I. The Current Situation

Revenue estimates state the changes in expected revenues associated with changes in the tax law. Thirty years ago, they were an afterthought in the process of formulating tax policy. Today, due to a series of laws designed to contain federal spend­ing,[1] revenue estimates occupy center stage-the numbers can make or break otherwise worthy tax initiatives.

Meanwhile, there have been some very signifi­cant improvements in revenue-estimating tech­niques.[2] As a result, revenue estimators are better positioned today to produce both dynamic scores that take economic feedback effects into account and improved distributional estimates that show the impact of tax changes by income class. Other changes include the use of linked spreadsheets, more powerful computers, and multi-year tax models using longitudinal data.

Despite these developments, the revenue-esti­mating process is nearly as much a mystery today as it was when estimates were much less important policy desiderata. Now and then, a ray of sunshine gets thrown on the estimating pro­cess, but in general, it remains shrouded in secrecy.

This paper examines the pros and cons of this ongoing secrecy and concludes that it is time to consider some modest changes that, on balance, will make the estimating process more transpar­ent. Enhanced transparency, in turn, should make it possible to begin addressing other reve­nue-estimating issues that demand attention.[3]

Transparency Differences Between Agencies

The staffs of three federal agencies prepare revenue estimates on a regular basis. They are, specifically:

• The OTA Staff. The Revenue Estimating Staff in the Office of Tax Analysis (OTA) in the U.S. Department of the Treasury[4] has about 13 financial economists who are supported by a separate eight-person modeling and com­puter applications staff.[5] In addition, OTA's individual, business, and international staffs assist with revenue estimates in their areas of expertise.
• The JCT Staff. The revenue-estimating staff of the congressional Joint Committee on Taxation (JCT)[6] has about 16 economists and senior economists, as well as five supporting statisti­cal and computer specialists.[7]
• The CBO Staff. The Budget Analysis Division in the Congressional Budget Office (CBO) has approximately 45 analysts in four operating units who prepare spending projections and cost estimates for legislation. In addition, the CBO's Tax Analysis Division has about 11 tax analysts, as well as managers and supporting staff, who estimate future tax receipts and pre­pare studies of the U.S. tax structure.[8]

Although there are three revenue-estimating staffs, only two estimates are typically prepared with respect to a given tax proposal: one by Trea­sury and the other by either the Joint Committee or the Congressional Budget Office. As explained in more detail later in this paper, revenue estimates from the Joint Committee relate to individual and corporate income taxes, Social Security and Medi­care taxes, estate and gift taxes, and excise taxes. Budget Office revenue estimates relate to tariffs, offsets, and user charges.

OTA and JCT (or CBO) revenue estimates become important at different stages of the legislative pro­cess. Treasury estimates matter when the President's budget is being prepared or when a tax proposal is being readied by Treasury for submission to Con­gress. Later, during the consideration of a tax bill, Joint Committee estimates become critical, espe­cially if the tax-writing committees are required to hit congressionally mandated budget targets. Still later, if Treasury recommends a veto, Treasury esti­mates will be used to advise the President.

With respect to estimating transparency, there are some important differences between OTA, the JCT, and the CBO. Of the three, the Congressional Budget Office is the most willing to make informa­tion public regarding its revenue-estimating meth­odology. The reasons for this are discussed later in this paper.

In contrast, the OTA staff at Treasury is very secretive. As one former senior economist in the current Bush Administration observed of the Trea­sury staff, "They are discouragingly non-transpar­ent even to folks within Treasury."

Similar comments abound on Capitol Hill with respect to the JCT estimating staff. However, the JCT staff has been consistently more open than Treasury in some significant ways, which will be described below.

What Does Transparency Mean?

This paper focuses primarily on the technical process of preparing revenue estimates by assem­bling data and making assumptions and calcula­tions at the staff level. It focuses only secondarily on the process by which staff estimates are reviewed and sometimes changed by senior staff and political officials.

Transparency in the technical revenue-estimat­ing process at the staff level means different things to different people. Depending on how transpar­ency is defined, individuals can easily find them­selves on opposite sides of the debate over whether greater transparency is a good idea.

The following are some of the possible compo­nents of the transparency concept as applied to the revenue estimates prepared by federal agencies:

Generalized Transparency

1. Publishing written explanations describing how, in general, an agency prepares revenue estimates. Variations on this theme include publishing focused explanations relating to specific topics such as capital gains estimating methodology.
2. Sponsorship of occasional conferences or sem­inars at which revenue estimators describe their work, provide illustrations, and solicit comments from peers.
3. Publication of source data-such as the Inter­nal Revenue Service's Statistics of Income[9] or the IRS tax return "public-use file"-that can be used by private organizations to make their own revenue estimates.
4. Publication of retrospective analyses of the accuracy of prior revenue estimates in the lim­ited instances[10] in which this is possible.

Specific Transparency

5. Preparation and release of written explanations describing how a specific revenue estimate was done.
6. Release of the assumptions that were made by an estimator when doing a specific revenue estimate, including the behavioral assumptions (i.e., expected changes in taxpayer behavior consequent on a change in taxation).
7. Listing the sources of data that were used in making a specific revenue estimate, including any links to publicly available data.
8. Description of the spreadsheets or simulation models used in doing a specific estimate, including any use of specialized software.
9. Release of the spreadsheet or computer simula­tion model output files underlying a specific revenue estimate, preferably in electronic form, and any off-model adjustments made to the output files.
10. Release of the non-privileged elements of the simulation models themselves.
11. Release of the actual data used in making a specific revenue estimate to the extent that this can be done without jeopardizing taxpayer pri­vacy or business confidentiality.

Which of these components we include in our definition of transparency depends on our goals. If the goal is to permit "accuracy checking" through full replication and validation of a spe­cific revenue estimate, our definition of transpar­ency should include all 11 of these general and specific components. If, in contrast, the goal is to acquire a general understanding of how revenue estimates are done, the components listed under "Generalized Transparency" are sufficient.

The goals of most requests for revenue-esti­mating transparency fall between these two extremes. For some individuals, it is sufficient to know how estimates are prepared and to be assured that the estimates are honest and reliable. For them, generalized transparency is usually adequate. But for scholars, researchers, and tax professionals, more than generalized disclosure is needed, and at least some of the components of "Specific Transparency" should be made available to these individuals.

