|
Article Excerpt
Tradable permit programs allow firms greater flexibility in reducing emissions than command-and-control regulations and encourage firms to use low cost abatement options, including small-scale modifications to capital equipment. This paper shows that firms have extensively modified capital equipment in the Nitrogen Oxides Budget Trading Program, which covers power plants in the eastern United States. The empirical strategy uses geographic and temporal features of the program to estimate counterfactual emissions, finding that modifications have reduced emission rates by approximately 10-15 percent. The modifications would not have occurred under command-and-control regulation and have reduced regulatory costs.
1. INTRODUCTION
Over the last 20 years, regulators have instituted tradable permit programs to reduce emissions from the electric power sector. In the earlier approach, command-and-control, regulators set emission rate standards and mandated specific technologies. Under a tradable permit program, on the other hand, firms must submit permits to cover emissions and they may sell excess permits or purchase permits in the event of a shortfall.
One of the motivations for using permit programs is that they create stronger incentives for innovation and technology adoption than command-and-control regulation (e.g., Milliman and Prince, 1989 and Jung, Krutilla and Boyd, 1996). (1) Consistent with the theory, recent empirical research has found evidence for technological improvements in the Acid Rain Program (ARP) for sulfur dioxide (e.g., Popp, 2003, Lange and Bellas, 2005 and Keohane, 2005). (2)
The empirical work has focused on technological improvements to large-scale, add-on, capital equipment, such as scrubbers. But economists have widely recognized that a potential benefit of permit programs is that they encourage small-scale changes to existing capital equipment, referred to as modifications (e.g., Burtraw, 1996). For example, in the nitrogen oxides (N[O.sub.x]) permit program for U.S. power plants, firms may adjust the timing and location of air injection in the boiler to reduce the combustion temperature and emissions. Firms will use such modifications if the cost is less than the permit price. By setting an emission rate standard, on the other hand, command-and-control does not create such incentives. Therefore, modifications could reduce compliance costs of a permit program, relative to command-and-control. There is anecdotal evidence of modifications in the ARP (Ellerman et al., 2000) and a N[O.sub.x] tax in Sweden (Isaksson, 2005), but there is no quantitative evidence of how important modifications are in reducing aggregate emissions or compliance costs.
The lack of empirical work on modifications largely reflects two difficulties. First, modifications are seldom observed in the available data. Second, because a firm's decision to use a modification depends on a power plant's characteristics, many of which are unobserved, there is rarely a valid control group from which counterfactual emissions can be estimated.
This paper shows that the major U.S. N[O.sub.x] permit program, the Nitrogen Oxides Budget Trading Program, has caused modifications that significantly reduce emissions, but which were not widely used under command-and-control. Section 2 describes the modifications in more detail, which include operating modifications to the combustion equipment. The empirical analysis overcomes the first challenge by focusing on boilers at which the firm does not install any capital equipment after the program begins. Therefore, all else equal, a decline in emission rates must be due to modifications.
The second empirical difficulty is addressed by using the geographic and temporal coverage of the program to construct a control group that is observably quite similar to the treatment group. The permit program caps summer emissions (May-September) and mainly includes power plants. The program first covered the Northeast, from 1999-2003, and expanded in 2004 to include the Midwest and Southeast. The empirical strategy is to compare the emission rates of northeastern boilers with the emission rates of boilers in the Midwest and Southeast when only the northeastern boilers are in the permit program (1999-2003). A key feature of the sample is that units in the control group have the same compliance strategy in 2004-2006, i.e., no additional abatement equipment, as the treatment group.
