Oxford Review of Economic Policy, Volume 26, Number 2, 2010, pp.

225252

Cap-and-trade: a sufficient or necessary condition for emission reduction?

Michael Hanemann*
Abstract Influenced by the success of emission trading in the US for sulphur dioxide (SO2), some economists have argued for an upstream, economy-wide cap-and-trade scheme as the primary tool for achieving the required reduction in greenhouse gas (GHG) emissions. This paper addresses that argument and concludes that cap-and-trade will need to be accompanied by complementary regulatory measures. While it is a necessary component in a climate mitigation programme, it is unlikely to be sufficient by itself to accomplish the desired emission reductions. The paper reviews the evidence on how SO2 emissions were reduced and the extent to which actual emission trading was responsible for the reduction as opposed to other innovations. It also identifies differences between the past regulation of SO2 and other air pollutants and the challenges presented by the regulation of GHG emissions. What actually happened in the US with SO2 emission trading deviated in several significant respects from what would be predicted based on the conventional theoretical analysis. While there was a dramatic reduction in SO2 emissions, it occurred because of several factors, some of which are unlikely to apply for GHG emissions, and others of which argue for an activist regulatory policy by the government as a complement to the functioning of an emissions market for GHGs. Keywords: climate change, emission trading, pollution control JEL classification: H23, Q54, Q58

I. Introduction
The success of allowance trading for sulphur dioxide (SO2) in the US under the 1990 Amendments to the Clean Air Act (CAA) has been widely seen as creating a new paradigm for government regulation of the environment.1 Instead of the heavily prescriptive

University of California, Berkeley, e-mail: [email protected] I am extremely grateful to Cameron Hepburn and Giles Atkinson for very helpful comments. 1 As will become evident, this paper deals with domestic rather than international climate mitigation policy, and its focus is the US and, to a lesser extent, the EU, but not developing countries. doi: 10.1093/oxrep/grq015 The Author 2010. Published by Oxford University Press. For permissions please e-mail: [email protected].

226

Michael Hanemann

command-and-control regulation that marked the first two decades of pollution control regulation in the US, the new approach emphasizes the role of price signals as incentives for changing polluter behaviour. One way to create such incentives is through emissions taxes, such as the SO2 and carbon taxes introduced in the Scandinavian countries and elsewhere. In the US context, however, the introduction of new taxes is seen as politically infeasible. Hence, cap-and-trade (emission trading) has emerged as the preferred approach in the US.2 Compared to the prior regulatory approach, this involves a more minimalist role for the government. The government sets emission caps for individual polluters, distributes emission allowances according to those caps, creates a mechanism to monitor the emissions, and establishes a procedure whereby polluters turn in allowances to cover their emissions. Having thus established the parameters of a market, the government then sits back and allows events to proceed on their own. Polluters reduce their emissions or buy allowances. Private intermediaries come along. A market evolves. Emissions are reduced. This is what happened in the US with SO2. Under Title IV of the 1990 Amendments to the CAA, emission trading was instituted in two phases. Phase I, lasting from 1995 to 1999, covered the largest and dirtiest generating units. Starting in 2000, Phase II extended the coverage to virtually all fossil-fuel power plants in the US. The total SO2 emissions from the generating units covered by Phase I had been 9.4 m tons in 1980 and 8.7 m tons in 1990. In Phase I these units SO2 emissions were capped at 5.5 m tons; by 1999, their actual emissions were 3.5 m tons, a reduction of almost 60 per cent compared to their 1990 emissions. With Phase II, the total SO2 emissions from the units covered in that phase had been 17.3 m tons in 1980 and 15.7 m tons in 1990. These units emissions were capped at a declining rate, starting at 10 m tons in 2000 and declining to 8.95 m tons in 2010, remaining fixed at that level thereafter. In 2008, these units emissions were capped at 9.5 m tons, but their actual emissions were 7.6 m tons, a reduction of almost 52 per cent compared to their 1990 emissions. The empirical success with SO2 reduction was not unexpected by economists. The economic explanation dates back to Crocker (1966), Dales (1968), and Montgomery (1972). Several theoretical propositions about emission trading have been demonstrated. A capand-trade system achieves the set reduction in aggregate emissions at a minimum total cost. The economy-wide equilibrium is independent of the initial allocation of allowances. The outcome of emissions trading is identical to the outcome with an emissions tax when the tax rate is set equal to the clearing price in the emissions marketequilibrium prices are the same economy-wide. Because the economy-wide equilibrium prices are independent of the initial allocation of emission permits among individual producers, the point of regulation makes no difference. Whether there is an upstream cap (e.g. the point of regulation is the point of entry of fossil fuels into the economy) or a downstream cap (the point of regulation is the end user of the fossil fuels, or the end user of energy derived from fossil fuels), the ultimate economic outcome is the same.3 In this theoretical analysis, market prices are the drivers of behavioural change. When a price is placed on a pollutant through a cap-and-trade system (or a tax), the prices of pol2 As Convery (2009) points out, the European Unions Emissions Trading Scheme for CO2 (EU ETS), introduced in 2005, arose as the result of the European Commissions failure to introduce an effective EU-wide carbon tax. 3 The equivalence result assumes that there are no deviations from perfect competition along the supply chain. It would not necessarily hold if, for example, the railroads which deliver low-sulphur coal have market power or the rates charged by pipeline companies are regulated.

Cap-and-trade: a sufficient or necessary condition for emission reduction?

227

luting commodities rise, reducing the demand for those commodities and raising the demand for non-polluting substitutes. In turn, this lowers input demands by producers of the polluting commodities, reducing those prices and passing the cost burden backward along the supply chain. At the same time, there is increased demand for inputs that reduce the generation of emissions. The price changes radiate throughout the economy, inducing a suite of price-driven demand and supply responses in sectors both upstream and downstream of the sectors that are capped. Once the government establishes the parameters of the market, price signals take over and do their work. Thus, both economic theory and the empirical experience of emission trading for SO2 allowances appear to support the concept of a minimalist role for the government in the regulation of pollution. However, in this paper I argue that what actually happened in the US with SO2 emission trading deviated in several significant respects from what would be predicted based on the theoretical analysis outlined above. While there was a dramatic reduction in SO2 emissions, it occurred because of several factors some of which are unlikely to apply for greenhouse gas (GHG) emissions, and others of which argue for an activist regulatory policy by the government as a complement to the functioning of an emissions market for GHGs. Thus, emission trading is a necessary but not a sufficient condition for an effective climate policy. The remainder of the paper is organized as follows. Section II reviews the evidence on how SO2 emissions were reduced from 1995 onwards and the extent to which actual emission trading was responsible for the reduction as opposed to other innovations. I argue that other innovations played a major role, and these were due largely to the fact that the 1990 CAA Amendments functioned as a performance standard rather than the technology standard previously existing under the Clean Air Act. Section III examines the trading that did occur under Title IV and notes the occurrence of significant over-compliance and banking of emission reduction, non-arms length transactions, and non-participation in the permit market. In light of this empirical evidence, section IV discusses the possibility that some firms behaviour in reducing emissions may have been influenced not just by the price of permits in the SO2 market but also by the cap set on their individual emissions. Section V examines the differences between the past regulation of SO2 and other air pollutants and the challenges presented by the regulation of GHG emissions. Given these differences, section VI examines the evidence regarding whether emission trading will be effective in promoting the degree of technological innovation that will be required for GHG abatement. Section VII makes the case for governments to go beyond emission trading and become more engaged with GHG abatement by adopting various complementary policy measures. Section VIII offers some brief conclusions.

II. How SO2 emissions were reduced

The experience with SO2 trading has played a crucial role in the US debate on climate change policy, both in individual states and at the national level. It is regularly cited by proponents of emission trading as an important reason for applying this to GHGs.4 It is useful, therefore, to review exactly how SO2 emissions were reduced in the US and what was the role of emission trading.
4

For example, Stavins (2007, 2008).

228

Michael Hanemann

Ellerman and Dubroeucq (2004) estimate that about 85 per cent of the reduction in SO2 emissions between 1994 and 2002 was associated with a reduction of emissions at individual generating units (i.e. cleaning up old plants) as opposed to switching generation from highto low-emitting units. In other words, firms responded to prices largely by reducing emissions from their own plants, rather than trading allowances between firms. Burtraw (1996) called this response cost savings without emissions trading. Some of this reduction was due to the retrofitting of scrubbers, but about half or more of the total reduction in SO2 emissions was due to two other factors: (i) a switch from high- to low-sulphur coal, and (ii) a substantial reduction in capital and operating costs for both new and existing scrubbers. Neither of these was expected prior to the introduction of trading. Coal switching, in part, involved the introduction of western low-sulphur coal to power plants in the Midwest. This was a consequence of railroad deregulation in 1976 and 1980, which gave railroads greater operating freedom. In addition, there were what Ellerman (2003) calls [o]ther cost-reducing changes that might be termed innovations in the other major coal production regions. In the Midwest and Northern Appalachians, lower-sulphur coal came to be mined as a result of newly opened mines, changes in mining practices, and increased sulphur removal in coal preparation plants. There was also innovation with regard to scrubbers. The price of new scrubbers dropped by nearly half between 1989 and 1994. The operating efficiency increased, the average removal efficiency of new scrubbers rising from about 85 per cent before 1992 to about 95 per cent by 1995. Also, there was an increase in the utilization of the retrofitted scrubbers. As a result, there was a substantial reduction in the cost of using scrubbers. Another factor leading to cost savings without emissions trading was that the act of giving plant operators the freedom to choose between buying emission allowances, installing a scrubber, changing the dispatch order, and switching fuels created an element of competition among the suppliers of inputs to abatement that was not formerly present. By itself, the enhanced competition contributed to a reduction in abatement costs. As Ellerman (2003) puts it, [w]hat emerges from the experience with Title IV is that [abatement] costs are lower for reasons beyond the ability to trade emission reductions among sources. The cost reductions were a form of induced innovation. This did not involve the emergence of fundamentally new technologies but rather improvements in operating practices and innovation in the application of existing technologies. One should also note what did not happen. Despite increased demand, the prices of inputs used for the abatement of SO2 in electricity production did not risethey fell. And there was virtually no discernible increase in the price of electricity as a result of allowance trading for SO2. The total value of electricity sales in 2000 was $233 billion, while the total cost of compliance with the Phase II programme that year is estimated to have been about $1.5 billion (i.e. about 0.6 per cent of sales). Moreover, the retail price of electricity fell continuously throughout this period from (in 2000$) 9.7 cents per kWh in 1982 to 8.05 cents in 1990, 7.66 cents in 1994, and 6.81 cents in 2000.5 Thus, although SO2 abatement may have raised the cost of electricity production by about 0.6 per cent in 2000 compared to pre-Phase I, the retail price of electricity fell by more than 10 per cent between 1994 and 2000.There also was no obvious reduction in the production of electricitynet generation grew from 3,247 billion kWh in 1994 to 3,802 billion kWh in 2000 (an increase of 17 per cent).

