of rivers which were classified under the earlier scheme as unpolluted, and under the later scheme as good quality. A second method was to use an index based on the total length of rivers classified in the two worst classes under each classification.
This second method was also used by Hope et al (1992) in their pilot environmental index for the UK for the years 1980 to 1988. Choosing between these different indices is difficult. The first method assumes that it is the change in the length of the highest water quality which is most relevant in assessing the impacts. The second assumes that it is changes in the length of lowest water quality which are important.
The first method ignores importance of transfers between the lower quality classes, and the second method ignores the importance of transfers between the upper quality classes. In the absence of any clear preference criteria we decided to use an average of the two indices, rebased to 1972 and applied to the assumed costs in that year.
The OECD has estimated public expenditures on water pollution in various years (OECD 1991, p59). In the mid-seventies, for instance, public expenditure on water pollution in the UK was reckoned at around $40 per capita at 1980 prices. However, these expenditures represent defensive government expenditures (which are already excluded from the index) rather than the costs of degradation, and it is not appropriate to subtract these costs in the index.
There are some estimates of the costs of damage from water pollution in Germany and in the Netherlands. Schulz (1986) estimated annual costs in Germany in 1985 for damage to groundwater and to freshwater fishing amounting to $3 billion or around œ2.6 billion (1990 prices).
For the Netherlands, Opschoor (1986) estimates costs of between $100 and $300 million. The estimates used by Daly and Cobb for costs of water pollution in the US in 1972 suggest a much higher figure of $15.3 billion (1972 dollars) which corresponds to a total of around œ36 billion (1990 pounds).
The higher estimate for the US is due in part to the greater quantity of economic activity in the US relative to Germany, and this is obviously one of the factors which influences the extent of water pollution. The other factors which might contribute to differences in damage costs are a) technical factors such as the relative material intensity of industry and the efficiency and cleanliness of production, b) patterns of industrialisation and c) geographical factors.
The influences of these last two factors are obvious here if geographical factors such as land scarcity or water availability cause industrialisation to intensify in specific localities. Quantifying the influence of these factors is difficult however.
For the purposes of this exercise we have therefore made an assessment of costs for the UK using a simple scaling of costs in the USA on the basis of respective GDPs. Thus the cost estimate for the US, scaled according to the difference in economic activity levels in the two countries, yields an estimate of around œ4.2 billion (1990 pounds) for UK water pollution costs in the year 1972.
Thus essentially, our estimate of water pollution costs takes a cost equivalent to the same percentage of GDP attributed to water pollution costs (in 1972) as the cost for the US assumed in the Daly and Cobb work.
This column attempts to account for the negative services flowing from the environmental damage caused by air pollution in each accounting period. This factor is typical of the kind of cost which lies outside the production boundary of the traditional national accounts, and is a significant omission from conventional indicators such as the GDP.
The original Daly and Cobb ISEW used emissions of three priority pollutants - sulphur dioxide, nitrogen oxides and particulates - to provide an index of air pollution and used this index to provide a time series of cost data on the basis of a cost estimate of $30 billion for damages from air pollution in the year 1970.
Since both particulate emissions and sulphur emissions have decreased significantly over the scenario period as a result of clean air measures and fuel switching, the results suggested a significant reduction in air pollution costs towards the end of the period.
Jackson and Marks (1994) criticised this method for two main reasons. Firstly, they argued that the choice of pollutants - emissions of which all decreased over the period in question - was non-representative, and ignored the impact of certain other air emissions - such as volatile organic compounds and carbon monoxide - which tended to increase (see for example: DoE 1992, Chapter 2).
These kinds of pollutants are contributors to photochemical smogs for instance which have been responsible for some considerable deterioration in air quality in towns and cities during the past decade. Secondly, the index on which costs are based is divorced from the costs associated with the damages for individual pollutants.
For instance, many of the damage costs cited in Daly and Cobb (1989, p430) were for damages associated with acid emissions (SO2 and NOx). But the indexed costs are determined on the basis of the total of acid emissions and particulates.
This means, in particular, that the benefits of reduced particulate emissions are (incorrectly) attributed to the total emissions index. A better way to account for the costs would be to account separately for the costs of each type of emission and then to sum these costs.41
For the UK we have therefore adopted a slightly different methodology to estimate the costs associated with air pollution. First, we compiled time series data on atmospheric emissions of five key pollutants: particulates (black smoke), sulphur dioxide, nitrogen oxides, carbon monoxide, and volatile organic compounds.
We have then multiplied emissions of each pollutant by an estimate of the marginal social costs of that pollutant to obtain the costs of each kind of air pollution in each year. The total negative services flowing from air pollution in each year is taken to be the sum of these costs.
Figure 6 shows an index of emissions from each type of air pollutant over the period 1950-1996. As we remarked above, different kinds of pollutants display markedly different trends over time. Sulphur dioxide and black smoke show a considerable decline in emissions levels. Nitrogen oxides emissions increase slightly; and VOCs and carbon monoxide show considerable increases, particularly towards the end of the period.
The data on which these time series are based are taken from a variety of sources. The former Warren Spring Laboratory (Leech 1991) compiled time series data on each of the chosen pollutants for the years 1970 to 1989.
Some of this data - most notably the data relating to VOCs has subsequently been revised in the light of better knowledge by the Department of the Environment. Revised data and data for the later years of the study are taken from the Digest of Environmental Statistics (DETR 1997, eg).
In order to extend the series prior to 1970, we have used constant (1970s) emissions factors applied to fuel consumption trends data (DEn 1990, 1993; DTI 1995, 1997) to account for fuel-related emissions of the chosen pollutants. For all the pollutants except VOCs, fuel-related emissions account for over 90% of the total emissions, so this regression provides a reasonable estimate of emissions from these pollutants between 1950 and 1970.
For VOCs the main non-energy related sources are from forests (assumed constant by Warren Spring between 1970 and 1990, and by us prior to 1970) and from industry, processes and solvents (for which we have extrapolated (for 1950 to 1970) using a linear regression through the data for subsequent years. Allocating costs to air pollution is by no means a straightforward exercise.
Since many of the effects of air pollution lie outside the market, and those which lie inside the market are difficult to quantify, there is considerable uncertainty inherent in the allocation of actual damage costs. There are other ways in which economic costs are allocated to environmental externalities, and a burgeoning literature describing these different methods.42
The Dutch approach to sustainable national income (Hueting et al, 1992) proposes the use of control costs: costs based on achieving sustainability targets for emissions of specific pollutants.
Damage cost estimates of air pollution do exist for countries other than the UK. Daly and Cobb use a cost of $30 billion (1972 dollars) for the USA for the year 1970. This estimate is based on relatively disaggregated estimates for specific kinds of damage.43
It is not clear however, that these cost estimates are applicable even on a pro-rata basis to the emission of air pollution in this country, for a number of reasons related to the different land structure, climate, and population densities of the two countries.
Some estimates have been made of the costs of air pollution for the (former) Federal Republic of Germany and for the Netherlands, and these sums might be considered more compatible with conditions in the UK. For Germany, a recent study estimated total air pollution damages in 1985 to amount to around $19 billion in 1985 dollars (Schulz, 1986).
A Dutch study estimated annual air pollution damages in the Netherlands in 1986 to amount to between $0.5 and $0.8 billion. Even taking into account the differences in emissions levels in the two countries, however, these sums represent costs which differ by a factor of around five or six, illustrating considerable disparity in cost estimates. 44
Recently, considerable attention has been given to the task of determining externality 'shadow costs' for use in determining appropriate levels of investment in the dispatch of energy technologies (Baumann and Hill 1991, CEC 1995, Hohmeyer et al 1997, Pace 1990, Tellus 1991). This interest in costing environmental externalities for electricity dispatch is more than academic.
