Establishing a Bureau of Environmental Statistics

More data collection and analysis would greatly enhance our ability to set policy and measure its effectiveness.

In 2003, Rep. Doug Ose (R-Calif.) proposed the Department of Environmental Protection Act, which would elevate the Environmental Protection Agency (EPA) to a cabinet department and create within it a Bureau of Environmental Statistics (BES). Although cabinet status for the EPA may have symbolic or organizational advantages, the creation of a BES could prove to be the most meaningful portion of the bill, as well as an important development for future environmental policymaking.

Noting the weakness of available data describing the environment, in comparison to data available to other agencies in their own respective purviews, the bill would authorize the proposed BES to collect, compile, analyze, and publish “a comprehensive set of environmental quality and related public health, economic, and statistical data for determining environmental quality . . . including assessing ambient conditions and trends.”

Why do we need another bureaucratic agency collecting statistics? The overarching reason is that we simply do not have an adequate understanding of the state of our environment. In many cases, the network of monitors measuring environmental quality is insufficient in geographic scope. For example, in many cases our knowledge of national air quality is based on a few monitors per state; our knowledge of water quality is even weaker.

Of course, this easy answer begs the further question of why we need a better understanding of the state of our environment. There are at least three returns that will result from collecting environmental data, each of which could pay greater dividends with reorganization and investment.

The first return from a BES would be to improve our monitoring and enforcement of environmental standards. Environmental standards in the United States generally fall into one of three types: standards for production technology or other behavior, emissions, and ambient concentrations. Technology standards prescribe that a specific technology or technique be used in the production process (for example, a specific type of equipment at a factory or plowing practice for farmers). Emissions standards specify a maximum rate of pollution emissions from a source, per unit of time or output. When pollution emissions are concentrated at a discrete number of sources (power plants and large factories), both types of standards are fairly straightforward to enforce through inspections or monitoring. Ambient standards pose greater challenges. Ambient standards require that pollution, after dispersing from its source through the air and water, not surpass some specific level. For example, eight-hour average concentrations of ozone cannot exceed 0.08 parts per million on more than three occasions per year at any location. If concentrations do exceed this standard, they trigger technology-based standards and other rules for the region. In the case of air quality, counties and regions that fail to meet the ambient standards risk the loss of federal highway dollars, bans on industrial expansion, and mandatory installation of expensive pollution-abatement equipment.

The integrity of such a system depends on the network of monitors measuring ambient quality. Currently, many expensive environmental regulations, with serious consequences for businesses and local economies, are based on a limited monitoring network. One may well wonder if some areas that are above the ambient thresholds have escaped detection. At the same time, there is some evidence that other areas continue to be designated as noncompliant even when they seem to meet the ambient standards. Recent research by Michael Greenstone of the University of Chicago has shown that many counties remain in official noncompliance for sulfur dioxide standards, even though readings from the available monitors have shown compliance for many years. The catch-22 is that a county must prove compliance throughout its jurisdiction even if the monitoring network is inadequate to shed light on all areas. Both kinds of errors undermine the fairness and effectiveness of the system and could be reduced with a more extensive monitoring network.

The second return from more data collection would be to satisfy our natural desire to understand broad trends that affect our society and its welfare. It is because of such a desire that we first began to collect many of our national economic statistics, including the familiar measures of gross domestic product (GDP) and inflation. Yet ever since the origination of the GDP concept in A. C. Pigou’s seminal Wealth and Welfare (1912), it has been acknowledged that GDP is only a proxy and is not a perfect measure of welfare, because it omits many important components that do not pass through markets. Even then, the environment was acknowledged to be one of the important omissions. Since that time, we have invested enormous resources in improving measures of the market components of national well-being, but we have not proportionately broadened that effort to other components such as the environment.

As its tools and data develop, the BES could eventually arrange such data about the state of our environmental well-being into a three-level hierarchy. The first level would essentially be the raw data that we currently gather (such as ozone levels in the troposphere and dissolved oxygen levels in lakes), albeit with some reforms. Reforms are required because, to date, most ambient monitoring has been motivated by the first objective described: the assessment of ambient standards. Accordingly, we tend to focus on areas that are known or suspected problem areas. For example, 10 states account for more than half of our ozone monitors. The Los Angeles and Houston metropolitan areas alone, with some of the worst air quality in the nation, have about 45 and 35 ozone monitors respectively–more than most states.

This approach is perfectly reasonable from the standpoint of enforcement, but not for surveying the overall state of things. From that perspective, the data we have is biased: Precisely because it is gathered at the problem spots, it is not representative of other areas. Between two monitored problem spots (two cities, say), pollution concentrations are presumably much lower, but we cannot tell without a monitor, and simply averaging the concentrations from the two monitors would not be correct. From the standpoint of assessing the real state of things, we need something like a random spatially distributed sample of observations. A smart sample would still focus on cities and other areas where, because of topography and economic activity, pollution concentrations vary more widely across short distances. But it would still look very different from our current sample. With a wider network, a BES with a broad statistical mandate could develop a sampling scheme that balances both the enforcement and the surveying objectives.