Sources of Transparency Guidance

Revenue estimates are related more or less directly to several other types of government fore­casts. While each type of forecast can be distin­guished from the others, transparency practices developed with respect to one type are often instructive when determining the degree of trans­parency that is appropriate for other types. Accordingly, this paper takes into account the transparency practices that apply to all of the fol­lowing forecasts:

• Baseline Receipts Estimates. Baseline receipts estimates indicate the expected reve­nues under current law for a specified time period, such as five or 10 years.
• Baseline Economic Estimates. Baseline eco­nomic estimates forecast major economic vari­ables, including gross domestic product, unemployment, inflation, and interest rates, for a specified time period.
• Distributional Estimates. Distributional esti­mates, such as those prepared by the Treasury and Joint Committee revenue-estimating staffs, indicate the financial impact, by income class, of changes in the tax laws.[11]
• Cost Estimates. Cost estimates, such as those prepared by the Congressional Budget Office and Office of Management and Budget, project the fiscal effects of spending bills.
• Revenue Estimates. Revenue estimates (the subject of this paper) are forecasts indicating the expected change in tax payments resulting from a change in the tax law.

One must be cautious, however, when applying transparency practices developed in one of these forecasting areas to other types of forecasts or esti­mates. Baseline estimates, for example, are few in number and attract considerable public interest. In contrast, thousands of cost and revenue estimates are prepared each year, and the public is generally interested in only a few of them. This makes it important to think in cost-benefit terms about the amount of detail to be provided when making cost or revenue estimates public. In-depth documenta­tion is appropriate for baseline estimates, but run-of-the-mill revenue estimates ordinarily demand only brief explanations.

Nevertheless, if one keeps this caveat in mind, there are useful lessons to be learned for the reve­nue-estimating process from studying transpar­ency practices in all the areas listed above.

The Transparency Situation Today

No matter how transparency is defined, reliable, official information about the U.S. Treasury reve­nue-estimating process is insufficient to provide even a rudimentary understanding of that process.

In contrast, since 1990, the Joint Committee on Taxation has published a series of documents describing its revenue-estimating procedures, has held several revenue-estimating conferences open to private-sector invitees, and has established an advisory panel of private revenue-estimating experts. In almost all cases, however, the technical foundations for specific Joint Committee estimates remain shrouded in secrecy.

Meanwhile, thanks to a congressional disclosure mandate,[12] Congressional Budget Office estima­tors are relatively open regarding the basis for their baseline and cost estimates, and the staff in general has an accepting attitude toward public disclosure. But, although it is clearly better than the JCT or Treasury when it comes to transparency, profes­sional observers state that the CBO is definitely not laying out all the details that underlie its forecasts.

The Treasury Staff

Treasury has published almost nothing that offi­cially describes its revenue-estimating procedures. However, Treasury economists familiar with reve­nue-estimating issues have created a fairly signifi­cant body of unofficial, scholarly revenue-estimating research over the years, some of it writ­ten after the authors had left Treasury for other employment.[13]

About 15 years ago, the Treasury revenue-esti­mating staff compiled an internal document known as the "Revenue Estimating Handbook," a loose-leaf compilation of documents that was designed to introduce newly hired employees to their work.[14] This handbook was never published or released to the public. Nor has it been updated.

As a result, the estimating staff today uses a 13-year-old article from Tax Notes magazine as an introduction to the revenue-estimating process for newly hired staff. The article, entitled "The Reve­nue Estimating Process" was authored by Emil M. Sunley, a former Deputy Assistant Secretary for Tax Policy at Treasury, and Randall D. Weiss, a former Deputy Chief of Staff of the Joint Committee on Taxation.[15] By the time their article was written, both had entered the private sector.

Although Treasury has not published an official description of its revenue-estimating process, it has released an excerpt from its internal revenue-estimating "briefing book"[16] in response to a 1990 Freedom of Information Act request from Tax Ana­lysts, a nonprofit publisher. The Treasury briefing book discusses such topics as the use of revenue estimates, deadlines, general estimating methodol­ogy, the data and models used in revenue esti­mates, the conventions employed, and relations with outside estimators including the Joint Com­mittee on Taxation and "Lobbyists/Accounting Firms."

The briefing book discussion also sets forth[17] Treasury's rationale for limiting the amount of reve­nue-estimating information that is released to the public. The following is quoted from that discussion:

• "Estimates are often made on short notice and a great deal of experience and judgment is required."
• "Estimates frequently involve novel ideas where there is little or no data, no previous experience to draw upon, no studies to refer to, indeed nothing available but the estimator's past experience. Making this public will not result in better estimates and will hamper efforts to take advantage internally of the esti­mator's input."
• "Public release of detail allows interested parties to criticize those elements of the estimate favor­able [sic] to them and conveniently overlook issues on the other side. Only biased criticism results and the estimate is not improved."

Although Treasury has not produced any pub­lished, official description of its revenue-estimat­ing processes and is reluctant to release information about that process, two OTA Papers have touched on aspects of the subject in recent years.[18] The first, "U.S. Treasury Distributional Analysis Methodology" (OTA Paper 85), appeared in September 1999, shortly before Treasury reduced distributional analyses to "insignificant levels."[19] The second, "Defining and Measuring Marriage Penalties and Bonuses" (OTA Paper 82- Revised), appeared in November 1999 and it pro­vides insights into Treasury's estimating techniques relating to the consequences of filing tax returns as a couple rather than as individuals.

An earlier staff-originated resource is the "Com­pendium of OTA Papers," edited by OTA members Don Fullerton and Thomas S. Neubig and written after enactment of the Tax Reform Act of 1986.[20] That compendium sought, among other things, to make the Treasury revenue-estimating process more transparent. It included numerous articles explaining how Treasury prepared the revenue and distributional analyses required in connection with the 1986 Act.

A potentially significant Treasury revenue-esti­mating paper authored by John McClelland, a Financial Economist on the Treasury's Revenue Estimating Staff, was scheduled for presentation in the fall of 2003 at the annual meeting of the National Tax Association (NTA) under the work­ing title "Corporate Integration: Description of Revenue Estimating Methodology." However, for reasons that are not clear, it was not presented. If the paper had appeared, it would have described the issues involved in estimating the revenue loss attributable to the President's January 2003 divi­dend exclusion proposal. According to Treasury, the paper may appear at an unspecified later date as an OTA Paper.

As an executive branch agency, the Treasury Department is subject to the disclosure provisions of the Freedom of Information Act (FOIA). How­ever, with one exception, Treasury has not been required under the FOIA to release any informa­tion regarding specific revenue estimates.