A number of state utilities in the US have actually adopted some form of economic 'adder' or shadow cost for different pollutants when making planning decisions (Woolf 1992). Accordingly there have been a number of attempts to identify specific costs per tonne of emissions.
The theoretical and practical difficulties associated with incorporating shadow costs into decisions about the dispatch of electricity supply options have been exposed quite forcibly in places (Stirling 1992, eg). These difficulties do not go unacknowledged even by some proponents of such methods.
On the other hand, it has been argued by a variety of different observers (Hohmeyer, 1993; HCEC, 1992) that the inclusion of some measure of the environmental cost of conventional technologies could have a significant impact on the market penetration of the newer, cleaner technologies.
Nevertheless, the difference between cost estimates is illustrated graphically in Figure 7. Both estimates show increasing cost over the earlier years of the period (1950-1965), but significantly different results over the later years. These differences arise essentially because different costs are associated with the same pollutant in the different estimates, and the different pollutants show different trends in emissions.
In the Tellus estimate, the dominating cost elements are those associated with pollutants whose emissions are increasing. Taking the Pace estimates however, places greater monetary emphasis on sulphur dioxide and particulates.
Since emissions of both these pollutants have fallen considerably over the scenario period, the overall costs of air pollution using the Pace estimates decline between 1950 and 1996, wheras those using the Tellus estimates increase.
The difficulties of deciding on the appropriate costs to use are compounded by differences in methodology between the two different estimates. The Pace figures are essentially based on a review of the literature on damage costs.
The Tellus figures are based on the control cost method of monetarisation. For the purposes of this exercise, we have decided to use an average of the two costs. The trend in the average shows rising costs over the first half of the period, and generally falling costs over the second half of the period. By the end of tion are slightly lower than they were at the beginning of the period.
The costs associated with noise pollution are amongst the most difficult to assess, because actual measurement of noise on a nationwide level is difficult to undertake, data on trends in noise are scarce, and response to noise is subjective.
Some trends data exist both in the measurement of noise levels, and in subjective responses to noise. For example, activity levels (air transport movements) at the two major international airports have increased considerably over the period 1950 to 1996.
At Gatwick, air traffic movements increased by almost 50% between 1980 and 1990 (DoE, 1992, Chapter 12). Although reported figures on population within specified aircraft noise exposure contours show some decline since 1980, partly as a result of improved design of jet engines, and air traffic control procedures, it is to be expected that changing the noise contour (eg towards a lower threshold) would have a significant effect on that trend.
Equally, data on road traffic noise displays conflicting trends. Improved engine design, and more stringent regulations have certainly had some effect on noise pollution levels in recent years. On the other hand, traffic volume has increased considerably (DoE 1990; DETR 1997).
Generally speaking, complaints about noise have risen steadily during the 1980s. In particular complaints received about noise from roadworks, construction and demolition have almost tripled (DoE 1992, p174).
Although this does not necessarily mean that actual noise levels have increased at the same rate, there are enough indications about increased noise levels generally (even below nuisance thresholds) as a result of music reproduction, traffic volumes and so on to suggest that recent increases in complaint levels have some reflection in actual noise levels, in spite of the introduction of improved noise regulation under legislation such as the Control of Pollution Act in 1974, the Civil Aviation Act of 1982, the Local Government Act of 1982, and various recent European Directives.
For the purposes of the exercise here, none of the data on noise levels or complaints about noise provides a genuine basis for calculation. However, Maddison et al (1994) report that two measures of noise levels increased in the UK between 1986-91 at an average of 0.38% and 0.69% per year respectively, giving an average increase of 0.54% a year.
In addition, a recent survey based on Geographical Information System analysis (CPRE, 1995) shows that tranquil areas 45 in England declined over the period 1960-90 at an annual average of 0.48%. We have therefore assumed that noise levels have increased fairly continuously at an average rate of 0.5% per annum over the period under discussion. 46
When it comes to assessing the costs of noise pollution, the difficulties of measurement are compounded by difficulties of subjectivity. Some assessments of the costs of noise pollution have been made. Maddison et al (1994) have estimated that the costs associated with traffic noise in the UK in 1993 were of the order of œ2.6 billion.
Schulz (1986) estimated the cost of noise pollution damage in Germany in 1985 at $11.6 billion (œ11.7 million in 1990 sterling). This figure seems rather high by comparison with the figure of $4 billion (in 1972 dollars equivalent to œ9 billion in 1990 pounds sterling) quoted by Daly and Cobb for the USA in their original index, given the relative sizes of the USA and Germany.
On the other hand, it is not inconceivable that costs from noise pollution depend on population density rather than on population per se, or even on levels of economic activity. Clearly, the absence of an accepted rationale to account for noise costs, and the subjective nature of noise nuisance, lead to great uncertainty in any cost estimate. For the purposes of this exercise we have decided, conservatively, to use the Maddison estimate for traffic noise in the UK - œ2.3 billion in 1990 prices - and indexed this accordingly over the scenario period.
The original ISEW included an assessment of the costs associated with the loss of wetlands in the US. Wetlands serve a variety of functions including nutrient cycling and storage, groundwater storage and purification, storm protection, wildlife habitats, and commercial outlets based on the habitat function, and recreational opportunities (Turner and Jones 1991). Daly and Cobb assumed a progressive loss of wetlands in the US from about the middle of the 19th century whose cost gradually increased and accumulated.
Jackson and Marks (1994) followed this procedure in their earlier ISEW study, estimating that in the UK by 1990 these costs had accumulated to œ863 million (in 1990 prices). They noted, however, that consideration of wetlands as a principal category of land loss is a particularly North American viewpoint. In Europe, most of the wetlands were lost a long time ago.
But there are a number of other kinds of land type which have come increasingly under threat as a result of advancing urbanisation during the last half a century. For example, uncultivated heathlands and moorlands have been declining steadily over the period in question. They suggested, therefore, that it would be more appropriate for a European ISEW to take a broader approach to the loss of natural habitats.
In the revised UK-ISEW we have drawn on relatively limited data on the extent of loss of various habitat types over time, and even more limited data on the value of these habitats, to estimate time series costs for such losses.
The total estimated area of key natural habitats 47 in the UK in 1930 has been estimated at just over 2 million hectares. By 1980, well over half that area had been lost, and by the mid-1990s, the remaining areas of these habitats amounted to just over 530 thousand hectares, almost a quarter of the area in 1930 (Wynne et al 1995).
In estimating the cost of these losses we have chosen a figure of œ2,000 per hectare (in 1990 pounds sterling). This figure is based on two separate sources. Firstly, it is close to the average price paid per hectare by the Royal Society for the Protection of Birds to purchase a 1200 hectare site for just over œ2.4 million in 1996.
Secondly, it is at the low end of a range of values (from œ1,529 to œ5,703) derived from a willingness-to-pay study relating to the natural habitat areas in the South Downs (Willis and Garrod 1994) If we assume that each of these costs represents the discounted future value of the welfare associated with preserving such sites, then we can also take the same value as the discounted future loss of welfare associated with not preserving them.
A time series of costs for the loss of habitat has been constructed by computing the cumulative loss since 1930 for each year of the study, and multiplying this by the cost estimate of œ2,000 per hectare. In each accounting period therefore, the adjustment made for loss of natural habitats is equal (in principle at least) to the discounted future loss of welfare arising from the loss of land up to that point.
This interpretation is in line with Weitzman's (1976) suggestion for interpreting a welfare measure as "the present discounted value of future consumption". 48
This column is designed to reflect the loss of sustainable productivity from agricultural lands in two ways: firstly, it makes an assessment of the farm land lost to urbanisation; secondly it estimates the losses of productivity of agricultural land through deterioration in soil quality. Neither of these factors is entirely straightforward to estimate, in part, because data are scarce and often subject to severe limitations or gaps.
