1. Pollution policy with imperfect information
chapter 7
Pollution policy with imperfect information

2. Concepts
Risk and uncertainty: often used to characterise various situations in which less than complete information is available.
Risk: usually taken to mean situations in which some chance process is taking place in which the set of possible outcomes is known and probabilities can be attached to each possible outcome. However, it is not known which possible outcome will occur.
Uncertainty: usually taken to mean situations in which the set of possible outcomes is known but probabilities cannot be attached to each possible outcome.
Radical uncertainty: circumstances in which it would not be possible even to enumerate all the possible outcomes.

3. Difficulties in identifying pollution targets in the context of limited information and uncertainty
Much of the discussion of efficiency-based pollution targets in Chapter 5 implicitly assumed that the policy maker was well informed, and so either knew – or by an investment of resources could discover – the relevant cost and benefit functions.
But this assumption is often not plausible
The environmental agency will often not know with certainty the costs and/or benefits of pollution (or equivalently the costs and benefits of pollution abatement).
Where non-convexities are present, it is not sufficient to know the values of such things near the current position of the economy; they have to be known across the whole range of possibilities.
For stock pollutants, stock effects and spatial considerations imply that the appropriate functions vary from place to place and from time to time.

4. Limited information and uncertainty arises from:
The costs involved in acquiring, collating, validating and processing information, implying that census-data are likely to be prohibitively expensive.
Sampling error, associated with the use of sampling methods and statistical inference from the sample data.
The data collected may not properly represent what the investigator is seeking to obtain.
Abatement costs: those who possess relevant information may have incentives not to truthfully reveal it.  
Difficulties in indentifying and evaluating the benefits of pollution abatement (i.e. the benefits of avoided damages).
Scientific knowledge about pollution impacts is far from complete, and arguably can never be complete because of the stochastic and complex nature of ecosystem functioning.
Valuation of environmental services is beset by a host of theoretical and practical problems, and there is little consensus about the validity of current valuation techniques.

5. Some methodological issues
Relevant costs and prices (on both benefit and cost estimation sides) needed for evaluation should be those that correspond to a socially efficient outcome; these may bear little relation to observed costs and prices where the economy is a long way from that optimum.
Difficulties are compounded by ‘second-best’ considerations.
Limited information and uncertainty do not simply mean that decisions should be taken in the same way (but have less ‘accuracy’) as under conditions of full information.  

6. Sustainability-based approaches to target setting and the precautionary principle
Even taking the perspective of an economist, one would be very reluctant to rely exclusively on efficiency-based targets given the difficulties identified in the previous slides.
It seems sensible to at least give some weight to alternative approaches to pollution policy that explicit address limited information and uncertainty.
Non-economists are generally suspicious of driving policy on what are perceived as narrowly economic grounds and are critical of the importance that is often attached to efficiency by economists in thinking about pollution targets.
Natural scientists, environmentalists and ecologists typically regard stability and resilience as being more fundamental objectives.
These objectives – on the one hand, population stability and/or ecosystem resilience, and on the other hand, maximisation of net economic benefits – are not necessarily mutually contradictory.
Much of environmental economics (and, more so, ecological economics) consists of an attempted synthesis of the two.

7. Precautionary principle
In a world of certainty (and so complete predictability) taking precautions would be unnecessary. But in a stochastic or complex environment where outcomes are not certain, where processes are incompletely understood, or where non-linearities of various kinds are thought to exist, some form of ‘playing safe’ is sensible.
The precautionary principle – in some of its guises at least – can be thought of as a hybrid criterion.
It tries to bring together efficiency, sustainability, ethical and ecological principles, into a bundle that can inform target setting.
Of course, in trying to do several things at the same time, it runs the risk of not doing any of them particularly well.
But the approach is now being widely advocated.
If efficiency and sustainability criteria yielded identical policy recommendations, their relative importance would not matter. But analysis suggests they do not.