That exception relates to a case brought by the American Society of Pension Actuaries in 1990. In that case, the Society sought release of the Trea­sury's "assumptions and calculations" underlying an estimate that a shift in pension-related enforce­ment practices would generate $666 million in new revenue. The U.S. District Court for the Dis­trict of Columbia ordered release of the relevant documents because:

the estimate they produced became part of the President's Budget. By estimating that the IRS could generate $666 million in revenue by shifting its enforcement practices, the government adopted as its official policy the calculations and assumptions underlying this estimate…. Because [those] documents…contain analytic backup for the government estimate, they "embody the agency's effective law and policy" and therefore are not protected [from disclosure]….[21]

The government did not appeal the Pension Actu­aries decision, but that hardly constitutes acquies­cence in its result. Currently, the Treasury is resisting-or at least delaying action on-a Free­dom of Information Act request[22] filed by Tax Ana­lysts for documents underlying the Treasury's $364 billion cost estimate relating to the President's Janu­ary 2003 dividend exclusion proposal.[23]

Oversight of the Treasury revenue-estimating pro­cess is the responsibility of the managers who direct the Office of Tax Analysis in the Office of the Deputy Assistant Secretary for Tax Policy (Tax Analysis). They have the professional skills needed to keep tabs on the revenue-estimating staff, which is one of five under their direction,[24] and they provide expect­able, routine guidance. Individual revenue estimates are reviewed by these managers, as are other papers and memorandums generated by the OTA staff.

In the past 15 years, there have been two exter­nal reviews of the Treasury revenue-estimating process by other executive branch agencies. Alan Greenspan of the Federal Reserve Board con­ducted the first around 1990. The second, in 2001, was undertaken by the President's Council of Economic Advisers and apparently resulted from concern by then-Council member Laurence Lindsey that Treasury might be underestimating the beneficial effects of tax cuts. Neither of these estimating reviews was ever publicly announced, nor have the results been published.

With these two exceptions, there is no evidence that the estimating staff has been subjected at any point within the past generation to an in-depth, internal examination of its procedures, techniques, and estimating consistency (as contrasted with routine supervision).

Public oversight of the Treasury's revenue-esti­mating process is not currently possible as a sup­plement to any official, internal oversight that may occur. As outlined earlier, Treasury provides almost no official information about how it does revenue estimates, and none at all about the basis for particular estimates. Given that fact, effective monitoring by outsiders is currently impossible. For this reason, interested members of the public are unable to perform "trust-but-verify" reviews of the Treasury revenue-estimating process.

The Joint Committee Staff

The Joint Committee on Taxation has been engaged in some form of revenue estimating almost since its inception in 1926. When the 1974 Budget Act became law, a decision was made to enhance the JCT's revenue-estimating capability. Treasury's individual income tax estimating model was made available to the JCT soon thereafter. Fol­lowing the hiring of two former Treasury revenue estimators, the JCT gained access to Treasury's cor­porate income tax model as well.[25]

The JCT revenue-estimating process operated completely out of the public eye until the late 1980s, at which point the Joint Committee on Taxation began to publish a series of documents describing its revenue-estimating processes, thus providing a welcome amount of generalized transparency.

The Capital Gains Confrontation. The first of these documents appeared in March 1989.[26] In it, Joint Committee Chief of Staff Ronald A. Pearlman briefly discussed why the revenue-estimating staffs at Treasury and the Joint Committee had produced widely divergent revenue estimates relating to a proposal by the Administration of President George H. W. Bush to reduce capital gains tax rates.

On March 6, 1990, the Senate Finance Commit­tee held a hearing[27] to investigate the reasons for these differences in the Treasury and Joint Com­mittee estimates regarding the Administration's gains tax proposals. Treasury had estimated that, by spurring gains realizations, the Administration proposal would increase federal tax receipts by $12.5 billion for the fiscal years 1990-1995. Meanwhile, the Joint Committee estimated that the proposal would reduce federal revenues by $11.4 billion in the same time frame.

The hearing revealed a number of unresolved disparities in the estimating assumptions made by the two staffs. It also made clear that the principal reason for the $23.9 billion disparity in the esti­mates lay in relatively small differences in the underlying behavioral assumptions used by the Treasury and Joint Committee staffs.

In conjunction with the Senate Finance Com­mittee's March 1990 hearing, the Joint Committee published its Explanation of Methodology Used to Estimate Proposals Affecting the Taxation of Income from Capital Gains.[28] That report described the revenue-estimating process in general terms and provided specific detail with respect to the Joint Committee's approach to gains estimates. It then analyzed the differences between the Joint Com­mittee and Treasury approaches to the capital gains issue, including the differing assumptions about the likely behavioral responses to a cut in gains rates.

Treasury was also active in publishing materials relating to capital gains revenue-estimating issues. Starting in May 1989, it released a series of OTA Papers detailing several different methodological and data approaches to estimating the effects of gains tax changes.[29]

No one can read the JCT's 1989 and 1990 docu­ments on the gains tax estimation issue and the Treasury's 1989 OTA papers and March 6, 1990, Senate Finance testimony on the same subject[30] without acquiring a better understanding of both the revenue-estimating process in general and the importance of assumed behavioral responses in that process.

The 1989-1990 gains tax confrontation was "destructive," in the view of one Joint Committee observer, because it created considerable acrimony between the Treasury and Joint Committee reve­nue-estimating staffs, which theretofore had gener­ally worked cooperatively on estimating chores. At the same time, however, it demonstrated the abil­ity of full disclosure to dispel suspicion and dis­trust resulting from conflicting revenue estimates.

Descriptions of Methodology. Two other Joint Committee revenue-estimating reports also appeared in the early 1990s. Discussion of Revenue Estimating Methodology and Process, published in 1992, provided a definitive general overview of the procedures used by the JCT when making revenue estimates, including distributional analyses.[31] Part II provided answers to common questions asked by Members of Congress, who are the principal consumers of JCT estimates. Those questions and answers cover a wide variety of issues. The Q&A functions, in effect, as an estimating procedures manual for the JCT.

The following year, the Joint Committee pro­duced a document dealing with distributional analysis and methodology.[32] It has become known as "the Red Book." Although thoughtfully written, only a single distributional table was prepared using its methodology.[33] Thereafter, the proposed methodology was not used again because the JCT estimating staff failed to reach an ongoing consen­sus about how to implement it.[34]

A more recent description of the JCT's overall estimating methodology is set forth in May 2002 testimony before the House Ways and Means Over­sight Subcommittee by Lindy Paull,[35] who was then Chief of Staff of the Joint Committee. This tes­timony provides a brief but useful JCT revenue-estimating overview and history, and includes an "Attachment" that lists 15 Joint Committee "Reve­nue Estimating Methodological Publications" that appeared between 1990 and 2001. However, the main focus of the Paull testimony is on the feasibil­ity of incorporating the macroeconomic effects of tax law changes into JCT revenue estimates. That topic is discussed in the next section.