Nevertheless, it is clear that some account should be taken of these costs. The costs of agricultural land lost to urbanisation are needed to offset the higher value added achieved through urban land gains and reflected in personal consumption. The loss of productivity through soil deterioration is an important future cost relevant in particular to the question of sustainability of agricultural practice.
We have estimated the loss of agricultural land using statistics on agricultural land use changes in England between 1947 and 1986 compiled by the Ministry of Agriculture, Food and Fisheries, and presented in a report prepared for the Council for the Protection of Rural England (CPRE, 1992).
In the absence of reliable data on recent trends, we assume that the loss of land in England and Wales since 1986 has remained constant. Losses for the rest of the UK were extrapolated on the basis of total populations.
In calculating the costs of these losses we have used a capitalised value of just under œ1,500 per acre - equivalent to œ3,700 per hectare. Our estimate of the value of agricultural land is therefore higher than the market value of land. 49
Like Daly and Cobb (1989) we assume that the underlying value of land exceeds the market value today, because the market value reflects the fact that productivity can be increased in the short term by the application of energy and nitrogen fertilisers. "Since our aim is to calculate sustainable economic welfare, we have chosen a figure that represents the value of land as if cheap energy sources had already been depleted." (op cit, p435).
We have also made an attempt to estimate the costs of loss of agricultural productivity as a result of soil deterioration. This task is extremely difficult for several reasons. Firstly, the factors contributing to loss of soil productivity are various and complex. For instance, crop management practices inherently alter the natural nutrient cycles, and can contribute to a loss of soil productivity by removing nutrients from the soil.
This loss of productivity can be masked in the short-term by the addition of chemical fertilisers, but these fertilisers (and other chemical inputs such as pesticides) contribute to soil depletion themselves in the longer run. Secondly, the actual ecological effects of agricultural practices are obscured by the existence of natural effects.
For example, one of the specific effects of poor crop management is the increased risk of erosion as soil structure breaks down. But erosion itself is a natural process, and certain levels of erosion could be expected even in the absence of agriculture. It is almost impossible to identify what these 'natural' erosion rates might be.
Erosion depends on wind, on temperature, on rainfall, on vegetation cover, and a number of other ambient conditions in addition to the structure of the soil itself. Some of these ambient conditions are themselves altered as a result of anthropogenic activities.
Finally, in calculating the loss of productivity of soils, we are primarily interested here in calculating those losses which result from the future impacts of current activities. Clearly, it would not be appropriate to subtract lost present-day agricultural production, since this is already reflected (at least in principle) in lower personal consumption.
Rather, we are attempting here to provide some measure of losses in future production as a result of changes to the soil resulting from current activities. This task is considerably more difficult, and requires in particular that we take account of the reversibility and irreversibility of impacts.
Evans (1993) has estimated that the cost of lost agricultural production from resulting from annual erosion rates in the UK in the mid-1980s was œ2.76 million. Assuming that maybe two thirds of this erosion can be counted as resulting from anthropogenic practices and using a 10% discount rate for the capitalisation of agricultural land, 50 implies that productivity losses resulting from erosion account for around œ20 million each year.
We have assumed that this rate of soil deterioration has remained more or less constant over the time frame of this study, but we have started our analysis with an estimate of cumulative productivity loss of œ500 million by 1950. 52
The depletion of non-renewable natural resources represents a loss of natural capital, and consequently a reduction in future consumption possibilities. Generally speaking, these losses lie outside the boundary of the national accounts, but there is now widespread agreement that some account should be taken of them in measures of sustainable welfare.
In the original ISEW, Daly and Cobb examined various different methods of accounting for the depletion of non-renewable resources, including El Serafy's model for relating 'true income' X to total receipts R from mineral production (net of extraction costs) via the rate of discount r and the number of years n to depletion. 52
Having applauded this model as "the best attempt we have seen to come to grips with the proper method of accounting for depletion of non-renewable resources or 'natural capital'", Daly and Cobb point out a number of difficulties in actually determining appropriate values both for the number of years to depletion and the total value of receipts.
They then give some reasons for taking instead the total value of receipts from mineral production as the appropriate set aside, including the argument that this would be the result of applying a zero discount rate in the El Serafy formula.
Critics argued (Cobb and Cobb, 1994) that the assumption of a zero discount rate leads to absurd results, and also that this method does not actually represent an account of the value of changes in the stock of natural capital assets, since new resources may be discovered, the value of the stock itself changes over time and so on.
In response to these criticisms Daly suggested a different way of dealing with the depreciation of natural capital by calculating "amount of rent from resource production that should be reinvested in a process to create a perpetual stream of output of a renewable substitute for the non-renewable resource being depleted".
In the revised index, Cobb and Cobb therefore devised a depletion allowance based on assigning a replacement value to every barrel of oil equivalent of energy resources consumed over the period.
This replacement value was designed to reflect the cost of replacing each barrel of oil equivalent of energy consumed with renewable energy resources. The replacement cost in 1988 was taken to be $75 dollars per barrel (or equivalently around œ49 per barrel in 1990 pounds).
The replacement cost was assumed to have increased at an annual rate of 3% to account for the increasing costs of supplying each marginal unit of energy. In the earlier scenario years the assumed replacement cost was much lower at around œ16 per barrel (1990 pounds).
The assumed replacement cost of œ49 per barrel of oil equivalent seems extremely high, when compared against the actual costs of renewable energy technologies today, some of which are already competitive with conventional fossil supplies (Jackson 1997, eg).
In defence of their assumptions about cost growth, Cobb and Cobb point to the cost escalations in drilling for oil particularly through the 1970s and early 1980s. They also highlight three critical features which might lead to cost escalation.
The first is the inherent tendency for marginal supply of a commodity to become more expensive, the second is the relatively low energy ratio (energy output to energy input) of certain renewable energy technologies, and the third is the potential influence of land prices on biomass resources in the context of increasing population and food demand.
Can these cost escalations justify the use of a replacement cost several times greater than the present cost of renewables at the margin? Jackson and Marks (1994) argued that it was difficult to justify a 3% per annum increase in costs on the basis of cost escalations in the fossil fuel industry. But they pointed out that cost escalation in this example was completely different from conventional fossil fuel cost escalation. Several points have to be considered.
Firstly, the assumed œ49 replacement cost per barrel is for a system supplying 100% of the energy in the economy, so it is not directly comparable with the actual marginal costs for supply from renewable resources today. Indeed the assumed cost of œ16 per barrel for a system supplying 100% of the primary energy demand in 1950 (which represents around two-thirds of primary energy demand in 1990) seems if anything extremely conservative.
In spite of the fact that renewable energy technologies have come down in price dramatically over the last decade, the costs of completely replacing fossil fuels with renewable energy resources in 1950 would probably have been astronomical.
For the purposes of this exercise, we have therefore decided to use the same 1988 cost per barrel, and the same cost escalation factor of 3% as used in the revised US-ISEW. Statistics on consumption of primary fuels (coal, oil, gas, nuclear) in the UK since 1950 were taken from the Economic Trends Annual Supplement (ET, 1993, Table 14) and from the Digest of UK Energy Statistics for later years (DTI 1995a & 1997).
In the original Daly and Cobb ISEW, an adjustment was made to account for the long-term environmental damage incurred through energy consumption.
These damages result mainly from the build-up of greenhouse gases in the atmosphere from anthropogenic combustion of fossil fuels, but Daly and Cobb argued that it would also be appropriate to account for the long-term costs associated with the use of nuclear fuels, on the grounds that the "social cost of leaving behind a mountain of spent fuel rods and reactors requiring decommissioning is perhaps as great as that imposed by climate change" (Cobb and Cobb, 1994).
