Informed Public Economic Policy
Public policy making in the United States is a large and complex undertaking. In 1997, new legislation amounted to some 2,691 pages and changes in regulations required 68,530 pages of the Federal Register . Much of this activity is devoted to economic concerns, and the laws and regulations are means to many different economic and social objectives. The link between policies and economic objectives is often difficult to anticipate. For this reason, policy makers often seek to inform their understanding of the effects of policy changes before changes are put into effect.
As the issues confronting society and government have become more complex, so too has the study of effective economic policy. In the last 50 years, the increased involvement of government in economic policy has been accompanied by more detailed collection of data, expanded production of information of many kinds, and intensified economic analysis of the effects of alternative policies. In fiscal 1996, some 29 agencies of the federal government carried out significant economic information collection or research in support of their mission, at a cost of $182 million.1 The Economic Research Service (ERS) of the U.S. Department of Agriculture (USDA), in its previous manifestation as the Bureau of Agricultural Economics, is one of the oldest agencies of the federal government whose primary mission is providing economic research and information in support of
both public- and private-sector decision making. In fiscal 1996 its budget was nearly $53 million.
To understand and evaluate the function of any such federal agency, it is necessary to first appreciate the reasons for public economic policy in a representative democracy with a market economy. These reasons drive the policy making agenda. The provision of research and information in support of public economic policy is itself an economic activity. Whether the right amount of research and information will be provided by the private sector, and, if not, how government should intervene, is itself a very interesting question of public economic policy. The answer depends on the relevant characteristics of this economic activity. Much of this report is an analysis of these characteristics and their implication for how the production of research and information in support of public economic policy should be organized. The themes set out in this chapter reappear subsequently in this report in the detailed consideration of the specifics of the ERS.
The Nature of Public Economic Policy
Governments regularly intervene in market economies in representative democracies when the conditions that are necessary for markets to produce an efficient and equitable distribution of resources do not exist. Although specific reasons for intervention are many and vary with times and issues, many interventions can be ascribed to one of several kinds of actual or perceived failures of markets to produce efficient outcomes. Sources of these failures include natural monopolies, externalities, public goods, barriers to information, ill-defined property rights, and considerations of equity
Markets can lead to inefficient production levels in a particular industry, if it is technically most efficient for the good produced in that industry to be provided by a single firm. This will happen if the cost per unit of production continues to go down as more of the good is produced. For example, it was technically inefficient for more than one railroad to provide service to a local community, in most cases. The market outcome is a monopoly firm, which then charges a higher price than would be charged in a competitive industry. Consumers will demand less, and too little of the good will be produced. The usual solution is that the industry is regulated, or even owned, by the government, with the objective of producing a more efficient volume of the good. In the case of nineteenth-century railroads, the Interstate Commerce Commission was created to oversee freight rates. The cost of information sometimes continues to go down as more of it is produced by the same organization. For example, the Census Bureau produces the large national censuses and several household surveys. These products are widely used in the private sector as well as for public policy making.
Markets can lead to inefficient production levels because the decisions of an individual or an organization cannot be excluded from affecting the economic interests of other individuals or organizations. This kind of failure is called an externality. For example, if a large hog-raising enterprise is set up in a rural community, there may be consequent air and water pollution that adversely affects nearby residents. Various farming activities may degrade water quality if carried out in sensitive watersheds. In such cases, government intervention may be able to transfer costs borne by others to decision makers, thereby leading to a more efficient resource allocation. This may take the form of defining property rights: for example, if it is established that the rural community owns the water rights, then the community and the hog-raising enterprise may come to terms about water quality and compensation.
If a commodity can be consumed by many individuals, and consumption by one does not reduce the opportunity for consumption by others, then the commodity in question is said to be nonrival and is a public good. An important class of examples is new information that contributes to the productivity in an industry, with many competitors, like farming. If it is not possible to charge each individual for his or her consumption of the commodity, the commodity is nonexcludable. Commodities that are both nonrival and nonexcludable are pure public goods. Although pure public goods are demanded by many individuals, it is difficult to get individuals to pay for them voluntarily. If others pay, then a given individual who does not contribute toward provision of the good is still able to consume the good in exactly the same way and extent as if they had contributed. National defense, and in many instances important new information, are leading examples of pure public goods. Too little of a pure public good will be produced because it is difficult to get individuals or firms to pay. Government may intervene by producing the good directly, for example by providing education about new farming practices or statistics on prices and production. It might intervene indirectly by redefining property rights to embody the information in an excludable good, for example by creation of a patent system. Since an informed public is essential to the performance of democratic institutions, and since many forms of information are pure public goods, the potential for underproduction of information in a market-oriented democracy is public policy issue of the highest magnitude.