There was a spike in electricity prices in 2001, but they then fell back to the range 6.916.99 cents in 20024.

Cap-and-trade: a sufficient or necessary condition for emission reduction?

229

It also appears that the cap on SO2 emissions had virtually no effect in promoting either energy conservation or the use of renewable sources of electricity. Net generation from renewables rose by only about 6 per cent from 337 billion kWh in 1994 to 357 billion kWh in 2000. One provision of Title IV created the Conservation and Renewable Energy Reserve, under which bonus allowances worth 300,000 tons of SO2 were set aside to be allocated to utilities for energy efficiency and renewable energy development. However, little use was made of these allowances; as of February 2002, only about 49,000 of these allowances had been awarded (Vine, 2003). Thus, the reduction in SO2 emissions was only partly the result of price changes radiating from the electricity sector because the cap on emissions had raised the cost of producing electricity. Induced innovation within the electricity sector itself and its immediate suppliers played a more important role, whether in the form of market evolution associated with increased competition, or operational innovations by coal producers, railroads, boiler operators, and scrubber manufacturers and operators. Electricity generators and their suppliers were motivated to find new ways of responding to the implicit price placed on SO2 emissions. This should not be seen as implying that the introduction of emission trading for SO2 was unimportant or of little economic benefit. Even setting aside the substantial reduction in SO2 emissions mandated, the manner in which the emission reduction was regulated was a significant improvement compared to the prior regulatory regime established under the 1970 Clean Air Act and the 1977 CAA Amendments.6 Instead of mandating a particular abatement technology, Title IV established a performance standard for SO2. The emissions limit established under Phase II of the SO2 trading programme was the same as the prior NSPS limit, but the allowance trading was vastly more flexible. Installation of a scrubber was no longer required. A power plant operator became free to use low-sulphur coal, vary the dispatch order, substitute emissions among facilities, and/or purchase emission allowances in order to comply with the emissions limit. The reduction in the costs of a scrubber has already been noted. One factor that may have contributed is the flexibility provided by Title IV with regard to redundant scrubber capacity (Burtraw, 1996). Before the 1990 CAA Amendments, scrubber systems usually included a spare module to maintain low emission rates when any one module became inoperative. The alternatives afforded by Title IV lessened the need to install a spare module, thereby reducing the capital cost of a new scrubber. A performance standard is inherently superior to a technology standard because it affords firms the flexibility to attain their emission limit in whatever is the most cost-effective manner for them. The existing discussion of the SO2 trading programme has focused largely on differences in the marginal cost of SO2 abatement among different electricity generating units, and the resulting cost savings if emission reduction can be shifted among plants. This is how the gains from emission trading have been modelled in the theoretical literature: emission trading is efficient because it reallocates abatement from plants with high to low marginal costs. This is essentially a static analysis. The marginal abatement cost curves are
6 Under the prior regime, power plants in existence when the regulations implementing the 1970 CAA became effective in 1971 had to meet emission rate limits imposed by State Implementation Plans (SIPs), which the individual states were required to develop to ensure compliance with National Ambient Air Quality Standards (NAAQS) for six criteria air pollutants, including SO2. New power plants constructed after that date were regulated more stringently. They were required to meet the New Source Performance Standard (NSPS) which, after 1977, required the installation of a scrubber in new power plants, ruling out any reliance on low-sulphur coal alone.

230

Michael Hanemann

assumed to be fixed over time, and the focus is on annual emissions, the marginal cost of annual abatement, and the cost-minimizing re-allocation of annual abatement. In effect, this is a long-run analysis of where abatement should optimally occur. But, there are several other ways in which costs could be lowered because of the flexibility afforded by emission trading. One flexibility is the ability to reallocate emission reductions over time, which is enhanced by the fact the unused SO2 emission allowances are bankable. This gives companies the ability to control the timing of investments in new lumpy capital in the most opportune manner. If interest rates are low, for example, a company can install new abatement equipment before it is fully needed, and either bank or sell the excess emission reduction; or, a company can postpone abatement investments until more favourable financial circumstances arise, buying emission allowances in the interim. There are also gains associated with short-run, cost-minimizing operating flexibility as opposed to the long-run, cost-minimizing reallocation of emission reduction. One form of operating flexibility is the ability to respond to short-run fluctuations in operating conditions on, say, a weekly, daily, or hourly time scale by reallocating abatement among different plants, either internally within a firm that operates multiple plants or externally via the emissions market. A related source of operating flexibility concerns the firms ex ante uncertainty regarding what its emissions will be during the relevant regulatory period. A power plant operator cannot know what the demand for electricity will be next week, let alone for the rest of the year, and he cannot be sure of future prices for fuels of different sulphur content. So, how can he know what his emissions will be during the balance of the regulatory period? However, SO2 and other emissions markets are designed so that emission generation and compliance are asynchronous: emissions occur during a particular period (the calendar year, for SO2.) while the emissions permits must be turned over at the end of the period, or shortly thereafter. Thus, emissions permits have a convenience value (Burtraw, 1996) which physical abatement equipment lacks: if there is a shortfall on 31 December, say, it is possible to acquire additional permits at the last minute, but it is too late to lower the years emissions by installing a scrubber.7 Short-run operating flexibility and long-run reallocation of emissions reduction are clearly benefits arising from the existence of emission markets, but they are distinct phenomena and they serve different purposes. Operating flexibility, while economically valuable, does not necessarily lead to a long-run reallocation in emission reduction. For example, changing the dispatch order is a short-run fix to reduce emissions, but with growth in power demand it is not necessarily a substitute for the installation of new, low-emission generation capacity in the long run. Short-run flexibility and long-run reallocation of abatement are likely to lead to different patterns of participation in the emissions market. If a firm decides on a long-run strategy of abatement, one would expect to find it consistently selling permits in the market. If a firm decides to rely on buying permits as a strategy because this is cheaper than abatement, one would expect to find it consistently appearing as a buyer. However, if a firm varies between buying and selling permits, that could be consistent with use of the emission market for short-run flexibility rather than for long-run reallocation of abatement. At this point, it is an open question as to how much of the participation in the SO2 emission market was motivated by long-run versus short-run considerations. Nevertheless, the fact, noted above, that
7 In addition, the purchase of permits is reversible (the permits can be resold), while the installation of a scrubber is irreversible; the difference in reversibility gives rise to an option value for permits (Chao and Wilson, 1993).

Cap-and-trade: a sufficient or necessary condition for emission reduction?

231

85 per cent of the reduction in SO2 emissions between 1994 and 2002 was associated with a reduction of emissions at individual generating units rather than with switching generation from high-emitting to low-emitting units suggests a somewhat limited role for long-run reallocation of abatement via the SO2 emissions market.

III. The nature of SO2 trading

In Hanemann (2009), I review the literature on the functioning of the SO2 allowance market. Three distinctive features stand out. The first feature is over-compliance in emission reduction and the banking of allowances. In Phase I, 30 per cent of all allowances distributed during 19959 were banked; equivalently, the reduction in emissions was about twice what was required to meet the Phase I cap (Ellerman and Montero, 2007). Between 2000 and 2005 the bank was drawn downalbeit at a fairly modest rateto cover emissions in excess of the annual allotment of new allowances. In 2007 and 2008, there was again some overcompliance in emission reduction, and the allowance bank started to grow. In an analysis of banking between 1995 and 2002, Ellerman and Montero (2007) concluded that, contrary to the general impression of excessive banking, the amount of banking that occurred during this period was actually quite efficient based on economic fundamentals. At the time, Ellerman and Montero projected that an efficient bank would decline to about 3 m tons in 2008. This is about half the amount actually banked in 2008.8 It would seem that one cannot rule out the possibility of over-compliance in emission reduction beyond that called for by pure economic efficiency. The second feature is a significant degree of what Kreutzer (2006) calls autarky. This is when firms do not comply with their cap by purchasing allowances in arms-length transactions; they either reduce emissions to stay within the cap or, if there are excess emissions, they draw on their own past banked allowances or on allowances available from other units that they control. One measure of this is the extent to which, when firms submit allowances to the Environmental Protection Agency (EPA) to cover their emissions, these are allowances that had originally been allocated to them, rather than allowances obtained from someone else. Kreutzer (2006) finds that, between 1997 and 1999, about 70 per cent of the allowances retired each year were redeemed by the same unit to which they had originally been allocated, and only about 30 per cent were originally allocated to another unit.9 Between 2000 and 2003, the proportion of allowances redeemed by the same unit to which they had originally been allocated was lower, about 60 per cent, but still substantial. Third, even when allowances are transferred, this may not be an arms-length transaction since the same corporation may own multiple plants with multiple boilers, each with its own allocation of allowances. In its annual summary of allowance transaction data, the EPA distinguishes between what it calls economically significant transactions (i.e. between
8 Their analysis was conducted prior to the proposed tightening of emission limits under the Clean Air Interstate Rule in 2005, which strongly affected spot prices and could also have affected banking. The rule was set aside in July 2008. Risk aversion, not included in their analysis, may also be a factor. 9 In a personal communication with EPA staff, I have been told that, until 2006, account identity numbers represented individual boilers. Therefore, the same unit here means the same boiler.

232

Michael Hanemann

economically unrelated parties), and transactions between related entities.10 In 2007, for example, the EPA deducted 8.9 m allowances from sources accounts to cover their emissions that year. In addition, nearly 4,700 private allowance transfers moving roughly 16.9 m allowances of past, current and future vintages were recorded in the EPA allowance Tracking System. About 9.1 m (54 per cent) were transferred in economically significant transactions. The other 46 per cent were transfers between related entities. The large proportion of transfers between related parties has been a constant feature of the SO2 allowance market. Finally, while there clearly were significant cost savings from allowance trading compared to what would have happened under a command-and-control approach, the evidence suggests that the allowance market was not perfectly efficient and that not all gains from trade were successfully exploited. Carlson et al. (2000) find that, in the first years of Phase I, there were some differences in marginal abatement costs among facilities, and absolute compliance costs could have been reduced further with additional trades. Ellerman et al. (2000) reach a similar conclusion. In addition, plant-level studies of production and abatement efficiency by Coggin and Swinton (1996) and Swinton (2002, 2004) indicate that some plant owners did not take full advantage of the allowance market; they controlled emissions when it would have been cheaper to purchase allowances.

