7.1.2.1 Production planning

The organization of the production processes in a number of interrelated industries that buy and supply materials and goods among themselves may be described in terms of a production function Y (see, for example, Morishima, 1976), giving a measure of net total output,

$Y=Y(\{M_i,L_i,R_i\}).$

Here M_i, L_i, and R_i stand for the machinery, labor, and raw materials required by the ith process, and the net output z_j of one of the goods entering somewhere in the hierarchy of production processes may be assumed to be given by a linear expression,

$z_j=\sum _i{b}_{ji}{x}_i,$ (7.1)

(7.1)

where x_i is the rate of the ith process and {b_ji} is a set of coefficients. The entire collection of process rates, {x_i}, constitutes a production plan. It is seen that the conglomerate production function Y involves a weighting of the individual goods (index j), which will normally be described in different units. Equation (7.1) describes both inputs and outputs of goods, prescribing that those z_j that are inputs are given a negative sign and those that are outputs are given a positive sign.

A production plan aiming at maximizing the total output, as specified by Y, for given inputs, makes little sense, unless the weighting of different goods is properly chosen. The choice of this weighting may be taken to transform the measure of output into a measure of profit. If the monetary value (price) assigned to each good z_j is denoted p_j (in monetary units per unit quantity), then the net profit, y, may be written

$y=\sum _j{p}_j{z}_j.$ (7.2)

(7.2)

The production plan {x_i} may now be chosen, through (7.1) and (7.2), so as to maximize the total profit y. The method may be generalized to include all sectors of society, including the public sector.

Allocation of the net total profit to the individual processes that may be performed by different enterprises (with different owners in a capitalist economy) presents another problem. A “rational” (but not necessarily socially acceptable) distribution might be to return the total profit to the individual processes, in the same ratios as the contributions of the processes to the total profit. The prices of the individual goods defined in this way are called shadow prices. Thus, if all the prices p_j appearing in (7.2) are taken to be the shadow prices, then the net total profit y becomes zero. In practice, this would rarely be the case, and even if individual prices within the production hierarchy are at one time identical to the shadow prices, then changes in, for example, raw material costs or improved methods in some of the processes will tend to make y non-zero and will lead to frequent adjustments of shadow prices.

This behavior emphasizes the static nature of the above methodology. No dynamic development of production and demand is included, and changes over time can be dealt with only in a “quasi-static” manner by evaluating the optimal production plan, the pricing, and the profit distribution each time the external conditions change and by neglecting time delays in implementing the planning modifications. Dynamic descriptions of the time development of non-static economies have been formulated in qualitative terms, e.g., in the historical analysis of Marx (1859) and in the approach of Schumpeter (1961).

The quasi-static theory is particularly unsuited for describing economies with a long planning horizon and considerations of resource depletion or environmental quality. By assigning a monetary value to such externalities or external diseconomies, a dynamic simulation calculation can be carried out, defining the basic economic rules and including time delays in implementing changes, delayed health effects from pollution, and the time constants for significant depletion of nonrenewable resources, either physical depletion or depletion of raw materials in a given price range. Rather than planning based on instantaneous values of profit, planning would be based on suitably weighted sums of profits during the entire length of the planning horizon (see, for example, Sørensen, 1976a). The determination of values to be ascribed to externality costs is discussed in section 7.3 on life-cycle analysis.

Furthermore, the applicability of the simulation approach is limited because of the uncertainty inherent in time integration over very long time intervals, using economic parameters that are certainly going to be modified during the course of time or replaced by completely different concepts in response to changes in external conditions as well as major changes in the organization and value systems of societies. Examples of problems that defy description in terms of conventional economic theories with use of simulation techniques may be the socioeconomic costs of possible climatic impact from human (profit-seeking) activities, the global contamination resulting from nuclear warfare, and the more subtle aspects of passing radioactive waste to future generations. Such problems, it would seem, should be discussed in a general framework of social values, including the values associated with not restricting the possibilities for future generations to make their own choices by restricting their access to natural resources or by forcing them to spend a considerable part of their time on our leftover problems rather than on their own.

The quantification of, for example, environmental values and their inclusion in economic planning still suffer from the lack of a suitable theoretical framework. It has been suggested that contingent evaluation should be employed, implying that funds are allocated to environmental purposes (reducing pollution from given production processes, setting aside land for recreational purposes, etc.) to an extent determined by the willingness of people (salary earners) to accept the reduction in salary that corresponds to the different allocation of a given amount of profit (see, for example, Hjalte et al., 1977; Freeman et al., 1973). It is clearly difficult to include long-term considerations in this type of decision process. In many cases, the adverse environmental effects of present activities will not show up for many years, or our knowledge is insufficient to justify statements that adverse effects will indeed occur. Only recently have public debates about such issues started to take place before the activity in question has begun. Thus, the discussion has often been about halting an ongoing activity for environmental purposes, and the argument against doing so has been loss of investments already made. It is therefore important to evaluate environmental problems of production or consumption in the planning phase, which implies that, if environmental costs can be evaluated, they can be directly included in the planning prescription [such as (7.1) and (7.2)] as a cost of production on the same level as the raw materials and machinery (Sørensen, 1976a).

Another method that is in fairly widespread use in present economies is to implement government regulations, often in the form of threshold values for pollution levels (emissions, concentrations, etc.) that should not be exceeded. If, then, for a given production process, the emission of pollutants is proportional to the output, the regulation implies a government-prescribed ceiling on the output z_j of a given product in (7.2). This is similar to the ceilings that could be prescribed in cases of a scarce resource (e.g., cereal after a year of poor harvest) or other rationing requirement (e.g., during wartime). Such an environmental policy will stimulate development of new production processes with less pollution per unit of output, so that the output of goods may still be increased without violating the pollution limits.

The problem is, however, that government-enforced limits on pollution are usually made stricter with time, because they were initially chosen too leniently and because additional long-term adverse effects keep appearing (an example is radiation exposure standards). Thus, the socioeconomic effects of environmental offenses are constantly underestimated in this approach, and industries and consumers are indirectly stimulated to pollute up to the maximum permitted at any given time. By contrast, the approach whereby environmental impact must be considered earlier, in the production-planning phase, and whereby governments may determine the cost of polluting to be at such a high value that long-term effects are also included, has the effect of encouraging firms and individuals to keep pollution at the lowest possible level.

7.1.2.2 Distribution problems

Classical market theory assumes completely free competition between a large number of small firms. There are no monopolies and no government intervention in the economy. Extension of this theory to include, for example, a public sector and other features of present societies, without changing the basic assumption, is called neoclassical theory. According to this view, the distribution of production on the firms, and the distribution of demand on the goods produced, may attain equilibrium, with balanced prices and wages determined from the flexprice mechanism (this is called a Walras equilibrium).

In neoclassical theory, it is assumed that unemployment is a temporary phenomenon that occurs when the economy is not in equilibrium. Based on observations in times of actual economic crises occurring in capitalist countries, Keynes (1936) suggested that in reality this was not so, for a number of reasons. Many sectors of the “late capitalist” economies were of the fixprice category rather than the flexprice one, wages had a tendency to be rigid against downward adjustment, the price mechanism was not functioning effectively (e.g., owing to the presence of monopolies), and the relation between prices and effective demand was therefore not following the Walras equilibrium law. Since prices are rigid in the fixprice economy, full employment can be obtained only when demand and investments are sufficiently high, and Keynes suggested that governments could help stimulate demand by lowering taxes and that they could help increase economic activity by increasing public spending (causing a “multiplier effect” due to increased demand on consumer goods) or by making it more attractive (for capitalists) to invest their money in means of production. The distinction between consumer goods and capital goods is important, since unemployment may exist together with a high, unsatisfied demand for consumer goods if capital investments are insufficient (which again is connected with the expected rates of interest and inflation). This is a feature of the recent (2008+) financial crisis, which was caused by allowing nonproductive sectors, such as the financial sector, to attract a very high fraction of the total investments in society (in itself detrimental) and then to grossly mismanage the assets.

In a socialist economy, allocation of resources and investments in means of production is made according to an overall plan (formulated collectively or by representatives), whereas in a capitalist society the members of the capitalist class make individual decisions on the size of investment and the types of production in which to invest. Then it may well be that the highest profit is in the production of goods that are undesirable to society but are still demanded by customers if sales efforts are backed by aggressive advertising campaigns. It is in any case not contradictory that there may be a demand for such goods, if more desirable goods are not offered or are too expensive for most consumers. Yet relatively little is done, even in present mixed-economy societies, to influence the quality of the use of capital, although governments do try to influence the level of investments.

Unemployment is by definition absent in a socialist society. In a capitalist society, the effect of introducing new technology that can replace labor is often to create unemployment. For this reason, new technology is often fiercely opposed by workers and unions, even if the benefits of the machinery include relieving workers from the hardest work, avoiding direct human contact with dangerous or toxic substances, etc. In such cases, the rules of capitalist economy clearly oppose improvements in social values.

7.1.2.3 Actual pricing policies

The mechanism of price adjustments actually used in a given economy is basically a matter of convention, i.e., it cannot be predicted by a single theory, but theories can be formulated that reflect some features of the price mechanism actually found in a given society at a given time. A mechanism characteristic of present growth-oriented economies (both socialist and capitalist) is to adjust prices in such a way that excess profit per unit of good is reduced, whenever an excess demand exists or can be created, in order to be able to increase output (Keynes–Leontief mechanism).