Daly and Cobb assigned a damage value of $0.50 (1972 dollars) to each barrel of oil equivalent of the non-renewable fuels (coal, oil, gas and nuclear) consumed. The damage costs were assumed to grow cumulatively over time (starting from 1900), reflecting the fact that most greenhouse gases have long atmospheric residence times, and continue to contribute to environmental damage long after they have been emitted.
The authors themselves admitted that the figure of $0.50 per barrel of oil equivalent is largely arbitrary, but defended themselves on the grounds that it would be wrong to ignore major issues for lack of an accepted methodology for dealing with them.
In the manner of a post hoc justification of this method, Cobb and Cobb (1994) compared their costs against an estimate made by Cline (1992) that the annual damage from global warming in 2025 would be around $120 billion (1990 prices).
Cobb and Cobb pointed out that if Cline's figure was right then the "accumulated condition of damage in 2025 (from which annual damage would 'flow') would be around $1.2 trillion". They argued that this served as "indirect confirmation of the reasonableness" of their own estimate.
The major problem with this approach - which was also followed by Jackson and Marks in the earlier version of the UK-ISEW - is the arbitrary way in which a charge is calculated, a fact which has been seized on by critics as evidence that the adjustment itself is unwarranted (Atkinson 1995, eg). For the updated UK-ISEW we have therefore decided to revise the method of accounting for long-term environmental damage.
The new approach focuses specifically on the long-term damage associated with greenhouse gas emissions. The basic idea is assign to each tonne of emissions from 1900 onwards a 'marginal social cost' which reflects the total (discounted) value of all the future damage arising from that tonne of emissions.
The total costs arising in each year are calculated by multiplying this marginal social cost by the carbon emissions in that year. The of the index are calculated by accumulating the costs from 1900 up to that year.
A variety of estimates exist of the marginal social cost of carbon emissions in the 1990s, ranging from a little over $5 per tonne of carbon to more than $120 per tonne of carbon (IPCC 1996, Table 6.11). Fankhauser (1994), for example, provides estimates ranging from $6.2 per tonne to $45.2 per tonne, depending on the discounting procedure chosen, with a middle of the range estimate of $20.3 per tonne of carbon.
For the purposes of this study, we have converted an estimate of œ11.4 (1990 prices) corresponding to the mid-range Fankhauser estimate. Since, this is towards the lower end of the range of estimates identified by the IPCC, we regard this choice as relatively conservative.
We use this estimate as the marginal social cost of carbon emissions in 1990. Since the marginal social cost of a tonne of carbon emissions depends, at any point in time, on the accumulation of carbon in the atmosphere, it is clear that this cost is not constant over time. In earlier years, when less carbon had been emitted into the atmosphere, the marginal social cost of a tonne of carbon would have been lower than it was in 1990. Conversely, in later years the marginal social cost can be expected to rise above its 1990 level.
For the purposes of this exercise we have therefore constructed a time series of costs based on two assumptions: firstly that the marginal social cost in 1990 is equal to œ11.4 per tonne, and secondly that the marginal social cost in any year is proportional to the cumulative carbon emissions from the year 1900 up to that point in time. 53
Statistics on energy consumption in the UK since 1950 were taken from the Economic Trends Annual Supplement (ET, 1993, Table 14) and from the Digest of Energy Statistics (DTI 1995a, 1997) for later years. In constructing the time series data between 1900 and 1950 we have used statistics for energy consumption taken from Lisner (1985). Primary fuel consumption has been converted to carbon emissions using standard emission coefficients (DTI 1995b, eg).
In each year, the contribution to future damages has been calculated by multiplying the total annual carbon emission by the marginal social cost for that year. Since these future damages accumulate, year on year, the associated costs are accumulated through the index.
The total accumulated cost of long-term damage from greenhouse gas emissions in 1990 was œ66.6 billion. By comparison, Azar and Holmberg (1995) have provided a method for calculating the accumulated generational environmental debt associated with climate change. Applying this method to the UK, reveals a generational environmental debt in 1990 of œ276 billion.
Using the earlier Daly and Cobb methodology for long-term environmental damage would have resulted in an accumulated cost in 1990 of œ106 billion (1990 prices). Our revised estimate is therefore to be regarded as relatively conservative compared to these other estimates.
On the other hand, it should be remembered that we have constructed only a limited calculation of the long-term damages associated with energy consumption. In particular, we have accounted only for carbon dioxide emissions (neglecting emissions of methane and nitrous oxide for example) and we have taken no account of the long-term damage from the nuclear industry. 54
We gave detailed consideration to the view expressed by some critics that only the annual contributions to future damages should be counted in each year. Hamilton and Atkinson (1996) have constructed a Genuine Savings Index, in which they account for the marginal social costs associated with various types of air pollution. Interestingly the marginal social cost they attributed to carbon emissions - also based on the Fankhauser study - is œ12 per tonne, slightly higher than ours.
But the major difference between that study and ours is their contention that the only costs relevant to their index are those attributed to emissions in that year. Thus, they treat the annual damages from sulphur dioxide (for example) on exactly the same basis as the future damages from carbon emissions.
It is clear, however, that there are some fundamental differences between these two kinds of emissions, and the associated damage costs. The damage costs associated with sulphur dioxide emissions represent present-day social costs incurred in the accounting period as a result of activities carried out in the accounting period.
By contrast, the damage costs associated with carbon emissions are (essentially) the discounted value of losses in welfare suffered by a number of future generations. At the very least, this difference in the concept of damage cost between different kinds of pollutants needs to be recognised. If there is a 'correct' way of accounting for each of these different kinds of losses, it depends upon the interpretation given to a specific index.
From the perspective of the endeavour in this paper to provide a measure of sustainable welfare, we are drawn once again to the interpretation 55 that such a measure should reflect the present discounted value of future welfare. It is clear that, all other things being equal, paying off the long-term damages associated with activities from a single accounting period (say 1990) would not reduce the present value of future welfare losses to zero.
The accumulated debt of the past would still represent a very real loss to the future. The present value of that future loss is the very least that an index of sustainable economic welfare should measure. As a sensitivity, however, we investigate in section 4 (below) the impact on the index of accounting only for the losses associated with emissions from a single accounting period.
The costs associated with ozone depletion were incorporated into methodology during Cobb and Cobb's (1994) revision of the ISEW, following a criticism by Eisner that not all long-term environmental damage is related to energy use. The authors were "particularly concerned that [they] had omitted a damage estimate from the cumulative release of chlorofluorocarbons [CFCs] into the atmosphere".
Measurements over the Antarctic have shown dramatic changes in the thickness of the ozone layer since the late 1970s. The first warnings that CFCs could have a negative impact on the ozone layer were sounded in the early 1970s, but until then global production had increased unchecked and it was not until 1987 that the Montreal Protocol governing the production and consumption of CFCs came into force.
The US-ISEW accounted for the costs of ozone depletion by applying a unit cost of $15 (1972 prices) for each kilogramme of US production of each of CFCs 11 and 12. Cobb and Cobb explain that this method "amounts to the same as assuming that each individual in the US would demand about $960 (1972 prices) in 1985 to compensate for the risks involved in producing and having produced CFCs.
Or it may be thought of as the amount that would have to be set aside to compensate for future generations for having made their planet less habitable". As with the case of long-term environmental damage from energy consumption, it is assumed that the damage costs are cumulative, reflecting the present value of future welfare losses from the stock of environmental damage.
Jackson and Marks (1994) suggested that this column should account for the trend in all of the Montreal-listed CFCs 11, 12, 113, 114 and 115. Since production of CFCs 113, 114, and 115 was increasing even when production of CFCs 11 and 12 had begun to decline, they argued that by taking account of all five CFCs a more realistic picture of changes over time would be obtained.
Jackson and Stymne (1996) addressed the question of whether the costs of ozone depletion should be based on consumption rather than production of CFCs. Some countries (Sweden, for example) produce no CFCs at all, and yet by consuming them they are clearly contributing to future environmental damage, for which they are in some sense accountable.