Barriers to Information
Markets can lead to inefficient production levels if individuals and firms do not have common information. If some individual or firm is ill-informed relative
to another, it may find it difficult to profit from exchange with the better-in-formed party, and there may be no exchange at all. A fruit grower knows what pesticides he applied to his product, but if the retailer and the consumer do not, then they may have to take costly measures to reduce the risk of illness; alternatively, the grower and the retailer may find it costly to communicate information about pesticides to consumers in a useful way. Government intervention may take the form of food safety regulations.2
The public good character of primary and secondary data, and analyses based on these data, can lead to asymmetric information when there are many small participants on one side of a market for a commodity and a few large ones on the other. Each large participant realizes some return from investing in data collection and analysis, but on the other side of the market there is negligible return from these activities for my one participant. In this situation, the level of production of the commodity is very likely to be too high or too low. Government intervention to provide data and analyses can then bring about a more nearly optimal level of production of the commodity. Political support for this intervention may be grounded in considerations of equity, as has been the case historically in many agricultural markets.
Markets function only in the context of well-defined property rights. As technology changes, issues of property rights continually emerge, and a primary function of government is to establish property rights appropriate to the state of technology. For example, advances in molecular biology have greatly accelerated the development of new strains of crops, including those resistant to disease and infestation. What are the intellectual property rights associated with the new strains? Advances in communication, including cellular telephones, cable television, electronic mail, and Internet services, have greatly multiplied the uses of the electromagnetic spectrum. In view of these changes, what form should media ownership take? As the impacts of industry and agriculture on the environment become better understood, new kinds of property—for example, an upper atmosphere undepleted of ozone—take on value, and questions of ownership come to the fore. Only governments can provide the constantly changing infrastructure of property fights essential to the efficient functioning of private markets.
Governments often intervene in markets deliberately to change the distribution of income and wealth. The intervention may be motivated by broad consensus about an appropriately equitable distribution of income, or it may reflect the give and take of particular interests with respect to specific issues. But the distribution of income is invariably changed by any government economic intervention. Understanding, much less accurately anticipating, the distributional and other effects of economic interventions is a difficult undertaking, even when conducted by a disinterested and skilled party. For example, crop price support programs have had their ultimate impact on the value of land, not on the earned incomes of farmers, and at the same time they have adversely affected soil conservation by creating disincentives for crop rotation. The greater is the uncertainty about the distributional implications of existing and proposed interventions, the more likely it is that specific groups will claim or fear widely varying consequences.
Government economic intervention is undertaken for a wide variety of other policy purposes, as well. For example, in the United States as well as other industrialized centuries, there are long-standing programs to maintain or improve the welfare of rural populations. In an earlier era, programs for this purpose included rural free delivery and rural electrification in the United State. Today, these programs include measures to increase the flow of information to rural areas, for example through Internet access, and federal partnerships with nonprofit organizations providing assistance to rural areas.
Why Government-Sponsored Research?
Government supports many research programs. In 1997, federal support for nondefense research and development was $28 billion (National Science Foundation, 1998a). Most of this expenditure is in basic and applied physical and biological sciences. Some federal agencies, for example the Agricultural Research Service of USDA, conduct this sort of research primarily within the agency; others, like the National Science Foundation, primarily sponsor research in the academic sector. This is an important economic intervention. It occurs primarily because the outcome of basic scientific research is usually a pure public good; the outcome of applied scientific research can be, as well. The key economic property of the research outcome is whether or not it is excludable. Breakthroughs in basic mathematics and the development of an improved surgical technique are
examples of nonexcludable public goods. The discovery of a new drug and the design of a new computer chip are public goods, but they are excludable through the patent system. Given the large number of corn producers in the United States, basic development of hybrid corn earlier in this century was a pure public good, and it was not until the 1930s that replication of inbred lines and distribution become excludable and thus primarily a function of private-sector firms. A vertically integrated producer of poultry with sales in the billions of dollars will have incentives to undertake both basic development and subsequent marketing of a patented, excludable commodity.
