that are growing in importance, as the Members of this Committee have already stated in their comments. Cost analysis, economy and efficiency auditing, contributions to improving financial management and information systems, performance auditing and selective program evaluation are crucial functions, particularly in an era of stringent budgeting, changing programs, and the re-engineering of processes that is going on across the government, as well as, I might add, in the private sector.

The point to be emphasized is that when major changes are occurring particularly when operational decisionmaking is being decentralized-auditing, evaluation, and performance measurement become enormously important. Congressional staff and executive agency resources for these activities are declining and apparently will continue to decline. The panel found that GAO at its best is a vital resource to Congress, to government, and to the public in general. Within the context of changes currently being made and contemplated, that importance is likely to increase.

Our panel did conclude that GAO "gotcha" reports—that is, narrow findings suitable for headlines that a public agency is doing something wrong, or could do something better-are often not useful. Historically, they do not lead to management improvement. In most cases, they do not provide specific practical agendas for institutional change or clear estimates of the costs of recommended changes. Requests for GAO work for the purpose of creating scenarios for Committee hearings critical of agencies, without comprehensive review of comparative costs and benefits of alternatives, do not contribute much to the constructive improvement in government. And I know that the members of the panel would endorse the comments that have been made by Members of this Committee that the need for change relates as much to Congress as to GAO. Second, objectivity on the part of GAO and broad perception of its ability to carry out competent, fair, and independent analysis are its most important assets. The panel recognized that in recent years there has been considerable criticism about GAO's work, including concerns about its objectivity and impartiality. The panel found no evidence of any deliberate partisan bias in GAO's work. The panel made several recommendations aimed at assuring the continuing objectivity and independence of that work.

Congress should not, in our judgment, lose sight of the fact that GAO is the National Government's central auditing agency and that all democratic industrial governments have such an agency. The United States, through its foreign aid programs and in its aid to developing countries and the countries emerging from communism, is urging they should create such agencies.

GAO's status as the government's central audit agency for audit and evaluation requires bipartisan trust. The panel's recommendations for serving that objective include: Sharing GAO's strategic planning and priority setting with its congressional oversight committees and others; its adherence to professional audit standards, including obtaining comments from audited agencies on all its reports—and I emphasize all; its adherence to professional program evaluation methods for studies in that category; and restraint on the part of congressional Committee leadership and individual Members in requesting GAO to undertake studies that will draw

it into partisan issues that lack a solid factual base for empirical judgment.

Like most private sector businesses engaged in reorganization and re-engineering, GAO should define and focus resources on its core activities, on those activities not provided by other organizations and to which its legislative mission gives priority. If it is to maintain its objectivity and serve its core mission, GAŎ should not be expected to take positions on broad policy issues subject to political debate. Should taxes be cut? Should taxes be raised? Should government grow? Should government shrink? Should the Federal Government's responsibilities shrink and State and local government grow? These are not issues that belong on the agenda of an audit agency. But GAO should build the capacity to assess how agencies are performing their identified public missions and how they are dealing with changes emerging from policy decisions, and also to assess the impact of changes on both program costs and results-no small set of tasks, I might say.

Third, the panel did not address the question which I understand-and it has been reinforced this morning-is now a central concern of this Committee, namely, the impact on GAO of a very substantial additional cut in its budget. The panel did note, however, that GAO was in the midst of a major downsizing at the time of our study. The size of GAO's staff has been relatively stable at about 5,100 from fiscal year 1979 through fiscal year 1992. It has downsized from 5,062 staff in fiscal year 1992 to 4,900 in 1993 and 4,500 in 1994. Consequently, immediate and significant additional reduction in GAO resources could adversely impact the organization's ability to carry out its statutory mission and respond to congressional requests for GAO services.

The panel found that Congress-through its Committees and individual Members-has requested from GAO a level of work that has severely taxed the agency's resources. GAO has tried to respond to every request from congressional Committees and Members of Congress that falls within its general issue areas and does not unduly duplicate other work-often including requests generated by congressional staff of which the Members themselves may not be fully aware. In addition, a considerable number of projects emerge from issues identified in prior GAO studies and from the internal GAO strategic planning process which are usually supported by a congressional request letter.