Jackson and Stymne argued that by imposing the costs of environmental damage on the producers rather than the consumers, the consumers "would seem to be getting a polluting product at a cut-down price - one which does not reflect the environmental costs of using the product".
They therefore suggested that the appropriate way of allocating the costs associated with ozone-depleting substances between countries was on the basis of the quantity consumed in each country. This indicates that the costs associated with ozone depletion may have been overestimated in some earlier ISEW studies.
In this paper, we have adopted the argument made by Jackson and Stymne, and estimated the costs associated with ozone depletion on the basis of the UK consumption of the Montreal-listed CFCs. World production data for CFCs 11, 12, 113, 114 and 115 has been taken from data compiled by the Alternative Fluorocarbons Environmental Acceptability Study for the years 1979 to 1985 (AFEAS, 1991a&b).
We have made some simple linear regression assumptions about annual production prior to 1979, to deduce the annual world production for each of these CFCs over the earlier period of the index. We have then calculated the total ozone depletion potential (ODP) associated with world production using figures presented to UNEP by the International Union of Pure and Applied Chemistry. 56
UK production in 1986 was 10.6% of world production, according to data reported to UNEP under the requirements of the Montreal Protocol (UNEP 1992).
For the earlier years of the series, we have indexed UK production on the ratio of EU to global production. Consumption for the years up to 1985 was esn data for the UK for the years 1986 to 1995 were obtained directly from a Department of the Environment report (DoE 1996), and converted to ODP using the same figures as for production. For 1996, it was assumed that the Montreal Protocol had come into effect and consumption had ceased.
Converted to 1990 pounds sterling, the Cobb and Cobb cost per kilogram of CFC-11 and CFC-12 was œ34. Since we are also including the other Montreal-listed CFCs in our analysis, the total costs implied by unit costs of œ34 per kilogram of consumption of CFCs would be greater than those assumed on the basis of a direct comparison with the US work.
Rather than accept this methodological increase in cost, we have calculated an equivalent cost per kilogram (about œ30 per kilo) by weighting the Cobb and Cobb cost by the ratio of the cumulative consumption of CFCs 11 and 12 to the cumulative consumption of all the listed CFCs. The total cost of ozone depletion in each year is calculated by multiplying the accumulated consumption up to that year by this equivalent cost.
The Hicksian notion of income demands that account should be taken of changes in the net stocks of capital over time. Economic consumption which depletes capital cannot be regarded as sustainable. The conventional calculation of GDP incorporates a measure of investment expenditure as gross fixed capital formation. Daly and Cobb (1989) argued that this inclusion neglects two important factors.
Firstly, it omits to account for the depreciation of capital. Secondly, some account needs to be taken of the increases (or decreases) in the need for capital to provide for increases (or decreases) in the workforce.
The authors argued that "one element of economic sustainability is constant or increasing quantities of capital available for each worker" (op cit p 442). Column V is therefore a measure of the net increase (or decrease) in both public and private capital after subtracting an appropriate increase in the requirement dictated by increases in the labour force. 58
In the revised index Cobb and Cobb conceded that "our inclusion of government capital in net capital growth was inconsistent with our exclusion of the services from that capital elsewhere" and consequently, in the revised index, they included only private capital growth.
For the UK-ISEW we have more or less followed the revised methodology, with the exception that we have included public sector corporations in the accounts as well as net changes in the stock of private capital.
In the UK, the exclusion of public sector capital would imply a particular accounting problem over the 1980s, because of the transfer of capital from public hands to private hands during a series of privatisations of public corporations initiated by the former Conservative governments.
Essentially, this transfer of capital would appear (deceptively) as a considerable increase in net capital growth if the index only accounted for changes in private capital. For the UK index therefore, we have included capital stocks from public corporations in addition to capital stocks from the private sector.
Changes in the capital stock were taken from the UK National Accounts (UKNA, various years, Table 14.7). In calculating the workforce capital requirement, we have used figures on the workforce taken from the Economic Trends Annual Supplement (ET various years) appropriately adjusted to take account of changes in the definition of unemployment over the study period. 59
As with the Daly and Cobb index, a five year rolling average has been used to smooth out year on year fluctuations in the account. Even so, Figure 9 reveals significant changes in the net capital stock over the period. In particular, there was a large increase in capital investment during the 1960s and 1970s culminating in net capital growth of more than œ30 billion by 1976/7.
Thereafter however, there has been a sharp fall in capital investment, with a period of significant net capital depreciation during the mid 1980s and again in the 1990s. These changes will be seen to have a noticeable impact on the overall index.
In line with the methodology carried out by Daly and Cobb, we have calculated the net international position by taking the net transactions in assets and liabilities from the Economic Trends Annual Supplement (ET various years) and changing the sign.
The resulting figure then represents the difference between UK investment overseas (+ve) and overseas investment in the UK (-ve). Given rather large year on year fluctuations in this measure of net international position, whose significance is not clear in the context of long-term sustainability, we have again used a five-year rolling average figure in the index.
The total index of sustainable economic welfare is calculated in column X by adding columns D to H, subtracting columns I to U and adding in columns V and W.
Per capita ISEW has been calculated from column X using UK population statistics from the Annual Abstract of Statistics (AAS, various years).
The gross domestic product has been taken from the Economic Trends Annual Supplement (ET various years, Table 2).
Per capita GDP has been calculated from column Z by using the population statistics from the Annual Abstracts of Statistics (AAS various years).
AAS, various years, Annual Abstract of Statistics, Central Statistical Office,
HMSO, London.
AFEAS, 1991a. Chlorofluorocarbons 11 and 12 - annual production for the years
1931 - 1975 and annual production and sales for the years 1976 - 1990, prepared
by Grant Thornton for the Alternative Fluorocarbons Environmental Acceptability
Study, November 1991, Washington, DC.
AFEAS, 1991b. Chlorofluorocarbons 113, 114, and 115 - cumulative production
through 1979 and annual production and sales for the years 1980 - 1990, prepared
by Grant Thornton for the Alternative Fluorocarbons Environmental Acceptability
Study, November 1991, Washington, DC.
Anderson, V., 1991. Alternative Economic Indicators, Routledge, London.
Atkinson, A., 1983. The Economics of Inequality, 2nd edition, reprint 1995,
Oxford University Press, Oxford.
Atkinson, A., 1970. On the measurement of inequality, Journal of Economic Theory,
vol 2, pp244-263.
Atkinson, G., 1995. Measuring Sustainable Economic Welfare: a critique of the
UK ISEW, CSERGE working paper, GEC95-08.
Atkinson, G., and Pearce, D., 1996. Indicators to change policy, in New Economy,
vol 3(1).
Azar, C., and Holmberg, J., 1995. Defining the generational environmental debt,
Ecological Economics, vol 14, p7-19.
Baumann A., and Hill, R., 1991. External Costs/Benefits of Energy Technologies
- development of a methodology, Final Report on Contract JOUR-0020-UK(CH), Newcastle-upon-Tyne
Polytechnic, June 1991.
Becker, G.S., 1975. Human Capital. A theoretical and Empirical Analysis, with
Special Reference to Education. Second Edition 1980. The University of Chicago
Press, Chicago and London.
Blundell, R., Browning, M., and Meghir, C., 1994. Consumer Demand and the Life-Cycle
Allocation of Expenditures, Review of Economic Studies, vol 61, p57-80.
Castaneda, B., 1997. An Index of Sustainable Economic Welfare (ISEW) for Chile,
Institute for Ecological Economics, University of Maryland, Solomons, MD.
CEC, 1995. Assessment of the External Costs of the Coal Fuel Cycle, ExternE
Project, Joule Programme, Commission of the European Communities, Brussels.
