|Project Status Reports:|
For the Year 1999
Objective: This project examines productivity in the paper industry, combining plant-level production and cost data, pollution emissions, and demographic data to calculate a broadly defined measure of productivity at each plant, incorporating the benefits from reduced pollution exposure. These data allow us to compare the costs and benefits of pollution regulation in a productivity framework. We use data on the production and abatement technologies used at each plant to help explain differences in productivity across plants, as well as changes over time.
Progress Summary: Most of the second year of the grant was devoted to completing the plant-level database. We used various industry directories to gather data on each plant. The plant-level database identifies the products produced by the plant, and the plant's overall production capacity. It includes information on the production technology in use at the plant, especially whether or not it has its own pulping facility, what type of pulping processes it uses, and the pulping capacity. The database has information on water usage, discharges, and treatment facilities. Because we have directory data from many years (going back to 1960), we can infer something about the plant's age from when it first enters the directory. In addition, we have the plant's name and address, and information about its corporate ownership.
We developed linkages from these data to a number of additional databases, giving further information. We linked our plant list to several EPA databases, providing sources of data on water pollution (PCS), air pollution (AIRS), and toxic waste (TRI). We also calculated each plant's exact location (taken from EPA data sets and from commercially available mapping software), and used it to link the dataset to demographic data from the Census of Population, giving us the population density near plant. This enabled us to calculate the number of people that would be exposed to local air and water pollution from the plant.
This extended database allowed us to achieve a major goal of our research, by giving us a measure of the benefits associated with reductions in pollution at each plant. For some pollutants, particularly air pollution, we have measures of current emissions and of the efficiency of the control equipment in use at the plant, so we can project what emissions at the plant would have been in the absence of the control equipment. For other pollutants, we base our estimated pollution reductions on an engineering study that calculated reductions in pollution at a typical paper mill that could be attributed to compliance with the Clean Air and Clean Water Acts, separately for different types of pulping processes. We combined the estimated pollution reduction at each plant, the nearby population density, and dollar measures of the benefits from reducing exposure to pollutants to come up with an overall measure of the benefits from reduction in pollution at each plant.
Our first research paper using the benefits data ("Spatial Efficiency of Pollution Abatement Expenditures") tests whether the differences across plants in their pollution abatement costs can be explained by differences in the benefits from their pollution reductions. An efficient allocation of abatement costs should have plants where emission reductions are especially valuable (either because of an especially dirty production technology or especially high population densities nearby) being required to do more pollution abatement. Thus, high-benefit plants should have correspondingly higher pollution abatement costs. We combined our benefits data with plant-level pollution abatement expenditures data from the Census Bureau's Pollution Abatement Costs and Expenditures survey (working at the Census Bureau), and compared the costs and benefits. We found that plant-level costs were related to the benefits from abatement, as expected, with higher-benefit plants having higher abatement costs. However, the differences in abatement costs across plants were not as sensitive to differences in benefits as we would have expected if the allocation of expenditures were perfectly efficient. This could be due to a desire on the part of regulators to have uniform standards for different plants in the same industry, regardless of their location, but suggests that there are some efficiency costs to achieving this uniformity of standards.
Future Activities: We will finish updating the "benefits from pollution abatement" numbers, incorporating more complete plant-specific information. For water pollution, we will consider the incremental impact of the plant's discharges on the receiving waterway. For air pollution, we will give some consideration to the interactions between different pollutants in the atmosphere. Once these numbers are updated, we will revise our "Spatial Efficiency" paper (we are scheduled to present an updated version at a Harvard University seminar in Spring 2000).
We also will use the updated benefits numbers in two other papers. One will be a sort of plant-level benefit-cost analysis, using an expanded measure of productivity. This will incorporate the dollar benefits from pollution abatement as one of the "outputs" from the plant, potentially offsetting the costs from pollution abatement, which are already counted in traditional productivity measures. We can see how many plants have benefits from pollution abatement that exceed their costs, and also aggregate the data across plants to provide an overall benefit-cost analysis for the industry.