Dynamic Scoring. Since the appearance of the Red Book more than 10 years ago, most of the rev­enue-estimating information released by the Joint Committee on Taxation has focused on the issue of "dynamic scoring," which is the phrase used to describe the incorporation of macroeconomic effects into the assumptions underlying revenue estimates.

The principal macroeconomic aggregates that relate to dynamic scoring are gross domestic prod­uct, aggregate investment, interest rates, the over­all price index, and the total level of state and local taxes.[36] In ordinary revenue estimates, these "baseline aggregates," which are supplied to the JCT by the CBO, are assumed to remain unchanged by proposed tax law changes. Hence, any "feedback effects" on revenue attributable to changes in these aggregates are excluded from the revenue estimate. For that reason, these estimates are sometimes called "standard" or "static" esti­mates. In contrast, a dynamic revenue estimate relating to a tax proposal includes both the pro­posal's initial revenue effects and the feedback effects on revenue attributable to the likely changes in the baseline economic aggregates due to enactment of the proposal.[37]

Dynamic scoring involves more than merely considering the behavioral effects of tax changes. Revenue estimators at Treasury and the JCT have always taken behavioral effects into account[38] in revenue estimates whenever they had some basis for predicting those effects. As a former JCT chief of staff recently pointed out, the JCT has never prepared truly "static" revenue estimates that dis­regarded behavioral effects. Instead, he states, the JCT now prepares both "standard" estimates (which take behavioral effects into account when­ever possible) and "dynamic" scores (which take both behavioral effects and macroeconomic feed­back effects into account).

Use of dynamic scoring has been urged by sup­ply-side proponents of cuts in tax rates, who wish to take into account the beneficial effects of tax cuts on subsequent economic activity and thus on tax revenues.[39] Static or standard scoring, in their view, overstates the likely revenue losses attribut­able to tax cut proposals by omitting subsequently generated revenue attributable to the additional economic activity induced by the tax cut. Those who view dynamic scoring less enthusiastically point to the uncertainties associated with properly modeling feedback effects.[40]

The year 1995 was a watershed in the Joint Committee's approach to dynamic scoring. The year began with an unusual joint hearing con­ducted on January 10 by the House and Senate Budget Committees.[41] The purpose of the hearing, in the words of House Budget Committee Chair­man John R. Kasich (R-OH), was "to consider the role that static versus dynamic scoring plays in promoting or hindering accurate assessment of the costs and benefits of policy options before the Congress."[42] The hearing was inspired by Republi­can demands that dynamic scoring techniques be applied to evaluate the likely economic effects of the "Contract With America" that formed the Republican legislative agenda for the first 100 days of the 104th Congress.

The hearing featured a distinguished panel of witnesses from the private sector[43] in addition to representatives of the JCT and CBO.[44] The wit­nesses, whose views spanned the political spec­trum, generally agreed that, in theory, revenue estimates would be more accurate if they included the feedback effects of tax changes on the basic macroeconomic aggregates. Except for Dr. Martin Feldstein of Harvard, they also agreed that there were serious practical and theoretical hurdles to overcome before macroeconomic feedback effects could be included in revenue estimates.[45]

The Senate Finance Committee held a hearing on January 24, 1995, that produced similar con­clusions regarding the feasibility of using dynamic revenue-estimating techniques. As a result, Finance Committee Chairman Bob Packwood (R- OR), announced on January 24, 1995, that reve­nue estimates would remain "static" for the time being. Packwood said the message he had received from witnesses was "be careful, don't go wild."[46]

Subsequent to these hearings, the JCT continued to work on the dynamic scoring issue. These efforts came to fruition on May 8, 2003, when a floor state­ment by House Ways and Means Committee Chair­man William M. Thomas (R-CA) made public a detailed description by the Joint Committee of the steps it had taken to provide a dynamic estimate of the revenue costs of H.R. 2, which contained Presi­dent George W. Bush's 2003 tax cut proposals.[47] This was the Joint Committee's first full-blown dynamic analysis. Its release came in response to the January 7, 2003, adoption of House Rule XIII.3.(h)(2),[48] which required a "macroeconomic impact" analysis of major tax legislation.

That Joint Committee floor-statement explana­tion discussed the three economic models used in preparing the JCT's dynamic estimate, described the estimators' assumptions regarding federal fis­cal and monetary policy, outlined the behavioral response assumptions made by the estimators, and set forth the estimators' simulation results including the probable changes in economic growth, labor supply, and employment. It also described the likely budgetary effects of the pro­posals under review and listed the data sources used by the estimators.

Subsequently, on December 22, 2003, the Joint Committee published an expanded version of its May 2003 analysis of the macroeconomic effects of H.R. 2.[49] The December report also contained a history and description of the Joint Committee's macroeconomic modeling work and a description of the dynamic scoring activities of its "Blue Rib­bon Advisory Panel."

Transparency Efforts. The JCT squarely addressed the issue of revenue-estimating trans­parency in a May 18, 1995, press release,[50] in which the committee announced that it was taking steps to "open up the estimating process so that the public has a better understanding of the assumptions and procedures used…." The press release also said that the Joint Committee would establish an advisory board "to improve the esti­mating process and estimating methodology." The release added that the JCT would begin to include "information about the behavioral assumptions" in its estimates, provide a "description of significant potential macroeconomic effects" of tax proposals, and make public "aspects of its primary [estimat­ing] models which do not include confidential tax­payer information."

The Joint Committee's May 18 press release was accompanied by two substantially identical letters, also dated May 18, from Kenneth J. Kies, Chief of Staff of the JCT, to Bill Archer, House Ways and Means Committee Chairman, and Bob Packwood, Senate Finance Committee Chairman.[51] In addi­tion to the changes outlined in the press release, those letters stated that the JCT would begin to meet with "reputable outside economists" to dis­cuss the feasibility of actually "incorporating mac­roeconomic effects into our estimates" as contrasted with simply describing them. The let­ters also pledged to monitor the accuracy of the estimating process "by selecting several estimates to study over a period of years following enact­ment" and to institute a new inventory system "so that we can more efficiently process and monitor the progress of pending estimate requests."