Gathering such basic data would be only the first step in improving our understanding of the state of our environment. Eventually, to obtain a complete picture, we would want to move to the second level in the hierarchy of our understanding: aggregate indices of the environment or of environmental systems. What is the state of air quality, taken as a whole? The state of riparian or forest ecosystems? Although meaningful, such questions raise still more. How do we define the limits of an ecological system in the dimensions of space, media, and species? What are the measurable indicators of its health? How can these various measures be aggregated into a single index? Mirroring the entrepreneurial beginning of our economic indices, a number of groups are pursuing these questions. A leading example is the H. John Heinz III Center for Science, Economics and the Environment, which produced the 2002 report, The State of the Nation’s Ecosystems. A BES that improved and centralized our collection of basic data would undoubtedly contribute to these efforts. But it would also be in a position to advance them to the next step of an official government index.

The playbook of strategies with which we might attack environmental problems is limited by lack of information.

In the long run, with an additional congressional mandate, the bureau could aspire to a third level of assessment: collaborating with the Bureau of Economic Analysis, which produces the GDP, to produce a “green GDP.” An idea that is actively being pursued by European countries, a green GDP would advance Pigou’s original vision by more closely approximating our overall welfare. It would net out reductions in our natural resources and “natural capital assets” in the same way that net GDP currently accounts for depreciation in the stock of human-made capital. It could also account for services provided by the environment–protection of human health, enhancement of outdoor recreation, flood protection, and so forth–in the same way that GDP accounts for the services provided by market goods. The services would be valued at people’s marginal willingness to pay, analogous to the price paid for market goods, using the type of data routinely collected today for benefit-cost analyses.

Designing better policies

The third type of return from more data collection would be in our ability to design better public policies for the environment. Currently, our ability to design policies that properly balance environmental quality with other objectives, or that attain environmental objectives in the most efficient and effective manner, is hampered by inadequate information. This knowledge gap is more meaningful than a mere shortage of beans for bean counters. It manifests itself in every stage of policy design and evaluation.

Looking in the rearview mirror, we do not know in many cases whether existing policies have been effective, which makes it difficult to assess what remains to be done. Looking forward, we often find that the playbook of strategies with which we might attack environmental problems is limited by lack of information. Sometimes, the lack of information creates practical problems for implementing and enforcing a strategy. For example, recent thinking about the control of water pollution has focused on the total maximum daily load (TMDL) of pollution, from all sources, entering water bodies that violate ambient standards. Theoretically, this is a sound approach for two reasons. First, it is firmly grounded in the reality of water pollution problems, which, with large point sources well regulated, increasingly have their source in disparate urban and agricultural runoff. Second, it can increase the flexibility and efficiency of pollution control by considering all the sources of pollution and concentrating on the most cost-effective targets. But it is data-intensive. Like any ambient standard, it would require a sufficient monitoring network. But it would also require an inventory of pollution sources (point sources, roads, farms, and other land uses), their levels of pollution, and models of the transport of their pollution to the water body. It is difficult to imagine pulling off such a policy without a great deal of investment in data collection and analysis.

In other cases, the lack of information makes it difficult to anticipate the effects of a policy, creating political uncertainties. For example, the 1990 Clean Air Act Amendments ushered in the large-scale use of markets to limit pollution: Total sulfur dioxide emissions from power plants are capped at a certain level, and utilities can trade permits representing the right to pollute under this cap. This system has proven to be a highly cost-effective way to reduce air pollution nationally, but one outstanding question is whether it might allow pollution to concentrate in particular areas. Without a more thorough monitoring network, it is impossible to know whether these so-called hot spots are a serious problem. The consequence is hesitation in further use of this potentially effective policy instrument.

Although collecting and analyzing such physical data would probably need to be the first priority of a BES, better economic and social data related to the environment would also improve our ability to design and evaluate policies. The last comprehensive survey of expenditures for environmental abatement and mitigation was in 1990. Without such data, we cannot have a good sense of the aggregate cost of our environmental policies.

A centralized database of estimates of the benefits of various environmental improvements would also be useful. New research, much of it sponsored by the EPA, continues to estimate the “services” provided by the environment to people, in the form of protected health, enhanced recreational opportunities, flood protection, and so forth. Much of it also estimates the monetary value of such services for use in benefit-cost analysis. Other researchers and government agencies routinely refer to this research for estimates that can help gauge the impact of new policies, but in doing so they must comb through the vast and disparate literature each time. A centralized database or library of the research would prevent much duplication of effort. It would also be a necessary source of data for any efforts to compute a green GDP.