Important parameters in an empirical description of these relations are the price flexibility,

${\eta }_i=\frac{\mathrm{\Delta }{p}_i/p_i}{\delta d_i/d_i},$ (7.3)

(7.3)

and the demand elasticity,

${\epsilon }_i=-\frac{\mathrm{\Delta }{d}_i/d_i}{\mathrm{\Delta }{p}_i/p_i},$ (7.4)

(7.4)

of the ith good or sector. Here p_i and d_i represent price and demand, and δd_i is an exogenous change in demand, whereas Δd_i is the change in demand resulting from a change in price from p_i to p_i+Δp_i. The change in output, Δz_j, being indicated by an exogenously produced change in demand, Δd_i, is

$\mathrm{\Delta }{z}_i=\mathrm{\Delta }{d}_i+\delta d_i=(1-{\eta }_i{\epsilon }_i)\delta d_i.$ (7.5)

(7.5)

The use of the indicators η_i and d_i for predicting the results of changing prices or demand is equivalent to a quasi-static theory, whereas a dynamic theory would require η_i and ε_i to be dynamic variables coupled to all other parameters describing the economy through a set of nonlinear, coupled differential equations (i.e., including feedback mechanisms left out in simple extrapolative approaches).

The large number of assumptions, mostly of historical origin, underlying the present theoretical description of pricing policy (actual or “ideal”) does suggest that comparative cost evaluations may constitute an inadequate or at least incomplete basis for making major decisions with social implications. Indeed, many important political and business decisions are made in bold disregard of economic theory. This is relevant for the discussion of energy systems. A statement like, one system is 10% cheaper than another one, would be seen in a new light in cases where the combination of ingredients (raw materials, labor, environmental impact, availability of capital) determining the prices of the two systems are different. There might be a choice between changing the production process to make a desired system competitive or, alternatively, making society change the rules by which it chooses to weight the cost of different production factors.

7.1.4.1 Private and social interest rate, intergenerational interest

The concept of interest is associated with placing a higher value on current possession of assets than on promised future possession. Using the monetary value system, a positive interest rate thus reflects the assumption that it is better to possess a certain sum of money today than it is to be assured of having it in 1 year or in 20 years from today. The numerical value of the interest rate indicates how much better it is to have the money today, and if the interest has been corrected for the expected inflation, it is a pure measure of the advantage given to the present.

For the individual, who has a limited life span, this attitude seems very reasonable. But for a society, it is difficult to see a justification for rating the present higher than the future. New generations of people will constitute the society in the future, and placing a positive interest on the measures of value implies that the “present value” (7.7 below) of an expense to be paid in the future is lower than that of one to be paid now. This has a number of implications.

Nonrenewable resources left for use by future generations are ascribed a value relatively smaller than those used immediately. This places an energy system requiring an amount of nonrenewable fuels to be converted every year in a more favorable position than a renewable energy system demanding a large capital investment now.

Part of the potential pollution created by fuel-based energy conversion, as well as by other industrial processes, is in the form of wastes, which may be either treated (changed into more acceptable substances), diluted, and spread in the environment, or concentrated and kept, i.e., stored, either for good or for later treatment, assuming that future generations will develop more advanced and appropriate methods of treatment or final disposal. A positive interest rate tends to make it more attractive, when possible, to leave the wastes to future management, because the present value of even a realistic evaluation of the costs of future treatment is low.

For instance, any finite cost of “cleaning-up” some environmental side-effect of a technology (e.g., nuclear waste) can be accommodated as a negligible increase in the capital cost of the technology, if the clean-up can be postponed sufficiently far into the future. This is because the clean-up costs will be reduced by the factor (1+r)ⁿ if the payment can wait n years (cf. the discussion in section 7.1.2 of proposed economic theories for dealing with environmental side-effects, in connection with “production planning”).

It may then be suggested that a zero interest rate be used in matters of long-term planning seen in the interest of society as a whole. The social values of future generations, whatever they may be, would, in this way, be given a weight equal to those of the present society. In any case, changing from the market interest rate to one corrected for inflation represents a considerable step away from the interest (in both senses) of private capitalist enterprises and toward that of society as a whole. As mentioned above, the interest rate corrected for inflation according to historical values of market interest and inflation, and averaged over several decades, amounts to a few percent per year. However, when considering present or presently planned activities that could have an impact several generations into the future (e.g., resource depletion, environmental impact, including possible climatic changes), then even an interest level of a few percent would appear to be too high to take into proper consideration the welfare of future inhabitants of Earth. For such activities, the use of a social interest rate equal to zero in present value calculations does provide one way of giving the future a fair weight in the discussions underlying present planning decisions.

One might argue that two facts speak in favor of a positive social interest rate: (i) the capital stock, in terms of buildings and other physical structures, that society passes to future generations is generally increasing with time, and (ii) technological progress may lead to future solutions that are better and cheaper than those we would use to deal with today’s problems, such as waste disposal. The latter point is the reason why decommissioned nuclear reactors and high-level waste are left to future generations to deal with. However, there are also reasons in favor of a negative social interest rate: environmental standards have become increasingly more strict with time, and regulatory limits for injection of pollutants into both humans (through food) and the environment have been lowered as function of time, often as a result of new knowledge revealing negative effects of substances previously considered safe. The problems left to the future may thus appear larger then, causing a negative contribution to social interest, as it will be more expensive to solve the problems according to future standards. The use of a zero social interest rate may well be the best choice, given the uncertainties.

However, it is one thing to use a zero interest rate in the present value calculations employed to compare alternative future strategies, but quite another thing to obtain capital for the planned investments. If the capital is not available and has to be procured by loans, it is unlikely that the loans will be offered at zero interest rate. This applies to private individuals in a capitalist society and to nations accustomed to financing certain demanding projects by foreign loans. Such borrowing is feasible only if surpluses of capital are available in some economies and if excess demand for capital is available in others. Assuming that this is not the case, or that a certain region or nation is determined to avoid dependence on outside economies, then the question of financing is simply one of allocation of the capital assets created within the economy.

If a limited amount of capital is available to such a society, it does not matter if the actual interest level is zero or not. In any case, the choices between projects involving capital investments, which cannot all be carried out because of the limited available capital, can then be made on the basis of zero interest comparisons. This makes it possible to maximize long-term social values, at the expense of internal interest for individual enterprises. Social or “state” intervention is necessary in order to ensure a capital distribution according to social interest comparisons, rather than according to classical economic thinking, in which scarcity of money resources is directly reflected in a higher “cost of money,” i.e., positive interest rate.

The opposite situation, that too much capital is available, would lead to a low interest rate, as has been seen during the first decade of the 21st century and is still prevailing. A low short-term market interest signals that an insufficient number of quality proposals for investment are being generated in society, and this may explain (but not justify) why financial institutions during this period made unsecured loans to unworthy purposes, such as excess luxury consumption and substandard industrial or commercial product ideas. Declining quality of education in the richest countries may be invoked as an explanation for the lack of worthy new investment ideas, or it may be a result of the many technological advances during the preceding decades having generated larger-than-usual profits, i.e., capital to be reinvested.

7.1.4.2 Net present value calculation

The costs of energy conversion may be considered a sum of the capital costs of the conversion devices, operation and maintenance costs, plus fuel costs if there are any (e.g., conversion of nonrenewable fuels or commercially available renewable fuels, such as wood and other biomass products). Some of these expenses have to be paid throughout the operating lifetime of the equipment and some have to be paid during the construction period, or the money for the capital costs may be borrowed and paid back in installments.

Thus, different energy systems would be characterized by different distributions of payments throughout the operating lifetime or at least during a conventionally chosen depreciation time. In order to compare such costs, they must be referred to a common point in time or otherwise made commensurable. A series of yearly payments may be referred to a given point in time by means of the present value concept. The present value is defined by

$P=\sum _{i=1}^n{p}_i{(1+r)}^{-1},$ (7.7)

(7.7)

where {p_i} are the annual payments and n is the number of years considered. The interest rate r is assumed to be constant and is ascribed only once every year (rather than continuously), according to present practice. The interest is paid post numerando, i.e., the first payment of interest would be after the first year, i=1. Thus, P is simply the amount of money that one must have at year zero in order to be able to make the required payments for each of the following years, from the original sum plus the accumulated interest earned, if all remaining capital is earning interest at the annual rate r.

If the payments p_i are equal during all the years, (7.7) reduces to

$P=p_{(i)}{(1+r)}^{-1}({(1+r)}^{-n}-1)/({(1+r)}^{-1}-1),$

and if the payments increase from year to year, by the same percentage, p_i=(1+e)p_i−1 for i=2, …, n, then the present value is

$P=p_0{(1+r)}^{-1}\frac{{((1+e)/(1+r))}^n-1}{(1+e)/(1+r)-1}.$ (7.8)

(7.8)

Here e is the rate of price escalation (above the inflation rate if the average inflation is corrected for in the choice of interest r, etc.) and p₀ is the (fictive) payment at year zero, corresponding to a first payment at year one equal to p₀(1+e).

The present value of capital cost C paid at year zero is P=C, whether the sum is paid in cash or as n annual installments based on the same interest rate as used in the definition of the present value. For an annuity-type loan, the total annual installment A (interest plus back-payment) is constant and given by

$A=C(1+r)\frac{{(1+r)}^{-1}-1}{{(1+r)}^{-n}-1},$ (7.9)

(7.9)

which, combined with the above expression for the present value, proves that P=C. When r approaches zero, the annuity A approaches C/n.