The other paper will examine the connections between changes in technology and reduction in pollution over time. This will use the plant-level production and abatement technology information, identifying changes in technology at particular plants, and seeing whether it is linked to changes in pollution emissions at the plant. Of particular interest will be comparing the magnitudes of emission reductions arising from shifts in technology with reductions arising from the use of traditional end-of-pipe abatement methods.
Shadbegian R, Gray W. Spatial efficiency of pollution abatement expenditures. Presented at the Western Economic Association Annual Meeting, June 1999.
Objective: This research project examined productivity in the paper industry, combining plant-level production and cost data, pollution emissions, and demographic data to calculate a broadly defined measure of productivity at each plant, incorporating the benefits from reduced pollution exposure. These data allowed for a comparison of costs and benefits of pollution regulation in a productivity framework. The data also allowed for addressing several related questions: How are differences in the costs of pollution abatement across individual plants related to differences in the benefits from pollution reduction at those plants? Have environmental concerns influenced investment decisions, including the choice of production technology and the overall level of investment? Have firms tended to shift production away from plants with especially serious environmental problems? What determines a plant's environmental performance, measured either by pollution emissions or by compliance with regulations? The dataset created for this research also has been used in other research projects, and will continue to provide research benefits in the future.
Summary/Accomplishments: Much of the effort under this grant was spent in preparing a large database containing plant-level data on 758 U.S. pulp and paper mills (including some mills that currently are not operating, and many small mills that do not produce their own pulp). These data came from various industry directories, providing information on the products produced by the plant and the plant's overall production capacity. They include information on the production technology in use at the plant, especially whether or not they have their own pulping facility, what type of pulping process(es) they use, and the pulping capacity. The dataset has information on water usage, discharges, and treatment facilities. It also includes some information on investment in both production and abatement capital. Because we have directory data from many years (going back to 1960), some plants' ages can be inferred from when they first entered the directory, though most mills in our data were in operation in 1960. In addition, the plant's name, address, and information about its corporate ownership are included.
Linkages were developed from these data to a number of additional databases, using the plants' names and addresses. On the regulatory side, the plant list was linked to several EPA databases, providing sources of data on water pollution (PCS), air pollution (AIRS), and toxic waste (TRI). Each plant's exact location (based on data from the EPA datasets and from commercially available mapping software) also were calculated, and were used to link in demographic data from the U.S. Census, giving the population density near the plant. The plant location also was used to link each plant to data from the closest available meteorological station.
A key accomplishment of this research was the calculation of pollution abatement benefits, particularly for air pollutants. Assistance in this area was provided by Jonathan Levy of the Harvard School of Public Health, an expert in atmospheric chemistry and quantifying the impacts of air pollutants on health. He helped develop a model of the impact of air pollution from a specific plant based on the predicted health impact of those pollutants on the surrounding population, particularly the mortality effects of particulates. The necessary calculations were incorporated in a spreadsheet that was used to calculate benefits per ton of reductions in specific air pollutants, incorporating local meteorological conditions near the plant (average wind speed and mixing height), the increased mortality risk, the size of the affected population, and estimates of the dollar value of a "statistical life" (for which an estimate of $4.8 million was used, based on an EPA analysis of existing studies). These per-ton benefits then were multiplied by the estimated reductions in plant emissions achieved at the plant to yield a measure of total benefits per plant, based on the levels of current emissions and of the efficiency of the control equipment in use at the plant.
The benefits from reductions in water pollution were calculated, using a less sophisticated model. The estimated pollution reductions were based on an existing engineering study that calculated the reductions in water pollution at a typical paper mill that could be attributed to compliance with the Clean Water Act. This study distinguished between different types of pulping processes, and reported pollution reductions per unit of output. Data on each plant's capacity and pulping process were used to estimate its reductions in water pollutant discharges. These reductions were multiplied by a national-level estimate of the benefits per ton of water pollution reduced, then adjusted by the population density near the plant to provide a plant-specific estimate of benefits. It was hoped that a more detailed model of water pollution could be used that incorporated plant-specific data such as differences in stream flow and water quality, but access to the model was not possible. There are some indications that the model may be used in the near future, and at that point our calculations will be revised to incorporate the new information.