8. Safe Minimum Standard
The precautionary principle can be thought of as proposing a lexicographic approach to setting targets.
We regard one criterion (in this case sustainability) as being of overriding significance, and require that any target do as well as possible in terms of this measure.
If this leaves us with more than one option, then other desirable criteria can be employed (perhaps in a hierarchical way too) to choose among this restricted choice set.
Alternatively, a constraint approach could be adopted: pollution policy should in general be determined using an efficiency criterion, but subject to an overriding sustainability constraint.
Chapter 13 explains the notion of a safe minimum standard (SMS) of conservation.
When applied to pollution policy, the adoption of a strict version of the SMS approach entails that threats to survival of valuable resource systems from pollution flows are eliminated.
A modified SMS would eliminate the pollution flow, provided that so doing does not entail ‘excessive cost’.

10. Taxes Permits t* P* t* P* MC MC PH MCH MCH PL MC MC MCL MCL
Emissions, M M PH
 DEPENDABILITY
The differing way in which abatement cost uncertainty affects taxes and marketable permits is illustrated in Figure 7.1.
In the upper half of the figure, a single aggregate marginal abatement cost function (MC) is drawn, assumed to be known by the authority. Tax and permit regimes are identical in outcomes.
In contrast, abatement cost uncertainty is represented in the lower half of the diagram by showing three different realisations of marginal abatement costs.
These three curves can be thought of as three drawings from a probability distribution that describes the whole set of possible outcomes.
It is evident that quantity-based controls can have very different impacts from price-based controls.
In general, uncertainty about abatement costs translates into uncertainty about the quantity of abatement in a price system (such as emissions taxes or emissions abatement subsidies).
It translates into uncertainties about prices or costs under a quantity-control system.
For example, the aggregate marginal abatement cost and the marginal abatement costs for individual firms will be uncertain under a non-tradable permits system, or the equilibrium permit price, P, and aggregate marginal abatement cost will be uncertain under a marketable permits system. t* P*
MCH
MCH PL MC MC
MCL
MCL ML M* MH M M
L*(= M*)
Figure 7.1 A comparison of emissions taxes and marketable emissions permits when abatement costs are uncertain

11. Establishing a ‘baseline’ against which the efficiency losses from errors due to uncertainty can be measured.
The efficient target, M*, is that level of emissions which equates the marginal cost of emissions abatement (MC) and the marginal damage of emissions (MD).
The shaded area represents the total net social benefit that would be generated at that level of emissions.: the maximum net benefit available.
Efficiency losses from uncertainty have in mind are those in which emissions are at any level other than M*, and so attained net benefits fall short of their maximum level.
Figure 7.2 Target setting under perfect information MD t* MC
The costs associated with (unknowingly) selecting the wrong target are ‘efficiency losses’ or ‘welfare losses’.
It is important to be clear about what kind of loss we have in mind here. Figure 7.2 helps to fix ideas.
This establishes a ‘baseline’ against which the efficiency losses from errors due to uncertainty can be measured. The efficient target, M*, is that level of emissions which equates the marginal cost of emissions abatement (MC) and the marginal damage of emissions (MD). The shaded area in Figure 7.2 represents the total net social benefit that would be generated at that level of emissions. This is the maximum net benefit available. The efficiency losses we have in mind are those in which emissions are at any level other than M*, and so attained net benefits fall short of their maximum level. M*
Emissions, M