The present value of running expenses, such as fuels or operation/maintenance, is given by (7.7), or by (7.8), if cost increases yearly by a constant fraction. It is usually necessary to assign different escalation rates e for different fuels and other raw materials; for labor of local origin, skilled or unskilled; and for imported parts or other special services. Only rarely will the mix of ingredients behave like the mix of goods and services defining the average rate of inflation or price index (in this case, prices will follow inflation and the escalation will be zero). Thus, for a given energy system, the present value will normally consist of a capital cost C plus a number of terms that are of the form (7.7) or (7.8) and that generally depend on future prices that can at best be estimated on the basis of existing economic planning and expectations.

7.1.6.1 Energy analysis

For energy conversion systems, the energy accounting is of vital interest because the net energy delivered from the system during its physical lifetime is equal to its energy production minus the energy inputs into the materials forming the equipment, its construction, and its maintenance. Different energy supply systems producing the same gross energy output may differ in regard to energy inputs, so that a cost comparison based on net energy outputs may not agree with a comparison based on gross output.

The energy accounting may be done by assigning an energy value to a given commodity, which is equal to the energy used in its production and in preparing the raw materials and intermediate components involved in the commodity, all the way back to the resource extraction, but no further. Thus, nuclear, chemical, or other energy already present in the resource in its “natural” surroundings is not counted. It is clear that energy values obtained in this way are highly dependent on the techniques used for resource extraction and manufacture. Energy analysis along these lines has served as a precursor for full life-cycle analyses of energy systems.

For goods that may be used for energy purposes (e.g., fuels), the “natural” energy content is sometimes included in the energy value. There is also some ambiguity in deciding on the accounting procedure for some energy inputs from “natural” sources. Solar energy input into, for example, agricultural products is usually counted along with energy inputs through fertilizers, soil conditioning, and harvesting machinery, and (if applicable) active heating of greenhouses, but it is debatable what fraction of long-wavelength or latent heat inputs should be counted (e.g., for the greenhouse cultures), which may imply modifications of several of the net energy exchanges with the environment. Clearly, the usefulness of applying energy analysis to different problems depends heavily on agreeing on the definitions and prescriptions that will be useful in the context in which the energy value concept is to be applied.

Early energy analyses were attempted even before sufficient data had become available (Chapman, 1974; Slesser, 1975; Hippel et al., 1975). Still, they were able to make specific suggestions, such as not building wave energy converters entirely of steel (Musgrove, 1976). More recently, a large number of energy systems have been analyzed for their net energy production (see, for example, the collection in Nieuwlaar and Alsema, 1997).

It should be said, in anticipation of the life-cycle approach discussed in section 7.3, that, if all impacts of energy systems are included in a comparative assessment, then there is no need to perform a specific net energy analysis. Whether a solar-based energy device uses 10% or 50% of the originally captured solar radiation internally is of no consequence as long as the overall impacts are acceptable (including those caused by having to use a larger collection area to cover a given demand).

In order to illustrate that net energy is also characterized by time profiles that may differ for different systems, Fig. 7.3 sketches the expected time profiles for the net energy production of two different types of energy conversion systems. They are derived from cost estimates in Sørensen (1976b). For both systems, the initial energy “investment” is recovered over 2–3 years of energy production.

The prolonged construction period of some energy systems does present a problem if such systems are introduced on a large scale over a short period of time. The extended delay between the commencement of energy expenditures and energy production makes the construction period a potentially serious drain of energy (Chapman, 1974). This problem is less serious for modular systems, which have short construction periods for each individual unit.

7.2.1.1 National economy

In mixed or socialist economies, a production may be maintained even if it yields a loss. One reason may be to maintain the capability of supplying important products when external conditions change (e.g., maintaining food production capacity even in periods when imported foods are cheaper) or to maintain a regional dispersion of activity that might be lost if economic activities were selected strictly according to direct profit considerations. This may apply to regions within a nation or to nations within larger entities with a certain level of coordination.

Because energy supply is important to societies, they are likely to be willing to transfer considerable funds from other sectors to ensure security of energy supplies and national control over energy sources and systems.

International trade of raw materials and intermediate and final products has now reached very substantial dimensions. Often, quite similar goods are exchanged between nations in different parts of the world, with the only visible result being increased demand for transportation. Of course, all such activities contribute to the gross national product, which has been increasing rapidly (even after correction for inflation) over recent decades in most countries in the world.

A measure of a nation’s success in international trade is its balance of foreign payments, i.e., the net sum of money in comparable units that is received by the nation from all other nations. Since the worldwide sum of foreign payment balances is by definition zero, not all nations can have a positive balance of payments, but they could, of course, all be zero.

An important factor in foreign trade is each dealer’s willingness to accept payments in a given currency. To some extent, currencies are simply traded like other commodities, using basically the flexprice method for fixing exchange rates. It is also customary in international trade that governments are called upon to provide some form of guarantee that payments will be made. In international finance, the “trade of money” has led to an increasing amount of foreign loans, given both to individual enterprises and to local and national governments. A positive balance of current foreign payments may thus indicate not that a surplus of goods has been exported, but that a large number of loans have been obtained, which will later have to be returned in installments, including interest. In addition to the balance of current payments, the balance of foreign debts must also be considered in order to assess the status of an individual nation. The origin of international finance is, of course, the accumulation of capital in some countries, capital that has to be invested somewhere in order to yield a positive interest to its owners.

Assuming that all nations try to have a zero or positive balance of foreign payments, their choice of, for example, energy systems would be made in accordance with this objective. For this reason, direct economic evaluation should be complemented by a list of the import values of all expenses (equipment, maintenance, and fuels, if any) for the energy supply systems considered. Import value is defined as the fraction of the expense that is related to imported goods or services (and which usually has to be paid in foreign currency). The import value of different energy technologies depends on methods of construction and on the choice of firms responsible for the individual enterprises. It is therefore heavily dependent on local and time-dependent factors, but is nevertheless a quantity of considerable importance for national planning. In many cases, the immediate flexibility of the import value of a given system is less than that implied by the above discussion, because of the time lags involved in forming a national industry in a new field or, depending on the number of restrictions placed on foreign trade, because of the existence of well-established foreign enterprises with which competition from new firms would be hard to establish. Examples of calculations of import values for different energy technologies, valid for the small nation of Denmark, may be found in Blegaa et al. (1976).

7.3.1.1 Treatment of import and export

Many of the available data sources include indirect impacts based on an assumed mix of materials and labor input, taking into account the specific location of production units, mines, refineries, etc. This means that an attempt has been made to retrace the origin of all ingredients in a specific product (such as electricity) in a bottom-up approach, which is suited for comparison of different ways of furnishing the product in question (e.g., comparing wind, coal, and nuclear electricity).

In the case of the analysis of an entire energy system, the data for specific sites and technology, as well as specific countries from which to import, are not the best suited. Especially for the future scenarios, it seems improper to use data based on the current location of mines, refineries, and other installations. One may, of course, average over many different sets of data, in order to obtain average or “generic” data, but the selection of future energy systems should not depend sensitively on where, for example, our utilities choose to purchase their coal this particular year, and therefore a different approach has to be found.

One consistent methodology is to consider the energy system of each country being studied a part of the national economy, so that, if that country chooses to produce more energy than needed domestically in order to export it, this constitutes an economic activity no different from, say, Denmark’s producing more LEGO blocks than can be used by Danish children. Exports are an integral part of the economy of a small country like Denmark, because it cannot and does not produce every item needed in its society and thus must export some goods in order to be able to import other ones that it needs. Seen in this way, it is “their own fault” if manufacturing the goods they export turns out to have more environmental side-effects than the imports have in their countries of origin, and the total evaluation of impacts should simply include those generated as part of the economy of the country investigated, whether for domestic consumption or export. Similarly, the impacts of goods imported should be excluded from the evaluation for the importing country, except, of course, for impacts arising during the operation or use of the imported items in the country (for example, burning of imported coal).

A consistent methodology is thus to include all impacts of energy production and use within the country considered, but to exclude impacts inherent in imported materials and energy. If this is done for a Danish scenario (Kuemmel et al., 1997), the impact calculation related to energy for the present system turns out not to differ greatly from one based on the impacts at the place of production, because Denmark is about neutral with respect to energy imports and exports (this varies from year to year, but currently, oil and gas exports roughly balance coal imports). However, for other countries, or for future systems, this could be very different, because the impacts of renewable systems are chiefly from manufacture of conversion equipment, which consists of concrete, metals, and other materials, of which 30%–50% may be imported.

The argument against confining externality calculations to impacts originating within the country is that this could lead to purchase of the most environmentally problematic parts of the system from countries paying less attention to the environment than we claim to do. Countries in the early stages of industrial development have a tendency to pay less attention to environmental degradation, making room for what is negatively termed “export of polluting activities to the Third World.” The counterargument is that this is better for the countries involved than no development, and that, when they reach a certain level of industrialization, they will start to concern themselves with the environmental impacts (examples are Singapore and Hong Kong). Unfortunately, this does not seem to be universally valid, and the implication of environmental neglect in China, India, and countries in South America extends outside their borders, as in the case of global warming, which is not confined to the country of origin or its close regional neighbors. Still, from a methodological point of view, the confinement of LCA impacts to those associated with activities in one country does provide a fair picture of the cost of any chosen path of development, for industrializing as well as for highly industrialized countries.

The problem is that a large body of existing data is based on the other methodology, where each energy form is viewed as a product, and impacts are included for the actual pathway from mining through refining and transformation to conversion, transmission, and use, with the indirect impacts calculated where they occur, that is, in different countries. In actuality, the difficulty in obtaining data from some of the countries involved in the early stages of the energy pathway has forced many externality studies to use data “as if” the mining and refining stages had occurred in the country of use. For example, the European Commission study (1995c) uses coal mined in Germany or England, based on the impacts of mining in these countries, rather than the less well-known impacts associated with coal mining in major coal-exporting countries. It has been pointed out that this approach to energy externalities can make the LCA too uncertain to be used meaningfully in decision processes (Schmidt et al., 1994). Recent product LCAs looking at soft-drink bottles and cans clearly suffer from this deficiency, making very specific assumptions regarding the place of production of aluminum and glass and of the type of energy inputs to these processes (UMIP, 1996). If the can manufacturer chooses to import aluminum from Tasmania or the United States instead of, say, from Norway, the balance between the two types of packing may tip the other way.