There are other reasons to support the creation of a BES. A BES would facilitate one-source shopping for members of Congress, agency administrators, and the public, who currently must navigate a maze of agencies to construct a picture of the nation’s environment. In addition, an independent BES might lend more credibility–a sense of objectivity–to our environmental statistics, giving the public a commonly accepted set of facts from which to debate policy, much as the Bureau of Labor Statistics (BLS) and the Bureau of Economic Analysis (BEA) have done for economic statistics.

Indeed, our previous experience with these economic agencies provides important lessons on which we can build when establishing a BES. First, we must admit that statistics can be politically controversial. During World War II, for instance, industrial wages were linked to changes in the Consumer Price Index (CPI). At the same time, the CPI began to move out of synch with the popular perception of price changes, recording much lower inflation rates than people experienced in their everyday lives, largely because it missed the deterioration of quality in the goods that were selling at modestly increasing prices: Eggs were smaller, housing rental payments no longer included maintenance, tires wore out sooner, and so forth. The result was political uproar, with protests on the home front from organized labor. In the end, a lengthy review process, with representatives from labor, industry, government, and academic economists, resolved the issue.

The importance of regular reviews

Although environmental statistics will probably never hit people’s pocketbooks as directly as did the CPI, they can get caught in the crossfire between business and environmental groups. Building in a regular external review process would help keep the peace during such moments. Crises aside, external reviews would ensure that a BES is balanced and objective, in both fact and perception, and help improve its quality over time.

Indeed, the regular external reviews of the CPI have raised points that would be of value to a future BES. Some are academic questions about sampling and analyzing data that could be addressed within the bureau. Others, such as the need for data sharing, may require congressional action from the beginning. In our economic statistics, there is substantial overlap between information collected for the U.S. Census (housed within the Department of Commerce), the unemployment statistics and the CPI (collected by the BLS), and the GDP (collected by the BEA). To address this concern, Congress recently passed the Confidential Information Protection and Statistical Efficiency Act, which allows the three agencies to share data and even coordinate their data collection.

Similar data-sharing issues would arise regarding environmental statistics. Currently, environmental statistics are collected not only by the EPA but also by the Departments of Agriculture, Interior, Energy, and Defense, and in some cases by multiple bureaus and agencies within these departments. Even some of the economic statistics collected by the Census Bureau, BEA, and BLS would overlap in a complete picture of environmental statistics. Coordination across these agencies–and in some cases consolidating tasks into the new bureau–would be essential for generating the best product without duplication of effort.

An additional insight gained from looking back on our experience is that economic statistics now play a much larger role in our economy and in economic planning than originally envisioned. Most generally, they have been used as a scorecard for the nation’s well-being, a basis for leaders to set broad policy priorities (to stop inflation or spur growth), and a basis for the public to assess its leaders. At a more detailed level, they now fit routinely into the Federal Reserve’s fine-tuning of the economy. Finally, through indexing of wages and pensions, tax brackets, and so on, the CPI automatically adjusts many of the levers in the economic machine.

A centralized database of estimates of the benefits of various environmental improvements would also be useful.

One could imagine environmental statistics playing each of these roles. First, despite their current weaknesses, environmental statistics already help us keep score of our domestic welfare. Second, they increasingly could be used to adjust policies. Initially, they may serve as early warning signals for problems approaching on the horizon (or all-clear signals for problems overcome). Later, as the data develop and policies evolve to take advantage of them, they may even be used in fine-tuning: On theoretical drawing boards, economists have already designed mechanisms that, based on regularly collected data, would dynamically adjust caps for pollution levels or annual fish catches. The only thing missing is the data with which to make such mechanisms possible.

A final lesson learned is that high-quality statistics cannot be collected on the cheap. We currently spend a combined $722 million annually on data collection for the U.S. Census (excluding special expenditures for the decennial census), the BLS, and the BEA, and more than $4 billion each year for statistical collection and analysis throughout the federal agencies. During the past three years, these budgets have increased at annual rates of approximately 6.5 percent and 9.7 percent, respectively. Nevertheless, these efforts are widely considered to be well worth the cost.

By comparison, the current budget of $168 million for environmental statistics seems small. Consider that in 1987 (the last year for which comprehensive data are available) the annual private cost of pollution control was estimated to be $135 billion, and that government spends $500 million a year for environmental enforcement. With approximately 2 percent of our GDP at stake in these expenditures, and the welfare of many people, a top-notch set of environmental statistics seems long overdue.

Your participation enriches the conversation

Respond to the ideas raised in this essay by writing to [email protected]. And read what others are saying in our lively Forum section.

Cite this Article

Banzhaf, H. Spencer. “Establishing a Bureau of Environmental Statistics.” Issues in Science and Technology 20, no. 2 (Winter 2004).

Vol. XX, No. 2, Winter 2004