Chernick, P., and Caverhill, E., 1989. The Valuation of Externalities From Energy
Production, Delivery and Use, a Report to the Boston Gas Company, December 1989.
Church, J., Koudra, M., and Murgatroyd, L., 1996. Where have all the hours gone?
- measuring time use in the UK, Statistical News, vol 111 (Spring 1996).
Cline, W.R., 1992. The Economics of Global Warming, Institute for International
Economics, Washington DC.
Cobb C., and Cobb, J., 1994. The Green National Product, University of Americas
Press, Lanham, Md.
Cobb, C., Halstead, E., & Rowe, J., 1995. The Genuine Progress Indicator- summary
of data and methodology, Redefining Progress, Washington, DC.
CPRE, 1992. The Lost Land -land use change in England 1945 -1990, Council for
the Protection of Rural England, London.
CRPE, 1995. The Tranquil Areas Maps, Council for the Protection of Rural England,
London.
Dalton, H, 1920. The Measurement of the Inequality of Incomes. The Economic
Journal, vol 30, p348-361.
Dasgupta, P., and Mèler, K-G., 1991. The environment and emerging development
issues, proceedings of the annual world bank conference on development economics.
World Bank. Washington, DC.
Dasgupta, P., KrisstrÜm, B., and Mèler, K-G., 1994. Current Issues in Resource
Accounting in Current Issues in Environmental Economics, Johansson, P., KrisstrÜm,
B., and Mèler, K-G., (eds), Manchester University Press.
Daly, H., and Cobb, J. 1989. For the Common Good - redirecting the economy towards
community, the environment and sustainable development. Beacon Press, Boston.
DEmp, various years, New Earnings Survey, Department of Employment, HMSO, London.
DEn, 1993. Digest of UK Energy Statistics 1992, Department of Energy, HMSO,
London.
Diefenbacher, H., 1994. The Index of Sustainable Economic Welfare in Germany,
in Cobb and Cobb 1994.
Diefenbacher, H., and S. Habicht-Erenler, 1991. Wachstum und Wohlstand: neuere
Konzepte zur Erfassung der Sozial- und Umweltvertrèglichkeit, Metropolis-Verlag,
Marburg.
DoE, 1996. UK Use and Emissions of Selected Halocarbons, Department of the Environment,
HMSO, London.
DoE, 1992. The UK Environment, Department of the Environment, HMSO, London.
DoE, 1990. Digest of Environmental Protection and Water Statistics, 1990, Department
of the Environment, HMSO.
DETR, 1997. Digest of Environmental Statistics, Department of Environment, Transport
and the Regions, HMSO.
DoT, 1993. Road Accidents GB - The Casualty Report, Department of Transport,
August 1993, Table 2, column 7, HMSO, London.
DoT, 1992. Environmental Appraisal: a review of monetary evaluation and other
techniques, Transport and Road Research Laboratory Contractor Report 290, Department
of Transport.
DoT, 1989. Roads for Prosperity, Transport White Paper, HMSO, London.
DoT, various years, Transport Statistics GB, Department of Transport, HMSO,
London.
DTI, 1997. Digest of UK Energy Statistics, Department of Trade and Industry,
HMSO, London.
DTI, 1995a. Digest of UK Energy Statistics, Department of Trade and Industry,
HMSO, London.
DTI, 1995b. Energy Projections for the UK - energy use and energy-related emissions
of carbon dioxide in the UK 1995-2020, Energy Paper 65, Department of Trade
and Industry, HMSO, London.
EC, 1992. Towards Sustainability - a European Community programmes of policy
and action in relation to the environmental and sustainable development, Commission
of the European Communities, COM(92) 23 final, volume II, Brussels, March 1992.
Eisner, R. 1988. Extended Accounts for National Income and Product, Journal
of Economic Literature, vol.26, pg.1611-1684.
Eisner, R, 1985. The Total Income System of Accounts, Survey of Current Business.
Vol.65. No.1. United States Department of Commerce/Bureau of Economic Analysis.
Ekins, P., 1992. Wealth beyond Measure - an atlas of new economics, Gaia Books,
London.
ET, various years, Economic Trends - Annual Supplement 1993, Central Statistical
Office, HMSO, London
ET, 1993. 'The effects of taxes and benefits on household income, 1990', Economic
Trends No 471, Appendix 4, Table 2, January 1993, Central Statistical Office,
HMSO, London.
ET, 1992. Economic Trends, No 469, November 1992, Central Statistical Office,
HMSO, London.
ET, 1991. 'The effects of taxes and benefits on household income, 1988', Economic
Trends No 449, March 1991, Central Statistical Office, HMSO, London.
Evans, R., 1993. Soil Erosion and its Impacts in England and Wales, Friends
of the Earth, London.
Fankhauser, S. 1994. The Social Costs of Greenhouse Gas Emissions- an expected
value approach, The Energy Journal 15(2), pp157-184.
FES, various years, Family Expenditure Surveys, Central Statistical Office,
HMSO, London.
Friedman, M. 1957. A theory of the consumption function. National Bureau of
Economic Research. General Series 63. Princeton, N.J.
Gershuny J., and Jones, S., 1987. The changing work/leisure balance in Britain
1961-1984, Sociological Review Monograph 33.
Goodman, A., Johnson, T., and Webb, S. 1997. Inequality in the UK, Oxford University
Press, Oxford.
Goodman, A., and Webb, S. 1994. For Richer, for Poorer - the changing distribution
of income in the UK 1961-1991. Institute for Fiscal Studies, London.
Guenno, G., and Tiezzi, S., 1996. An index of sustainable economic welfare for
Italy, Feem Newsletter vol 2, p16-21.
HCEC, 1992. Renewable Energy, Fourth Report of the House of Commons Energy Committee,
Session 19he Smoking Epidemic - a prescription for change, Health Education
Authority, London.
Hamilton, K., and Atkinson, G., 1996. Air Pollution and Green Accounts, Energy
Policy, vol 24(7), p675-684.
Hicks, J.R. 1939. Value and Capital, Oxford University Press, Oxford and New
York.
Hohmeyer, O., 1993. Renewables and the Full Costs of Energy, Chapter 17 in Jackson
1993.
Hohmeyer, O., Oittinger, R. and Rennings, K., (eds) 1997. Social Costs and Sustainability,
Springer-Verlag, Berlin.
Hope, C., Parker, J., and Peake, S., 1992. A pilot environmental index for the
UK in the 1980s, Energy Policy, vol 20(4), April 1992, p335.
Hueting, R., Bosch, P. and de Boer, B., 1992. Methodology for the Calculation
of Sustainable National Income, Statistical Essays No M44, Central Bureau voor
de Statistiek, SDU Publishers, The Hague.
IPCC, 1996. Climate Change 1995 - Economic and Social Dimensions of Climate
Change, Working Group III report to the IPCC, Cambridge University Press, Cambridge.
Jackson, T., 1997. Power in Balance - energy challenges for the 21st Century,
Friends of the Earth, London.
Jackson, T., 1996. Material concerns - pollution, profit and quality of life,
Routledge, London.
Jackson, T., 1993. Renewable Energy - prospects for implementation, Butterworth-Heinemann,
Oxford.
Jackson, T., and Marks, N., 1998. Consumption, Sustainable Welfare and Human
Needs - with reference to UK consumers expenditure 1954-1994, Ecological Economics,
forthcoming.
Jackson, T., and Marks, N., 1994. Measuring Sustainable Economic Welfare: a
pilot index for the UK 1950 - 1990. Stockholm Environment Institute/New Economics
Foundation, London.
Jackson, T., and Stymne, S., 1996. Sustainable Economic Welfare in Sweden: a
pilot index 1950-1992, Stockholm Environment Institute, Sweden.
