Data Quality

The Data Quality Appropriations Rider: New Procedures and Information Disclosure

by Sidney Shapiro

Background

Federal agencies increasingly are seeking to fulfill their statutory missions by disseminating information, particularly through the Internet, about the entities, products and topics within their purview. Programs like the Toxic Release Inventory (TRI), an annual, national compilation of chemical releases issued by the Environment Protection Agency (EPA), create political and economic pressure on firms to improve their performance, such as by reducing toxic exposures beyond the amounts required by existing regulations. Other programs empower individuals to alter their market activity in a manner that reduces their risk. The crash worthiness ratings issued by the National Highway Safety Administration (NHTSA) illustrate this potential. More broadly, information disclosure satisfies the public’s right to know about potential hazards.

In 2001, Congress passed a two-paragraph provision buried in an appropriations bill that requires agencies to ensure and maximize the quality of information that they disseminate and to establish an error correction process. Congress also gave the Office of Management and Budget (OMB) the power to issue guidelines to agencies about how to implement the requirement. Rep. Jo Ann Emerson (R-MO) sponsored the rider without legislative hearings, committee review, or debate. Representative Emerson reportedly acted at the behest of Jim Tozzi, a former OMB-official who runs the corporate sponsored Center for Regulatory Effectiveness. As far as can be determined, few, if any, other members of Congress knew of the appropriations rider at the time they voted for it. In February 2002, OMB issued instructions telling agencies how to implement the legislation. After seeking public input, agencies adopted permanent procedures to implement the rider in October 2002.

What People are Fighting About

The appropriations rider requires OMB and agencies to establish procedures to ensure the quality of information disseminated by government. While ensuring high-quality information is a worthy goal, procedural requirements have an important side effect – they slow down the government’s capacity to act and, if they are sufficiently burdensome, they can bring government to a standstill. As a result, the benefits of imposing additional procedures have to be balanced against the consequences to the public of delaying agency action. In the case of data quality, the goal of ensuring the quality of information has to be reconciled with the substantive mission of an agency and the role of disseminated data in the implementation of that mission. Unfortunately, Congress gave little indication of how these competing goals were to be reconciled. It defined none of the key terms of the rider, and left no legislative history. Just as nature abhors a vacuum, an ill-defined statute invites government officials and interested parties to interpret it to serve their own ends.

This vaguely worded data quality rider has the potential to harm the environment and to reduce health and safety protections far beyond its modest length. There is a crucial distinction between incomplete data and poor quality data. For example, an excellent study of the adverse health effects of heightened blood lead levels can be incomplete with respect to the hazards of heightened levels of lead in the air if the rates of transfer between airborne lead and blood lead are poorly understood. The absence of knowledge about air-to-blood transfer rates might be interpreted by some, however, to mean that the scientific evidence about the hazards of airborne lead to human health is of poor quality. If the quality requirements of the data quality rider are interpreted to make it harder to take protective action when data is incomplete, this will mark a huge shift in American environmental policy, which since the 1970s has relied upon the principle that we should base policy on the best available evidence. In fact, one of the most successful protective actions ever taken was EPA’s 1973 decision to phase out the lead content in gasoline, when the air-to-blood transfer rate was not completely understood. Policy decisions should take into account the quality of the evidence as part of the process of deciding what to do, but it is often wise to act before all the answers are in.

Striking an appropriate balance between protecting the public and the environment and improving the quality of information is therefore a complex matter. Refusing to act until there is more information about a risk can result in substantial harm to vital public purposes. The danger is that data quality will become a goal in and of itself, rather than a means to ensure the most effective protection of individuals and the environment under existing circumstances.

Industry and trade organizations are already expansively interpreting the rider, arguing that it provides an open-ended remedy to them for government information that they believe to be of insufficient quality. Environmental and other public interest groups interpret the legislation in a narrow manner, doubting that such sweep can be read into the statute’s plain text, especially given the absence of any concrete guidance from Congress concerning the scope of the legislation. The courts will eventually have to rule on the scope of the appropriations rider. Even if the rider is not finally interpreted to prohibit basing action on incomplete information about the extent of a risk, the dangers posed by the rider are real. It takes years for an agency to promulgate a significant regulation because regulators must first complete time-consuming and laborious analyses of the potential impacts of proposed rules. This type of delay is so common it has its own name—rulemaking “ossification”. The appropriations rider creates the potential that the disclosure of data and other information will also become ossified.

Industry groups can be expected to use the new procedures in a strategic manner to slow, or even stop, the release of information that is embarrassing or politically inconvenient to them. The vague nature of the legislation invites such challenges. Moreover, the rider provides no deterrents for multiple, meritless complaints. Thus, the legislation will force agencies to devote considerable time and effort to administering the mandatory review process for reviewing data before information is disseminated and to the error-correction process after the information is disclosed. The only way an agency can avoid these burdens is to decline to make public information about the environmental and health and safety risks, which agencies may do at least some of the time to save resources. Thus, the public will likely receive less information in a less timely fashion.

CPR's Perspective

The Center supports efforts to ensure that data and information disseminated to the public are of high quality. This objective, however, must take into account the impact of data quality activities on an agency’s substantive mission and the role of disseminated data in the implementation of that mission. Nothing in the language, structure, or history of the appropriations rider evidences any considered congressional judgment to alter any agency’s substantive mandates. A balanced approach to implementation of the rider would include the following aspects.