I might add, in parentheses, that the panel noted that the percentage figures on the share of congressionally requested versus self-initiated work is of uncertain meaning.

Although not designed to reduce costs or to downsize GAO, a number of the panel's recommendations could, we believe, improve GAO's efficiency and effectiveness, because they emphasize on preserving and improving quality and cost-effectiveness by making GAO's work processes more open, visible, and efficient. One category is the group of recommendations that would open up scrutiny of GAO work priorities to public and congressional discussion. Another group of recommendations would alter internal GAO work processes in ways that would shorten the time needed to complete work while simultaneously improving the quality of the producta comment not unrelated to the charts on display here.

Before summarizing these recommendations, it is important to point out that some of the panel's recommendations may actually require, in the short run, additional expenditures for certain types of resources, for example, increased capability in computer hardware and software for analysis and communications. While GAO is making progress in these areas, it is not up to what might be called "state of the art." Another example is the panel's recommendation that GAO strengthen its skills in cost analysis and financial management systems. At present, GAO does not have all the capability needed for sophisticated cost/benefit analysis. In the long run, addition of these resources and capabilities will save money, but they may not in the short run.

Finally, in order to open up GAO's internal processes and contribute to its efficiency and effectiveness, the panel recommends that GAO establish clear "terms of reference" for all its projects. I believe, by the way, this is at the heart of our recommendations. What do we mean by TORS? TORS for a GAO project would outline clearly, in writing, objectives and research questions, the scope of the work, general methods and sources to be applied, staff skills, costs, and time needed to complete the work. GAO staff, managers, and requesters would agree on the TORS before the job is approved and the work begins.

This is not a new invention, by the way. TORS are extensively used in other parts of the government as well as in the private sector. Studies and other services contracted for by Federal, State, and local governments from the private and non-profit sectors have long required specific terms of reference in the bidding or the procurement process. Business-to-business contracting for research and development and other consulting services also requires clear benchmarks, cost and schedule estimates. The TORS identify the product that the organization is providing at an agreed-upon price.

The panel concluded that effective use of TORS would have three major impacts: Improving the efficiency and effectiveness of GAO operations; linking its work agenda more effectively to congressional priorities; and giving higher visibility to all of GAO undertakings.

First, better definition of jobs and the teams assigned to them up front will improve quality, substantially reduce review and rework after drafts are prepared, decrease time and costs of jobs, and give job teams clear benchmarks to meet. They will also better define sources and methods, thereby avoiding later criticisms of weak or biased research.

Second, linking GAO's work agenda more effectively to congressional agendas-the use of TORS as recommended by the panel would produce substantial improvement. To elaborate on one, each job would have a cost tag attached before it was authorized. This is a significant contribution because of the fact that the demand for free goods is often unlimited and, therefore, there needs to be some effort to control that.

The panel did reject the recommendation that there be established some kind of voucher system which Congress would use in ordering reports, but we did suggest that a great deal more attention should be devoted to getting at cost questions relevant to GAO's undertakings.

The panel considered but did reject the idea of a bipartisan congressional board to provide prior review for requests for GAO work. The volume of work required is simply too great for a prior review by a special board; however, we do suggest that GAO establish a peer review system by external reviewers, but that there not be a pre-job review by a congressional Committee, which would probably end up being done by staff.

In conclusion, the Academy panel, members of which began from quite different perspectives on GAO, produced a consensus report. The panel concluded that GAO, an institution created almost 75 years ago, is essential to functioning democratic government. While the panel does not see a need for any major legislative changes, it found opportunities and needs for improvements in processes and perspectives in both GAO and Congress. Now it is up to Congress to ensure that such improvements are made and that GAO's capacity to perform its basic mission is preserved and even enhanced.