The above-mentioned data were brought to the U.S. Census Bureau's Boston Research Data Center (BRDC), where they could be combined with confidential plant-specific data from the Census Bureau. The Census data included the Longitudinal Research Database (LRD) containing information on the production and costs components for productivity calculations, as well as the Pollution Abatement Costs and Expenditures (PACE) survey, with information on capital expenditures and operating costs for pollution abatement. In earlier research at the BRDC, the LRD and PACE data were linked to create datasets covering the 1979-1990 period with two samples of plants: a set of 116 plants that had complete productivity data in the LRD and at least some years of PACE data, and a subset of 68 plants that had PACE data from all years. It was planned to expand these samples, getting as close as possible to a complete sample of plants in the industry, and extending the data period into the 1990s. Unfortunately, a dispute between the U.S. Census Bureau and the Internal Revenue Service (IRS) over use of tax data in the LRD led to an extended moratorium on new projects with Census data. Research at the BRDC was permitted to continue, but was constrained to use of the existing LRD sample (though we were allowed to link in the new outside datasets we had created).
This constraint had two major consequences for this research work. First, it limited the set of plants in the analyses to the relatively large plants with continuous LRD data. This meant that the distribution of productivity growth and the effect of the adjustments for pollution abatement benefits, for a truly representative set of plants, could not be examined. Second, it limited the regulatory data in the analyses to those available in the 1980s. This effectively focused the detailed regulatory analyses on air pollution, for which some 1980s data exist (though we were able to calculate some measures of benefits for water pollution abatement, as described above). In recent months, an agreement has been reached between the U.S. Census Bureau and IRS that makes it possible for new projects to proceed. It has been proposed, and approval has been received from the U.S. Census Bureau for a new multiyear BRDC project that includes productivity related work with the 1990s data (it also includes work on the new EPA project on "Corporate Environmental Performance and the Effectiveness of Government Interventions"). It is anticipated that some of the analyses carried out in this project would be extended in the future.
A central research question for this project concerned the impact of incorporating the benefits from pollution abatement on the calculation of productivity measures for paper mills. In our first research paper using the benefits data ("Spatial Efficiency of Pollution Abatement Expenditures"), the magnitudes of the benefits and the costs of pollution abatement were compared for both air and water pollution. The annual benefits of air pollution reduction greatly exceeded the annual costs for an average plant: $16.1 million vs. $0.8 million. The benefits for water pollution reduction also exceeded the costs for an average plant, but by a much closer margin: $2.2 million vs. $1.6 million. Results from past research on the relationship between reported pollution abatement costs and productivity in paper mills has suggested that the reported abatement costs could be sizably understated, with $1.00 in abatement costs being associated with about $1.80 in productivity reductions. Adjusting the abatement costs for this effect would result in costs exceeding benefits for water pollution reduction, although the benefits for air pollution reduction would still greatly exceed the adjusted abatement costs.
The dominance of the benefit numbers by air pollution effects, specifically the mortality effects of particulates, is not new to this study. Still, it points out an apparent inequality in the regulation of air and water pollution: air pollution reductions appear to have the greater dollar-valued benefit, but water pollution reductions get more regulatory attention, as measured by abatement spending. This may be attributable to the greater visibility of water pollution with its localized impact, while air pollution spreads for hundreds of miles and is correspondingly less obvious to the plant's neighbors. An alternative explanation for the discrepancy might arise from some errors in calculating the relative benefits for air and water pollution, though it would take a substantial error to reverse the seven-fold difference in benefits.
When calculations of productivity growth were adjusted to account for the benefits and costs of pollution abatement, a similar result was found, with adjustments for the benefits of abatement (for air and water pollution combined) having a substantially larger impact on productivity growth than adjustments for abatement costs. Traditional productivity measurement includes the inputs used for pollution abatement in measures of total inputs, tending to understate productivity growth. One could either subtract the pollution abatement inputs from total inputs, or add the pollution abatement benefits to the calculation of total output?in either event increasing measured productivity growth. In this case, the benefits adjustment had twice as large an impact on the average productivity growth rates for the 1972-1990 period: 0.26 percentage points per year vs. 0.13 percentage points per year. To provide a benchmark for comparison, the measured paper-industry productivity growth rate during the same period was only 0.93 percentage points per year, so the adjustment would increase measured productivity growth by more than one-quarter.