Jacobs, M., 1991. The Green Economy - environment, sustainable development and
the politics of the future, Pluto Press, London, Chapter 18, p 222 et seq.
Leech, P., 1991. UK Emissions of Air Pollutants 1970-1989, LR 926 (AP), Warren
Spring Laboratory, Stevenage.
Lisner, T., 1985. Economic Statistics 1900 - 1983: UK, USA, France, Germany,
Italy, Japan, Economist Publications, London.
Lutz, M. and K. Lux, 1988. Humanistic Economics - the new challenge, Bootstrap
Press, New York.
Maddison, D., D. Pearce., O. Johansson., E. Calthrop., T. Litman and E. Verhoef,
1996. Blueprint 5. The True Costs of Road Transports. Earthscan Publications
Ltd, London.
MAFF, 1993. Prices of agricultural land in England: 3 months and 6 months ended
31st March 1993, Stats 120/93, Government Statistical Service, London.
Mèler, K-G., 1995. Welfare indices and the environmental resource base. Beijer
International Institute of Ecological Economics, Stockholm.
Mèler, K.G. 1991. National Accounts and Environmental Resources. Beijer Reprint
Series No.4. Beijer International Institute of Ecological Economics, Stockholm.
Manzini, E., 1993. Values, Quality and Sustainable Development - the role of
the cultural factor in the re-orientation of the system of production and consumption,
Chapter 18 in Jackson, 1993a.
Martinez-Alier, J., 1987. Ecological Economics, Blackwell Scientific, Oxford.
Max Neef, M.A., 1991. Human Scale Development - conception, application and
further reflections, The Apex Press, New York.
Mitchell, C., J. Sweet, and T. Jackson, 1990. A study of methane leakage from
the UK natural gas distribution system, Energy Policy, vol 18(9). MLG, various
years, Ministry of Labour Gazette, HMSO, London.
MoT, various years, Passenger Transport in Great Britain, Ministry of Transport,
HMSO, London.
Murgatroyd, L., and Neuburger, H., 1997. A Household Satellite Account for the
UK, Economic Trends, October 1997, p63-71.
Nordhaus, W.D. 1992. Is Growth Sustainable? Reflections of the Concept of Sustainable
Economic Growth. Paper prepared for International Economic Association Varenna,
Italy. October 1992.
Nordhaus W., and Tobin, J., 1972. Is Growth Obsolete? in Economic Growth, Fiftieth
Anniversary Colloquium, National Bureau of Economic Research, Columbia University
Press, NY.
NTS, various years, National Travel Survey, HMSO, London.
OECD, 1991. Environmental Indicators - a preliminary set, Organisation for Economic
Co-operation and Development, Paris.
Oegema, T., and Rosenberg D., 1995. A Pilot ISEW for the Netherlands 1950-1992.
Instituut voor Milieu - en Systeemanalyse, Amsterdam.
Opschoor, J.B., 1986. A Review of Monetary Estimates of Benefits of Environmental
Improvements in the Netherlands, OECD Workshop on the Benefits of Environmental
Policy and Decision-making, October 1986, cited in Pearce et al 1989.
Pace 1990. Environmental Costs of Electricity, Pace University Center for Environmental
Legal Studies, prepared for the New York State Energy Research and Development
Authority and the US Department of Energy, Oceana Publications.
Patterson, K., 1992. The Service Flow from Consumption Goods with an Application
to Friedman's Permanent Income Hypothesis, Oxford Economic Papers, July 1992,
Oxford.
Pearce, D., 1993. Economic Values and the Natural World, Earthscan, London.
Pearce, D., and Ulph, D., 1995. A social discount rate for the United Kingdom,
CSERGE working paper GEC 95-01, CSERGE, London.
Pearce, D., A. Markandya, and E. Barbier, 1989. Blueprint for a Green Economy,
Earthscan, London.
Pearce, D., and Turner, R.K., 1990. Economics of Natural Resources and the Environment,
Harvester Wheatsheaf, London.
Samuelson, 1961. The Evaluation of Social Income: capital and wealth formation,
in The Theory of Capital, proceedings of an IEA conference, Lutz and Hague (eds),
St Martin's Press, New York, cited in Weitzman 1976.
Schultz, T.W., 1971. Investment in Human Capital. The Free Press, New York
Schulz, W., 1986. A survey on the status of research concerning the evaluation
of benefits of environmental policy in the Federal Republic of Germany, OECD
workshop on the Benefits of Environmental Policy and Decision-Making, Avignon,
France, 1986.
Schwartz, J., and Winship, C., 1980. The Welfare Approach to Measuring Inequality,
in Schuessler, F., (ed), Sociological Methodology 1980, Jossey-Bass, San Francisco.
SNA, 1993. System of national accounts, 1993. Prepared under the auspices of
the Inter-Secretariat Working Group on National Accounts;Studies in methods
- United Nations. Series F; 2:Rev.4. New York.
Stirling, A.. 1992. Regulating the Electricity Supply Industry by Valuing Environmental
Effects: how much is the emperor wearing?, Futures, vol 24, December 1992, pp
1024-1047.
Stockhammer, E., Hochreiter, H., Obermayr, B. and Steiner, K., 1997. The index
of sustainable economic welfare (ISEW) as an alternative to GDP in measuring
economic welfare. The results of the Austrian (revised) ISEW calculation 1955-1992.
Ecological Economics. Vol.21, p19-34.
Taylor, M.C., 1990. Cost of Vehicle damage resulting from road accidents, Transport
and Road Research Laboratory, Department of Transport, London.
Tellus, 1991. Environmental Benefits of DSM in New York: Long Island Case study,
proceedings of the Demand-Side Management and the Global Environment Conference,
Arlington, Virginia, April, 1991.
Thurow, L.C., 1975. Education and economic equality, in The Inequality Controversy:
schooling and distributive justice, edited by D.M. Levine and M.J Bane, Basic,
New York.
Turner, R.K., and Jones, T., 1991. Wetlands: market and intervention failures
- four case studies, Earthscan, London.
UKNA, various years, United Kingdom National Accounts, Central Statistical Office,
HMSO, London.
UN, 1993. Integrated Environmental and Economic Accounting, Interim version.
Handbook of National Accounting, Series F, No.61. Dept. for Economic and Social
Information and Policy Analysis, Statistical Division, New York, 1993.
UNEP, 1992. The Reporting of Data by the Parties to the Montreal Protocol on
Substance that Deplete the Ozone Layer, report of the Secretariat to the 4th
Meeting of the Parties to the Montreal Protocol,, UNEP/OzL.Pro.4/6, 26th August
1992.
Vaze, P., and Balchin, S., 1996. The Pilot United Kingdom Environmental Accounts,
Economic Trends, September 1996, HMSO.
Weitzman, M.L,. 1976. On the Welfare Significance of National Product in a Dynamic
Economy. The Quarterly Journal of Economics. Vol.90, p156-162.
Willis, K., and Garrod, G., 1994. The ultimate test: measuring the benefits
of ESAs, in Whitby, M., (ed), Incentives for Countryside Management: the case
of environmentally sensitive areas, CAB International, Cambridge, pp179-217.
Woolf, T., 1992. Accounting for the Environmental Externalities of Electricity
Production - a summary of US practice, Association for the Conservation of Energy,
London.
Wynne, G., Avery, M., Campbell, L., Gubbay, S., Hawkswell, S., Juniper, T.,
King, M., Newbery, P., Smart, J., Steel, C., Stones, T., Stubbs, A., Taylor,
J., Tydeman, C., and Wynde, R., 1995. Biodiversity Challenge - an agenda for
Conservation in the UK, 2nd edition, RSPB, Sandy.
16 One of the implications of this is that regional differences in income
do not necessarily reflect equivalent differences in welfare. For example, where
local amenities are high, firms do not have to offer high wages to attract workers,
so that incomes may be lower, without adverse impacts on welfare.