IV. What caused the reduction in SO2 emissions?

The standard theoretical model on which most of the economic analysis of emission markets is based (Montgomery, 1972), is a purely static model. A firm minimizes total cost, which consists of abatement cost plus the cost of procuring emission permits (or minus the revenue from selling permits). Production and abatement technology are given. There is no effective distinction between capital and operating costs. To the extent that capital costs are involved, there is implicitly an instantaneous adjustment of the capital stock, with no allowance for a time lag associated with the turnover in capital. Whether there is a tax or a cap on emissions, firms equate the marginal cost of abatement to the tax or the market price of emission permits. The introduction of a tax or a cap on emissions causes the firms costs to rise. The price increase is transmitted upstream and downstream from the polluting sector, inducing movements along demand and supply curves throughout the economy. It is the price signal that induces a reduction in emissions. The cap on the individual firms emissions has no economic significance per seonly the aggregate cap on emissions affects the outcome, because it determines the equilibrium price in the market for emission permits. As indicated above, what actually happened with SO2 does not exactly match these stylized facts. No price signal was transmitted to electricity users downstream. All of the emission reduction occurred as the result of actions taken by the railroads and the electricity producers themselves. Rather than movements along demand and supply curves, the main mechanism of adjustment was shifts in those curves induced by changes in railroad market structure and unexpected operational innovations in electricity production. The indication is that these innovations went beyond the ability to trade.

The EPA comments that transfers between economically unrelated parties are arms length transactions and are considered a better indicator of an active, functioning market than transactions among the various facility and general accounts associated with a given company (US EPA, 2008, p 12).
10

Cap-and-trade: a sufficient or necessary condition for emission reduction?

233

Then, what motivated the innovations? The market price of SO2 emissions permits may have had some influence but this can hardly be the entire explanation, given that many firms did not resort to the permit market to meet their obligation to the EPA. As noted above, there was a significant degree of autarky. Many firms kept their emissions within their cap and/or relied on unused permits from previous years which they had previously banked. In addition, a substantial number of firms relied on internal transfers of permits rather than arms-length market transactions. Moreover, the evidence of autarkic behaviour by firms is not limited to the SO2 marketthere is similar evidence from other pollutant trading schemes. With the US lead trading programme for gasoline, for example, Kerr and Mare (1998) found a significant amount of non-arms length trading, in the form of internal transfers among different refineries owned by the same company. In their data, 67 per cent of the quantity of lead bought was bought within the same company, and 70 per cent of the lead sold was sold within the same company. Two recent studies of firms confronting the EUs Greenhouse Gas Emission Trading System (EU ETS) also provide evidence of some predisposition to autarky. In a survey of the 500 largest companies worldwide conducted in 2004, 31 per cent of the German firms, 38 per cent of the UK firms, and 61 per cent of the French firms expressed no interest in the EU ETS, due to begin in January 2005, or stated that they will not participate at all (Pinkse, 2008). A survey of Swedish firms in 2006 found that 46 per cent of firms thought they would handle a potential allowance deficit by reducing emissions internally rather than purchasing allowances (Sandoff and Schaad, 2009). In a different context, a recent experimental study of allowance trading by Goeree et al. (2009) finds evidence of a predisposition to autarky when allowances are grandfathered, although not when they are auctioned. Loss aversion is another possible factor that Kreutzer (2006) suggests might have come into play and promoted autarky in the SO2 market. A firm that does not sell allowances when it should do so forgoes a gain; a firm that sells allowances when it should not do so, and then has to buy them back at a higher price, suffers a loss which may weigh more heavily than a gain of the same magnitude. If there is a preference for autarky, or more generally a wish not to bother with the complexity of the market, then the caps on individual firms emissions would themselves drive the reduction in those emissions. The individual caps have, in fact, been identified by some commentators as a major reason for the reduction in SO2 emissions.11 The caps could also drive behaviour if managers have a preference for complying with the law and simply choose to keep emissions within the limit set for them by the EPA. Evidence of a compliance norm has been found in other contexts where firms responses to regulation have been studied. Braithwaite and colleagues interviewed and observed firm managers in a variety of industry sectors, including coal mining, nursing home operations, and pharmaceutical production. 12 In an analysis of those studies, Ayres and Braithwaite (1992) concluded that, in many cases, managers were motivated by social responsibility and sought to comply with the law. Gunningham et al. (2003) found evidence of a compliance norm in their investigation of pulp and paper mills response to environmental regulation in the US, Canada, Australia, and New Zealand. They found that management style was a key explanatory variable, and they identified five styles which they characterized as laggards, reluctant compliers, committed compliers, environmental strategists, and true believers.

11 12

For example, Malloy (2002, p. 549), citing US EPA (2001); also Driesen (1998, p. 318). Braithwaite (1984, 1985) and Braithwaite et al. (1990), cited by Malloy (2003, note 91).

234

Michael Hanemann

Are there any economic reasons why their individual caps would influence firms behaviour? The conventional economic model of emission trading based on cost minimization by a polluting firm holds that the firms optimal emissions are independent of its cap (Montgomery, 1972). However, this result follows directly from the assumption that the marginal cost of obtaining permits through emission trading is constant; it does not hold when the marginal cost varies. Thus, if there are transactions costs that depend nonlinearly on the number of emissions permits purchased, the optimal level of emissions will not be independent of the cap (Stavins, 1995). Nevertheless, transactions costs may not be a sufficiently significant factor in the SO2 market. There are many brokers and other market intermediaries who are active in that market and, in fact, trading costs are considered to be low (Joskow et al., 1998). If one introduces risk aversion into the conventional model of cost minimization, this would cause optimal emissions to depend in part on the firms cap. For example, the cap influences emissions if there is uncertainty regarding whether a permit transaction will be approved which varies nonlinearly with the magnitude of the transaction (Montero, 1997). There can also be uncertainties about future emissions, about future allowance prices, and/or about future programme regulations which would make a firms optimal emissions depend on its cap. The conventional economic model of emission trading represents the firm as a single, unitary decision-maker, with a single objective, namely profit maximization. This has been characterized as the black-box model of the firm (Malloy, 2002). There is an alternative view of the firm as an organization with a multiplicity of actors and certain distinctive internal features. In economics, this view goes back to Coase (1937) and was importantly developed by Williamson (1975). This view also is the central focus of the organizational behaviour literature, where it received a powerful stimulus from Cyert and March (1963). Viewing the firm as an organization opens up additional explanations for why firms might be influenced by the cap set on their emissions. It has been pointed out by von Malmborg (2008) that the decision to stay within ones emissions cap or resort to purchasing emission permits from the market is isomorphic to the make or buy decision discussed by Coase (1937) and analysed further by Williamson (1975, 1981) using a transactions cost approach. In this approach, the question of when firms choose to produce a good or service themselves as opposed to buying it on the market is analysed in terms of three main factors: (i) the degree of uncertainty/complexity inherent in the transaction, (ii) how frequently the transaction is to be made (the transaction density), and (iii) the cost of transaction-specific investments. If a transaction is seen as complex and is marked by uncertainty, which Williamson suggests is commonly the case, it is mainly the other two factors that determine how the firm proceeds. Applying Williamsons logic to the circumstances of a firm enrolled in the EU ETS programme, von Malmborg concludes that, if transaction density and transaction-specific costs are both seen as low, the firm will choose to buy emission permits; if both are seen as high, the firm will engage in in-house emission reduction (the hierarchical solution); if transaction density is seen as high while transaction-specific costs are low, the firm will resort to a bilateral solution, which in this case could be participating in a joint implementation (JI) or clean development mechanism (CDM) project; and, if transaction density is seen as low while transaction-specific costs are high, the firm will resort to a third-party solution, which in this case could be investment in a fund such as the World Bank Prototype Carbon Fund. For electric utilities covered by the SO2 trading programme, transaction density is likely to be seen as high. If transactionspecific costs are high, this argues for the hierarchical solution of in-house emission reduction.

Cap-and-trade: a sufficient or necessary condition for emission reduction?

235

If transaction-specific costs are low, in the absence of a JI or CDM alternative the best bilateral solution might be emission reduction in another facility owned by the same company (trade with a related entity). Thus Williamsons logic could provide an explanation for the observed prevalence of autarkic behaviour in the SO2 market. The behavioural view of the firm as a system for allocating and coordinating organizational resources, such as capital, personnel, and information, can also lead to the conclusion that the emissions cap may be more influential on firm decision-making than the market price of emission permits. Within the firm, there are multiple actors and, quite possibly, multiple objectives. In a typical manufacturing establishment, there is a manager responsible for the purchase of energy services, a manager responsible for product design, and perhaps a manager in charge of product pricing, as well as the CEO. These people have different responsibilities, they face different incentives, and consequently they may not all respond to a given price signal in an identical manner. The consequence is that the firm as an entity may fail to exploit some of the profitable opportunities that are available to it.13 By contrast, a downstream cap on GHG emissions could have more impact on the firms decision-making. Compare an upstream cap versus a downstream cap on, say, the emissions associated with new model vehicles manufactured by Ford for sale in California.14 The upstream cap raises the price of gasoline, which affects both Ford as a consumer of fuel, and also Fords customers. The downstream cap affects Ford more directly, because it limits what new model vehicles Ford can sell. It is not necessarily the case that the same decision-makers within Ford are mobilized to deal with the consequences of the two emissions caps, or that the same corporate response emerges. Because the downstream cap is more direct and visible, it makes abatement more salient for senior managers and is likely to attract their attention more strongly than the price signal triggered by an upstream cap.15 With senior management more engaged, the downstream cap can have a larger, and more rapid, impact on what cars Ford designs and on how it prices them than the upstream cap. Malloy (2002, 2003) has argued that the behavioural view of the firm has important implications for the design of environmental regulatory policies, and casts doubt on some of the conventional arguments for market-based price incentives. Theoretical studies of how firms respond to regulatory incentives, he writes, typically fail to consider the role of attention. Instead, they assume at the outset that all stimuli are created equal in terms of their ability to garner firm attention. However, he argues that the capacity of any organization to process information and to detect and respond to stimuli is necessarily limited: attention
13 There is robust empirical evidence in the organization learning literature and business strategy literature of such failures. They have also been documented in the pollution context by a number of researchers, including DeCanio (1998), King and Lenox (2002), Saele et al. (2005), and Muthulingam et al. (2009). 14 This is how the Pavley Bill (AB 1493) functions. Enacted by California in 2002 and implemented by the Air Resources Board in 2004, it requires a 30 per cent reduction by 2016 in the overall emissions of GHGs of the fleet of new vehicles sold by each manufacturer in California. Implementation required a waiver of the CAA by the US EPA which was withheld under President Bush. The waiver was granted in June 2009. In September 2009, the US EPA and Department of Transportation announced a new national standard for GHG emissions from light- and mediumduty vehicles very similar to Californias standard. 15 Gillingham (2009) has also suggested that, because of considerations of salience, a downstream cap may induce a larger behavioural response from firms than an upstream cap or a carbon tax. However, his argument is slightly different: he suggests that the signal from the price of carbon in an emissions market with downstream caps may be more salient for firms than the signal from an increase in the price of energy resulting from an upstream cap or a tax. He attributes no influence on the firms emissions to the cap itself.