The two types of analysis are illustrated in Fig. 7.7. The method in common use for product LCA, where imbedded impacts are traced through the economies of different countries, not only seems problematic when used in any practical example, but also is unsuited for the system analysis that decision-makers demand. In contrast, one may use a conventional economic approach, where import and export are specific features of a given economy, deliberately chosen in consideration of the assets of the country, and calculate LCA impacts only for the economy in question in order to be consistent.

7.3.1.2 What to include in an LCA

The history of LCA has taken two distinct paths. One is associated with the energy LCAs, developed from chain analysis without imbedded impacts to analysis including such impacts, as regards both environmental and social impacts. The history and state of the art of energy LCA are described in Sørensen (1993a, 1996a). The main ingredients suggested for inclusion in an energy LCA are listed below. The other path is associated with product LCA and has been pushed by the food and chemical industries, particularly through the SETAC (Consoli et al., 1993). SETAC tends to consider only environmental impacts, neglecting most social impacts. The same is true for the prescriptions of standardization institutions (ISO, 1997, 1999), which mainly focus on establishing a common inventory list of harmful substances to consider in LCA work. Both types of LCA were discussed methodologically around 1970, but only recently have they been transferred to credible calculations, because of the deficiencies in available data and incompleteness. Many absurd LCA conclusions were published from the late 1960s to the late 1980s. The present stance is much more careful, realizing that LCA is not, and cannot be made, a routine screening method for products or energy systems, but has to remain an attempt to furnish more information to the political decision-maker than has previously been available. The decision process will be of a higher quality if these broader impacts are considered, but the technique is never going to be a computerized decision tool capable of replacing political debate before decisions are made. This is also evident from the incommensurability of different impacts, which cannot always be meaningfully brought to a common scale of units. To emphasize this view of the scope of LCA, this section lists impacts to consider, without claiming that the list is inclusive or complete.

The impacts contemplated for assessment reflect to some extent the issues that at a given moment in time have been identified as important in a given society. It is therefore possible that the list will be modified with time and that some societies will add new concerns to the list. However, the following groups of impacts, a summary of which is listed in Table 7.1, constitute a fairly comprehensive list of impacts considered in most studies made to date (Sørensen, 1993a):

7.3.1.3 Choosing the context

When the purpose of the LCA is to obtain generic energy technology and systems evaluations (e.g., as inputs into planning and policy debates), one should try to avoid using data depending too strongly on the specific site selected for the installation. Still, available studies may depend strongly on location (e.g., special dispersal features in mountainous terrain) or a specific population distribution (presence of high-density settlements downstream relative to emissions from the energy installation studied). For policy uses, these special situations should be avoided if possible and be left for the later, detailed plant-siting phase to eliminate unsuitable locations. This is not normally a problem if the total planning area is sufficiently diverse.

Pure emission data often depend only on the physical characteristics of a given facility (power plant stack heights, quality of electrostatic filters, sulfate scrubbers, nitrogen oxide treatment facilities, etc.) and not on the site. However, dispersion models are, of course, site dependent, but general concentration versus distance relations can usually be derived in model calculations avoiding any special features of sites. The dose commitment will necessarily depend on population distribution, while the dose–response relationship should not depend on it. As a result, in many cases a generic assessment can be performed with only a few adjustable parameters left in the calculation, such as the population density distribution, which may be replaced with average densities for an extended region.

The approach outlined above will serve only as a template for assessing new energy systems, as the technology must be specified and usually would involve a comparison between different new state-of-the-art technologies. If the impact of the existing energy system in a given nation or region has to be evaluated, the diversity of the technologies in place must be included in the analysis, which would ideally have to proceed as a site- and technology-specific analysis for each piece of equipment.

In generic assessments, not only do technology and population distributions have to be fixed, but also a number of features characterizing the surrounding society will have to be assumed, in as much as they may influence the valuation of the calculated impacts (and in some cases also the physical evaluation, e.g., as regards society’s preparedness for handling major accidents, which may influence the impact assessment in essential ways).

7.3.1.4 Aggregation issues

Because of the importance of aggregation issues, both for data definition and for calculation of impacts, this topic is discussed in more detail. There are at least four dimensions of aggregation that play a role in impact assessments:

The most disaggregated studies done today are termed “bottom-up” studies of a specific technology located at a specific site. Since the impacts will continue over the lifetime of the installation, and possibly longer (e.g., radioactive contamination), there is certainly an aggregation over time involved in stating the impacts in compact form. The better studies attempt to display impacts as a function of time, e.g., as short-, medium-, and long-term effects. However, even this approach may not catch important concerns, as it will typically aggregate over social settings, assuming them to be inert as a function of time. This is, of course, never the case in reality, and in recent centuries, the development with time of societies has been very rapid, entailing also rapid changes in social perceptions of a given impact. For example, the importance presently accorded to environmental damage was absent just a few decades ago, and over the next decades there are bound to be issues that society will be concerned about, but which currently are considered marginal by broad sections of society.

Aggregation over social settings also has a precise meaning for a given instance. For example, the impacts of a nuclear accident will greatly depend on the response of the society. Will there be heroic firemen, as in Chernobyl, who will sacrifice their own lives in order to diminish the consequences of the accident? Has the population been properly informed about what to do in case of an accident (going indoors, closing and opening windows at appropriate times, etc.)? Have there been drills of evacuation procedures? For the Chernobyl accident in 1986, the answer was no; in Sweden today, it would be yes. A study making assumptions on accident mitigation effects must be in accordance with the make-up of the society for which the analysis is being performed. Again, the uncertainty in estimating changes in social context over time is obvious.

Aggregation over sites implies that peculiarities in topography (leading perhaps to irregular dispersal of airborne pollutants) are not considered, and that variations in population density around the energy installation studied will be disregarded. This may be a sensible approach in a planning phase, where the actual location of the installation may not have been selected. It also gives more weight to the technologies themselves, making this approach suitable for generic planning choices between classes of technology (e.g., nuclear, fossil, renewable). Of course, once actual installations are to be built, new site-specific analyses will have to be done in order to determine the best location.

In most cases, aggregation over technologies would not make sense. However, in a particular case, where, for example, the existing stock of power plants in a region is to be assessed, something like technology aggregation may play a role. For example, one might use average technology for the impact analysis, rather than performing multiple calculations for specific installations involving both the most advanced and the most outdated technology. Thus, in order to assess the total impact of an existing energy system, one might aggregate over coal-fired power stations built at different times, with differences in efficiency and cleaning technologies being averaged. On the other hand, if the purpose is to make cost-benefit analyses of various sulfur- and nitrogen-cleaning technologies, each plant has to be treated separately.

In a strict sense, aggregation is clearly not allowed in any case, because the impacts that play a role never depend linearly or in simple ways on assumptions of technology, topography, population distribution, and so on. One should, in principle, treat all installations individually and make the desired averages on the basis of the actual data. This may sound obvious, but in most cases it is also unachievable, because available data are always incomplete and so is the characterization of social settings over the time periods needed for a complete assessment. As regards the preferences and concerns of future societies, or the impacts of current releases in the future (such as climatic impacts), one will always have to do some indirect analysis, involving aggregation and assumptions on future societies (using, for example, the scenario method described in Chapter 6).

One may conclude that some aggregation is always required, but that the level of aggregation must depend on the purpose of the assessment. In line with the general characterization given above, one may discern the following specific purposes for conducting an LCA:

For licensing of a particular installation along a fuel chain or for a renewable energy system, clearly a site- and technology-specific analysis has to be performed, making use of actual data for physical pathways and populations at risk (as well as corresponding data for impacts on ecosystems, etc.). For the assessment of a particular energy system, the elements of full chain from mining or extraction through refining, treatment plants, and transportation to power plants, transmission, and final use must be considered separately, as they would typically involve different locations. A complication in this respect is that, for a fuel-based system, it is highly probable that over the lifetime of the installation fuel would be purchased from different vendors, and the fuel would often come from many geographical areas with widely different extraction methods and impacts (e.g., Middle East versus North Sea oil or gas, German or Bolivian coal mines, open-pit coal extraction in Australia, and so on). Future prices and environmental regulations will determine the change in fuel mix over the lifetime of the installation, and any specific assumptions may turn out to be invalid.

For the planning type of assessment, in most industrialized nations it would be normal to consider only state-of-the-art technology, although even in some advanced countries there is a reluctance to applying known and available environmental cleaning options (currently for particle, SO₂, and NO_x emissions, in the future probably also for CO₂ sequestering or other removal of greenhouse gases). In developing countries, there is even more of a tendency to ignore available but costly environmental mitigation options. In some cases, the level of sophistication selected for a given technology may depend on the intended site (e.g., close to or far away from population centers). Another issue is maintenance policies. The lifetime of a given installation depends sensitively on the willingness to spend money on maintenance, and the level of spending opted for is a matter to be considered in the planning decisions. The following list enumerates some of the issues involved (Sørensen, 1993a):

Impact assessments suitable for addressing these issues involve the construction of scenarios for future society in order to have a reference frame for discussing social impacts. Because the scenario method has normative components, in most cases it would be best to consider more than one scenario, spanning important positions in the social debate occurring in the society in question.