17 See Jackson and Marks 1998 for a more detailed discussion both of this issue
and of alternative models of welfare.
18 The Gini coefficient is calculated by taking the area between a curve of
actual income distribution and the line of perfect (ie equal) distribution.
The coefficient takes values between 0 and 1, with 0 representing perfect distribution.
19 Such as Sweden (Jackson and Stymne 1996) or the Netherlands (Oegema and Rosenberg
1995).
20 In a thought experiment, Atkinson (1983) suggests that e can be thought of
as the proportion of transfered income that it would be acceptable to lose (in
transfer costs) when distributing income from a richer to a poorer person. The
transfer would only be acceptable if the net welfare benefit remained positive:
ie the gain for the poorer person was higher than the loss to the rich person
plus the transfer costs.
21 Schwartz and Winship (1980) point out however, that values of e less than
zero violate important principles of welfare, and that for such values the index
should be thought of as "the proportionate increase in social decadedence attributable
to inequality".
22 To see this, note that y=Y/P where P is the total income population and Yi
= yi.pi.P is the total income in the ith group. W then reduces to Si Yi = Y.
23 For 1996 this was checked against decile income data from Economic Trends,
March 1997. The indexed calculation gave a lower value for the Atkinson index
than the Economic Trends data. Since the two data sources are not necessarily
comparable, we adopted the more conservative estimate.
24 Actually, this work yields a mean value of 0.83 for the elasticity of the
marginal utility of consumption over time (Pearce and Ulph 1995, based on Blundell
et al 1994), rather than over different income groups.
25 This is inaccurate to the extent that different income groups allocate different
proportions of their income to net savings, rather than to expenditure. It is
beyond the scope of this paper to investigate any consequences of this inaccuracy.
26 This advantage becomes a disadvantage, however, when it comes to providing
an aggregate index of welfare.
27 In general, we have not considered the welfare value of `leisure' within
the ISEW, although it has been pointed out elsewhere (Max Neef, 1991), that
both leisure and 'idleness' do contribute to human well-being.
28 This may be an overestimate since the last decade has seen an increase in
commercial child care. However, the assumption is conservative in the sense
that it increases the contribution from domestic labour over the later years
of the study, and therefore inflates the ISEW over what it might otherwise have
been.
29 This makes the value of domestic labour calculation compatible with labour
force statistics - used in calculating net capital growth adjustment requirements.
30 Typically, the male wage rate for other cleaners was 40% (1970) to 20% (1988)
higher than the equivalent female rate. The 'home & domestic' female rate was
between 1-10% higher than the female 'other cleaners' rate.
31 We have valued male and female domestic labour identically, even though there
has certainly been some discrimination within the labour market over the period
of the study.
32 See the sensitivity analysis in section 4 below.
33 Data on health expenditures comes from AAS 1955 (Table 48), 1961 (Table 41),
1966 (Table 39), 1974, 1982, 1991 (Table 3.3); data on education expenditures
is taken from Table 43 in AAS 1972; Table 36, 1974; Table 3.2, 1983; Table 3.2
1986; Table 3.2, 1991. Prior to 1963, the expenditure on further education was
estimated from the total expenditure on education using the average percentage
of total expenditure on further education (calculated as 24.3%) during the years
1963 to 1987.
34 Friedman 1957, cited in Patterson 1992.
35 Patterson, personal communication, 1993.
36 In principle, of course, a more detailed analysis might also identify reduced
obsolescence in certain categories of durable goods, although anecdotal evidence
points to the contrary assumption. It should be noted here that not all 'premature'
replacement of durable goods can be regarded as unwanted obsolescence, since
some replacement arises from technological upgrading. In Patterson's model,
these kinds of technological improvements would be reflected in the calculation
of capital gains and losses.
37 These are published as an appendix to the Family Expenditure Survey (FES,
various years) for the years 1954 and 1980-1996, and are available on microfiche
for interim years.
38 It should be noted that both the DoT data and the MoT data on expenditures
referred only to Great Britain. Our costs will therefore be underestimates to
the extent that they exclude the costs of commuting in Northern Ireland.
39 Converted to 1990 pounds sterling using the consumer expenditure inflator
gives costs of œ2,459 and œ826 respectively.
40 At the time of writing the figures for 1996 were not yet available, and have
been estimated to be the same as the 1995 figures.
41 A further objection levelled at the US methodology by some of its critics
(see Cobb and Cobb, 1994) was that the index should measures losses (or gains)
in services from the stock of clean air, rather than emissions per se. In order
to accommodate this objection, the revised US index used ambient air quality
levels (as an indicator of the stock of clean air) rather than emissions levels.
The difficulty with this approach is that it incorrectly assumes that the damages
incurred by emissions into the atmosphere all relate to ambient air quality.
But acid emissions, for example, are implicated in environmental damage to soils,
to water, to plant-life and to buildings. A further problem is that ambient
air quality measurements are necessarily site specific, and the question of
an appropriate index on a national basis is confused by the need to make an
appropriate weighting of measurements from different sites.
42 See for example: Pearce et al 1989, Pearce and Turner 1990, Baumann and Hill
1991, DoT 1992, Woolf, 1992.
43 Since the Daly and Cobb analysis does not actually distinguish in the index
between the effects of different kinds of damage this level of disaggregation
is somewhat supernumerary.
44 Using SO2, NOx and Black Smoke emissions as an indicator of air pollution
in the two countries, the cost estimates given are equivalent to average costs
(in 1990 pounds) of œ644 and œ4,118 per tonne of emissions, respectively for
Netherlands and Germany.
45 Defined as being a given distance away from major roads, flight paths and
power lines.
46 Daly and Cobb assumed a 3% per annum growth in noise pollution from 1950
to 1972, and 1% per annum reduction thereafter.
47 The habitats for which sufficient estimates of extent over time are available
are native pine woodlands, wetlands, grazing marsh, unimproved neutral grasslands,
lowland heathland, limestone pavements and upland semi-natural woods.
48 See footnote 12 above.
49 By comparison, published data on land prices (MAFF, 1993) puts the price
of agricultural land (land only with vacant possession) at around œ3,000 per
hectare towards the end of 1992 and the beginning of 1993. 50 The level of discount
rate is not explicitly identified in Daly and Cobb's work, but implicitly they
seem to adopt (p435) a 10% discount rate when discussing land values. For the
sake of argument, we have also adopted this discount rate here.
51 This cumulative cost is rather conservative given that erosion in some areas
has been relatively severe since deforestation occurred on a large scale in
the UK, but assumes that erosion rates prior to 1950 were considerably lower
than today.
52 The El Serafy formula is X/R = 1-1/(1+r)n+1. The 'set-aside' to account for
resource depletion is then given by R-X.
53 It can be shown that this method leads to marginal social costs which are
very similar to those which would be derived by extrapolating back in time on
the basis of Fankhauser's assumption that in each (future) decade the marginal
Implicitly included in Daly and Cobb's estimates although excluded from the
much higher Azar and Holmberg estimate.
56 According to the standard classification CFC is taken to have an ODP of 1.
The ODPs for CFCs 12, 113, 114, and 115 are (respectively) 1.03, 0.8, 0.75 and
0.42.
57 In 1986 consumption was 56% of production. During the years from 1986 to
1990 the average figure was 45%.
58 In fact, five year rolling averages of the increase in the labour force and
the increase in capital stock have been used to smooth out year on year variations.
59 The specific effects of these definitional changes were the need a) to subtract
20,000 from the workforce figures from 1980 onwards b) to add 160,000 from 1983
onwards and c) to add a further 50,000 from 1986 onwards (cf p109 of AAS, 1991a).
Introduction | Replacing GDP | ISEW explained | Make your own ISEW | View the results | International examples | How ISEW terms are calculated
