The Effect of Reporting Thresholds on the Validity of TRI Data as Measures of Environmental Performance: Evidence from Massachusetts

This paper examines the effect of reporting thresholds on the use of TRI data for both time-series and cross-sectional policy analysis. Facilities are only required to report releases to TRI if their use of a chemical exceeds some threshold. This creates an incentive for facilities to reduce their use of a listed chemical to a level just below the reporting threshold. However, this does not necessarily represent a real improvement in environmental performance, as the facility’s release level may remain largely unchanged. The State of Massachusetts has expanded the disclosure requirements under TRI in ways that better allow for assessment of the measurement error induced by reporting thresholds. This paper utilizes data from the Massachusetts Toxics Use Reduction Act (TURA) to estimate bounds on the degree of bias introduced by the reporting thresholds in both time series and cross-sectional analyses for Massachusetts’ facilities. In particular, this paper asks the questions: (1) How much of the observed decline in reported releases is artificially created by the existence of the reporting thresholds?; and (2) How much would our rankings of facilities change if we could account for releases by facilities that are below the reporting threshold? The results of the analysis suggest that the bias introduced by the reporting thresholds may be significant. Up to 40 percent of the observed decline in TRI releases in Massachusetts may be attributed to truncation at the reporting thresholds. In addition, quartile rankings of facilities based on reported releases may be in error 45 percent of the time when behavior around the reporting thresholds is not taken into account. Because the TURA data are in many ways identical to the TRI data, the fact that truncation bias is a large concern in the TURA data suggests that this bias may be important for the national TRI data as well. * Bennear is an Assistant Professor of Environmental Economics and Policy at the Nicholas School of the Environment and Earth Sciences, Duke University. She can be reached at [email protected] or at 919-613-8083. The author is grateful to Cary Coglianese, Nolan H. Miller , Cody Rice, Robert N. Stavins and participants in the Harvard Environmental Economics Seminar and the Environmental Protection Agency’s Corporate Environmental Behavior and the Effectiveness of Government Interventions Workshop.