Government-sponsored data collection began with the constitutionally mandated decennial census in 1790, and the earliest collection of agricultural data by the federal government was undertaken by the Patent Office in 1840. Data are manifestly public goods: use of data by one party in no way diminishes its usefulness to other parties. They need not be pure public goods, because they can be excluded. For example, econometric consulting firms provide data in electronic form under licensing agreements that prohibit further disclosure of the data. This excludability of data, a comparatively recent development, was much costlier in the nineteenth century when USDA began systematic data collection. Similarly, if data are collected by literally going door to door, it is more efficient for one collector to gather all the information from that door, than for many collectors to go to the same door. With electronic communication, there can be many data collectors, and the case for a natural monopoly weakens. Data collection by a disinterested party can ensure that all parties have access to information in a way that collection by an interested party cannot. This does not necessarily imply that government must intervene. In many industries, the best data are collected by parties under contract to industry-wide associations. In some cases, firms in the industry do not trust each other or their association, and data collection is contracted to a trusted government agency.
Data collection at USDA evolved naturally into monitoring and reporting, as detailed in Chapter 3. The line between data and the interpretation of data is subjective, but by the 1920s the BAE was providing forecasts of commodity prices. Monitoring and reporting began at a time when, by comparison with the present day, information was very limited, sophisticated financial markets for the purpose of conveying information scarcely existed, and there was widespread misunderstanding of the role information in a modern economy.3 The case for nonexcludability of monitoring and reporting, as well as an asymmetric, informational disadvantage for farmers, was stronger in the 1920s than it is today, Government agencies today collect data extensively, but the monitoring and reporting function is often left to the private sector. For example, the Forest Service of USDA sells timber tracts at auction and keeps detailed records of sales and subse-
quent logging of the tracts. But it does not provide the some sort of monitoring and reporting that ERS provides in its Situation and Outlook reports.4 That function is fulfilled by a private firm that sells a detailed newsletter to mills in the industry.5
Why Research in Economics?
The past two centuries have seen by far the greatest advancement of living standards in human history. The development of new technology has been essential for these advances, but technology alone has not been sufficient. The invention of the steam engine had its effect on living standards through the development of new industrial processes and rail and sea transportation, all of which required much more than technological advances. The economic success of rail transportation, for example, also required innovations in property rights, the development of a large supporting economic infrastructure, and eventually substantial modifications of the market economy itself. The history of these changes can be read in many Supreme Court decisions of the late nineteenth century. A generation later, a similar process of institutional change accompanied the economic impact of the automobile. The process is being repeated today with innovations in information technology. Realizing the full potential of these innovations is entailing changes in work habits, in definitions of intellectual property, and in the education system.
The changes that must follow scientific and engineering innovations for these advances to contribute to human well-being are primarily economic changes, broadly defined. The need for economic knowledge grows out of the need for institutional change and improvements in institutional performance, driven by technical progress. The knowledge required includes definitions of property rights and an understanding of the ways in which markets will succeed and will fail—in the ways just considered—as society accommodates, and exploits new technical knowledge. Technical knowledge, and its application, are becoming more and more widely available: software is written in India, shoe manufacturing takes place in Malaysia, and light aircraft are produced in Brazil. In a world of freer trade and shared information, the relative success of nations depends as much on the successful adaptation of their economic systems to technical innovations as it does on the output of their research laboratories (Porter, 1990).
The systematic, analytical study of the economic aspects of scientific progress has been essential to the rapid advancement of American living standards in the twentieth century (Ruttan, 1984:552):
These reports are described at the start of Chapter 5.
Government provision of information to rural areas could be undertaken in an effort to subsidize and thereby maintain small rural businesses and thus rural populations. In such a situation, this kind of information provision might be compared with other subsidies for the same purpose, for example, tax incentives for firms to locate in rural areas.