That, Mr. Chairman, concludes my summary of our report. I do urge the Committee to ask us any questions they wish, and also, if they have not had the chance, to take a look at the full report. Thank you very much, sir.

Chairman ROTH. Well, thank you, Mr. Campbell.

In the NAPA report, you state that no deliberate bias was found in GAO and its work, and you also state that Members of Congress are often to blame for making requests which put GAO's reputation as unbiased, objective evaluator at risk.

One situation that concerned me when it arose was when GAO recommended that taxes be raised in the so-called transition reports. Were the transition reports self-initiated? And if the answer is yes, is this a case of bias?

As I understand your testimony-and with that, I agree GAO should avoid policy decisions, whether you are for or against taxes, as you say, whether you are for or against or think government should be larger or smaller. Those are policy decisions to be made by the duly elected. But my concern is that at least for a period through these transition reports, there did seem to be some movement into policy areas. I mentioned specifically the question of taxes.

So I will go back: Were these transition reports self-initiated?

Mr. CAMPBELL. Yes, the reports were self-initiated. We did, in fact, review the transition reports, and in our report to the Committee, we do discuss that specific issue.

The one thing that had been pointed to most frequently was the recommendation that perhaps the deficit could be dealt with only with some combination of expenditure cuts and revenue enhancements. One can argue whether that was a partisan comment or not, but certainly, in our recommendation about avoiding the kind of policy advocacy that, to some degree, they have been accused of, we recommended that there be some backing away from purely policy recommendations and more effort devoted to true evaluation and auditing, recognizing that that is not an easy line to draw.

Chairman ROTH. No, but it is a very important line to draw as well as you can, because I think, again, we are all here to strengthen GAO. I think we are all in agreement that GAO performs a critically important function, and the real question is how can we

strengthen that and ensure that its mission is that which we need for the 21st Century.

One of your recommendations, and I think it is an important one, if you are going to have bipartisan support across the board, from liberals to conservatives, is that it does its best to stay out of policy opinion. Now, you are right; the line sometimes is difficult to draw.

In trying to evaluate the work of GAO, it depends primarily on financial benefits and recommendations implemented. Both of these measurements are produced by GAO and, of course, not independently derived.

To what extent can this Committee and the Congress depend on these statistics as true measurements of the value of GAŌ work? Are there any better alternative methods?

Mr. CAMPBELL. I think I can say that the panel would agree that there is some softness in the numbers that are stated relevant to potential savings and the like. One recommendation that we make which is relevant to that point is that GAO establish an external peer review system, not before the studies but after studies are completed. The panel recommended that three or four major areas of GAO work be selected each year for peer review, with panels of outside experts convened to assess a sampling of recently completed GAO reports and then report to GAO and share the results with Congress on what they find (quality, methodology, sources, presentation, as well as practicality, usefulness and value of the products).

I understand that GAO is quite supportive of doing this, and I would urge on this Committee that they urge GAO to do it, because it seems to me it is an opportunity to have an independent examination on a regular basis that would deal with questions like reliability of cost estimates and the like.

Chairman ROTH. Well, let me move on. GAO, of course, is not the only agency providing oversight of Federal expenditures of the Federal workforce. We have created the inspectors general, the chief financial officers, OMB and CBO, all of which oversee and review the spending of Washington.

My question is: What relationship should GAO have with these other oversight bodies? In other words, with these other independent auditors, does the financial auditing become less important? And we talk about reviewing performance and goals. How do you see that developing?

Mr. CAMPBELL. Well, let me make a couple responses and then ask my colleagues if they wish to respond to that, because that deals with a very major problem of government.

May I say that when I headed a Federal agency, one had the sense that everybody is engaged in oversight of what you are doing, whether it is the inspectors general or subcommittees up here on the Hill or the media or GAO. There is just an unbelievable amount of oversight going on-which, by the way, I would argue often stands in the way of getting the kind of entrepreneurial and independent action from the executive side which is often advocated. So I would suggest that any examination this Committee would make of the piling on of oversight which has gone on in this government would be a very useful undertaking.