Another issue is the emergence of new technologies that may play a role during the planning period considered. Most scenarios of future societies involve an assumption that new technologies will come into use, based on current research and development. However, actual development is likely to involve new technologies that were not anticipated at the time the assessment was made. It is possible, to some extent, to analyze scenarios for sensitivity to such new technologies, as well as for sensitivity to possible errors in other scenario assumptions. This makes it possible to distinguish between future scenarios that are resilient, i.e., do not become totally invalidated by changes in assumptions, and those that depend strongly on the assumptions made.

In the case of energy technologies, it is equally important to consider the uncertainty of demand assumptions and assumptions about supply technologies. The demand may vary according to social preferences, as well as due to the emergence of new end-use technologies that may provide the same or better services with less energy input. It is therefore essential that the entire energy chain be looked at, not just to the energy delivered but also to the non-energy service derived. No one demands energy, but we demand transportation, air conditioning, computing, entertainment, and so on.

The discussion of aggregation issues clearly points to the dilemma of impact analyses: those answers that would be most useful in the political context often are answers that can be given only with large uncertainty. This places the final responsibility in the hands of the political decision-maker, who has to weigh the impacts associated with different solutions and in that process to take the uncertainties into account (e.g., choosing a more expensive solution because it has less uncertainty). But this is, of course, what decision-making is about.

7.3.1.5 Monetizing issues

The desire to use common units for as many impacts in an LCA as possible is, of course, aimed at facilitating the job of a decision-maker wanting to make a comparison between different solutions. However, it is important that this procedure does not further marginalize the impacts that cannot be quantified or that seem to resist monetizing efforts. The basic question is really whether the further uncertainty introduced by monetizing offsets the benefit of being able to use common units.

Monetizing may be accomplished by expressing damage in monetary terms or by substituting the cost of reducing the emissions to some threshold value (avoidance cost). Damage costs may be obtained from health impacts by counting hospitalizations and workday salaries lost, replanting cost of dead forests, restoration of historic buildings damaged by acid rain, and so on. Accidental human death may, for example, be replaced by the life insurance cost.

Unavailability of data on monetization has led to the alternative approach of interviewing cross-sections of the affected population on the amount of money they would be willing to pay to avoid a specific impact or to monitor their actual investments (contingency evaluations, such as hedonic pricing, revealed preferences, or willingness to pay). Such measures may change from day to day, depending on exposure to random bits of information (whether true or false), and also depend strongly on the income at the respondents’ disposal, as well as competing expenses of perhaps more tangible nature. Should the monetized value of losing a human life (the “statistical value of life,” SVL, discussed below) be reduced by the fraction of people actually taking out life insurance? Should it be allowed to take different values in societies of different affluence?

All of the monetizing methods mentioned are clearly deficient: the damage caused by not including a (political) weighting of different issues (e.g., weighting immediate impacts against impacts occurring in the future), and the contingency evaluation by doing so on a wrong basis (being influenced by people’s knowledge of the issues, by their available assets, etc.). The best alternative may be to avoid monetizing entirely by using a multivariate analysis, e.g., by presenting an entire impact profile to decision-makers, in the original units and with a time sequence indicating when each impact is believed to occur, and then inviting a true political debate on the proper weighting of the different issues. However, the use of monetary values to discuss alternative policies is so common in current societies that it may seem a pity not to use this established framework wherever possible. It is also a fact that many impacts can meaningfully be described in monetary terms, so the challenge is to make sure that the remaining ones are treated adequately and do not “drop out” of the decision process.

The translation of impacts from physical terms (number of health effects, amount of building damage, number of people affected by noise, etc.) to monetary terms (US$/PJ, DKr/kWh, etc.) is proposed in studies like the European Commission project (1995a,b) to be carried out by a study of the affected population’s willingness to pay (WTP) for avoiding the impacts. This means that the study does not aim at estimating the cost to society, but rather the sum of costs inflicted on individual citizens. The concept of WTP was introduced by Starr (1969). The WTP concept has a number of inherent problems, some of which are:

The outcome of an actual development governed by the WTP principle may be inconsistent with agreed social goals of equity and fairness, because it may lead to polluting installations being built in the poorest areas.

The accidental deaths associated with energy provision turn out in most studies of individual energy supply chains, such as the European Commission study, to be the most significant impact, fairly independently of details in the monetizing procedure selected. We therefore deal with choice of the monetized value of an additional death caused by the energy system in a little more detail below. At this point, it should only be said that our preference is to work with a monetized damage reflecting the full LCA cost of energy to society, rather than the cost to selected individual citizens.

7.3.1.6 Chain calculations

This section outlines the use of LCA to perform what is termed a chain calculation above. The procedure consists in following the chain of conversion steps leading to a given energy output, as illustrated in Fig. 7.4, but considering input from, and outputs to, side-chains, as exemplified in Fig. 7.5. Chain LCAs are the equivalent of product LCAs and usually involve specific assumptions about the technology used in each step along the chain. For example, the immediate LCA concerns may be emissions and waste from particular devices in the chain. However, before these can be translated into actual damage figures, one has to follow their trajectories through the environment and their uptake by human beings, as well as the processes in the human body possibly leading to health damage. The method generally applied to this problem is called the pathway method.

The pathway method consists of calculating, for each step in the life cycle, the emissions and other impacts directly released or caused by that life-cycle step, then tracing the fate of the direct impact through the natural and human ecosystems, e.g., by applying a dispersion model to the emissions in order to determine the concentration of pollutants in space and time. The final step is to determine the impacts on humans, on society, or on the ecosystem, using for instance dose–response relationships between intake of harmful substances and health effects that have been established separately. The structure of a pathway is indicated in Fig. 7.8.

Consider as an example electricity produced by a fossil-fuel power station using coal (Fig. 7.9). The first step would be the mining of coal, which may emit dust and cause health problems for miners, then follows cleaning, drying, and transportation of the coal, spending energy like oil for transportation by ship (here the impacts from using oil have to be incorporated). The next step is storage and combustion of the coal in the boiler of the power station, leading to partial releases of particulate matter, sulfur dioxide, and nitrogen oxides through the stack. These emissions would then have to be traced by a dispersion model, which would be used to calculate air concentrations at different distances and in different directions away from the stack. Based upon these concentrations, inhalation amounts and the health effects of these substances in the human body are obtained by using the relation between dose (exposure) and effect, taken from some other study or World Health Organization databases. Other pathways should also be considered, for instance, pollutants washed out and deposited by rain and subsequently taken up by plants, such as vegetables and cereals that may later find their way to humans and cause health problems.

For each life-cycle step, the indirect impacts associated with the chain of equipment used to produce any necessary installation, the equipment used to produce the factories producing the primary equipment, and so on, have to be assessed, together with the stream of impacts occurring during operation of the equipment both for the life-cycle step itself and its predecessors (cf. Fig. 7.5). The same is true for the technology used to handle the equipment employed in the life-cycle step, after it has been decommissioned, in another chain of discarding, recycling, and re-use of the materials involved.

In the coal power example, the costs of respiratory diseases associated with particulates inhaled may be established from hospitalization and lost workday statistics, and the cancers induced by sulfur and nitrogen oxides may be similarly monetized, using insurance payments as a proxy for deaths caused by these agents.

Generally, the initiating step in calculating chain impacts may be in the form of emissions (e.g., of chemical or radioactive substances) from the installation to the atmosphere, releases of similar substances to other environmental reservoirs, visual impacts, or noise. Other impacts would be from inputs to the fuel-cycle step (water, energy, and materials like chalk for scrubbers). Basic emission data are routinely collected for many power plants, whereas the data for other conversion steps are often more difficult to obtain. Of course, emission data, e.g., from road vehicles, may be available in some form, but are rarely distributed over driving modes and location (at release), as one would need in most assessment work.

Based on the releases, the next step, calculating the dispersal in the ecosphere, may exploit available atmospheric or aquatic dispersion models. In the case of radioactivity, decay and transformation have to be considered. For airborne pollutants, the concentration in the atmosphere is used to calculate deposition (using dry deposition models, deposition by precipitation scavenging or after adsorption or absorption of the pollutants by water droplets). As a result, the distribution of pollutants (possibly transformed from their original form, e.g., from sulfur dioxide to sulfate aerosols) in air and on ground or in water bodies will result, normally given as a function of time, because further physical processes may move the pollutants down through the soil (eventually reaching ground water or aquifers) or again into the atmosphere (e.g., as dust).

Given the concentration of dispersed pollutants as function of place and time, the next step along the pathway is to establish the impact on human society, such as by human ingestion of the pollutant. Quite extended areas may have to be considered, both for normal releases from fossil-fuel power plants and for nuclear plant accidents (typically, the area considered includes a distance from the energy installation of a thousand kilometers or more, according to the European Commission study, 1995c). Along with the negative impacts there is, of course, the positive impact derived from the energy delivered. In the end, these are the impacts that will have to be weighed against each other. Finally, one may attempt to assist the comparison by translating the dose– responses (primarily given as number of cancers, deaths, workdays lost, and so on) into monetary values. This translation of many units into one should only be done if the additional uncertainty introduced by monetizing is not so large that the comparison is weakened (see Fig. 7.10). In any case, some impacts are likely to remain that cannot meaningfully be expressed in monetary terms. The impacts pertaining to a given step in the chain of energy conversions or transport may be divided into those characterizing normal operation and those arising in case of accidents. In reality, the borderlines between often occurring problems during routine operation, mishaps of varying degree of seriousness, and accidents of different size are fairly blurry and may be described in terms of declining frequency for various magnitudes of problems.