At the same time, however, the two May 18 let­ters cut back on some of the broad language in the JCT press release. For example, the letters indi­cated that most of the information listed in the JCT release would be prepared only if "requested by the Member of Congress submitting the revenue esti­mate request" and thus would not be available to the public unless the Member decided to release it.

Later in 1995, Joint Committee Chief of Staff Kies authored a "Viewpoint" article in Tax Notes maga­zine,[52] in which he declared flatly that "the public knows too little about how the Joint Committee on Taxation prepares revenue estimates." He went on to outline the initiatives then underway to begin "letting in the light on the estimating process while at the same time deflating the wrong and misleading statements about the estimating process."

The 1995 Joint Committee initiatives mark the high-water point for JCT efforts to "open up" the estimating process. Nearly 10 years later, it is not clear how much those initiatives accomplished to promote transparency.

For example, it does not appear that the JCT has initiated a project "to monitor the accuracy" of its estimating process by "selecting several estimates" for retrospective analysis subsequent to enact­ment. If it has done monitoring of that sort, the results have not been made public.

The status of several of the other initiatives is also murky. In particular, it is not clear whether Members of Congress regularly request and receive information about modeling techniques or macro­economic effects. However, in a few instances, according to former JCT officials, Members of Con­gress have asked about-and been briefed on-the behavioral assumptions underlying estimates.

In addition, former JCT officials also confirm that the Joint Committee has now and then made public the basic elements of its estimating models, mainly in response to requests from major accounting firms.

Unlike Treasury, the Joint Committee is not required to disclose documents under the Free­dom of Information Act, but it is subject to a dif­ferent-and perhaps more effective-disclosure regime because it has "535 Congressional bosses" who can request both revenue estimates and sup­plementary information about how they were done. In addition, by working through a sympa­thetic Member of Congress, the lobbying commu­nity can obtain detailed information about Joint Committee revenue estimates. This type of disclo­sure is sometimes called "selective transparency."

However, the existing Joint Committee restric­tions on disclosure of information relating to reve­nue estimates do have an impact on the general public, including students, academic researchers, and the press. Their access to specific revenue-esti­mating information depends to a large degree on how well-connected they are on Capitol Hill. Under existing selective transparency arrange­ments, if individuals lack political connections, they lack access.[53] Thus, outside review of the JCT revenue-estimating process-especially scholarly review-is frequently thwarted.

Advisory Boards. The advisory board announced in the Joint Committee's May 18, 1995, press release became known as the "Blue Ribbon Advisory Panel." It was composed of pri­vate-sector economists from both political parties and various political persuasions. Its status is murky, and it is not clear what work it has actually done. It appears that the panel has been largely inactive, except for three meetings in 2002 to con­sider dynamic scoring issues.

The Blue Ribbon Panel superseded an earlier Revenue Estimating Advisory Board, formed on April 18, 1989, at the request of Ways and Means Committee Chairman Dan Rostenkowski (D-IL), to consult with the Joint Committee revenue-esti­mating staff on econometric models, estimating methodology, and current academic research.[54]

From its inception, the Blue Ribbon Panel's principal focus has been to assist the Joint Com­mittee in moving toward use of dynamic analysis. However, until 2002, the panel was not regularly called upon for advice. In that year, in response to a request from Ways and Means Committee Chair­man William Thomas, the panel met three times to review the Joint Committee's dynamic scoring capability.

The original announcement of formation of the Blue Ribbon Panel envisioned use of the panel to review all aspects of the Joint Committee's revenue-estimating work, but this review has not occurred. Nor has the panel met since it finished its series of three 2002 meetings on dynamic scoring.

Opinions differ about the Blue Ribbon Panel's usefulness. A recent Joint Committee description of the panel's activities paints a favorable picture,[55] but a private-sector observer who is a member of the panel states that its work has been "directed entirely" to the issue of dynamic scoring and that, while it has been "useful and influential" with respect to that subject, it has "not taken up" any other topics. Another knowledgeable private-sector revenue estimator describes the panel as a semi-moribund body whose chief function seems to be to provide the JCT with a response when the commit­tee is challenged on the need for enhanced public oversight over the revenue-estimating process.

In 1996, the Joint Committee also convened a group of macroeconomic modelers who had expe­rience in forecasting the economic effects of major tax initiatives. This "macro group" was separate from the Blue Ribbon panel discussed above. The modeling group worked with the Joint Committee staff to ascertain why different forecasting or simu­lation models produced varying results. The results of this effort were made public in a sympo­sium held in January 1997.[56] The Joint Commit­tee then set to work to develop a macroeconomic equilibrium growth model (MEG). The MEG, along with other models, can be used to produce a dynamic analysis of tax proposals and to estimate the effects of specific tax proposals on different groups of taxpayers.

The JCT and Congress. Relations between Members of Congress and the JCT are a problem­atic element in the revenue-estimating transpar­ency picture. Congress, acting through the Joint Committee on Taxation, is in full charge of the JCT revenue estimators. Congress is the revenue esti­mators' employer and the party that has authority to oversee the estimating process, alter the rules, and implement changes.

Hence, if the Joint Committee is secretive about its revenue estimates, that fact reflects the will of Congress. Congress, acting through the JCT Chief of Staff, currently requires the revenue estimators to maintain confidentiality about leg­islative proposals and their related estimates. Even the existence of a request to the JCT for a revenue estimate is confidential,[57] and the resulting estimates are also confidential. Early in the history of JCT revenue estimating, the prac­tice emerged that estimates were released only to Members of Congress and that the Members could then release the estimates to the general public if they so desired.[58] Greater transparency for the JCT revenue-estimating process could threaten this custom; but even if this custom remains unchanged, it should nevertheless be possible to increase the degree of transparency for estimates that are part of a chairman's mark or scored legislation.

Members of Congress now and then hide behind delays[59] in the revenue-estimating process when they really do not want to support a specific tax proposal but want to appear to do so. "We can't move without a revenue estimate," they tell advo­cates of the proposal. In addition, when Members wish to block a proposal, they sometimes attack it by claiming it is based on flawed or biased revenue estimates. The Joint Committee estimators find it difficult to defend themselves when they are used as whipping boys in these ways.

Oversight. The Members of Congress who comprise the Joint Committee seldom meet, and at no point have they conducted a meaningful inves­tigation into the quality of the work of the JCT rev­enue-estimating staff. Instead, Members' interests in recent years have focused on questions relating to so-called dynamic scoring, which is the subject of a separate paper in this series. Beyond that, there has been no oversight of the JCT estimating process by the members of the Joint Committee.