12. tH t* Mt M* LH Emissions, M
Figure 7.3 Uncertainty about abatement costs – costs overestimated MD
Loss when
licenses used tH t*
MC (assumed)
Loss when
taxes used
Uncertainty about abatement costs
Uncertainty about abatement costs may result in an efficiency loss.
Suppose that the EPA knows the pollution marginal damage function (MD) but has to estimate the marginal emissions abatement cost function (MC), and will often make errors in doing so. Overestimation and underestimation of abatement costs will each lead the EPA to wrongly identify the efficient level of emissions, and so to an efficiency loss. But, as we shall see, the magnitude of that loss will differ depending on which instrument the EPA chooses to use.
Let us investigate the relative magnitudes of efficiency loss under an emission tax system and an emission licence scheme.
Figure 7.3 shows the case in which the marginal cost of abatement is overestimated.
Consider first an emissions fee. On the (incorrect) assumption that the marginal abatement cost curve is the one labelled ‘MC (assumed)’, the EPA imposes a tax at the rate tH (as opposed to its true efficiency level, t*). Firms will abate emissions as long as their actual (true) marginal abatement costs are below the tax, and so will emit at Mt, a rate less than the efficient level. The resulting efficiency loss is defined by the shortfall of net benefits at Mt compared with the maximum obtainable level at M*; this is indicated by the hatched area in the diagram.
Compare this efficiency loss with that which results from using an emissions licence system. Using incorrect information, the EPA believes the efficient target is LH (when in fact it should be M*). Incorrect information has led the regulator to pursue an insufficiently tight control. The efficiency loss is indicated by the solidly shaded area (corresponding to the surplus of marginal damages over marginal abatement costs for the excessive units of emissions).
MC (true) Mt M* LH
Emissions, M

13. t* tL LL M* Mt Emissions, M
Figure 7.4 Uncertainty about abatement costs – costs underestimated MD t* tL
MC (true)
Underestimation of abatement costs:
This is represented in Figure 7.4, in which the shapes and positions of the ‘true’ functions are identical to those in Figure 7.3 to allow direct comparison of the two diagrams. Now the assumed marginal abatement cost curve lies below its true position. Using similar reasoning to that given above, it can be seen that an emissions tax results in a loss (shown by the hatched area) that is greater than the loss associated with licences (the solidly shaded triangle).
An incorrectly estimated abatement cost function results in an efficiency loss. In the case we have investigated, irrespective of whether the error is one of over- or underestimation, the loss from using taxes exceeds that from using licences. However, this result depends on the manner in which we constructed the functions in Figures 7.3 and 7.4. Compare these with the cases shown in Figures 7.5 and 7.6. These are analogues of the two situations just investigated, but are drawn with a substantially flatter marginal damage curve. Once again, both instruments generate efficiency losses when mistakes are made about abatement costs. But the ranking of the two instruments is now reversed: the loss is larger with licences than with a tax.
MC (assumed) LL M* Mt
Emissions, M

14. tH t* Mt M* LH Emissions, M
Figure 7.5 Uncertainty about abatement costs – costs overestimated MD MD tH t*
MC (assumed)
MC (true) Mt M* LH
Emissions, M

15. t* tL M* LL Mt Emissions, M
Figure 7.6 Uncertainty about abatement costs – costs underestimated MD MD t* tL
MC (true)
MC (assumed) M* LL Mt
Emissions, M

16. General results for abatement cost uncertainty
What differentiates these two pairs of cases is the relative slopes of the MC and MD functions. We obtain the following general results:
When the (absolute value of the) slope of the MC curve is less than the slope of the MD curve, licences are preferred to taxes (as they lead to smaller efficiency losses).
When the (absolute value of the) slope of the MC curve is greater than the slope of the MD curve, taxes are preferred to licences (as they lead to smaller efficiency losses).

17. Figure 7.7 Uncertainty about damage costs – damages underestimated
MD (true)
MD (estimated) t
Uncertainty about pollution damages
The conclusions we reached do not carry over to uncertainty about damage costs.
In this case, the choice of quantity- or price-based instruments has no bearing on the magnitude of the efficiency loss arising from errors in estimating damage costs. The size of that loss is the same in each case, and so knowledge about the relative slopes of functions can give no information that would minimise such losses. This result is illustrated in Figure 7.7.
Given the estimated marginal damage function and the marginal cost function (assumed here to be correctly estimated), an EPA might set a tax at the rate t or a quantity control at the amount L. In either case, the level of realised emissions exceeds the efficient level M* and the efficiency loss associated with the erroneous target is shown by the shaded area in Figure 7.7.
MC (true) M* L
Emissions, M