Another question related to the productivity adjustment is whether discrepancies in productivity growth across plants can be attributed to differences in regulatory pressures. Did the poor performance of some plants on productivity measures come from a greater impact of regulation on those plants? We addressed this question in two ways. First, we compared the amount of variation in productivity growth across plants for both the adjusted and unadjusted productivity measures. In fact, the adjusted productivity measure showed a greater variation, indicating that there was some tendency for plants with better productivity performance to have larger pollution abatement benefits. This was consistent with our results in "What Determines Environmental Performance at Paper Mills? The Roles of Abatement Spending, Regulation, and Efficiency," where it was found that high-productivity plants tended to have lower emissions per unit of output.
The second comparison of the impact of adjusting measured productivity growth involved comparing the relative ranking of a plant in terms of its adjusted and unadjusted productivity levels and growth rates. The productivity adjustment usually had very little impact on a plant's relative productivity ranking, with slightly more of an impact on productivity growth rates than on levels. About 90 percent of all plants fell into the same quartile ranking for both adjusted and unadjusted productivity levels. About 80 percent of all plants fell into the same quartile ranking for both adjusted and unadjusted productivity growth rates. Thus, no evidence was found that discrepancies across plants in measured productivity could be explained by differences in pollution abatement benefits (or costs) across the plants.
The pollution abatement benefits data used to test whether differences across plants in their pollution abatement costs could be explained by differences in the benefits from their pollution reductions ("Spatial Efficiency of Pollution Abatement Expenditures"). A spatially efficient allocation of abatement costs was defined as one in which plants for which emission reductions were especially valuable (either because of an especially dirty production technology or especially high population densities nearby). Thus, high-benefit plants should have had correspondingly higher pollution abatement costs. It was found that plant-level costs were related to the benefits from abatement as expected, with higher-benefit plants having higher abatement costs. This suggested that regulators responded to differences in benefits when determining regulatory stringency. However, the elasticity of costs with respect to benefits seemed to be smaller than the value of one that would have been expected if the regulators' only goal was spatial efficiency. Given the multitude of regulatory goals, including preventing deterioration in pristine areas, maintaining uniform standards across plants in the industry, and reducing differences across states in regulatory stringency, this was not necessarily a surprising result, but did suggest that there were some efficiency costs to achieve uniformity of standards.
The connection was examined between a plant's productivity and the nature of its capital stock ("Capital Vintage, Efficiency, and Environmental Regulation"). Substantial impacts of pollution abatement costs on productivity levels were found and checked for differences in impact, based on the plant's vintage and production technology. Pulping mills seemed to be especially severely affected by regulation?the observed negative relationship between higher abatement costs and lower productivity levels was almost entirely due to pulping mills. Somewhat surprisingly, older plants seemed to be slightly less sensitive to abatement costs, perhaps due to their being "grandfathered" under many regulations.
In "Environmental Regulation, Investment Timing, and Technology Choice," it was tested whether environmental regulation affected investment decisions, using Census data for individual paper mills. New mills in states with strict environmental regulations chose cleaner production technologies. Differences in air and water pollution regulation across states also influenced technology choice (e.g., the most water-polluting technology was less common in states with stricter water pollution regulation). Examining investment allocation across existing plants, it was found that abatement and productive investment tended to happen in the same years, consistent with the high cost of shutting down a paper mill for renovations. However, plants with relatively high pollution abatement investment over the entire period invested significantly less in productive capital. The reduction in productive investment is greater than the increase in abatement investment, leading to lower total investment at high abatement cost plants. The magnitude of this impact was quite large, suggesting that a dollar of pollution abatement investment reduced productive investment by $1.88 at that plant. This seemed to reflect both environmental investment crowding out productive investment within the plant, and firms shifting investment towards plants facing less stringent abatement requirements. Estimates placing less weight on within-firm reallocation of investment indicated approximately dollar-for-dollar ($0.99) crowding out of productive investment.