236

Michael Hanemann

itself is a scarce resource that is directed and allocated.16 In a case study of the regulation of toxic emissions from dry-cleaning firms in Southern California, Malloy and Sinsheimer (2004) present evidence that many of these firms appear relatively insensitive to modest but non-trivial opportunities for increasing profit by adopting less costly and less polluting production technologies. The reason, they suggest, is that the novel technology loses out to the conventional technology in the competition for firm resources and managerial attention. Consequently, they argue that direct regulatory intervention can be more effective than incentive-based approaches. In summary, there are both economic and behavioural reasons why an emissions cap could turn out to be a more visible and salient trigger of change in a firms emissions than a price increase alone.

V. SO2 versus CO2

The obvious question is: could not emission trading work just as well for GHGs as it did for SO2? At first glance, GHGs seem a better candidate for emission trading than SO2 because there are no hot spots for GHGs: the environmental consequences of their emission in terms of climate change depend just on the aggregate volume of emissions, regardless of the location where the emissions occur.17 However, there are several other differences between SO2 and GHGs which make cap and trade unlikely to be as effective at accomplishing a large emission reduction for GHGs as it was for SO2. In this section we enumerate the differences. The implications for emission trading are discussed in the following two sections. First, the emission of GHGs, especially CO2, is more widely diffused throughout the economy than for SO2. About 70 per cent of SO2 emissions in the US in 1995 came from the burning of fossil fuel in electricity generation, but only 34 per cent of the GHG emissions in the US today come from electricity generation.18 Nearly as many GHG emissions (28 per cent) come from transportation.19 The reduction of SO2 emissions engaged a relatively small number of decision-makersjust a few hundred power companies accounted for the vast majority of SO2 emissions. However, narrowing the focus of CO2 reduction to the electricity generation sector alone would leave the majority of US emissions untouched hardly a satisfactory outcome. To accomplish a more extensive reduction in CO2 emissions requires somehow engaging with emissions from other sectors, including transportation and buildings, and also forestry and agriculture.20
Malloy (2002, p. 556). Saele et al. (2005) also identify lack of managerial attention as one of the factors preventing organizations from implementing profitable energy-saving measures. 17 The impact depends on the composition of GHGs emitted (with methane, for example, being more potent in the short run than CO2), but not the location. The consensus in the literature is that hot spots do not so far appear to have been a major problem with SO2 allowance trading (Swift, 2000; Burtraw et al, 2005). 18 Electricitys share of CO2 emissions alone is 39 per cent (US EPA, 2009). CO2 accounts for about 85 per cent of the 7.15 billion metric tons (tonnes) of CO2-equivalent GHG emissions in the United States in 2007; but methane, nitrous oxides, and other gases also contributed. About 94 per cent of the CO2 comes from the combustion of fossil fuels, with the rest from changes in land use (deforestation, etc.). The methane comes mainly from landfills and cows. 19 In California, by contrast, electricity (including electricity imported from out of state) accounts for 22 per cent of the CO2 emissions, and transportation accounts for 40 per cent of the emissions. 20 The discussion below addresses transportation and buildings, but not forestry or agriculture.
16

Cap-and-trade: a sufficient or necessary condition for emission reduction?

237

Second, unlike SO2, we cannot rely on existing technologies to achieve the requisite reduction in CO2 emissions. A 50 per cent reduction in SO2 emissions was accomplished virtually overnight without any major technological breakthrough. The emission reduction was accomplished through technologies with which power plants operators had long been familiar. By 1995, scrubbers were a mature technology. Burning low-sulphur coal in boilers designed for high-sulphur coal was a challenge for boiler operators and constituted a significant operational advance, but it was hardly a major technology breakthrough. By contrast, it will be necessary to develop new technologies to achieve the needed reduction in CO2 emissions from burning coal to generate electricity. There is no low- CO2 coal, and there is no end-of-pipe device, like a scrubber, that can be retrofitted in an existing coal-fired power plant. Carbon capture and storage (CCS) has not yet been practised at the scale of a commercial coal-fired generating plant. A handful of small-scale CCS pilot projects and demonstration projects are under way in the US and elsewhere. If all goes well, it is expected that a commercial version of this technology may become available by 2020.21 Construction of new plants would then take several years. However, for existing coal-fired power plants it is believed that retrofitting CCS would not be economically feasible because it would cost virtually as much as building a brand new plant (MIT, 2007). Working with existing power plants, one can reduce CO2 emissions by changing the dispatch order.22 CO2 can also be reduced by co-firing coal with biomass which, since it is renewable, has effectively zero carbon intensity. Until recently it was thought this could be done on a very limited scale (under 5 per cent of biomass). It is now believed this can be done with up to 15 per cent of biomass.23 This may improve further with experience, as happened with the use of low-sulphur coal. But still, it seems unlikely that co-firing with biomass will be as much of a boon as low-sulphur coal was for SO2 reduction. Consequently, unlike with SO2, the only way significantly to reduce CO2 emissions from existing coal-fired plants appears to be to operate them less. As noted earlier, renewable energy generation played no role in the reduction of SO2 emissions. With CO2, by contrast, there is a significant potential for non-fossil fuel electricity generation based on nuclear power and renewables, but it will take some time to realize this potential. Nuclear power plants based on existing designs are extremely expensive and take a very long time to construct, especially in the US. The capital cost is the most important factor determining the competitiveness of nuclear energy. The smallest nuclear power plant that can be built is typically larger than other power plants, making the nuclear option a much larger commitment of funds. With the increase in commodity prices around 2008, the construction cost of all power plants has greatly increased, but the increase has been dramatically larger for nuclear than for coal-fired plants. MIT (2003) had estimated the construction costs for nuclear power, excluding interest (i.e. the overnight cost), at $2,000/kW in 2002 dollars. Du and Parsons (2009) subsequently updated this estimate to $4,000/kW in 2007 dollars; other estimates are even higher.24 Moreover, the recent liberalization of the
Wall Street Journal, The Long Road: Carbon Capture and Storage, 22 February 2010. Electricity from natural gas has about half the CO2 emissions of coal, and electricity from oil is about half way between natural gas and coal. 23 I owe this observation to Dallas Burtraw. 24 By contrast, Du and Parsons (2009) estimate the construction cost of a coal-fired plant at $2,300/kW in 2007 dollars.
21 22

238

Michael Hanemann

electricity markets in the US and some other countries means that the risks of construction delays and cost overruns, and the risk of cheaper power becoming available from other sources, are now borne by plant suppliers and operators rather than consumers. This has made the economics of nuclear power less attractive for utilities. Efforts are under way to develop less costly new designs for nuclear power plants using multiple smaller reactors, but these technologies could take about a decade to be approved and put into operation. In the past decade there have been significant improvements in the technologies for generating electricity from wind and solar power. The cost of wind energy at a good site is now close to being competitive with the cost of electricity from coal and natural gas. Solar thermal power (concentrated solar power) is still more expensive than coal or natural gas, but is coming within range.25 However, wind and solar thermal face two difficulties in competing with coal and natural gasintermittency and location. Wind and solar thermal generate electricity only when the wind blows and the sun shines; both need some form of storage to be fully effective. While it is an active area for research, the storage problem does not yet have a definitive solution. Furthermore, many good wind and solar thermal sites are located a great distance from where electric power is consumed, and therefore need transmission in order to be effective. Construction of the needed transmission will take time and will be expensive.26 While wind and solar thermal hold promise for the future, their market shares are still very small. In the EU27, wind accounts for 5.5 per cent of electricity consumed, and all solar (photovoltaic plus thermal) accounts for 0.8 per cent. In the US, wind accounts for 1.9 per cent of electricity consumed, and solar accounts for less than 0.1 per cent. All forms of renewable electricity, including hydropower, geothermal, and biomass, as well as wind and solar, account for about 9 per cent of the electricity consumed in EU27, and 10.4 per cent in the US. The EU has adopted the target of a 20 per cent share for renewable energy by 2020, and there has been some discussion of a similar target for the US. Even with 20 per cent of electricity generated from renewables in the US and Europe, this would still leave the large majority of GHG emissions from electricity generation untouched. The above discussion highlights a third point of difference between SO2 and CO2the need for replacement of capital stocks in order to accomplish the reduction in CO2 emissions. With SO2, emission reduction was accomplished for the most part with existing capital assets: it was possible to retrofit existing power plants with scrubbers and/or to modify the combustion for low-sulphur coal. There was not a wholesale retirement of capital assets well before the end of their useful lives. By contrast, before there can be a major reduction in CO2 emissions from electricity generation, existing power plants will have to be replaced by nuclear energy or renewable energy or by new coal-fired plants using stateof-the-art technologies, such as supercritical combustion or integrated gasification combined-cycle, and designed from the beginning to incorporate CCS. 27 Similarly with transportation, the other major source of emissions, existing vehicle fleets will have to be
25 For example, Heal (2009), gives the cost of electricity from coal and natural gas as between 6 and 7 cents/ kWh; that of wind (leaving aside transmission cost and intermittency) as between 6 and 10 cents/kWh; and that of solar thermal as about 12 cents/kWh. 26 A US Department of Energy (DOE) report estimates that wind could contribute 20 per cent of US energy by 2030; for this to occur it would be necessary to construct over 12,000 miles of new transmission lines at a cost of approximately $20 billion (US DOE, 2008). 27 The situation is different in developing countries. In China, for example, it is estimated that over two-thirds of the electricity generating capacity in place by 2020 will have been built after 2005. But in the OECD, it is estimated that almost two-thirds of todays generating capacity will still be in operation in 2030 (International Energy Agency, 2003).

Cap-and-trade: a sufficient or necessary condition for emission reduction?

239

replaced before there can be a major reduction in CO2 emissions from transportation. The problem is that today we have the wrong types of power plants, the wrong types of motor vehicles (gas guzzlers rather than highly fuel-efficient smaller vehicles), and the wrong types of urban development, with urban sprawl and limited public transit. It is unlikely that the reduction in GHG emissions required by mid-century can be accomplished with existing capital stocks left mostly unscathed. The fourth difference concerns the role of energy conservation and energy efficiency. As noted earlier, conservation by energy users played no role in the reduction of SO2 emissions. By contrast, conservation by end users and energy efficiency will have a central role in the reduction of CO2 emissions, especially in the near term while we wait for new technologies to be developed and deployed. It will be necessary to promote a degree of behavioural change among energy users throughout a broad swathe of the economy, a far more daunting challenge than was faced when dealing with SO2 emission reduction. Given the differences between what occurred when SO2 emissions were reduced and what will be required in order for CO2 emissions to be reduced, the question arises whether emission trading will be as effective with CO2 as it was for SO2. As explained below, there are some reasons to doubt this.