To a considerable extent, the pathways of impact development are similar for routine and accidental situations involving injuries and other local effects, such as those connected with ingestion or inhalation of pollutants and with the release and dispersal of substances causing a nuisance where they reach inhabited areas, croplands, or recreational areas. The analysis of these transfers involves identifying all important pathways from the responsible component of the energy system to the final recipient of the impact, such as a person developing a mortal illness, possibly with considerable delays, such as late cancers.

7.3.1.7 Matrix calculations

A full systemic LCA calculation proceeds by the same steps as the chain calculations, but without summing over the indirect impact contributions from side-chains. All devices have to be treated independently, and only at the end may one sum impacts over the totality of devices in the system in order to obtain systemwide results. In practice, the difference between chain and system LCAs in work effort is not great, because in chain LCA calculations each device is usually treated separately before the chain totals are evaluated. There are exceptions, however, where previous results for energy devices in chains that already include indirect impacts are used. In such cases, the chain results cannot immediately be used as part of a systemwide evaluation.

Particular issues arise when LCA evaluations are made for systems considered for future implementation: there may be substitutions between human labor and machinery, linking the analysis to models of employment and leisure activities. In order to find all the impacts, vital parts of the economic transactions of society have to be studied. The total energy system comprises conversion equipment and transmission lines or transportation, as well as end-use devices converting energy to the desired services or products. Demand modeling involves consideration of the development of society beyond the energy sector.

More factual is the relation between inputs and outputs of a given device, which may be highly nonlinear but in most cases are given by a deterministic relationship. Exceptions are, for example, combined heat-and-power plants, where the same fuel input may be producing a range of different proportions between heat and electricity. This gives rise to an optimization problem for the operator of the plant (who will have to consider fluctuating demands along with different dispatch options involving different costs). A strategy for operation is required in this case before the system LCA can proceed. But once the actual mode of operation is identified, the determination of inputs and outputs is, of course, unique. The impact assessment then has to trace where the inputs came from and keep track of where the outputs are going in order to determine which devices need to be included in the analysis. For each device, total impacts have to be determined, and the cases where successive transfers may lead back to devices elsewhere in the system can be dealt with by setting up a matrix of all transfers between devices belonging to the energy system. In what corresponds to an economic input–output model, the items associated with positive or negative impacts must be kept track of, such that all impacts belonging to the system under consideration in the life-cycle analysis will be covered.

Once this is done, the impact assessment itself involves summation of impacts over all devices in the system, as well as integration over time and space, or just a determination of the distribution of impacts over time and space. As in the chain studies, the spatial part involves use of dispersal models or compartment transfer models (Sørensen, 1992), while the temporal part involves charting the presence of offensive agents (pollutants, global warming inducers, etc.) as a function of time for each located device as well as a determination of the impacts (health impacts, global warming trends, and so on) with their associated further time delays. This can be a substantial effort, as it has to be done for each device in the system, or at least for each category of devices (an example of strong aggregation is shown in Fig. 6.83).

The first step is similar to what is done in the chain analysis, i.e., running a dispersion model that uses emissions from point or area sources as input (for each device in the system) and thus calculating air concentration and land deposition as function of place and time. For example, the RAINS model is used to calculate SO₂ dispersal on the basis of long-term average meteorological data, aggregated with the help of a large number of trajectory calculations (Alcamo et al., 1990; Hordijk, 1991; Amann and Dhoondia, 1994).

Based on the output from the dispersion model, ingestion rates and other uptake routes are again used to calculate human intake of pollutants through breathing, skin, etc., and a model of disposition in the human body, with emphasis on accumulation in organs and rates of excretion, is used. The resulting concentration for each substance and its relevant depository organs is via a dose–response function used to calculate the morbidity and mortality arising from the human uptake of pollution. It is customary to use a linear dose– response function extending down to (0,0) in cases where measurements only give information on effects for high doses. This is done in the European Commission study, based on a precautionary point of view as well as theoretical considerations supporting the linear relation (European Commission, 1995a-f). The alternative, assuming a threshold below which there is no effect, is often used in regulatory schemes, usually as a result of industry pressure rather than scientific evidence.

Systemwide calculations are often interpreted as comprising only those components that are directly related to energy conversion. Sometimes this borderline is difficult to determine; for example, transportation vehicles cause traffic and thus link to all the problems of transportation infrastructure. A general approach would be to treat all components of the energy system proper according to the system approach, but to treat links into the larger societal institutions and transactions as in the chain LCA. In this way, the overwhelming prospect of a detailed modeling all of society is avoided, and yet the double-counting problem is minimized because energy loops do not occur (although loops of other materials may exist in the chains extending outside the energy sector).

7.3.3.1 LCA of greenhouse gas emissions

The injection of carbon dioxide and other greenhouse gases, such as methane, nitrous oxide, and chlorofluorocarbons, into the atmosphere changes the disposition of incoming solar radiation and outgoing heat radiation, leading to an enhancement of the natural greenhouse effect. Modeling of the Earth’s climate system in order to determine the long-term effect of greenhouse gas emissions has taken place over the last 40 years with models of increasing sophistication and detail. Still, there are many mechanisms in the interaction between clouds and minor constituents of the atmosphere, as well as coupling to oceanic motion, all of which only modeled in a crude form. For instance, the effect of sulfur dioxide, which is emitted from burning fossil fuel and in the atmosphere becomes transformed into small particles (aerosols) affecting the radiation balance, has been realistically modeled only over the last couple of years. Because of the direct health and acid-rain impacts of SO₂, and because SO₂ is much easier to remove than CO₂, emissions of SO₂ are being curbed in many countries. As the residence time of SO₂ in the atmosphere is about a week, in contrast to the 80–120 years for CO₂, climate models have to be performed dynamically over periods of some 50–200 years (cf. Chapter 2).

Figure 7.14 shows measured values of CO₂ concentrations in the lower atmosphere over the last 300 000 years. During ice-age cycles, systematic variations between 190 and 280 ppm took place, but the unprecedented increase that has taken place since about 1800 is primarily due to combustion of fossil fuels, with additional contributions from changing land use, including felling of tropical forests. If current emission trends continue, the atmospheric CO₂ concentration will have doubled around the mid-21st century, relative to the pre-industrial value. The excess CO₂ corresponds to slightly over half of anthropogenic emissions, which is in accordance with the models of the overall carbon cycle (Chapter 2; IPCC, 1996b). The ice-core data upon which the historical part of Fig. 7.14 is based also allow the trends for other greenhouse gases to be established. The findings for methane are similar to those for CO₂, whereas there is too much scatter for N₂O to allow strong conclusions. For the CFC gases, which are being phased out in certain sectors, less than 40 years of data are available. Both CO₂ and methane concentrations show regular seasonal variations, as well as a distinct asymmetry between the Northern and the Southern Hemispheres (Sørensen, 1992).

Despite the incompleteness of climate models, they are deemed realistic enough to permit the further calculation of impacts due to the climate changes predicted for various scenarios of future greenhouse gas emissions. This is the case despite the changes in the models underlying IPCC’s assessment work from the first report in 1992 to the most recent in 2013.

It should be mentioned that while the IPCC assessment is considered credible regarding the influence of greenhouse gases on the physical earth–ocean–atmosphere system for particular emission scenarios, the statements published by IPCC on life-cycle impacts and mitigation options have been increasingly politically biased since the 1997 change of the working group compositions from independent scientists to people selected by national governments.

The role of anthropogenic interference with the biosphere was a focus already of the 2^nd IPCC (1996b, Chapter 9) assessment, in addition to the temperature changes discussed in Chapter 2. Changed vegetation cover and agricultural practices influence carbon balance and amounts of carbon stored in soils, and wetlands may emit more methane, or they may dry out owing to the temperature increases and reduce methane releases, while albedo changes may influence the radiation balance. Figure 7.15 shows how carbon storage is increased owing to the combined effects of higher levels of atmospheric carbon (leading to enhanced plant growth) and changed vegetation zones, as predicted by the climate models (affecting moisture, precipitation, and temperature). Comprehensive modeling of the combined effect of all the identified effects has not yet been performed.

The anthropogenic greenhouse forcing, i.e., the net excess flow of energy into the atmosphere taking into account anthropogenic effects since the late 18th century, in 1995 was estimated at 1.3 W m⁻², as a balance between a twice as big greenhouse gas contribution and a negative contribution from sulfate aerosols (IPCC, 1996b, p. 320). The same value (1.3 W m⁻²) is quoted in the 2013 report for year 2000 (Prather et al., 2014). An example of estimated further change in the radiative forcing to year 2100 is shown in Fig. 7.16a, which also gives the non-anthropogenic contributions from presently observed variations in solar radiation (somewhat uncertain) and from volcanic activity (very irregular variations with time), for the scenario of future human behavior called A1B in the 4^th IPCC assessment (about 6 W m⁻² forcing; IPCC, 2007a). The additional scenarios of the 5^th IPCC assessment have net forcings ranging from 2.6 to 8.5 W m⁻² (Moss et al., 2010; Prather et al., 2014). Both the old and the new scenarios are of questionable quality. For example, none of the scenarios have more than a few percent of energy supply from solar or wind sources (Vuuren et al., 2011). In countries such as Denmark these sources already by 2016 provides close to 50% of the electricity demands. Also the treatment of the results of computer runs of the large-scale motion circulation and climate models (see section 2.5.2) have in the recent IPCC reports been presented as averages and spreads over many models. This pretends that all the models considered are of equal quality, despite the fact that they differ in resolution and in the effects included. Some effects of relevance to climate are included in one model while others are only in another model. There is no clear “best” model, and the dispersal between model results for year 2100 is quite large (IPCC, 2013).