18. Figure 7.8 Consequences of a threshold in the damages function
Total
damages D
The consequences of a threshold effect in the pollution damage function
Now assume that there is uncertainty about the location of the MC function, but that the pollution damage function is known and contains a threshold effect.
Can any conclusions be obtained about the best choice of instrument in the case? We go through an argument used in Hartwick and Olewiler (1998).
The total damage function contains two linear segments, with a discontinuity (‘threshold’) at emission level M. M
Emissions, M

19. Figure 7.8 (Panel (b))
The non-linearity in damages implies that a price-based policy has attractive properties where errors are not too large. However, when the estimation error goes just beyond some critical size, the efficiency loss can switch to a very large magnitude.
Marginal
damages
&
marginal (abatement) costs t2 MD
MC1
The total damage function in Figure 7.8(a) corresponds to the marginal damage function portrayed in Figure 7.8(b).
As with the total function, marginal damages again exhibit a discontinuity, although they are constant above and below that discontinuity (because of the linearity of the two segments of the total damage function).
The EPA knows the shape and position of the marginal damage function, and so is aware that a threshold exists at M = M. However, the EPA is uncertain of the location of the marginal abatement cost function.
Two of the many possible locations of the marginal abatement cost curve are labelled as MC1 and MC2.
Suppose that the EPA estimates that marginal costs are given by the curve MC1, and sets an emissions tax at the rate t1. If the EPA’s estimate of MC were true, this would yield the efficient level of emissions, M1. Even if that estimate were incorrect by a relatively small amount that tax rate would still generate an efficient level of emissions. More precisely, provided that the true value of MC is such that its intersection with MD is somewhere between M = 0 and M = M, the tax rate t1 would induce an efficient emission. Note, in contrast, that a quantity control would not have this attractive property.
However, suppose that the EPA had grossly underestimated marginal abatement costs, with the true function actually being MC2. Inspection of the diagram makes it clear that a tax rate set at the value t1 would lead to substantially excessive emissions. Efficient emissions would be M2 but realised emissions are M22 (with an efficiency loss shown by the heavily shaded area in Figure 7.8b).
If the tax rate had been set at t2, efficient emissions result if the true value of MC lies in the neighbourhood of MC2, but it imposes a massively deficient emissions outcome (here zero) if MC1 is the actual value. Overall we see that the non-linearity in damages implies that a price-based policy has attractive properties where errors are not too large. However, when the estimation error goes just beyond some critical size, the efficiency loss can switch to a very large magnitude. t1
MC2 M
M22 M1 M2
Emissions, M

20. Quantity controls
We leave the reader to explore the use of quantity controls.
You should find that if the EPA set a control at the quantity M1 or M2 (depending on which MC function it deems to be relevant), the likelihood of extremely large efficiency losses is reduced, but at the expense of losing some efficiency for relatively small errors in estimation.
Hartwick and Olewiler conjecture that a best policy in the case analysed in this section is one that combines a tax (price) control and an emissions (quantity) control.
They propose a tax equal to the lower value of the MD function, and an emissions limit equal to the threshold level. The tax bites – and generates efficient emissions – if marginal abatement cost lies in the neighbourhood of MC1. Where MC is sufficiently large to intersect MD in its upper segment, the quantity constraint bites. This composite policy does not eliminate efficiency losses, but it prevents them being excessively large.
The authors argue that such a combined policy is also prudent where uncertainty surrounds the position of the MD function.

21. General conclusions
Where functions are linear, and uncertainty relates to the marginal abatement cost (MC) function, then an EPA should prefer a quantity policy (licences) to an emissions tax if MC is flatter than MD, and an emissions tax to a licence system if the reverse is true, if it wishes to minimise the efficiency losses arising from incorrect information.
However, where uncertainty pertains to the MD function, knowledge of relevant slopes does not contain information that is useful in this way.
Once the existence of non-linearity and/or threshold effects is admitted, general results are harder to find. In some circumstances at least, combined tax–quantity-control programmes may have attractive properties.
The presence of uncertainty substantially weakens the general presumption in favour of incentive-based instruments over quantitative regulations that we developed in the previous chapter. They may be better in some circumstances but not in all.