VI. Emission trading and technological innovation

Since many of the technologies required for GHG reduction are still in their infancy, there is general consensus on the need for some form of research and development (R&D) policy to promote long-term technological change. Because information is a public good, there is an impaired ability to appropriate all the rents from an innovation, and this weakens private incentives to invest in R&D. Hence most economists agree on the need to address this market failure through some government-sponsored climate-oriented R&D policy as a complement to a carbon tax or an emission trading system. Schumpeter famously identified three stages in the process of technological change: invention, innovation, and diffusion. Invention is the first development of a scientifically or technically new product or process, which may involve both basic and applied research. Innovation is accomplished when the new product or process is commercialized, i.e. made available on the market. Diffusion is when the product or process comes to be widely used through adoption by many firms or individuals. In the case of climate change, invention and innovation are the core issuesthe development and commercialization of technologies that do not exist yet or, at best, are still experimental (e.g. CCS). In the pollution control literature there is some limited empirical evidence that more stringent environmental policies lead to invention, as reflected in increased patent activity. Lanjouw and Mody (1996) found a positive correlation between the stringency of environmental policy, measured in terms of pollution abatement expenditures, and the number of environmental patents granted in the US, Japan, and Germany. Using US manufacturing data, Brunnermeier and Cohen (2003) also found a positive relation between pollution abatement expenditures and counts of environment-related patents. In a more extensive international comparison, De Vries and Withagen (2005) analysed patent counts relating to SO2 abatement in the US and 13 European countries over the period 19702000 and found some evidence that strict environmental policies lead to more inventions.

240

Michael Hanemann

A less settled question is how the choice of policy instrument influences the rate of invention. It is sometimes asserted that, by rewarding emission reductions however they are accomplished, a carbon tax or an emission trading system would also provide broad incentives for inventions that lower the cost of emission reduction.28 The actual experience of the promotion of technological change under emission trading for SO2 is more ambiguous. While scrubber operating efficiency increased and scrubbers became cheaper, there does not appear to have been any fundamental change in scrubber technology. The data on patent counts relating to post-combustion SO2 control technology actually show a decline in the number of new patents starting around 1990, and steepening after 2001.29 Moreover, there was no obvious boost to other low-emission technologies for coal combustion, such as integrated gasification combined cycle (IGCC) which results in lower emissions of SO2, particulates, and mercury as well as improved combustion efficiency compared to conventional pulverized coal.30 Developments such as the burning of lowsulphur coal in boilers designed for high-sulphur coal and the increased operating efficiency of scrubbers were primarily refinements in operating practices rather than fundamentally new technologies.31 As noted above, the increased use of low-sulphur coal and the improvement in scrubber operating efficiency can have been due to the fact that the cap functioned as a performance standard rather than to the price signal created by emission trading. Thus, emission trading for SO2 (and also lead) promoted the adoption of technologies that were already available; in Schumpeters terminology, while there was diffusion, there was no invention and no innovation. With NOx, Popp (2006) examines patents over the period 19702000 and shows that, in the US, the rate of patents for Nox-control technologies reached a peak in 1990. However, this is evidence of the impact of regulatory stringency on invention, rather than of emission trading per se. The 1990 Clean Air Act was the beginning of binding regulation for existing stationary sources of NOx. The act significantly tightened limits for NOx emissions from power plants and, unlike the previous legislation in 1970 and 1974, applied those limits to existing as well as new power plants. But, emission trading for NOx came later. In addition to acid rain, ozone was a second focus of concern with NOx in the CAA Amendments. Seeing the need for a regional strategy for the Northeast Corridor to achieve compliance with the ozone standard, the CAA Amendments created the Ozone Transport Commission (OTC). In 1994, the states belonging to the OTC recognized that emission rate standards alone would not be sufficient for ozone compliance and agreed to implement a summertime

For example, Stavins (2007, p. 32). By 2004, the annual number of new patents was less than half the annual average for the period 19751989 (Taylor, 2008). Popp (2003) shows that there is a correlation between the increase in scrubber operating efficiency after 1990 and the cumulative stock of new patents issued after 1990. But it is not clear whether the specific focus of post-1990 patents related to operating efficiency, nor is it necessarily the case that the relationship found by Popp is a causal one. His explanatory variable, the cumulative stock of new patents, may function as similar to a time trend, in which case his regression shows merely that scrubber operating efficiency rose over time during the 1990s. 30 There are currently only two IGCC plants generating power in the US, which started operating in 1995 and 1996; several new IGCC plants are expected to come online in the US around 201220. 31 A similar conclusion applies to the US lead-refining programme which operated from 1982 to 1987. Kerr and Newell (2003) examine a technology known as isomerization used by refineries to replace octane when lead is restricted. The technology was introduced in the late 1960s. By 1980, the cost of isomerization had fallen by about 40 per cent. During the 1980s, the cost fell hardly at all. More than half the adoption of isomerization occurred after 1986, when the phase-out of lead trading had been set. In fact, the flexibility afforded by lead trading may have permitted some refineries to postpone the installation of isomerization from, say, 1985 to 1987.
29

28

Cap-and-trade: a sufficient or necessary condition for emission reduction?

241

trading programme for large sources of NOx, including both power plants and certain industrial facilities. The trading programme commenced operation in 1999. There was a small increase in the rate of patents for NOx control in 1994 (for NOx post-combustion patents) and 1997 (for NOx combustion modification patents), but in each case this was followed by a subsequent decline. Two papers have recently analysed inventions relating to climate change mitigation technologies. Glachant et al. (2009) employ a global patent data set covering 13 classes of technologies with significant potential for GHG emission reduction over the period 19782003. Between 1978 and 1997, patents relating to climate change grew at the same pace as all patents generally. Between 1998 and 2003i.e. after the Kyoto Protocol patents relating to climate change grew much faster than all patents generally. Moreover, the increase in climate-related patents occurred in countries which ratified the Kyoto Protocol but not in the US and Australia, which did not ratify the Protocol.32 One commentator characterized this finding as climate policy does wonders for your green-tech patent count.33 However, this again should be seen as evidence of the impact of regulatory stringency on invention, rather than of emission trading per se. The two categories with by far the largest annual number of patents between 1998 and 2003 were lighting and fuel injection. These were followed respectively by waste (recovery of heat from waste incineration, recovery of waste heat from exhaust gases, and production of energy from waste or waste gases), solar, building efficiency, and wind. Solar and wind patents may be related to investments by power companies that are regulated in Europe through the EU ETS, which was established by a directive in October 2003 and went into effect in January 2005. The other patent categories that experienced unusual growth between 1998 and 2003 are unlikely to be related to emission trading in the EU or elsewhere. It is noteworthy that, between 1998 and 2003, CCS had the smallest annual number of patents of all 13 categories.34 Johnstone et al. (2010) analyse a subset of the same data relating to renewable energy technologies involving wind, solar, geothermal, ocean, and biomass/waste in 25 countries over the period 19782003, Their focus is the effect on the rate of invention of alternative types of policy instruments. Since the data end prior to the institution of EU-ETS, emission trading per se is not one of the instruments considered, but the analysis does cover tradable renewable energy certificates (RECs) in connection with renewable portfolio standards. The econometric analysis shows that RECs induced inventions in wind power, while targeted subsidies, such as feed-in tariffs, induced inventions in solar power. The authors explain this by pointing out that a renewable portfolio standard is a performance standard which leaves electric utilities free to choose any renewable technology; therefore it promotes innovation in a technology such as wind that is close to competitive with fossil fuels. By contrast, solar is more expensive than fossil fuels and therefore requires subsidies in order to promote innovation. In short, the empirical evidence demonstrates that the pace of invention is influenced by the stringency of environmental regulation and perhaps also whether the regulation involves a performance standard, but not necessarily whether it involves emission trading per se.
Australia subsequently ratified it in December 2007. http://www.env-econ.net/2009/02/climate-policy-does-wonders-for-your-greentech-patent-count.html. 34 According to Glachant et al. (2009), annual patents relating to CCS increased sharply in the late 1980s, reaching a peak in 1992, but then fell for 5 years. Since 1997, the level of CCS patents has increased gradually, but in 2005 it was still below the 1992 record high.
33 32

242

Michael Hanemann

Montgomery and Smith (2007) reach a similar conclusion. Commenting that climate policy demands a policy prescription that is effective in stimulating future technology development without imposing inefficiently high costs in the near-term stages of the policy, they concluded that the ability of the cap-and-trade approach to perform in [this] manner has not been demonstrated by any of its previous applications, nor has it been explored adequately in modeling or in theory. There are at least two economic reasons why the price signal generated by emission trading might not be sufficiently powerful, by itself, to promote the type of technological innovation required for the reduction of GHG emissions. One reason, emphasized by Montgomery and Smith, is the problem of: dynamic inconsistency between what governments will announce as a future policy and what governments will subsequently be motivated to adopt as a policy. As a result of this dynamic inconsistency, efforts to address climate change by imposing caps or taxes in the near-term will fail to provide an adequate credible incentive for the R&D necessary to lower the cost of long-term reductions. Additionally, even if the R&D externality is being effectively addressed, implementation today of a cap or tax that will not become stringent until a later date will provide little or no supplemental benefit in the form of an announcement effect.35 Second, much of the existing theoretical literature on R&D and pollution abatement assumes that the firm which does the invention is the same as the firm which causes pollution and invests in abatement. In fact, however, this is generally false: the vast majority of the inventers are not the polluters but rather machinery suppliers and other outside sources.36 The fact that different parties engage in invention, innovation, and pollution abatement creates the possibility of what is known as a coordination problem (Rodrik, 1996); the coordination problem arises because each of the various actors does not know the others expectations and intentions. For an innovation to be successful, these expectations need to be coordinated, especially when the innovation is costly and capital intensive. The polluting firm must be reasonably confident it can recoup the investment if it invests in an expensive new production process that lowers GHG emissions. The equipment supplier, likewise, must be reasonably confident that there will be an industrial market before he invests in commercializing this product. And the venture capitalist must be reasonably confident that the product will be marketable if it is successfully commercialized before he finances the R&D. Given the long time lags and high degree of technological uncertainty, the coordination problem is likely to be especially severe for many GHG control technologies. It is quite possible, therefore, that the price signal from an emission trading programme by itself might not generate sufficient coordination; some additional policy instruments may be required.

Montgomery and Smith (2007, p. 328). The findings of Glachant et al. (2009) may, however, be an indication of some degree of announcement effect associated with the adoption of the Kyoto Protocol. 36 Lanjouw and Mody (1996) estimate that machinery suppliers were the source of about 80 per cent of the patents for the control of industrial air pollution, water pollution, oil spills, and the exploitation of non-fossil-fuel energy sources. Taylor (2008) shows that electric utilities and oil companies accounted for only about 18 per cent of the patents for SO2 control, while 82 per cent of the patents are held by research institutions and, especially, other entities.

35

Cap-and-trade: a sufficient or necessary condition for emission reduction?