The A1B scenario used in some of the illustrations in this book predicts a 1.6°C global average warming relative to 1990 by 2050 and a 3.0°C warming by 2100. Relative to pre-industrial times, the warming is about 0.75°C higher. The 42 models considered in the 5^th IPCC assessment give 2100 over 1995 average temperature increases between 0.6 and 8°C, plus several degrees spread between models, as shown in Fig. 7.16b that for some of the models is further extended to year 2300 (IPCC, 2013).

7.3.3.2 Direct health impacts of temperature changes

The impact of extreme temperatures on human survival is rarely a simple relation between high or low temperatures and specific diseases. Ambient temperature may furnish a small push on top of other causes, altering the outcome of a health condition. In this way, ambient temperatures may be a contributing cause for fatal outcome of a variety of diseases, particular in the areas of respiratory and cardiovascular disease and heat failures, myocardial infarction and adverse cerebrovascular conditions, having special affinity to individuals with particular dispositions and health histories. The risk associated with severe frost and its dependence on clothes and shelter is perhaps more well known, and so is the one associated with specific heat-wave episodes that are supposed to become more frequent and/or prolonged in a warmer and less stable climate induced by greenhouse gas emissions (IPCC, 2007b). Basu (2009) notes that the very young (below 4 years) and the old (with a gradual rise accelerating above age 75) are most vulnerable to heat waves.

Temperature changes also have indirect effects through altered requirements for energy management in shelters such as dwellings or workspaces in buildings (costs estimates in Sørensen, 2011). Additional allergy cases would be associated with increased levels of pollen and air pollution due to warming and would occur predominantly at lower latitudes, whereas asthma incidence is highest in humid climates and is expected to be enhanced at higher latitudes (IPCC, 1996a, Chapter 18).

With respect to exceptional upward temperature excursions, the US NRC (2010) presents a US heat-wave duration index defined as the average length (in days) of events. At 2°C average warming, the length index increases by about 8 days in Southern United States and by about 4 days in the Northern States. WHO (2004a) have surveyed the impacts of heat waves but find attribution of death of illness difficult. Epidemiological studies during recent heat-wave incidences are reviewed in Basu (2009). A general relation between maximum daily temperature and mortality has been noted both in Europe and the United States.

Figure 7.17 shows the relation between mortality and daily maximum temperature used as a basis for the calculation discussed below. It is constructed from specific data from a number of European sites (WHO, 2004a), including detailed data for Madrid (Diaz and Santiago, 2003). The inhabitants of Madrid suffer health problems at cold temperatures that are not as low as those causing trouble in most other parts of Europe, while their problems at high temperatures set in at a temperatures higher than in most other European locations. Regional studies have identified several other details of heat-related health impacts. For example, Checkley et al. (2000) finds a doubling of diarrhea cases in Lima (Peru) during an exceptional heat-episode in 1997–98, Ishigami et al. (2008) find 7%–20% elevated mortality per degree warming in a study comprising Budapest, London, and Milano, and Almeida et al. (2010) find about 2% increase in mortality per degree in Lisbon and Oporto. Clearly, the impacts of either cold or hot weather depend on how much time people spent outside at ambient temperatures, and on the quality of the shelter (such as homes, workplaces, and commercial buildings) where they spent the rest of their time. The fact that health problems in cold periods seem higher in Spain than in Norway may not necessarily reflect an adaptation to a certain climate, but could derive from Spanish buildings being less well insulated and otherwise prepared for severe cold spells than those in Norway.

The relative mortality in Fig. 7.17, called the mortality factor, is normalized (choice of unit “1”) in such a way that the Madrid mortality agrees with the observed one. The curves for different locations all have a similar shape, with increased mortality at low and high ambient temperatures, but they are displaced horizontally from each other, by up to about 10°C between Norway and Spain. The one used is an envelope, broadening the bottom part of the curves and using the low-temperature rise of the coldest region and the high-temperature rise of the hottest region. This makes sure that the health impact is not overestimated. The cause of the differences between locations is likely some kind of adaptation, obtained by populations that have lived in a given region for periods exceeding 1000 years. Newcomers usually have different mortality patterns, for many kinds of reasons. It is possible that the correlation between mortality and maximum temperatures averaged over 1 or 2 weeks before the death would be even stronger than the correlation between mortality and just one day’s maximum temperature. For cold spells, one might instead have used minimum temperatures, but these usually differ little from maximum temperatures during winter periods.

The global distribution of January and July daily maximum temperatures for the mid-21^st century are shown in Fig. 7.18, based on a high-quality circulation and climate model, the MIHR model of Hasumi and Emori (2004) and emissions from a scenario called “A1B,” one of the 2007 IPCC reference emission scenarios (Meehl et al., 2007). The maximum temperatures are not necessarily representing heat waves, although they occasionally do. Their magnitude and geographical distribution are similar to those found for average temperatures.

The annually averaged global distribution of the change in mortality factor for the mid-21^st century, obtained from the daily maximum temperatures exemplified in Fig. 7.18 by use of the relation in Fig. 7.17, is shown in Fig. 7.19.

The changes in mortality factors due to global warming vary quite strongly through the seasons, and Fig. 7.20 shows the monthly results for the difference between the mid-21^st century and the pre-industrial mortality factors. As seen, the effects of greenhouse warming are rarely simply beneficial or detrimental, but have seasonal variations representing increases or decreases in mortality. Still, the overall pattern is that of a division of regions of the world into some with predominantly positive effects and other ones with predominantly negative effects. There is excess mortality near the Equator during some parts of the year and reduced mortality at many latitude regions with subtropical and temperate climate, plus some increase in the mortality factor at very high latitudes. However the mortality factors have to be multiplied by the number of people exposed at a given location, and population densities change the picture because they are high at the low and middle latitudes, but low at the very highest latitudes.

The model thus uses UN projected year-2050 population data available for each grid cell, with a modification proposed by Sørensen and Meibom (2000) for city growth, but otherwise based on and maintaining averages from the middle of three United Nations projections for 2050 (CIESIN, 1997; United Nations, 1997, 2010). The total 2050 population is 9.3×10⁹, which is higher than the 8.7×10⁹ used in the IPCC A1B emission scenario, but perhaps more realistic, at least in case there is no major wars or other setbacks. The IPCC scenario is not made on a geographical area-basis and thus cannot be used directly for the type of study undertaken here. The Sørensen–Meibom model used takes actual population data and projects the future development on the basis of demographic and economic development, assuming a gradual shift in lifestyles, which will halt the previous trend toward larger and larger population densities in city centers. Instead, city activities are assumed spreading to neighboring grid cells once the population density exceeds a value of 5000 km⁻². Such a spread is in agreement with the average actual developments in many parts of the world, but not with the prestigious increase in high-rise buildings in certain newly rich cities. Because the temperatures in city centers can be 1°C–2°C higher than that outside the cities, the assumptions made in this respect will have an impact of the mortality estimates, probably overlooked by the current coarse-grid climate models.

The model now goes on to multiply the 2050 population by the excess death rate obtained by multiplying current death rates (taken as national figures as given by WRI, 2008) by the 2050 change in mortality factor (for which averages were given in Figs. 7.19 and 7.20). The area of the grid cells entering the geographical-information-system (GIS) calculation to get from the input population density to grid-cell population is for the MIHR model equal to the cosine-corrected latitude increment of 1.121283° times a longitude increment of 1.125°, or 15610.6×cos(j) km², with j being the latitude angle. When the mortality change in this way is folded with the assumed 2050 population densities, one obtains the predicted number of both additional deaths and avoided deaths shown in Fig. 7.21. The effect of greenhouse warming on the mid-21^st century temperature environments is the only one included in this modeling, departing from the current global mortality pattern and adding only this type of modification. There are many other developments that can change and in several cases is known to historically have changed the mortality (medical progress, healthier living or the opposite, etc.), and no attempt has been made to estimate the nature of these other changes from now to year 2050.

The striking conclusion from Fig. 7.21 is that that the numbers of extra lives taken or saved by the greenhouse warming are both very large. Although they are hidden behind medical causes to which deaths are often attributed, due to being only a contributing factor of anything from a few undisclosed percent to the comparatively rare identification as a main cause of death, the statistical approach taken is solid (due to the U-shaped effect shown in Fig. 7.17 being far beyond uncertainty) and the results therefore basically indisputable. The death certificates earlier in use are now being discontinued in many countries (except if criminal causes are suspected), because of their arbitrariness, where the certificate is sometimes signed by a doctor who does not know the deceased and writes say “pneumonia” instead of a more relevant underlying disease that would have been known to a doctor with longer acquaintance with the deceased. In any case multiple causes contributing various percentages to a death are nearly never mentioned in death certificates. Using as here overall mortality data rather than specific causes of death significantly increases the statistical reliability of the study, and at the same time removes the negative influence on the results, that incorrect cause-of-death assignments on death-certificates could otherwise have.

Nearly 4 million people will be affected each year by the climate change brought about by previous greenhouse gas emissions and changes in area use according to the A1B scenario. Of these, 2.2×10⁶ y⁻¹, living in the northern part of the globe plus New Zealand, the Andes and Southern Chile, are people surviving that would have died without the greenhouse warming, and 1.6×10⁶ y⁻¹, living in the Equatorial and southern part of the globe are people that would otherwise have survived but now die. The slightly larger number of deaths avoided by the greenhouse warming according to Fig. 7.21, compared to the number of deaths caused by it, does not allow the interpretation that the two “average out,” as they occur in different regions of the world. Furthermore, from a moral point of view, it is worrisome that the additional deaths occur in countries contributing only little to the greenhouse gas emissions, while the saved deaths occur precisely in the countries most responsible for the greenhouse emissions.