As a result, the task of internal oversight has been left almost entirely to the Joint Committee Chief of Staff, who is usually trained as a lawyer or public administrator, not as an economist. This increases the difficulty of supervising economists working in an arcane area. Further, the Chief of Staff has duties that go far beyond review of reve­nue-estimating practices. Hence, for routine reve­nue estimates, the Chief of Staff "can obtain only a rudimentary understanding of the staff's method­ology," according to one former JCT revenue esti­mator. Detailed scrutiny is necessarily left to the revenue-estimating staff itself. For more controver­sial estimates, the Chief of Staff meets with the estimating staff and asks questions. However, this type of review falls short of consistent, ongoing oversight and quality control.

Moreover, there is no indication that this lack of oversight has been remedied to any degree by cri­tiques from individual Members of Congress, lob­byists, advisory groups, or the general public. Lobbyists and the Members of Congress with whom they work may ask questions about a reve­nue estimate with which they disagree; but once an estimate has been completed, it is generally dif­ficult or impossible to change,[60] especially when legislative time constraints are severe.

Almost never do discussions between lobbyists and revenue estimators amount to a thoughtful critique of estimating procedures, methodologies, and consistency. Lobbyists usually want to pro­duce a specific result, not necessarily the right result. Now and then, a lobbyist will share sugges­tions and insights privately, but private sugges­tions hardly amount to a thorough investigation of the quality of the work of the JCT revenue-esti­mating staff. A critique of that sort has yet to be conducted.

Operating Guidelines. The JCT revenue-esti­mating process is not governed by any statutes or published regulations,[61] as is the CBO process,[62] but there are internal guidelines that are prepared by the committee itself.

An example is set forth in Section II(A) of the Joint Committee's Discussion of Revenue Estimation Methodologies and Process,[63] which, under the heading "Administrative Procedures," discusses such questions as how estimates are requested, how the JCT staff decides which requests to answer, how Members of Congress can obtain fur­ther information, the confidentiality of estimates, the content of estimates, the time periods to which estimates relate, the criteria for doing re-estimates, and the participation of "outside parties" in the estimating process.

Because these guidelines are internal rather than statutory, they can be changed by the JCT when change seems appropriate. Thus, for example, the Joint Committee could decide to expand the num­ber of revenue estimates it publishes on the Web so as to include substantially all of its revenue esti­mates,[64] just as the CBO now does for substan­tially all of its legislative cost estimates.

Who Is Served by the JCT? The JCT's revenue-estimating transparency obligations sometimes become muddled because it is not clear who should be served by the work of the JCT estimat­ing staff. Is the JCT staff working primarily for individual Members of Congress who "need a number" to facilitate introduction of tax legisla­tion, or should it be working primarily for the gen­eral public and satisfying the public's need to know about government operations?

A facile answer is that the JCT estimating staff should be doing both things simultaneously, but sometimes these dual goals come into conflict. For example, as outlined earlier, the JCT maintains strict confidentiality about revenue estimates pre­pared at the request of individual Members of Congress. When preparing those estimates, the JCT operates, in effect, as a confidential profes­sional adviser to the individual Member of Con­gress who requested the estimate. Upon completion, the estimate effectively becomes that Member's property, not the public's, so the estimate is released only if the requesting Member so authorizes. As a professional adviser, the JCT regards itself as being in much the same position, vis-à-vis a Member of Congress, as a private attor­ney or accountant is in relation to a client.

A variation on this position is the claim that dis­closure of revenue estimates would have a chilling effect on the sound advice that the JCT currently provides to Members of Congress and that revenue estimates should therefore be protected from disclo­sure just as "attorney work product" is protected from disclosure in an attorney-client situation.

One problem with this argument is that revenue estimates relate to proposals that have the potential to become public laws. Once a bill is introduced, a revenue estimate is not simply a private matter between "attorney" and "client." Instead, estimates are an important piece of information that the pub­lic needs to evaluate proposed public laws. Hence, even if estimates are kept confidential while legisla­tion is being developed, the JCT Web site should never lack a revenue estimate for an introduced bill. Secrecy for estimates should end when a tax bill is dropped into the legislative hopper.

A second problem with the attorney-client argument is that the existing practice of publishing the JCT's revenue estimates with respect to chair­men's marks and other scored legislation does not appear to have had any chilling effect on the frank­ness of the JCT's advice. Surely, if any "chill" were to occur, it would occur in these cases.

A change in the JCT's Member-centric orienta­tion is needed if the JCT is to achieve the same degree of revenue-estimating transparency as the Congressional Budget Office. Bringing about such a change will not be easy. Members of Congress clearly benefit from having a skilled professional staff to crunch numbers without having to release those numbers to the public if they prove embar­rassing. Under these circumstances, a Joint Com­mittee Chief of Staff will need strength, some courage, and the backing of supportive Ways and Means and Finance Committee chairmen to assert the public's right to know against built-in congres­sional pressures favoring secrecy.[65]

In any event, the attorney-client argument has little relationship to other estimating disclosure issues such as publishing the assumptions under­lying revenue estimates, making estimating models available to the public, and providing sensitivity analysis for estimates.

The Congressional Budget Office Staff

Unlike either Treasury or the Joint Committee on Taxation, the Congressional Budget Office operates under statutory rules that require it to provide "the basis for each…estimate" and to make "informa­tion, data, estimates, and statistics" obtained from other agencies available for "public copying."[66]

In obedience to this mandate, the CBO has developed a disclosure policy[67] that requires the agency "to disclose the basis for each budget and mandate cost estimate." The policy statement says this includes "disclosure of the critical assump­tions and analytic methodologies used to prepare the estimate." The statement also assures inter­ested persons that the CBO will supply "further details on request."[68] Plainly, this approach is dif­ferent from the disclosure regime at Treasury or the Joint Committee.

However, the CBO's basic estimating job is quite different from that of Treasury and the Joint Com­mittee. The CBO's main job is to prepare baseline economic forecasts involving the major economic variables: gross domestic product, unemployment, inflation, and interest rates. It also prepares base­line receipts estimates. These forecasts constitute the guideposts in terms of which congressional committees, including the JCT, can compute the effects of changes in the tax law.

The CBO also develops cost estimates for most legislation, and the underpinnings for those esti­mates are published on the Web at www.cbo.gov/ CESearch.htm. However, in the case of legislation involving income, estate and gift, excise, and pay­roll taxes, the applicable law requires the CBO to use the estimates provided by the Joint Committee on Taxation.[69] CBO revenue-related analyses are limited to proposals involving tariffs, offsets, and user fees. Proposals in these three areas are gener­ally of minor importance to most tax practitioners.