243

VII. The need for complementary measures

As noted earlier, not only was a reduction of more than 50 per cent in SO2 emissions accomplished virtually overnight, but the only policy action required of the federal government was the minimalist one of establishing the allowance trading programme under Title IV. The question arises whether the same will hold true for GHG emissionswill a minimalist policy of allowance trading alone secure the desired reduction in GHG emissions? For GHGs, the European Union has adopted a more comprehensive approach including, in addition to emission trading for CO2, measures to promote renewable energy; improvements in energy efficiency in buildings, household appliances, and industry; emissions standards for new passenger cars; and the reduction of methane emissions from landfills. Similarly, the GHG legislation adopted by the US House of Representatives in 2009 (HR 2454, WaxmanMarkey) has provisions relating to a federal renewable portfolio standard, energy efficiency standards for buildings, lighting, appliances, and new motor vehicles, CCS, performance standards for new coal-fired power plants, smart grid deployment, and R&D support for electric vehicle development along with emission trading for CO2. Among the US states, Californias programme to reduce GHG emissions, being copied by its five partner states in the Western Climate Initiative (WCI), involves a suite of complementary measures alongside emission trading, including efficiency standards for motor vehicles, appliances, and buildings; renewable energy standards; a low-carbon standard for transportation fuels; government procurement policies; and a performance standard for new long-term power contracts.37 However, whether there should be complementary measures to deal with GHGs, or simply emission trading alone, is a contested issue. In contrast to the WCI, the Regional Greenhouse Gas Initiative (RGGI), formed by ten north-eastern and mid-Atlantic states, relies solely on cap-and-trade covering power plants to limit CO 2 emissions. 38 While the KerryBoxer bill currently before the US Senate adopts a broad approach similar to that of HR 2454, the other bill before the Senate, the CantwellCollins bill, focuses more narrowly on emission trading for CO2, without any complementary measures,39 and adopts an upstream cap on fossil-fuel carbon as it enters the US economy. Also, many economists have advocated a cap-and-trade system for CO2, or a carbon tax, unaccompanied by complementary measures other than an R&D policy.40 As noted above, the actual experience with emission trading for SO2 and lead deviated in some significant respects from the standard economic model on which the economists recommendation relies. Moreover, as argued above, GHGs differ from SO2 and lead in several important dimensions that make cap and trade alone unlikely to be as effective at accomplishing a large emission reduction for GHGs as it was for them. The key fact about the past experiences with emission trading in the US is that all the emission reduction was accomplished by producers in the capped sector modifying their
For a description of the package of complementary measures adopted by California between 2002 and the present, which created a template for the other members of WCI, see Farrell and Hanemann (2009). In addition to the partner states, there are seven observer states, plus partner and observer provinces in Canada and observer states in Mexico. 38 At least 25 per cent of the allowances have to be auctioned, with auction revenues directed to strategic energy investments. 39 All of the allowances would be auctioned. Of the revenues generated by auctioning permits 25 per cent would be allocated to a Clean Energy Reinvestment Trust which could finance emission reduction activities. 40 For example, Stavins (2007), Metcalf (2007).
37

244

Michael Hanemann

production process or reformulating the product, without any substantial increase in production costs. The reduction in emissions was not due to price increases that choked back demand for the product. The adjustments were on the supply side, not the demand side. The same outcome is not likely to occur with GHGs. With electricity generation, unlike SO2, there is limited scope for reducing GHG emissions from existing fossil-fuel power plants. Therefore, the main focus will have to be on lowering the GHG emissions associated with new power plants. Renewable energy will not assume a large share of the power sectors in the US or EU overnight. The share of nuclear power in those countries will increase, but also not overnight. For political reasons, coal will continue to account for a substantial share of new power plant construction, making the timing of CCS deployment and regulatory pressure for high thermal efficiency combustion technologies important determinants of the pace of GHG reduction in the electricity sector. Since power plants are capital-intensive and long-lived investments, it is doubtful whether the price signal from an emissions market will, by itself, be sufficiently powerful to produce a decisive shift in the selection of future generating capacity away from fossil fuels or, in the case of fossil fuels, towards technologies embodying high thermal efficiency combustion and pre-equipped for CCS. Not only an R&D programme but also complementary measures, whether performance standards for new coal-fired plants, feed-in tariffs, purchase commitments, subsidies, or loan guarantees will surely be required. Moreover, whereas demand-side management played no role in the reduction of SO2 emissions, it seems clear that it will have to assume a significant role in order to attain 2020 GHG emission targets. Together, residential and commercial buildings account for 70 per cent of electricity consumption in the US. Approximately 38 per cent of the CO2 emissions in the US comes from the energy use associated with these buildings.41 The largest component of the CO2 emissions comes from the generation of the electricity used in them (71 per cent), but emissions also arise from the direct combustion of natural gas and petroleum, especially fuel oil. The major uses of this energy include space heating and cooling, lighting, and water heating.42 Thus, an effective strategy for GHG mitigation must identify options for reducing these emissions, including both reducing emissions from the current building stock and also tackling the buildings that will be constructed in the future. Some of the options for reducing GHG emissions entail higher costs and investments; but others, especially those focused on increased energy efficiency, could yield net savings. Major opportunities for improvements in end-use efficiency include space heating (especially residential), air conditioning, lighting (especially in commercial buildings), and water heating (especially in residences). There are several reasons why it is doubtful whether the price signal from an emissions market will, by itself, be sufficiently powerful to induce the improvements in end-use efficiency and the reduction in CO2 emissions. One factor is the sheer fragmentation of decision-making, especially in the residential sector. While about 500 firms control the operation of the electric power industry in the US, there are over 110 m occupied residential

Brown et al. (2005). Twenty-one per cent of the CO2 emissions result from residential buildings and 17 per cent from commercial buildings. In addition, industrial buildings account for another 5 per cent of CO2 emissions. 42 In the residential sector, 30 per cent of the energy consumed is for space heating, 12 per cent is for water heating, 12 per cent is for lighting, and 11 per cent is for air conditioning; the remainder goes for appliances, electronics, and other purposes. In the commercial sector, the breakdown is 21 per cent for lighting, 12 per cent for space heating, 9 per cent for air condition, and 8 per cent for office equipment; the rest goes for water heating, refrigeration, and other purposes (Brown et al., 2005).

41

Cap-and-trade: a sufficient or necessary condition for emission reduction?

245

housing units in the US. Of these, about 75 m are owner occupied and 35 m are renter occupied. In this context, the principalagent problem is a significant obstacle to investment in energy-efficiency improvements. The problem arises whenever the energy user does not pay for the cost of his energy use and/or does not choose the energy-using appliances. The latter is not limited to rentersit can also apply to owner-occupiers.43 First, in an existing home, the building insulation, the windows, the space-heating equipment, and perhaps, to a lesser extent, some other appliances are likely to be pre-determined and changeable only at substantial cost. Second, with the significant mobility of the US population, long-lived investments in improving energy-efficiency may seem a risky proposition to home owners who do not know how long they will continue to live in the home and are not sure that they will recoup the investment if they sell it.44 In addition to the principalagent problem, there are behavioural impediments to investment by individual end-users in energy efficiency. One impediment is the high initial capital costs of energy efficiency investments and the inability or unwillingness of consumers to finance these, even though there is a subsequent stream of operating cost savings. Empirical studies have demonstrated cases of high implicit individual discount rates,45 which could reflect high rates of time preference or, alternatively, the influences of illiquidity or uncertainty and risk aversion. Whatever the cause, this is not an instance of market failure. But, it is a market opportunity. The situation could be remedied if a market intermediary were to come forward who was satisfied with conventional rates of return, and was therefore willing to finance the up-front capital costs of energy-efficiency investments in return for a guaranteed share of the future stream of energy cost savings. There is a missing actor who could arbitrage the differential between the rate of return on energy efficiency investments and homeowners high individual discount rates. To plug this gap, California and 15 other states have now enacted legislation authorizing local governments to form Energy Financing Districts (EFDs) which use bond or other funds to finance residential or commercial energy-efficiency investments and are then repaid over a set number of years through a special assessment on the property tax bills of those property owners who elect to participate (Fuller et al., 2009). Another behavioural impediment for individual end-users is the lack of information about the potential opportunities for cost-saving energy-efficiency investments. Perhaps more important is their lack of attention to this potential, and its lack of salience for them. Salience is defined in psychology as the property of a stimulus that causes it to stand out and attract attention in its context (Fiske and Morling, 1996). This definition comprises three elements.
Davis (2009) compares appliance ownership patterns between homeowners and renters using US household level data and finds that, controlling for household income and other household covariates, renters are significantly less likely to have energy-efficient refrigerators, clothes washers, and dishwashers. In the US, leasing/rentals are a larger proportion of commercial than residential buildings, and principalagent problems arising from leasing/ renting are therefore likely to be more widespread for commercial property. 44 The incomplete capitalization of energy-efficiency improvements in house values is noted by Dubin (1993). It has been estimated that, in the US, the principalagent problem is a potential obstacle to energy efficiency affecting up to 25 per cent of the energy usage associated with residential refrigerators, 47 per cent of the energy usage associated with residential space heating, and 77 per cent of the energy usage associated with residential water usage (ACEEE, 2007). 45 For example, Hausman (1979) and Meier and Whittier (1983). It should be noted that the behavioural impediments to the adoption of energy-efficiency investments are not limited to householdsthey are also found in some businesses and other organizations. Muthulingam et al. (2009) find evidence that high up-front costs and a short planning horizon (economic short-termism) lead managers to overlook profitable opportunities for energy savings.
43