Throughout history, improvements in institutional performance have occurred primarily through the slow accumulation of successful precedent, or as a by-product of expertise and experience. Institutional change was traditionally generated through the process of trial and error much in the same manner that technical change was generated prior to the invention of the research university, the agricultural experiment station, or the industrial research laboratory. With the institutionalization of research in the social sciences it is becoming increasingly possible to substitute social science knowledge and analytical skill for the more expensive process of learning by trial and error.
The contributions of economic research are to identify the changes in institutions and physical and human capital necessary to exploit new technology and to identify the effects of these changes. It offers the opportunity to suggest the changes that are most likely to succeed, and thereby to reduce the costs of institutional innovation. Changes in institutions include modifications of government policy that address market failures and inequities in income distribution, as well as property rights and other aspects of the political infrastructure in which a market economy functions.
A particularly important contribution of economic research in a representative democracy is to identify in some detail the effects of changes in policy (or, for that matter, the effects of leaving policy unchanged in the face of changing technology) on different economic interests. It is rarely, if ever, the case that a change in policy will leave every interest in an improved or equivalent condition. But if those who gain do so enough that those who lose can in turn be compensated in such a way that no party loses ground, then a combination of a change in policy and redistribution maybe politically feasible. The identification of potential winners and losers beforehand is a contribution of economic research that can facilitate the political process.
Why Public-Sector Research in Economics?
The need for economic knowledge, driven by change in technology, is both private and public. Private firms allocate very substantial resources to the acquisition of economic knowledge in the pursuit of economic efficiency. This is especially the case for large enterprises, for example multinational firms that deal in a variety of legal and institutional environments. Markets for information have become extremely important components of the world economy, including not only securities but also derivatives of securities like futures and options contracts.
Addressing Difficult Issues
The largest and most difficult questions remain public, however. In a representative democracy, the establishment of property rights is inherently a public issue. Problems of market failure from environmental externalities alone demand
increasing public attention. Issues of income distribution are as important today as they have ever been. Economic knowledge has been developed and applied in addressing many of these questions. For example: the creation of property rights and subsequent allocation of part of the electromagnetic spectrum has made use of research on auction design (McAfee and McMillan, 1996). The institution of time-of-use pricing to delay the construction of new electricity-generating capacity has drawn on extensive controlled experiments carried out by econometricians (Aigner, 1981). Concepts of property rights have been extended and new market mechanisms have been developed as policy makers have balanced the amelioration of negative environmental externalities with economic efficiency. The creation of a market for sulfur dioxide emission licenses for electric power plants is one example. Another is the Conservation Reserve Program designed largely by ERS, in which farmers bid to take environmentally sensitive acreage out of production for 10-year periods.
Use of public economic research by one organization or individual does not diminish its value to anyone else. It is nonrival and therefore a public good. Many kinds of public economic research are also nonexcludable and are therefore pure public goods. For example, advances in computable general equilibrium models have made more timely and accurate anticipation of changes in taxes and tariffs possible, but new ideas as such cannot be patented, and these advances are therefore nonexcludable. Private markets will underproduce economic research because of its pure public good character. Economic information is also nonrival, but it may be excludable. For example, ERS regularly provides information of keen interest to various industries, as do industry newsletters, but given current technology this information is often excludable (for example, by limiting electronic access).
Uses of Private-Sector Research
In apparent contradiction, there is widespread production of economic knowledge by private interests on many economic questions. Any contemplated important change in public economic policy is likely to bring forth a host of studies originating in the private sector. On one hand, it is vital for private-sector interests to be able to identify impacts of contemplated policy change that may have gone unnoticed. On the other hand, many of the effects identified by private-sector interests are rent seeking: that is, a private-sector interest that stands to benefit substantially from a proposed change may support that change with evidence indicating that it will benefit some wider group. The claim may or may not be true, but in this situation the private-sector interest cannot be expected to provide evidence on who stands to lose.
The outcome of this sort of private-sector research is well understood and well anticipated by a simple analysis of market failure. To the extent that the gains or losses of a proposed policy change are spread out, with no single interest
much affected, no single interest will have an incentive to produce the economic knowledge. To the extent that the gains or losses are concentrated on a small collection of interests, the incentive to produce the knowledge increases. This situation has at least two very undesirable features. The first is that policy changes tend to reflect concentrated interests. The second is that this fact may emerge only well after the change has been made, if at all, in an environment of ongoing changes in diverse economic policies. This is not even policy experimentation: it is policy chaos.