There are some important differences between the CBO and the JCT when it comes to preparing tax-related revenue estimates. First, the JCT esti­mating staff has the advantage of immediate access to the tax attorneys on the JCT staff, who can help in significant ways to improve the accuracy of an estimate. The CBO lacks similarly prompt access to legal advice. "Having revenue estimators and tax attorneys working for the same boss helps to break down barriers" said one Washington think-tank economist familiar with both JCT and CBO esti­mating practices. "This is particularly true," he added, "with respect to issues such as international taxation and depreciation."

Second, the JCT estimating staff is often given much less turnaround time when preparing an esti­mate than is the CBO. JCT estimators are frequently called on to provide ballpark revenue estimates when Members of Congress are drafting legislation. In contrast, the CBO's Tax Division is hardly ever called on to provide ball-park estimates. Similarly, in the course of a single drafting session, tax legisla­tion may undergo numerous permutations, thus requiring multiple JCT revenue re-estimates. In contrast, the CBO is not usually required to pro­duce on-the-fly estimates and is therefore able to focus on developing more carefully honed figures.

The Joint Committee routinely provides the CBO with very little information about the bases for its revenue estimates relating to tax bills. As a result, the CBO remains largely in the dark, just like the general public, about the underpinnings for the Joint Committee's revenue estimates.

Because the CBO's basic job is to forecast the major economic variables, it finds transparency to be in its interest. Transparency is a way of achieving credibility for its estimates. "We have to convince the world that we have a sensible way of doing busi­ness," a senior CBO official recently said.

However, the CBO is not completely transpar­ent when making its baseline estimates. According to persons familiar with the baseline estimating process, the CBO's internal baseline forecast con­tains a great deal of detail that is of significant use to both CBO and Joint Committee revenue estima­tors but is never released, even to the House and Senate budget committees. It should be released, however, as a means of further enhancing the credibility of the CBO's estimates.

There is little downside for the CBO in making its baseline estimating process relatively transpar­ent, because its baseline estimates are unlikely to come under political pressures from lobbyists and special-interest representatives. "No one lobbies on the baseline," a Washington think-tank economist has observed. In contrast, when the CBO makes its estimates of the costs of non-tax legislation:

Lobbyists watch the process carefully, hoping to find analytic errors if a score goes against them-or to persuade analysts to adopt assumptions favorable to their cause before the analysis begins.[70]

Rudolph G. Penner, a former CBO director and the author of the article just quoted, continues:

More than once I had to hold the phone away from my ear when listening to a member of Congress who claimed that our scoring methods had destroyed some pet initiative that, but for our stupidity, would save the nation.

Fear of politicization has led to the physical and social isolation of the revenue-estimating staffs from other tax policy staffs at the JCT and Trea­sury. As a former Treasury estimator puts it, "A trade-off has been made at both JCT and Treasury to 'protect the revenue estimator,' which has had the partially intended effect of making the estima­tion process more secretive, and the staffs isolated and less accountable to their peers (but not less accountable to politicians and political appoin­tees). The current separation of revenue estimating from public tax policy analysis," he continues, "has prevented a lot of good economic analysis from being included in the public policy debate." He concludes that "The separation of the revenue esti­mating staff from the policy economists at both JCT and Treasury, and their general isolation from the rest of the economic and public policy com­munity, has been detrimental to the policy debate, the tax staffs, and the individual revenue estima­tors themselves."

Disclosure advocates should recognize the gen­uineness of agency concerns that greater transpar­ency for the revenue-estimating process will politicize it. At a minimum, these advocates need to consider how to protect the integrity of the pro­cess even though it becomes more open to scru­tiny by those who seek to influence it. In addition, protection for the estimators themselves, both from unwelcome publicity and from undue politi­cal pressure, is also very important.

The example of the Congressional Budget Office is instructive. It has had a succession of strong, capable directors, willing to protect their estimat­ing staffs from pressures brought to bear by out­siders.[84] The same is true at both Treasury and the Joint Committee, so this essential bulwark against politicization is in place. Written rules and prac­tices, modeled on those in place at the CBO,[85] would help to reinforce and solidify the protec­tions provided by strong tax administrators.

Argument #3: Lack of Staff Resources

Enhancing the transparency of the revenue-esti­mating process is a labor-intensive exercise. At bottom, it is no different from any other kind of publishing. Publishing typically involves writing, editing, revising, and classifying information and creating formats and indexing tools to facilitate access to that information by interested users. Enhancing transparency is not a job that can be done by estimating staffs that are already belea­guered and overworked. Transparency requires more staff time, more writing, more review, and a set of skills not commonly taught in economics classes. In addition, staff time will have to be spent evaluating and responding to comments from out­side groups.

Just how overworked are the Treasury and Joint Committee revenue-estimating staffs? And how much labor would be required to bring their work into the public domain? The short answer to both those questions is that no one really knows, not even the agencies themselves.

Accurate statistics are unavailable partly because there is no agreement about what constitutes a "revenue estimate." For example:

• Is guidance by an estimator in the course of a telephone call to be counted as a "revenue estimate"?
• Is an in-depth estimate requiring hundreds of hours of staff time to be given the same weight as an "easy" estimate that can be dashed off in the course of an afternoon?
• If an estimate regarding a specific tax proposal is rerun six successive times on the basis of six variations in a crucial assumption, does that count as one estimate or six estimates?
• If an estimate is redone multiple times to take into account successive changes in the lan­guage of a legislative proposal, does that con­stitute one estimate or many estimates?
• If a prior revenue estimate is rerun against an updated macroeconomic baseline[86] without other changes, does that count as one estimate or two?

Because answers to these questions are essen­tially arbitrary, it is difficult or impossible to pro­duce meaningful workload statistics relating to the revenue-estimating process.

However, no matter how the term "revenue esti­mate" is defined, it is clear that the number of esti­mates prepared by the Treasury and Joint Committee staffs has increased steadily over the years.[87] It is also clear that both staffs are sub­jected to unmanageable time pressures during the closing days of a congressional session and (in the case of Treasury) during the preparation of the annual federal budget. But periods when time pressures are unmanageable are juxtaposed with periods of calm after a legislative session concludes (or, in Treasury's case, after the President submits his budget to Congress). So, again, workload anec­dotes do not reveal the whole story.