246

Michael Hanemann

The stimulus has to be vivid enough to be noticed by the recipient; it has to be important in the light of her goals and interests; and whether it is vivid or important depends, in part, on the particular context. There is a large literature in political science, marketing, finance, and, now, fields such as behavioural economics and public finance demonstrating that salience influences consumer behaviour. For example, Chetty et al. (2007) analyse the effect of salience on consumers responses to prices in a supermarket. They find that when commodity taxes are included in the posted price that consumers see when shopping (and thus are more salient) they have a much larger effect on consumer demand. Similarly, Gallagher and Muehlegger (2008) show that more salient sales tax waivers given at the time of purchase have seven times as large an effect on hybrid vehicle purchases as income tax credits of an equivalent amount.46 In the terminology of Gabaix and Laibson (2006), such studies show that prices may often be shrouded attributes.47 The past experience with emission trading suggests that regulatory measures can increase the salience of emission reduction for decision-makers, thereby reinforcing the effects of the price signal emanating from emission trading. The second largest source of GHG emissions in the US is transportation which, as noted earlier, accounts for 28 per cent of emissions. Moreover, transportation is the fastest growing component of GHG emissions in the US.48 Highway vehicles account for 72 per cent of transportation energy use and CO2 emissions in the US. Within the highway mode, lightduty vehicles (passenger cars and light trucks) account for 75 per cent of the highway energy use (Greene and Schafer, 2003). There are more than 254 m registered passenger vehicles in the US. Thus, decision-making at the level of the individual end-user is tremendously fragmented, and subject to the same behavioural impediments as with energy use in buildings. The response to gas prices can shed some light on how passenger vehicle owners might respond to the price signal from an upstream tax or cap on carbon. There is no doubt that a very large price increase is likely to be both visible and salient to consumers. Thus, in the summer of 2008 there was a significant consumer response to the run-up in gas prices. Retail gas prices in the US rose from an average of about $2.92/gallon in JuneAugust 2007 to an average of $3.91 in JuneAugust 2008, an increase equivalent to a carbon price of about $113/CO2 ton. While gas prices rose by 34 per cent, vehicle miles travelled in the US fell during this period by about 4.6 per cent. The implicit short-run elasticity was therefore about 0.14. This is notably larger than the conventional estimate of the short-run price elasticity of demand for gasoline, which the Congressional Budget Office (2008) sets at 0.06. The higher elasticity in 2008 reflects the response to the large price increase.49 What would happen with the sort of carbon price envisioned in the current US debate on emission trading? The CantwellCollins bill in the Senate sets a price cap starting at $21 per ton of CO2 in 2012; the KerryBoxer bill sets a price cap of $28/ton in 2012. The latter figure corresponds to an increase of about 25 cents in the price of a gallon of gasoline. The current retail price of gasoline in the US in March 2010 is about $2.75/gallon. How substantial a reduction in gasoline consumption would be induced by a 9 per cent price
Chetty (2009) provides a number of additional examples. The literature in political science and marketing emphasizes the importance of salience for attributes of commodities or political issues as well as their price. Salience determines which attributes are considered by voters or consumers, and which ignored. 48 While the share is smaller in other countries, it is growing quite rapidly with the increased ownership of automobiles. 49 There is a parallel with the 1973 oil price embargo. Goel and Morey (1993) estimate that the price elasticity of demand in the US during the period 197380 was more than double the mean elasticity during 195273.
47 46

Cap-and-trade: a sufficient or necessary condition for emission reduction?

247

increase? Such an increase is well within the normal seasonal variation in the price of gasoline. In 2005, for example, before the major run-up in gas prices, the average weekly retail price of regular gasoline in the US varied from as low as $1.75/gallon, in the first week of January, to as high as $3.04, in the first week of September; the average for the year was $2.24. In that context, an increase of 25 cents per gallon due to a carbon price of $28/CO2 ton is unlikely to be visible, or salient, to most motorists. There is a striking contrast between the end-user decision-making on fuel use for passenger vehicles versus for commercial vehicles, commercial aviation, and freight transportation. For airlines and trucks, fuel is the second largest cost of operation after payroll; during 2008, it became the largest cost item. Therefore it is both visible and salient to them. Over the past decade or more, airlines and trucking companies appointed fuel managers whose task is to find and implement ways to raise fuel efficiency and reduce fuel use. For example, following the 2008 increase in fuel prices, airlines reprogrammed flight trajectories and flight paths to shave energy use, overriding decision making by individual pilots.50 Similarly, trucking companies have installed speed controllers in their trucks. Among the US corporations ranked most highly for their GHG reductions are the shippers DHL and UPS.51 UPS operates the largest private alternative-fuel fleet in the US. DHL has set a corporate goal to improve the CO2 efficiency of its worldwide operations by 10 per cent in 2012, relative to 2008, and by 30 per cent in 2020.52 In short, while a 9 per cent price change is neither visible nor salient to individual motorists, even a 2 per cent cost savings is salient to managers of large trucking operations and can induce a behavioural response from them. However, individual motorists account for at least 54 per cent of GHG emissions in the transportation sector. This necessarily limits what response can be expected to the price signal from emission trading if unaccompanied by complementary measures targeted at the transportation sector. Besides the end-users, two other actors influence the GHG emissions from transportation, namely the vehicle manufacturers and the producers of motor fuels. Fuel efficiency and the carbon intensity of motor fuels are likely to be salient for them. Compared to a 25 cent increase in the price of gasoline, it seems likely that more emission reduction would be accomplished with a cap on automobile manufacturers based on the emissions of their annual fleet of new vehicles, and a low carbon fuel standard (LCFS) for suppliers of motor fuels, both of which have been adopted by California.53 Because of limited attention, limited salience, and bounded rationality on the part of many individual end-users of energy in the transportation sector and, in addition, the principalagent problems in the building sector, it seems questionable whether the price signal from a carbon tax or an upstream cap on CO2 emissions will, by itself, motivate significant
50 For example, a recent story in the Wall Street Journal reports: By carefully planning flight paths and improving communication among air controllers, carriers and airports, major airlines should be able to reduce fuel consumption by up to 2 per centor many thousands of tonson flights between American and Europe (10 March 2010). 51 http://www.climatecounts.org/scorecard_sectors.php?id=28. 52 Besides truck operations, both companies are actively working to increase fuel efficiency and reduce GHG emissions in the operation of their aviation fleets. 53 The LCFS is sometimes characterized as a politically expedient substitute for a carbon tax which does not visibly affect fuel prices (Holland et al., 2009). However, its real significance is probably as a device targeted at the major oil companies to discourage them from making an irreversible commitment to the development of highly CO2-intensive oil from Canadian tar sands. Holland et al. criticize the LCFS using a model that has no fixed costs or investment requirements for low- or high-carbon fuels. They find that an optimal outcome can be obtained without a LCFS by just relying on emission trading. But, their analysis overlooks the irreversible investment decision facing oil companies.

248

Michael Hanemann

reductions in CO2 emissions in those sectors. Rather, complementary measures such as building efficiency standards, appliance efficiency standards, GHG emission limits for new vehicles, and an LCFS, taken in combination with a cap-and-trade programme, may induce more decisive actions to reduce emissions in those sectors.

VIII. Conclusions
While it is essential that there be a price on pollution emissions, the evidence from SO2 trading suggests that its success was due not only to the price signal that it created but also to the fact that it imposed a stringent but flexible performance standard on the firms that were regulated. Another part of its success was due to the fact that it impacted a small number of firmsfewer than 500 power companiesfor whom it was highly salient. By contrast, GHG emissions are widely diffused throughout the economy. A substantial reduction in GHG emissions can be achieved only through the actions of millions of firms and households, for many of whom the price signal generated by a carbon tax or an (upstream) cap-and-trade scheme is unlikely, by itself, to be visible or salient. To be sure, the downstream cap-and-trade scheme that I advocate will be more costly to implement, and it will necessarily be less complete, than a carbon tax or an upstream capand-trade scheme. In addition, some of the complementary regulatory measures may be costly in terms of administrative costs and also short-run economic efficiency before the requisite behavioural and technological changes are induced. Others, especially those targeted at improving energy efficiency and promoting technological innovation, are likely to be profitable and to increase economic welfare. The regulatory measures are required to overcome the impediments caused by bounded rationality and principalagent problems in end-using sectors such as transportation and buildings. Over time, attitudes and perceptions will surely change, and new technologies will emerge. Some of the complementary measures that are required now will become unnecessary, perhaps even counterproductive, and will need to be modified or eliminated. Moreover, emission trading may lead to some surprise outcomes, as occurred with SO2. Nevertheless, the empirical evidence suggests that, at present, an upstream cap-and-trade scheme unaccompanied by complementary measures is unlikely to be adequate to attain the reduction in GHG emissions desired by policy-makers.

References
ACEEE (2007), Quantifying the Effects of Market Failures in the End-use of Energy, American Council for an Energy-Efficient Economy, Washington, DC, report prepared for the International Energy Agency. Ayres, I., and Braithwaite, J. (1992), Responsive Regulation: Transcending the Deregulation Debate, Oxford, Oxford University Press. Braithwaite, J. (1984), Corporate Crime in the Pharmaceutical Industry, London and Boston, Routledge & Kegan Paul. (1985), To Punish or Persuade: Enforcement of Coal Mine Safety, Albany, State University of New York Press. Makkai, T., Braithwaite, V., Gibson, D., and Ermann, D. (1990), The Contribution of the Standards Monitoring Process to the Quality of Nursing Home Life: A Preliminary Report, Canberra, Department of Community Services and Health.

Cap-and-trade: a sufficient or necessary condition for emission reduction?

249

Brown, M. A., Southworth, F., and Stovall, T. K. (2005), Towards a Climate-friendly Built Environment, Pew Center on Global Climate Change. Brunnermeier, S. B., and Cohen, M. A. (2003), Determinants of Environmental Innovation in US Manufacturing Industries, Journal of Environmental Economics and Management, 45, 27893. Burtraw, D. (1996), The SO2 Emissions Trading Program: Cost Savings without Allowance Trades, Contemporary Economic Policy, 14(April), 7994. Evans, D. A., Krupnick, A., Palmer, K., and Toth, R. (2005), Economics of Pollution Trading for SO2 and NOx, Annual Review of Environment and Resources, 30, 25389. Carlson, C. P., Burtraw, D., Cropper, M., Palmer, K. (2000), SO2 Control by Electric Utilities: What are the Gains from Trade?, Journal of Political Economy, 108(6), 1292326. Chao, H.-P., and Wilson, R. (1993), Option Value of Emission Allowances, Journal of Regulatory Economics, 5, 23349. Chetty, R. (2009), The Simple Economics of Salience and Taxation, NBER Working Paper 15246, Cambridge, MA. Looney, A., and Kroft, K. (2007), Salience and Taxation: Theory and Evidence, NBER Working Paper 13330, Cambridge, MA. Coase, R. H. (1937), The Nature of the Firm, Economica, 4(16), 386405. Coggin, J. S., and Swinton, J. R. (1996), The Price of Pollution: A Dual Approach to Valuing SO2 Allowances, Journal of Environmental Economics and Management, 30(1), 5872. Congressional Budget Office (2008), Effects of Gasoline Prices on Driving Behavior and Vehicle Markets, Congressional Budget Office, Washington, DC. Convery, F. J. (2009), Origins and Development of the EU ETS, Environmental and Resource Economics, 43, 391412. Crocker, T. D. (1966), The Structuring of Atmospheric Pollution Control Systems, in H. Wolozin (ed.), The Economics of Air Pollution, New York, Norton, 6186. Cyert, R. M., and March, J. G. (1963), A Behavioral Theory of the Firm, Englewood Cliffs, NJ, Prentice Hall. Dales, J. H. (1968), Pollution, Property and Prices, Toronto, University of Toronto Press. Davis, L. W. (2009), Evaluating the Slow Adoption of Energy Efficiency Investments: Are Renters Less Likely to Have Energy Efficient Appliances?, Working Paper, University of California, Berkeley, Haas School of Business. DeCanio, S. J. (1998), The Efficiency Paradox: Bureaucratic and Organizational Barriers to Profitable Energy-Saving Investments, Energy Policy, 26(5), 44154. De Vries, F. P., and Withagen, C. (2005), Innovation and Environmental Stringency: The Case of Sulfur Dioxide Abatement, CentER Discussion Paper 2005-18. Driesen, D. M. (1998), Is Emissions Trading an Economic Incentive Program?: Replacing the Command and Control/Economic Incentive Dichotomy, Washington & Lee Law Review, 59, 289350. Du, Y., and Parsons, J. E. (2009), Update on the Cost of Nuclear Power, Working Paper CEEPR 09-004, Massachusetts Institute of Technology. Dubin, J. A. (1993), Market Barriers to Conservation: Are Implicit Discount Rates Too High?, in Proceedings of a POWER Conference: The Economics of Energy Conservation, University of California Energy Institute, 2133. Ellerman, A. D. (2003), Ex Post Evaluation of Tradable Permits: the US SO2 Cap-and-Trade Program, Working Paper CEEPR 03-003, Massachusetts Institute of Technology. Dubroeucq, F. (2004), The Sources of Emission Reductions: Evidence from US SO2 Emissions from 1985 through 2002, Working Paper CEEPR 04-001, Massachusetts Institute of Technology. Montero, J.-P. (2007), The Efficiency and Robustness of Allowance Banking in the US Acid rain Program, The Energy Journal, 28(4), 4771. Joskow, P. L., Schmalensee, R., Montero, J.-P., and Bailey, E. (2000), Markets for Clean Air: The US Acid Rain Program, Cambridge, Cambridge University Press. Farrell, A. E., and Hanemann, W. M. (2009), Field Notes on the Political Economy of California Climate Policy, ch. 5 of H. Selin and S. Van Deveer (eds), Changing Climates in North America, Cambridge, MA, MIT Press, 87109. Fiske, S. T., and Morling, B. A. (1996), Salience, in A. S. R. Manstead and M. Hewstone (eds), The Blackwell Encyclopedia of Social Psychology, Oxford, Blackwell.