Advantage of Prior Analyses
Public-sector research in economics complements but does not displace private research. Gathering evidence on likely outcomes before the fact, rather than after, affords the potential for large gains. First and most important, it provides the only alternative to carrying out the field experiment of actually making the policy change. For example: research on auction design is a very attractive alternative to uninformed experimentation with the rules for government auctions, for example in the sale of the electromagnetic spectrum and the leasing of productive cropland for conservation purposes. Controlled experiments in time-of-use pricing for electricity on a small scale indicated the pricing schedules that would best achieve energy conservation and environmental goals on the large scale. And the work of ERS in the 1980s provided the basis for prior assessment of important aspects of current world trade agreements on agriculture, whereas experimentation with actual policy change would have been extremely expensive.
Second, studying outcomes before rather than after the fact can identify gainers, losers, and the potential for redistribution from the former to the latter, thereby providing sound information from which essential political agreements can be struck. For example, ERS demand modeling made an important contribution in support of the Uruguay round (1986–1994) of trade negotiations under the General Agreement on Tariffs and Trade (GATT): it identified the important domestic gainers and losers in the principal agricultural exporting nations. This enabled policy officials to convey to Congress that domestic political interests who opposed the agreement were overstating their potential losses, and to make the case that the losses of losers from liberalized agricultural trade would be more than offset by gains of the winners.
A third attraction of economic analysis before the fact is that it suggests what should be monitored after the fact in order to evaluate the impact of the change in policy. This last contribution remains intact even if public-sector research on the issue is ultimately ignored in reaching the decision about policy change. The evidence gathered subsequently may have bearing on future policy questions and can be used to compare the consequences of policy change with predictions before the fact, thereby improving economic analysis. For example, ERS has long
monitored research expenditures and measures of farm productivity to facilitate the assessment of returns to research in agriculture. It maintains information on agricultural output and prices and models of supply, which have been used to identify the effects of proposed changes in subsidy programs during the consideration of major farm legislation that must be renewed about every five years.
Economic research is carried out in academic institutions, in private firms, and in government agencies (primarily federal). The product of research activities is new knowledge. A straightforward consideration of the market for this new knowledge typically explains the kinds of economic research conducted in these three sectors, as the following two examples illustrate.
Many assets, goods, and services are bought or sold by means of auctions. This is especially the case for assets, goods, and services acquired or divested by governments. Auctions can be conducted in a wide variety of ways: bids can be oral or written, the price in an oral auction may be ascending or descending, the price paid many or may not correspond to a highest or lowest bid, the seller or buyer may or may not announce a minimum or maximum acceptable price. To a buyer or a seller organizing an auction, the most important question is how to minimize or maximize (respectively) the price of the object being bought or sold.
The relation between the organization of the auction and the transaction price of the object depends on economic characteristics of the object and the potential buyers and sellers that transcend specific settings. Knowledge about this relation is a public good, because it is not diminished by its use, and it would be difficult to both apply this knowledge and exclude it from others. Thus it is not surprising that advances in the theory of auction design (Engelbrecht-Wiggans et al., 1983; Hirshleifer and Riley, 1992) have taken place almost entirely in the public sector. The theory identifies the economically relevant characteristics of an auction that in turn predict the consequences of alternative auction designs. Ascertaining these characteristics for a particular object being sold may or may not be a public good. For example, procurement of milk for public school lunches is a similar process in most states and school districts. Determination of the economically relevant characteristics is a public good, and no one school district has much incentive to carry out this research. Most of the work has been done by academics. In contrast, the features of the market for procurement of weapons by the U.S. Department of Defense are unique to that market, and the department has sponsored research to study these features and improve the design of its procurement auctions.