22. Information requirements: asymmetric information and incentive compatibility
Imperfect information puts restrictions on the ability of the EPA to devise ‘good’ targets and to attain them at least cost.
It also considerably complicates its choice of instrument because comparative advantages depend on the prevailing circumstances.
Moreover, limited information and uncertainty may prevent the EPA from knowing which circumstance actually pertains.
Faced with all this, there are strong incentives on the EPA to become better informed.
One would expect that it would invest in systems that deliver greater information. There are three ways that the EPA might do this:
undertake its own research to gather data;
build long-term institutional relationships with regulated businesses;
create reward structures that give firms incentives to reveal information truthfully to the regulator.
We shall focus on the last of these three.

23. Incentive compatibility
An instrument is incentive-compatible if the incentives faced by those to whom the instrument applies generate behaviour compatible with the objectives of the regulator.
In general, none of the instruments we have discussed so far has this property. Where polluters think that the numbers they report can influence the severity of regulation, they have an incentive to lie about the costs of complying with abatement targets.
This is true whether the instrument being used is command and control, emissions tax, abatement subsidy or a marketable permit scheme.
We illustrate two examples of such incentive effects.
If firms expect tax schemes to be used they have an incentive to understate abatement costs.
If they expect a marketable permit scheme, the incentive is to overstate these costs.
We also outline one possible instrument – a mixture of abatement subsidy and marketable permits – that is incentive-compatible.

24. If firms honestly report their actual abatement costs, L
If firms honestly report their actual abatement costs, L* permits are issued, allowing an efficient emissions level M*. In a competitive permits market the equilibrium permit price would be μ*. If firms overstate their abatement costs, the EPA incorrectly believes that the efficient target is MP, and so issues that number of permits (LP). Exaggerating abatement costs is better for firms than being truthful as more permits are issued, and so they incur lower real emission abatement costs.
Figure 7.9(a) Incentive effects with permit systems MD
P1 *
MC (reported)
P2
Case 1: Firms expect a permit system to be in operation
Suppose that firms expect the EPA to use a marketable permit system. Moreover, they believe that the total number of permits issued will be equal to what the EPA estimates to be the economically efficient level of emissions. Finally, firms realise that the EPA will make its choice of permit quantity only after firms have provided the EPA with information about their emissions abatement costs.
Figure 7.9(a) shows that it is in the interest of firms to exaggerate their marginal abatement costs (MC). If firms honestly report their actual abatement costs, L* permits are issued, allowing an efficient emissions level M*.
In a competitive permits market the equilibrium permit price would be μ*. If firms lie, and overstate their abatement costs, the EPA incorrectly believes that the efficient target is MP, and so issues that number of permits (LP). Exaggerating abatement costs is better for firms than being truthful as more permits are issued, and so they incur lower real emission abatement costs.
Note in passing a point that we will return to later, and which was alluded to earlier in remarking that the use of instruments will reveal useful information. The EPA expects the permit price to be μP1, although the actual price will turn out to be μP2 (because the true marginal abatement cost function is the demand curve for permits).
MC (true) M*
=L* MP
=LP
Emissions, M