First, both OMB and the agencies have assumed that the rider applies to the information that an agency references when it proposes a new or amended regulation. CPR believes that the better reading of the law is that, while the legislation applies to information in government reports and information posted on the Web, it does not apply to data relied upon in rulemaking. The rulemaking process already provides an adequate and fair procedural mechanism to determine the quality of data, and the application of the rider to rulemaking will slow down the rulemaking process without any offsetting advantages.

Close attention to the terms of the appropriations rider confirms that it does not apply to rulemaking. The rider, as mentioned earlier, requires agencies to create a new “administrative mechanism” to hear and resolve complaints about data quality. This means Congress intended the rider to apply to contexts where the dissemination of information is not already subject to an administrative mechanism to correct data problems. This would not include rulemaking because such a process already exists in rulemaking. Indeed, the rulemaking process provides more stringent procedures regarding the vetting of data than the appropriations rider, and the academic literature indicates that there are surprisingly few instances where agencies have relied on unreliable science among the thousands of public health and safety regulations promulgated annually. Since setting up another process would be superfluous or redundant, it has to be assumed that Congress had no such intention.

Second, the legislation requires that agencies ensure the “objectivity” of information without defining that term. According to OMB, information is “objective” when it is “accurate, reliable, and unbiased,” which requires the use of “sound statistical and research methods regarding scientific, financial, or statistical information.” OMB has defined “sound statistical and research methods” regarding the analysis of risks to human health, safety and the environment as the use of the principles applied by Congress to risk information used and disseminated pursuant to the Safe Drinking Water Act Amendments (SDWA) of 1996, and it ordered agencies to “either adopt or adapt” these principles. Agencies have chosen to “adapt” the SDWA principles, but even this approach may result in the application of the SDWA criteria in contexts where they may not be appropriate or helpful to fulfilling an agency’s mandate. CPR’s position is that an agency should not feel bound by a congressional prescription for the quality of scientific data employed in establishing regulations under the SDWA in determining the quality of information disseminated to the public in entirely different contexts.

Third, the legislation requires agencies to establish an “administrative mechanism” to allow persons to seek and obtain the correction of information “maintained and disseminated” by an agency that does not comply with its data quality guidelines. When information is objective, this requirement makes sense. Agencies should correct erroneous information when it is called to their attention. This requirement, however, is problematic in the context of presenting information about human and environmental risks. The characterization of such risks is a difficult and controversial process in part because it involves difficult subjective judgments. The need for such judgments arises because scientific information regarding risks is often incomplete and inconsistent. Thus, it is often difficult to say that a risk characterization is clearly “wrong,” given the degree to which assumptions, policy choices, and judgments are embedded into every step of the risk assessment process. In light of this reality, CPR believes that agencies are in compliance with the rider when they clearly disclose the manner in which information was derived and the assumptions on which it was based. Congress could not have intended agencies to withhold risk information until scientists resolve all, or even most, doubts about the extent to which a hazard is dangerous to people or the environment because that would end information disclosure about most risks. Industry and interest groups that disagree with an agency’s interpretation of data are free to inform the public of their own assumptions and to offer their own interpretations.

Fourth, CPR urges agencies to seek, and OMB to support, additional funding to carry out responsibilities under the legislation. Since the appropriations rider is an unfunded mandate from the agency’s perspective, compliance with it will siphon off agency resources from other activities, including the promotion of regulatory and information activities that protect the public and the environment. In order that data quality not become a zero-sum game, agencies should request from the Administration, if they are subject to OMB budget oversight, or from Congress, if they are not, additional funding to meet these new responsibilities. To the extent that agencies do not receive adequate funding for their data quality activities, they should prioritize their responses to data quality complaints and inform persons who file complaints when their complaint will be reviewed or acted upon, if the agency will be delayed in responding. In addition, agencies should save resources by dismissing data correction requests that are frivolous, duplicative of other requests, refer to issues that have been the subject of prior complaints that have been resolved, or that occur after reasonable time deadlines set for the submission of such claims.

Finally, the legislation is silent concerning whether someone who files a data quality complaint with an agency can obtain judicial review if the agency rejects that complaint. In the past, courts have generally refused to review information disclosure because the release of information does not determine legal rights or obligations, which is a prerequisite for judicial review under the Administrative Procedure Act (APA). Now, however, judges may undertake judicial review of data quality complaints because agency rejection of a complaint is a determination of a legal right. CPR’s position is that judicial review of data quality issues is inappropriate and unnecessary to ensure agency compliance with OMB’s directions. Judicial review is inappropriate because Congress failed to establish concrete standards under which such review could occur, which indicates that it did not intend that review be available. In addition, judicial review of claims under the Data Quality Act is inappropriate because there is no indication in this rider that Congress intended to provide a private right of legal action. Moreover, even if the courts find such standards exist, judicial review would be inappropriate unless a corporation or other entity could demonstrate that there were directly and immediately harmed by the agency’s action, which is unlikely in most cases. Finally, Congress assigned OMB the responsibility for monitoring agency compliance with the legislation, which indicates that judicial review is unnecessary to ensure such compliance.

In Sum

The dissemination of information is an important aspect of the government’s efforts to reduce the risk of injury to individuals and to the environment. In 2001, Congress passed a vague appropriations rider requiring OMB and agencies to establish procedures to ensure the quality of information that the government disseminates. Industry groups support an expansive interpretation of the rider that is likely to slow down the dissemination of information or even eliminate it in some cases. CPR supports efforts to ensure the high quality of information, but this objective must be carefully balanced to take account of the impact of data quality activities on an agency’s substantive mission and the role of disseminated data in the implementation of that mission.