Another examination of the impact of environmental regulation on allocation decisions came in "Do Firms Avoid Regulation by Shifting Production?" which examined the relationship between a firm's allocation of production across its plants and the environmental stringency faced by those plants. It was found that states with stricter regulations have smaller production shares at their paper mills, even after controlling for a variety of other state characteristics. This impact was concentrated on firms that are out of compliance with regulation. Firms with high compliance rates seemed to be less sensitive to regulation, and may even have had slightly larger production shares in stricter states. These results suggested that differences across firms in compliance were driven primarily by differences in compliance costs (economies of scale in compliance), rather than by differences in the benefits of compliance (maintaining the firm's reputation with customers and regulators).
The factors determining the environmental performance of paper mills also were examined. "When Is Enforcement Effective Or Necessary?" examined differences in plant-level compliance with air pollution regulation for U.S. paper mills during the 1980s. A variety of plant- and firm-specific characteristics were tested, along with measures of regulatory enforcement. Plants that included a pulping process, plants that were older, and plants that were larger were all less likely to be in compliance. In contrast, firm-level characteristics did not have much effect on compliance. Plants in violation of water pollution or workplace safety regulations were less likely to be in compliance with air pollution regulations than were plants in compliance with those other regulations. Measuring the impact of regulatory enforcement on compliance was complicated by the targeting of enforcement towards plants that are out of compliance. A negative relationship was found between enforcement levels and compliance levels, reflecting this targeting, but some positive impacts of enforcement on changes in compliance status also were found. Different types of plants did not seem to differ greatly in their responsiveness to enforcement, although there was some hint that those types of plants that tended to have lower compliance rates overall also tended to be less responsive to enforcement.
A more direct examination of performance, "What Determines Environmental Performance at Paper Mills? The Roles of Abatement Spending, Regulation, and Efficiency," examined the determinants of air pollution emissions at paper mills in the 1980s. Data were included on the plant's pollution abatement capital stock, productivity levels, and measures of local air pollution stringency. In simple cross-plant analyses, aggregate emissions (a weighted average of particulates and sulfur dioxide) per unit of output were significantly lower in plants with a larger air pollution abatement capital stock. A 10 percent increase in abatement capital stock appeared to reduce aggregate annual emissions per unit of output by about 6.9 percent, with similar impacts on particulates (6.3 percent) and sulfur dioxide (7.3 percent). Translating these impacts into dollars suggested a substantial return (although, of course, not a return that accrued to the paper mill making the investment): one dollar of abatement capital stock provided an annual return of about 75 cents in pollution reduction benefits. Local regulatory stringency also mattered: plants in nonattainment counties had an average of 43 percent lower emissions per unit of output when compared with plants in attainment counties, all else equal. More efficient plants (as measured by their productivity) had lower emissions, with a 10 percent higher productivity level associated with 2.5 percent lower emissions per unit of output. Having older or less productive equipment stock, as measured by the age and speed of their paper machines, had relatively little impact on emissions per unit of output.
Considered as a whole, the research conducted under this grant has provided new information in several areas, particularly concerning the relationship between productivity and environmental performance. Adjusting for the benefits caused by improvements in environmental performance substantially increased the measured productivity performance in the pulp and paper industry. This adjustment did not, however, reduce the discrepancy across plants in their productivity performance. Instead, the results suggested that plants that performed well in terms of measured productivity also performed well in terms of pollution abatement. Also, firms tended to shift production and investment away from their plants that experienced especially high abatement costs?perhaps explained by more efficient plants being better able to achieve greater pollution reductions with less abatement spending.