250

Michael Hanemann

Fuller, M. C., Kunkel, C., and Kammen, D. M. (2009), Guide to Energy Efficiency & Renewable Energy Financing Districts for Local Governments, University of California, Berkeley, Renewable and Appropriate Energy Laboratory. Gabaix, X., and Laibson, D. (2006), Shrouded Attributes, Consumer Myopia, and Information Suppression in Competitive Markets, Quarterly Journal of Economics, 121(2), 50540. Gallagher, K. S., and Muehlegger, E. (2008), Giving Green to Get Green: Incentives and Consumer Adoption of Hybrid Vehicle Technology, John F. Kennedy School of Government Faculty Research Working Paper RWP08-009, Cambridge, MA, Harvard University. Gillingham, K. (2009), Salience in Upstream versus Downstream Cap-and-Trade, paper presented at the 32nd International Association for Energy Economics International Conference, San Francisco, CA, 23 June. Glachant, M., Dechezlepretre, A., Hascic, I., Johnstone, N., and Meniere, Y. (2009), Invention and Transfer of Climate Change Mitigation Technologies on a Global Scale: A Study Drawing on Patent Data, Working Paper 343, Fondazione Eni Enrico Mattei. Goel, R. K., and Morey, M. J. (1993), Effects of the 1973 Oil Price Embargo, Energy Economics 3948. Goeree, J. K., Holt, C. A., Palmer, K., Shobe, W., and Burtraw, D. (2009), An Experimental Study of Auctions versus Grandfathering to Assign Pollution Permits, Institute for Empirical Research in Economics Working Paper No. 429, University of Zurich. Greene, D. L., and Schafer, A. (2003), Reducing Greenhouse Gas Emissions from US Transportation, Pew Center on Global Climate Change. Gunningham, N., Kagan, R. A., and Thornton, D. (2003), Shades of Green: Business, Regulation and Environment, Stanford, CA, Stanford University Press. Hanemann, M. (2009), The Role of Emission Trading in Domestic Climate Policy, The Energy Journal, 30(2), 73108. Hausman, J. A. (1979), Individual Discount Rates and the Purchase and Utilization of Energy-Using Durables, The Bell Journal of Economics, 10(1), 3354. Heal, G. (2009), The Economics of Renewable Energy, National Bureau of Economic Research Working Paper 15081. Holland, S. P., Hughes, J. E., and Knittel, C. R. (2009), Greenhouse Gas Reductions under Low Carbon Fuel Standards?, American Economic Journal: Economic Policy, 1(1), 10646. International Energy Agency (2003), World Energy Investment Outlook: 2003 Insights, Paris. Johnstone, N., Hascic, I., and Popp, D. (2010), Renewable Energy Policies and Technological Innovation: Evidence Based on Patent Counts, Environmental and Resource Economics, 45, 13355. Joskow, P. L., Schmalensee, R., and Bailey, E. M. (1998), The Market for Sulfur Dioxide Emissions, American Economic Review, 88(4), 66985. Kerr, S., and Mare, D. (1998), Transaction Costs and Tradable Permit Markets: The United States Lead Phasedown, Working Paper. Motu Economic and Public Policy Research Trust, New Zealand, available at SSRN: http://ssrn.com/abstract=1082596. Newell, R. G. (2003), Policy-induced Technology Adoption: Evidence from the US Lead Phasedown, Journal of Industrial Economics, 51(3), 31743. King, A., and Lenox, M. (2002), Exploring the Locus of Profitable Pollution Reduction, Management Science, 48(2), 28999. Kreutzer, J. (2006), Cap and Trade: A Behavioral Analysis of the Sulfur Dioxide Emissions Market, New York University Annual Survey of American Law, 62, 12562. Lanjouw, J. O., and Mody, A. (1996), Innovation and the International Diffusion of Environmentally Responsive Technology, Research Policy, 25, 54971. Malloy, T. F. (2002), Regulating by Incentives: Myths, Models and Micro-Markets, Texas Law Review, 80, 531605. (2003), Regulation, Compliance, and the Firm, Temple Law Review, 76, 451531. Sinsheimer, P. (2004), Innovation, Regulation and the Selection Environment, Rutgers Law Review, 57, 183233. Meier, A. K., and Whittier, J. (1983), Consumer Discount Rates Implied by Purchases of Energy-efficient Refrigerators, Energy, 8(12), 95762.

Cap-and-trade: a sufficient or necessary condition for emission reduction?

251

Metcalf, G. E. (2007), A Proposal for a US Carbon Tax Swap: An Equitable Tax Reform to Address Global Climate Change, The Hamilton Project Discussion Paper 200712, Washington, DC, Brookings Institution. MIT (2003), The Future of Nuclear Power, Massachusetts Institute of Technology, Cambridge, MA. (2007), The Future of Coal: Options for a Carbon-Constrained World, Massachusetts Institute of Technology, Cambridge, MA. Montero, J.-P. (1997), Marketable Pollution Permits with Uncertainty and Transaction Costs, Resource & Energy Economics, 20, 2750. Montgomery, W. D. (1972), Markets in Licenses and Efficient Pollution Control Programs, Journal of Economic Theory, 5, 395418. Smith, A. E. (2007), Price, Quantity, and Technology Strategies for Climate Change Policy, in M. E. Schlesinger, H. S. Kheshgi, J. Smith, F. C. de la Chesnaye, J. M. Reilly, T. Wilson, and C. Kolstad (eds), Human-induced Climate Change: An Interdisciplinary Assessment, Cambridge, Cambridge University Press, 32842. Muthulingam, S., Corbett, C. J., Benartzi, S., and Oppenheim, B. (2009), Managerial Biases and Energy Savings: An Empirical Analysis of the Adoption of Process Improvements Recommendations, Working Paper, University of California, Los Angeles, Anderson School of Management. Pinkse, J. (2008), The Implementation of Emissions Trading in Companies, in R. Antes et al. (eds), Emission Trading, Springer, 193208. Popp, D. (2003), Pollution Control Innovations and the Clean Air Act of 1990, Journal of Policy Analysis and Management, 22(4), 64160. (2006), International Innovation and Diffusion of Air Pollution Control Technologies: The Effects of NOx and SO2 Regulation in the US, Japan, and Germany, Journal of Environmental Economics and Management, 51, 4671. Rodrik, D. (1996), Coordination Failure and Government Policy: A Model with Applications to East Asia and Eastern Europe, Journal of International Economics, 40(12), 122. Saele, H., Naesje, P., Nordvik, H., and Hagen, O. (2005), What Prevents Organizations from Implementing Energy Saving Measures? Case Studies of Norwegian Public and Commercial Companies, European Council for an Energy Efficient Economy, 2005 Summer StudyWhat Works and Who Delivers, 10719. Sandoff, A., and Schaad, G. (2009), Does EU ETS Lead to Emission Reductions Through Trade? The Case of the Swedish Emissions Trading Sector Participants, Energy Policy, 37, 396777. Stavins, R. N. (1995), Transaction Costs and Tradeable Permits, Journal of Environmental Economics and Management, 29, 13348. (2007), A US Cap-and-Trade System to Address Global Climate Change, Hamilton Project Discussion Paper 200713, Washington, DC, The Brookings Institution. (2008), Addressing Climate Change With a Comprehensive US Cap-and-Trade System, Oxford Review of Economic Policy, 24(2), 298321. Swift, B. (2000), Allowance Trading and SO2 Hot Spots: Good News from the Acid Rain Program, Environment Reporter, 31(19), 9549, Washington, DC, BNA Inc. Swinton, J. R. (2002), The Potential for Cost Savings in the Sulfur Dioxide Allowance Market: Empirical Evidence from Florida, Land Economics, 78(3), 390404. (2004), Phase I Completed: An Empirical Assessment of the 1990 CAAA, Environmental and Resource Economics, 27, 22746. Taylor, M. R. (2008), Induced Technical Innovation under Cap-and-trade Abatement Programs, Working Paper, Goldman School of Public Policy, University of California, Berkeley. US DOE (2008), 20% Wind Energy by 2030: Increasing Wind Energy's Contribution to US Electricity Supply, US Department of Energy, DOE/GO-1020082567, Washington, DC. US EPA (2001), The United States Experience with Economic Incentives for Protecting the Environment, EPA-240-R-01-001, Washington, DC. (2008), Acid Rain and Related Programs 2007 Progress Report, US Environmental Protection Agency, Washington, DC. (2009), Inventory of US Greenhouse Gas Emissions and Sinks: 19902007, US Environmental Protection Agency, EPA 430-R-09004, Washington, DC.

252

Michael Hanemann

Vine, E. (2003), Opportunities for Promoting Energy Efficiency in Buildings as an Air Quality Compliance Approach, Energy, 28, 31941. Von Malmborg, F. (2008), Corporate Strategy and the Kyoto MechanismInstitutional and Transaction Cost Perspectives, in R. Antes et al. (eds), Emission Trading, Springer, 120934. Williamson, O. E. (1975), Markets and Hierarchies: Analysis and Antitrust Implications, New York, The Free Press. (1981), The Economics of Organization: The Transaction Cost Approach, American Journal of Sociology, 87(3), 54877.