A second example is the research that has provided the basis for time-of-use pricing of electricity. Because the generation and delivery of electricity has been a natural monopoly, it has been regulated by a public utility commission in each state. Demand for electricity varies systematically throughout the day and
throughout the year. Since electricity cannot be stored efficiently, generating capacity must be adequate to meet the maximum rather than the average demand for its use. However, by changing the price of electricity systematically throughout the day or the year, it is possible to alter the systematic variation in demand. Recognizing the advantages of studying policy changes before rather than after the fact (Aigner, 1981), the (then) Federal Energy Administration supported a series of experiments to quantify the relationship between the pricing policy and the demand for electricity. Constructing the econometric framework and sampling design for these studies raised similar methodological issues across states. Research addressing these issues was carried out predominately by academics, with findings published in research journals and therefore widely available. In applying the econometric framework and sampling design, issues specific to each state arose, since industry mix and seasonal demand for electricity vary widely by state (Aigner, 1981). This work was sponsored by individual state utility commissions, typically by contract to individuals or private-sector research organizations.
In both of these examples, the basic research produced knowledge that is close to a pure public good. The economics of auction mechanism design and the econometrics of electricity pricing experiments are in no way diminished when they are applied to yet another policy problem, and it is essentially impossible to exclude this knowledge from use by others. Indeed, centuries of experience have established that basic research is most successful when it is publicly available, so that those who do the same kind of research can criticize it and build on it in a timely fashion. Economic and many other kinds of research in the United States is based on a partnership between the federal government and academic institutions, one that explicitly recognizes the public good nature of basic and applied research and is designed to produce it efficiently.
This partnership is founded on two secondary marketplaces. In one, academic advancement and salaries are based on the contribution to knowledge as measured by peer evaluation, largely through the medium of peer-reviewed scientific journals. In the other, resources for carrying out research are provided by the federal government based on the prospects for their efficient use as measured by past performance and peer evaluation of proposals. The two secondary marketplaces are closely related. They provide a very strong set of incentives for creative, productive work that has made the United States the leader among all nations in economic and other research. It brings many of the brightest young scientists to the United States for training, and many of the best of these stay on and make further vital contributions.
In both the auction design and electricity pricing examples, the basic research conducted in the public-academic sector was used in private decisions as well as in the making of public policy. In both cases, the decision makers were aware that new research coming out of the public-academic sector could contribute to a better decision. They determined the bearing of this research on the
decision they faced, incorporating it into the decision itself. Establishing the link between basic research, on one hand, and a decision, on the other, is a critical step that typically demands substantial time, talent, and resources. If the link has previously been made by others in similar situations, then it is inefficient to repeat the process from the ground up each time. This gives rise to extensive markets in applied research in support of decision making, involving both private- and public-sector entities. In the private sector, it includes consulting firms and newsletters. In the public sector, it includes university professional schools and some of the activities of community colleges and vocational schools.
Effective Economic Policy Analysis
More novel circumstances are likely to demand greater sophistication in bringing economic knowledge to bear on the question at hand. To the extent that a situation is novel, the simple model of imitating others in similar situations is more likely to lead to a bad outcome, and the more critical it becomes to establish a solid economic basis for the decision. In the private sector, novel decisions are faced by firms in new industries, and markets provide substantial rewards for making such decisions effectively. In the public sector, decisions in novel circumstances must be taken by governments confronting the institutional aspects of technical progress for the first time. The U.S. federal government is in this position more often than any other public institution. But there is no market that provides timely rewards for the effective application of knowledge in these critical decisions. The only available substitute is a public policy that effectively links basic economic research to decision making.
The link between research and economic policy decisions is economic policy analysis: the disinterested prospective and retrospective evaluation of the economic and social implications of changes in public policy, and the effective communication of these evaluations to public policy makers. Effective policy analysis requires understanding proposed changes in public policy and the problems that drive those proposals, in order to frame the implications of changes as questions that have been, or can be, addressed in basic and applied economic research. At the level of federal policy, it is unlikely that the proposal at hand will have been addressed in the best way by previous applied research; in any event, knowing the body of applied research that comes closest is a demanding profession. Policy analysis does not emerge as the product or by-product of basic and applied research elsewhere or in economic research in support of private decisions. Effective and therefore timely policy analysis cannot be delegated to the academic sector, and evaluation that is both disinterested and relevant will not be carried out in the private sector.