The dataset prepared during this project already has been useful in other research projects at Clark University, and will continue to be useful to projects there, and elsewhere, in the future. One graduate student at Clark University, Bansari Saha, who was involved at various stages of the dataset creation as a research assistant, recently completed his dissertation using the data. This dissertation examined the relationship between the air and water pollution from a paper mill and the value of all houses located within 50 miles of the mill, using a sophisticated spatial econometric model. Evidence was found for sizable benefits of pollution reductions, as measured by increased value of the houses located near the mill. An ongoing research project at Clark University, for which Dr. Gray is serving as Principal Investigator, is studying "Industrial Restructuring and Corporate Risk Management" with funding from the National Science Foundation. This project is extending the plant-level dataset to include information on corporate restructuring and examining the impact of such restructuring on environmental performance. Finally, the researchers involved in this grant recently have received funding from EPA to examine "Corporate Environmental Performance and the Effectiveness of Government Interventions." This research will extend the dataset beyond the paper industry to include steel mills and oil refineries, and will examine the determinants of environmental performance in all three industries. As the dataset is expanded and organized, it will continue to be made available for use in other research projects, both at Clark and elsewhere.
The research results described above provide valuable information to environmental regulators responsible for calculating the economic implications of new pollution regulations. For example, it was found that more efficient plants have better environmental performance and lower abatement costs. If so, the economic impact of more stringent regulations is likely to disproportionately affect less efficient plants. If these less efficient plants were on the verge of exiting the industry anyway, the stricter regulation could lead to substantial shifts in economic activity. There is evidence that such shifting occurs for both production and investment across plants within a firm.
The results do not imply that stricter regulations are undesirable. After all, the estimates of the productivity implications of pollution abatement indicated that the benefits of regulation in the paper industry have exceeded the costs, especially for air pollution. These attempts to quantify the benefits from pollution abatement are important to inform the public debate about the optimal degree of environmental regulation, and to put benefits on a more equal footing with costs, which tend to be more easily quantified. Any such calculation of optimal regulatory decisions needs to incorporate both the benefits and the costs of regulations, recognizing the implications of firms' responses to those regulations.
Gray W, Shadbegian R. Environmental regulation, investment timing, and technology choice. Journal of Industrial Economics 1998;46(2):235-256.
Gray W, Shadbegian R. Capital vintage, efficiency, and environmental regulation. Presented at the Harvard University Kennedy School of Government, March 1998.
Gray W, Shadbegian R. Do firms avoid regulation by shifting production? Presented at the Allied Social Science Association Meetings, January 1998.
Gray W, Shadbegian R. Do firms avoid regulation by shifting production? Presented at the First World Congress of Environmental and Resource Economists Meeting, Venice, June 1998.
Gray W, Shadbegian R. Do firms avoid regulation by shifting production? Presented at the Indiana University Business Economics Workshop, October 1998.
Gray W, Shadbegian R, Levy J. Spatial efficiency of pollution abatement expenditures. Presented at the Western Economic Association Annual Meeting, June 1999.
Gray W, Shadbegian R, Levy J. Spatial efficiency of pollution abatement expenditures. Presented at the Environmental Economics and Policy Seminar, Harvard's John F. Kennedy School of Government, March 2000.
Gray W, Shadbegian R, Levy J. Spatial efficiency of pollution abatement expenditures. Presented at the National Bureau of Economic Research Environmental Economics Meeting, April 2000.
Gray W, Shadbegian R, Levy J. Spatial efficiency of pollution abatement expenditures. Presented at the University of Kansas, October 2000.
Gray W, Shadbegian R, Levy J. Spatial efficiency of pollution abatement expenditures. Presented at the International Atlantic Economic Society Conference, October 2000.
Gray W, Shadbegian R. Technology change, emissions reductions, and productivity. Presented at the Allied Social Science Associations Meeting, January 2001.
Gray W, Shadbegian R. What determines environmental performance at paper mills? The roles of abatement spending, regulation, and efficiency. Presented at the Western Economic Association Conference, June 2001.
Gray W, Shadbegian R. When is enforcement effective-or necessary? Presented at the Association of Environmental and Resource Economists Summer Workshop, June 2000.
Gray W, Shadbegian R. When is enforcement effective-or necessary? Presented at the National Bureau of Economic Research Summer Institute on Environmental Economics, August 2000.
Supplemental Keywords: regulatory impact, productivity, benefits analysis, pulp and paper industry, SIC 2611, SIC 2621, public policy.