25. With an emissions tax, firms have an incentive to understate their abatement costs.
Firms benefit because they emit more than if they told the truth (and so incur lower real abatement costs). Also, the tax rate is lower than it would be otherwise.
Figure 7.9(b) Incentive effects with an emissions tax MD * T
MC (true)
Case 2: Firms expect an emissions tax to be in operation
Now we suppose that firms expect the EPA to use an emissions tax system. Equivalently to Case 1, firms expect that the EPA will set what it believes to be the efficient tax rate only after firms have informed the EPA of their abatement costs. Figure 7.9(b) shows that firms have an incentive to understate their abatement costs. If abatement costs are reported truthfully, a tax rate of μ* is set, leading to emissions of M* (the efficient level). However, if firms understate their abatement costs the EPA incorrectly believes that the efficient tax rate is μT, and sets that rate. This results in a quantity of emissions MT1. Firms benefit because they emit more than if they told the truth (and so incur lower real abatement costs). Also, the tax rate is lower than it would be otherwise.
Note also that the EPA expects the quantity of emissions to be MT2 whereas in this scenario it will turn out to be MT1. Once again, this information will prove useful to the regulator. Indeed, whether a tax or a permit system is used, untruthful behaviour is revealed after the event to the EPA. The EPA observes a difference between what it expects the permit price to be and what it actually is (or between the actual and expected levels of emissions). Moreover, it will be able to deduce in which direction abatement costs have been misreported. So one possibility open to the EPA is to adopt an iterative process, changing the number of permits it issues (or tax rate) until there is no difference between actual and expected outcomes. But this may not be politically feasible, or it may involve large costs in making successive adjustments.
MC (reported)
MT1
MT2 M*
Emissions, M

26. An incentive-compatible instrument: Kwerel (1977)
Can an instrument be found which will encourage truthful behaviour and allow the EPA to achieve its objective?
We are looking for an instrument that creates an incentive to report abatement costs truthfully and which allows the EPA to achieve whatever target it wants in a cost-effective way.
Several schemes with such properties have been identified. One is proposed by Kwerel (1977).
The scheme involves a combination of marketable permits and subsidies on ‘excess’ emissions reduction.

27. Intuition
The scheme involves a combination of marketable permits and subsidies on ‘excess’ emissions reduction.
The costs that firms report have two effects:
they influence the number of permits issued
they also influence the subsidy received for excess emissions reduction.
The scheme balances these two influences so as to reward truthful reporting.

28. Mechanism
Kwerel’s scheme works in the following way. Firms are told that:
permits will be allocated through auction;
they will receive a subsidy for any emissions reduction over and above the number of permits they hold;
the subsidy rate will be set at the intersection of the marginal damage curve and the reported marginal abatement cost curve.
Given this information, firms are then asked to report their abatement costs, the subsidy is set accordingly, and the permit auction takes place.
The total cost of the scheme to all firms in the industry is equal to actual emission abatement cost plus the cost of acquiring permits less the subsidy payments received on any emissions reduction over and above the permitted amount of emissions.

29. Notation PCC = Abatement costs (area under MC curve)
We use the following notation: M = volume of emissions; L = number of permits made available to industry; P = price of permits; s = subsidy per unit of emissions reduction.
Then we can write an expression for pollution abatement costs (PCC) for the whole industry.
PCC = Abatement costs (area under MC curve)
+ Permit costs P • L
- Emissions reduction subsidy s • (L - M)
To demonstrate that this instrument is incentive-compatible, we compare the benefits to firms of being truthful with the benefits of (a) understating costs and (b) exaggerating costs.

30. Figure 7.10(a) An incentive compatible instrument and under-reporting costsMD P*
MC (true)
Case 1: Firms understate abatement costs (see Figure 7.10(a))
Understating causes permits to be scarce ( Lhat} rather than L*). The permit price is driven up to P hat, the level determined by true abatement costs. Hence total costs rise because (a) fewer emissions licences means it must do more abatement (shaded area) and it has to pay a higher permit price (hatched area). The combined area of these is larger the greater is the permit price. Note that the firms’ total costs are increasing in the permit price. It follows from this that firms’ costs are minimised when the costs reported are actual costs, which drives the permit price down to its lowest level.
MC (reported)
M*= L*
Emissions, M