For public policy support research to maintain its promise of reliably identifying the likely effects of changes in policy and institutions, it must have four characteristics. It must be of high quality. By definition, research of higher
quality provides a more reliable assessment of change than research of lower quality. Research that does not meet disciplinary and professional standards, or is grounded in poor or inappropriate data, runs the risk of being dismissed in favor of higher-quality research, including research carried out by stakeholders in the policy question at hand. Public policy support research must be relevant. Research that does not address the essential policy question will either be dismissed by policy makers or, in addressing the wrong question, may provide the wrong answer. It must be timely. Research undertaken prospectively may avoid the costs of social experimentation. Moreover, research that is conducted in anticipation of policy questions, and concluded before political lines are drawn is more clearly not beholden to specific interests than is research that is carried out after positions, including the position of a cabinet secretary, have been established. Public policy research must be credible. This characteristic derives from the first three. Any entity that regularly produces high-quality, relevant, timely research will enjoy a reputation for credibility, as well as independence from narrow parochial or political interests.
These characteristics can be maintained only if the distinction between policy analysis and the decision itself is maintained. Ideally, research in support of public policy sponsored by a federal agency will be the same, regardless of the political position of the president or the secretary, whereas the way in which that analysis is used will depend very much on those officials' political and policy positions. As Chapter 3 reveals, the history of the BAE and ERS amply demonstrates how destructive direct involvement in the political, problem-solving process is to policy analysis and its promise of improving public decisions. If policy analysis survives at all in this environment, then it is no more credible than the research of private interest groups. The more likely outcome is the departure of key members of the research group and its subsequent decline or dissipation. The decline of BAE prior to its dissolution in 1953 documents this consequence of mixing research and policy decisions or implementation. In the short run, the forces that would bend the factual outcome of policy analysis to political purposes are strong, but, in the long run, yielding to these forces renders policy analysis useless and policy decisions less well informed. In the long-term, open and credible policy analysis supports good decision making, can be a potent political and positive force, and can generate the political support necessary to protect and sustain credible research and analysis in government.
What Should We Expect from the Research Process?
The outcome of a research project is never clear when the project begins. The less routine is the problem addressed, the greater is the uncertainty about the outcome. Many research projects will have disappointing outcomes: new basic research may not exhibit the promise it seemed to hold; a key conjecture may turn out to be right but of little use in addressing the question; a new data set may turn
out to have the same limitations as a data set previously available. But in a smaller set of cases, the payoff may turn out to be high. For example, liberalization of agricultural trade in the Uruguay round of trade negotiations under the GATT agreement is estimated to have net benefits of up to $8 billion annually for the United States and $30 billion annually worldwide (Economic Research Service, 1989:Table 10). ERS had a small but vital role to play in the negotiations, and its annual budget is a very small fraction of the net benefits realized from the new agreement. Policy analysis, like research, shares an important characteristic with drilling for oil: failures often outnumber successes, but the gains from the successes exceed the costs of the failures.
For research in support of public policy, there are two further apparent risks. The first stems from the need to anticipate policy problems. In some instances, the time horizon for policy questions is clear. For example, the next round of World Trade Organization negotiations will begin in 1999, and trade issues in agriculture will be on the agenda. In other instances, it is not: hog waste may remain a local environmental problem, or it may emerge as a policy question at the federal level. It is inevitable that a group formed for public policy research support will anticipate some questions for which it is never asked to have problem-solving research available. Research in support of public policy is somewhat like investment in defense, made in anticipation of many contingencies, most of which may never arise. Sponsors of such efforts must accept this as a cost of the other successful applications of research.
Second, even an important, relevant contribution of policy analysis may not be reflected in the final policy decision. The decision is the outcome of a political process, and those who engage in policy support research are well aware that a proposed policy innovation well grounded in excellent research may not survive an eleventh-hour compromise.
These two features are risks from the narrow perspective of evaluating the utility of research with respect to the question that motivated it, but they are less risky from the appropriately broader perspective of national policy. Much of the research undertaken for policy support at the federal level will also provide useful support at the state and local levels, precisely because the questions it addresses emerge at those lower levels of government rather than as national policy issues. Indeed, the public good characteristics of research argue strongly for a federal role in support of problem-solving research even when used only by states and localities. The argument is all the more compelling with the devolution of federal programs to the states. Problem-solving research that ultimately does not affect policy is often still useful in retrospective evaluations; if the policy turns out to have addressed the problem poorly, then its day may come again.