Improvements to the simple model used here would use different mortality factors for every location or at least every region. However, such calculations are not presently possible, because temperature-mortality correlation data have been established only in selected areas (such as the ones mentioned above) and are not globally available. Furthermore, as mentioned, mortality may be influenced by many other factors, which adds uncertainty to use of data collected at even slightly different times. For instance, Donaldson et al. (2003) find that despite an increase in average temperature between 1971 and 1997, the overall mortality decreased in selected countries over the same period.

Another issue to consider in future modeling of the temperature effect is the timing of temperature events influencing mortality. For example, one could look for the correlation between mortality and temperatures a month or a week before each death and compare it with the correlation found between mortality and the temperature at the time of death.

Regarding the level of adaptation to temperature regimes at different locations, it would seem more likely that human populations may adapt to changes in mean temperature, as opposed to adaptation to extreme events.

Bosello et al. (2006) have previously looked at health impacts of warming based on data extrapolated from some specific investigations and they find for 2050 a decrease of deaths from cardiovascular diseases in all regions of the world, totaling 1.76×10⁶; an increase of respiratory diseases causing death in all regions, totaling 0.36×10⁶; and similarly an increase of deaths from diarrhea, totaling 0.49×10⁶. The order of magnitude thus agrees with the present study. However, details appear quite different. The geographical distribution cannot be compared, because Bosello et al. (2006) use aggregated regions with, e.g., India and China lumped together, and they have negative overall impacts only for the Middle East and the “rest of the world,” comprising what is left behind after detailing the “Annex 1” countries (UN jargon for countries classified as industrialized or economies in transition). Other early studies of greenhouse warming impacts by economists have as mentioned concentrated on impacts occurring in the United States (Cline, 1992; Frankhauser, 1990; Nordhaus 1994; Tol, 1995).

7.3.3.3 Impacts on agriculture, silviculture, and ecosystems

One major issue is the impact of climate change on agricultural production. Earlier evaluations (e.g., Hohmeyer, 1988) found food shortage to be the greenhouse impact of highest importance. However, the 1995 IPCC assessment suggested that, in developed countries, farmers will be able to adapt crop choices to the slowly changing climate, and that the impacts will be entirely in the Third World, where farmers will lack the skills needed to adapt. The estimated production loss amounts to 10%–30% of the total global agricultural production (Reilly et al., Chapter 13 in IPCC, 1996a), which would increase the number of people exposed to risk of hunger from the present 640 million to somewhere between 700 and 1000 million (Parry and Rosenzweig, 1993). However, there are also unexploited possibilities for increasing crop yields in developing countries, so the outcome will depend on many factors, including speed of technology transfer and development. The 2^nd IPCC assessment expected some 50 million deaths associated with migration (people running away from bad conditions) due to adverse climate events. The following IPCC assessments have added more aspects to the issue and further weakened the early view that warming would imply increased agricultural yields. Hot weather can damage certain crops, and so will dry spells and wildfires, insect attacks will increase, and water supply may become reduced, e.g., for irrigation (Ciais et al., 2005). Desalination of seawater is already in progress in many regions, using either solar energy or still in most cases fossil energy, causing more global warming along with the effort to mitigate its effects.

Fauna adapted to live in areas with a certain climate may try to migrate in order to find new habitats with the climatic conditions they prefer. This may involve displacement by over 1000 km in some cases and may not be possible due to human uses of the intermediate areas, with loss of diversity as a possible outcome (Wright et al., 2009; US NRC, 2010).

In addition to causing warming, ozone may have a direct effect on plant growth (Long et al., 2005; Challinor and Wheeler, 2008). Spruce forests in Sweden seem to be negatively influenced by the shortening of the frost period caused by greenhouse warming (Rammig et al., 2010), and Baltic Sea aquatic plant growth is weakened by changes in oxygen and salinity content in the water (Neumann, 2010). Oxygen is depleted by increased algal growth near shores, and associated cyanobacterial toxins can cause skin rashes and adverse gastrointestinal, respiratory, and allergic reactions (Kite-Powell et al., 2008; Stewart et al., 2006). The many impacts that greenhouse warming can have on agriculture, silviculture, and aquaculture need to be followed closely, but the current estimates of a yield reduction of some 15% already has serious implications for food supply and food prices.

7.3.3.4 Vector-borne diseases

A number of diseases caused by viruses or other microbes are transmitted by insects, called “vectors,” with typical ones being mosquitoes. The plasmodium falciparum parasite and anopheline mosquito varieties involved in the malaria transmission cycle are all sensitive to variations in temperature and humidity, and regions with strong seasonality or dryness in periods of the year crucial for mosquito breeding may experience considerable diminishment of malaria incidence. Other factors are of course lifestyle and technical means such as mosquito nets over beds or wire-netting on window and doors. Models have been constructed for describing the distribution and intensity of malaria transmission under assumptions of various greenhouse warming scenarios (Tanser et al., 2003; Lieshout et al., 2004; Rogers et al., 2002; Bouma, 2003). Current malaria deaths and disabilities (measured in terms of DALY’s, defined as life-shortening in years of not being able to live a normal, meaningful life) are shown in Fig. 7.22. Many attempts to prevent or to heal malaria have been disappointing or have led to drug resistance (notably reducing the usefulness of quinine, chlorochinine, and mefloquine), despite increased understanding of the genetic mechanisms underlying the disease (Olzewski et al., 2010). The earlier IPCC assessments listed spread of malaria to more regions as one of the largest impacts of greenhouse warming, but the actual development during the early 21^st century does not support this interpretation, maybe because of increased use of protection and improved living conditions. One should remember that malaria was widespread in Europe a few centuries ago, but largely disappeared along with general socioeconomic development.

Several other vector-borne or parasitic diseases have been monitored by the studies summarized by the IPCC assessments, including helminthiasis, dengue fever, schistosomiasis (bilharziasis), tryponospmiasis, Chagas disease (sleeping sickness), leichmaniasis (black fever), lymphatic filariasis (elephantiasis), and onchocerciatis (river blindness). All of these are considered to be influenced by greenhouse warming (Weaver et al., 2010; Hales et al., 2002; WHO, 2004b; Mathers and Loncar, 2006). Overall, considering the expected economic development in the regions currently affected, the World Health Organization expects these diseases to diminish in importance over the next decades (WHO, 2008, 2010).

7.3.3.5 Extreme events

Already at present, an increased frequency and severity of extreme events with possible connection to greenhouse warming (floods, landslides, storms, droughts, and fires) has been noted and has influenced insurance premiums. Some climate models claim to predict such increases (see discussion in Sillmann et al., 2013), but basically, the mechanism for many of these events would seem to involve interaction between atmospheric motion on small and large scales, which is not covered by the models used that has to omit local eddy circulation and chaotic air motion, as explained in section 2.3.1. The development in number of floods in two severity classes globally is shown in Fig. 7.23, death and disability caused by fires in Fig. 7.24, and an overview of some important natural and manmade extreme events is given in Fig. 7.25. There is no easy way to estimate the precise future frequency of any of these events, but global warming seems generally to lower the stability of climates.

7.3.3.6 Valuation of greenhouse warming impacts

As discussed earlier in this chapter, monetizing the physical impacts identified can be quite ambiguous due to the limitations of the concept of a “statistical value of life.” The study presented above predicts that additional deaths caused by greenhouse warming will predominantly occur in regions of low income, which in some studies has lead to setting the value of a life lost at a lower level than for the regions where lives according to the present study are abundantly being saved by the greenhouse warming. Clearly a number of ethical issues are involved in estimating the impacts of human activities carried out by precisely the group of people that stands to benefit from the green-house gas pollution, while negative effects are accumulated in populations with less affluence and a lesser possibility to control or mitigate the impacts.

Impacts on human societies and the natural environment involve health issues (deaths and DALYs), economic issues, degrading the environment, and affecting biological diversity. The key problem of valuing human lives lost is here handled by introduction of a “statistical value of life,” and Table 7.2 summarizes the studies made initiated by the 1995 European Commission study “ExternE” and several follow-up studies (US EPA, 2015; Viscusi and Aldi, 2003). The columns termed “European Standards” are the ExternE valuation expressed in € or US $ without correcting the amount for later inflation, and the two last columns give a purchasing power parity (PPP) and a GDP-adjusted conversion. The order of magnitude difference between the three methods of evaluating explains many of the disparities found in the literature and leaves a choice between evaluating lost lives everywhere by the standards of the rich countries or by the PPP measure. The GDP-adjusted valuation of statistical lives is clearly unacceptable.

Based on these rules of evaluation, the greenhouse warming effects identified above have been monetized. The results are presented in Table 7.3 and Fig. 7.26 (Sørensen, 2011), where an estimate for the entire 21^st century impact is obtained by multiplying the results for the mid-century by 100. It is seen that the large impact contributors are the decrease in food production (agriculture and fisheries) and the direct temperature impact on health, positive or negative. The temperature impact was deemed minor in previous IPCC estimates, and the malaria and other vector-transmitted diseases that dominated the previous IPCC estimates (see Sørensen, 2010) are in Table 7.3 not quantified, because the current stance is that they may not play any significant role a few decades ahead. Despite the large direct temperature impacts of both signs, the sum of all impacts included is negative and large, although it only equals about a third of the impacts found based on the previous IPCC assessment (Sørensen, 2010). As also discussed above, the presence of two large, partly canceling contributions does not diminish the human suffering from the negative part, because they occur in different regions of the world.