31. Figure 7.10(b) An incentive compatible instrument and over-reporting costs
Diagram shows the losses that firms make as a result of exaggerating abatement costs. The shaded area is the additional abatement costs, the hatched area is the additional price paid for permits. It can be seen that as MC (reported) goes towards MC (true), these losses disappear. The best that the firm can do is to be truthful! MD
MC (reported) P*
MC (true)
Case 2: Firms exaggerate abatement costs (see Figure 7.10 (b))
At first sight, exaggeration of abatement costs seems to be advantageous to firms: it increases allowed emissions (to L bar) and it increases the subsidy rate (to S bar). But there is another factor that dominates these considerations. The existence of the subsidy, S bar, puts a floor (a minimum level) on the permit price. That price cannot fall below S bar. For if it began to do so, firms would buy permits in order to receive the (higher-valued) subsidy payment from holding more permits. But if the permit price is equal to S bar, then the amount of permits actually bought will be M bar (even though a larger quantity L bar is available for purchase).
Figure 7.10(b) shows the losses that firms make as a result of exaggerating abatement costs. The shaded area is the additional abatement costs, the hatched area is the additional price paid for permits. It can be seen that as MC (reported) goes towards MC (true), these losses disappear. The best that the firm can do is to be truthful! M*
Emissions, M

32. Transactions costs and environmental regulation
In carrying out its responsibilities, an environmental protection agency necessarily incurs transaction costs. This is a generic term for a variety of costs that include:
acquiring relevant information;
creating, monitoring and enforcing contracts (of which one category is the EPA’s regulations);
establishing, implementing and revising the instruments it employs;
monitoring performance, and ensuring compliance.
Transactions costs do include the costs of the personnel and the structures an organisation puts in place that allow it to carry out its activities, and any equivalent costs imposed on other parties, including the regulated firms or individuals.
They do not include what are sometimes called the real resource costs of the controls – that is, the costs of pollution control equipment, higher fuel bills for cleaner fuel, more expensive exhaust systems and so on.
They also do not include any induced indirect costs that might occur such as loss of national competitiveness or increased unemployment.
Summing up all those costs gives the total compliance costs of environmental regulation.
Transactions costs are just one part – often a not insignificant part – of that overall total.

33. D C B A ZB ZA ZC Emissions abatement
Figure The net benefits of regulation
Marginal real resource cost of abatement + induced indirect costs + transactions costs
Marginal real resource cost of abatement + induced indirect costs D
Marginal real resource cost of abatement C
We assume that the marginal gross benefits of pollution abatement (the damages avoided) are correctly represented by the curve labelled as D in the picture.
Curve A represents the marginal real resource costs of pollution abatement. If there were no other costs, an efficient outcome would require ZA units of abatement.
There may also be induced, indirect costs, including impacts on unemployment and trade competitiveness. Adding these to the resource abatement costs, the composite cost curve B is obtained, with a correspondingly lower efficient abatement level, ZB.
Note that Figure 7.11 assumes that these induced indirect impacts are adverse to the economy in question. It is possible, though, that they may be beneficial. If the induced effects were beneficial rather than adverse overall, curve B would lie to the right (rather than to the left) of curve A.
Finally, curve C adds in transactions costs to the previous two categories of costs. The efficient abatement level, taking all relevant items of information into consideration, is Zc.
This does provide a useful way of thinking about instrument selection. Suppose that a choice of abatement target had already been made. That would then fix a particular point along the abatement axis that we are committed to reach. If we were comparing the relative merits of two instruments, we might construct two versions of Figure 7.11, one for each instrument. The preferred instrument would be the one that has the lower total cost of achieving that particular target. Even if one instrument is superior in terms of real resource cost of abatement, it need not be superior when induced effects and transactions costs are also considered.
There is another interesting inference to be drawn from Figure The notion that an EPA could devise an efficient target first, without consideration of instrument to be used, and then choose an instrument to best attain that target, may not be sensible. To see why, return to the idea that we might choose between instruments by constructing alternative versions of Figure 7.11, one for each instrument. It is evident that if the cost functions differ, so might the efficient abatement level. The independence of targets from instruments does not seem to go through.
Gross marginal benefits of abatement B A ZB ZA ZC
Emissions abatement