There is increasing interest in the United States and other countries in the use of information disclosure programs as potential substitutes for, or complements to, conventional command-and control or market-based environmental policy instruments. Much of this interest can be attributed to the apparent success of the Toxics Release Inventory (TRI) program, which requires large manufacturing facilities to report publicly their annual releases of certain chemicals. Since the inception of the TRI program in 1986, reported releases of over 300 regulated chemicals have fallen by more than 45 percent. However, since release data is only available for facilities that are required to report to TRI, one cannot determine what reductions in releases would have occurred in the absence of TRI and, therefore, one cannot infer how much of this 45 percent reduction is the direct result of the TRI reporting requirements. The primary purpose of the proposed research is to better isolate the causal effect of TRI on facility-level outcomes. To accomplish this, it is necessary to recognize that the production of toxic releases at a facility is directly linked to production of output at the facility. While we cannot observe releases for facilities that are not required to report to TRI, we can observe output levels for both facilities that are required to report and facilities that are not required to report.
This framework can be used to develop a natural experiment for testing the effects of TRI on production technologies and productivity. We will estimate production functions separately for facilities that are required to report to TRI 60 and facilities that are not required to report to TRI and empirically test for differences in the production technologies employed and the productivity of the two groups of facilities. This will help determine whether public disclosure of environmental performance information changes the way facilities do business. In addition, this project will provide a better understanding of the ways in which public disclosure affects facility decision-making. There are several pathways through which reporting of environmental information might lead to pollution reduction, including: green consumerism, green investing, community pressure, the threat of future regulation, and organizational limitations of the firm.
We will construct indictors for each of the pathways, estimate the effect of these pathways on changes in the production functions used by reporting and non-reporting facilities, and compare these effects with predictions from a theoretical model of facility-level response to information disclosure requirements. A critical element in the analysis is the construction of a facility-level dataset that contains information on environmental variables, economic variables, and variables that proxy for the four different pathways. We will create this dataset by linking publicly available data from the Toxics Release Inventory, trade organizations, environmental organizations, and other sources with confidential facility-level economic data from the Longitudinal Research Database (LRD) and the Pollution Abatement Costs and Expenditures (PACE) Survey for the years 1982 to 1999. In addition to providing a clearer understanding of the causal relationship between TRI and facility-level economic outcomes, the creation of this unique dataset will provide opportunities to evaluate and enhance Census data. The Census collects detailed data on the products produced, and the materials used in production, at the 7-digit SIC level (8-digit NAICS level beginning in 1997). However, many facilities do not report data at this level of detail. Instead they report data on products and materials at a higher level of aggregation. Because several of the chemicals that are reported under the TRI are also 7-digit SIC categories (or 8-digit NAICS categories) on the materials and product trailers, we can evaluate whether facilities are better able to report detailed data to Census on their usage of these specific chemicals if they are required to track that data to comply with the TRI. We will also evaluate potential improvements in data reporting that could be obtained if the connections between TRI reported chemicals and Census product and materials categories are enhanced.