A comparison of emission rates across regions could still be biased by unobserved regional shocks to emissions. In particular, other N[O.sub.x] regulations resulted in downward emissions trends before and during the permit program. The fact that the program covers summer emissions provides a convenient framework to implement a difference-in-difference estimator that controls for regional shocks. Specifically, I compare the summer/non-summer difference in emission rates for northeastern boilers with the corresponding difference for the control group. (3)
The main results are that firms have used modifications to reduce emission rates by approximately 10-15 percent under the program, and that the modifications appear to have been widely adopted in the estimation sample. The results have a number of implications. Because ex ante predictions of the program's cost did not consider modifications (e.g., Burtraw et al., 2001), these cost estimates were too high. Harrington, Morgenstern and Nelson (2000) find that ex ante cost estimates have been higher than ex post estimates for many environmental regulations, and the results in this paper suggest that modifications may explain some of this trend. Furthermore, modifications may have contributed to unexpectedly low permit prices in recent years, and similar modifications should occur when the program expands in 2009 under the Clean Air Interstate Rule. Finally, and more generally, the paper provides evidence of the importance of the greater flexibility of permit programs than command-and-control, which has long been emphasized in the theoretical literature, but has not been demonstrated empirically.
This paper builds on the literature on environmental regulation and technology in the electricity sector. (4) Popp (2006b) uses patent data from 1970-2000, and finds that the 1990 Clean Air Act Amendments (CAAA) caused an increase in patents for combustion and post-combustion N[O.sub.x] controls (Section 3 describes these technologies). He does not compare command-and-control with the permit program, however.
Popp (2006a) shows that the N[O.sub.x] permit program and the stock of N[O.sub.x]-related patents encouraged the installation of combustion and post-combustion controls. There are two major differences between that paper and this one. First, this paper focuses on modifications, rather than technology adoption via new capital investment. Second, Popp does not estimate how much the technologies reduce emissions.
Although past research has not systematically investigated the flexibility of installed capital, there is qualitative evidence of such changes in Acid Rain Program (ARP) and from a N[O.sub.x] emissions tax in Sweden. Several examples have been noted in the ARP, including modifications to plants that allow different types of coal to be blended in order to reduce the average sulfur content of the coal (Ellerman, et al., 2000 and Burtraw, 2000). As Section 3 describes, the technologies used to reduce N[O.sub.x] emissions are much different from sulfur dioxide technologies, so the outcome in the ARP provides little indication of how effective are modifications at reducing N[O.sub.x] emissions. Isaksson (2005) uses reported costs and emissions to estimate marginal abatement cost curves under a N[O.sub.x] emissions tax in the early 1990s that affected several sectors. Many firms used modifications to reduce emissions, but the empirical analysis does not quantify the effect of modifications on emissions. This paper is complementary, and uses a different strategy to estimate counterfactual emissions and to provide evidence for U.S. power plants, which are much more likely to use coal than the Swedish firms.
More broadly, electricity price regulation creates incentives for adopting pollution abatement technology (Ishii, 2004 and Lange and Bellas, 2007). Wholesale competition for electricity began in the 1990s in many parts of the country, while state regulators continued to set prices for plants in other regions. The effect of a permit program on pollution technology adoption may depend on the electricity price regulatory regime, which I investigate below.
The paper proceeds as follows. Section 2 provides an overview of recent N[O.sub.x] regulations for coal-fired power plants. Section 3 describes modifications in more detail. Section 4 discusses the empirical strategy, while Section 5 describes the data sources and presents summary statistics. Section 6 reports the empirical results, including both parametric and non-parametric estimates of the effect of modifications on emissions, and Section 7 concludes.
2. RECENT N[O.sub.x] REGULATIONS FOR POWER PLANTS
2.1 The 1970 Clean Air Act and the 1990 Clean Air Act Amendments
This section outlines the recent chronology of N[O.sub.x] regulations for electric power plants. Nitrogen oxides are precursors to acid rain and ground-level ozone. Under the Clean Air Act (CAA), the Environmental Protection Agency (EPA) created air quality standards for N[O.sub.x] and ozone that each county must attain. Power plants create N[O.sub.x] during fossil fuel combustion, which takes place in the boiler. The CAA required new boilers to meet certain emission standards, but did not regulate emissions from boilers that began operating before 1970.
By the end of the 1980s, many counties continued to exceed the ozone standard. The CAA did not affect pre-1970 boilers, which emitted a large share of total N[O.sub.x] emissions. Therefore, the 1990 CAAA attempted to reduce ozone levels by regulating the older boilers. Emission rates of all...
|