When asked, Treasury states that full transpar­ency for revenue estimates would require it approximately to double the size of its existing revenue-estimating staff. This would involve increasing that staff from 13 to 26 professionals, plus support staff. That estimate sounds unrealisti­cally high, particularly in light of the Congres­sional Budget Office's experience in implementing transparency initiatives. Nor is it clear how much of this additional staff would be needed to do work-such as quality control and documentation of estimates-that ought to be done already (but often is not).

However, suppose Treasury is right and that full transparency for its revenue-estimating process would require it to double the size of its estimating staff. That would cost $2 million-$4 million per year. This seems a relatively small price to pay, given the crucial role that revenue estimates play in the tax legislative process.

In summary, advocates of transparency need to recognize that enhancing the transparency of the Treasury and JCT revenue-estimating processes is going to require additional staff resources. The benefits of greater transparency, though intangible, are likely to be well worth those additional costs.

Argument #4: Confidentiality of Returns

A major objection to disclosure of the data used to make revenue estimates (as contrasted with merely listing data sources) is that the Internal Revenue Code strictly prohibits revealing tax return data and imposes both civil and criminal penalties for violations of that rule.[88] Since Trea­sury and JCT revenue estimates rely heavily on tax return data, the prohibition on disclosure of return information makes it impossible to release the pre­cise data used to make Treasury and JCT estimates.

For example, there is probably no way in which the data underlying so-called rifle-shot tax legisla­tion-i.e., legislation designed to benefit only one (or at most a few) individuals or firms-could ever be released. Release of the data underlying such estimates would almost certainly jeopardize taxpayer confidentiality. However, most revenue estimates relate to broad-based legislation, not to "rifle shots."

Fortunately, the Statistics of Income Division within the IRS Office of Research, Analysis, and Statistics prepares a special compilation of tax data known as the "public-use file," which contains return data that have been appropriately homoge­nized to protect taxpayer privacy. These are not precisely the same data used by Treasury and the JCT when making estimates, but they are as simi­lar as possible given the need to preserve tax return privacy and confidentiality.

The availability of the public-use file has been an essential element in development of private accounting firm and think-tank revenue-estimat­ing capabilities over the past 30 years.[89] In addi­tion, the public-use file provides an example of how government efforts to enhance transparency can support and spur helpful private-sector initia­tives in the revenue-estimating area.

However, there are serious problems with the public-use file.

First and most important, the public-use file contains only individual tax return data. A public-use file containing at least a limited amount of small- and mid-size business return data (both corporate and non-corporate) should also be made available. This suggestion is discussed later in this paper.

Funding for production of the IRS public-use file is currently very limited. More resources are essential, especially if a business return public-use file is to be produced. It is up to Treasury to request those resources.

Lack of timeliness is another serious problem. In recent years, there have been two- to three-year delays in production of the public-use file.[90] Some of these delays were apparently attributable to lack of publication funds. Publication has also been delayed by IRS concerns that the growth of pri­vate-sector computational power, the increasing availability of cross-matching financial data, and developments in record-linking mathematical models might make it possible to obtain individual tax return information from the public-use file, especially in the case of high-income individuals.

The problems associated with delays in pro­duction of the public-use file are compounded when government revenue estimators base their estimates on updated or corrected public eco­nomic data that have not yet been published by the compiling agency. Use of not-yet-published data is entirely appropriate when such data are available to government estimators, since the goal is to produce the most accurate possible revenue estimates; but to avoid public confusion, use of not-yet-published data should be mentioned in any estimating memorandums that accompany revenue estimates.

IRS concerns about possible disclosure of indi­vidual tax return information in the public-use file have been made even more acute by the growing use of "longitudinal" data-i.e., data covering more than a single year-in the revenue-estimat­ing process. Using longitudinal data helps to make revenue estimates more accurate than those pro­duced using only annual, cross-sectional data.

However, the availability of longitudinal panel data produces a concentration of information for one individual, and thus makes it easier to identify individual taxpayers. This fact will make it increas­ingly difficult to disclosure-proof the public-use file if longitudinal data are included, as they should be.

These concerns have led the IRS Statistics of Income Division to hire outside contractors to improve the "disclosure-proofing" of the public-use file. Release of the file was significantly delayed while the contractors finished their work. The IRS needs to prevent those delays from recurring.

The public-use file also omits data that could be included without jeopardizing taxpayer privacy, such as age distribution data. Private estimators are unhappy about these omissions, which seem unjustified on privacy or other grounds.

Argument #5: A Reduced Role for Judgment

Good professional judgment is a crucial, appro­priate, and required component of revenue esti­mating. It is what makes revenue estimating an art rather than a science.[91] Opponents of disclosure fear that enhanced transparency would make esti­mators less free to exercise independent profes­sional judgment, especially in cases in which they cannot readily document the basis for their beliefs and assumptions. Instead, they fear, estimators will tend to base their estimates on more easily jus­tified mechanical standards.

Typically, the judgmental part of a revenue esti­mate relates to the behavioral component (the "what will the consequences be" part) rather than the static component (the "what does one deduce from the current data" part). Professional judg­ment is necessarily involved when estimating the likely behavioral response to a tax change. Some­times judgment is required because little or no empirical evidence is available. In other instances, ample evidence is available,[92] but judgment is required to interpret it.

In these cases, adopting a mechanical alternative to the use of judgment is simply not feasible. There is nothing wrong with the use of judgment in these instances, provided the judgment is dis­closed and explained. Government regulators use judgment all the time. Making behavioral judg­ments public will assure greater consistency across revenue estimates with respect to behavioral responses, and that is all to the good. If the judg­ment is sound, it will be able to withstand public scrutiny.

Argument #6: Delay

Greater transparency will inevitably slow the production of revenue estimates, both because each step in a transparent estimating process is likely to be documented with care (rather than with informal notes or not at all) and because additional review is likely to be given to publicly released revenue estimates and documentation. Providing additional staff will ameliorate but not necessarily resolve this problem.

Some amount of delay may have to be regarded as the price of additional transparency. Policymak­ers will have to plan accordingly. Offsetting this cost, however, is a clear benefit: Revenue esti­mates, once they are released, will be more accu­rate because they will have been reviewed with care so they can withstand the light of day.

Argument #7: Confusing the Debate

Opponents of transparency for the revenue-esti­mating process argue that revealing the underpin­nings for a revenue estimate might distract attention from the proposal itself and transform the debate on the proposal into an argument over revenue-estimating methods.

If there are grounds for suspecting that a reve­nue estimate is unsound, this might certainly occur. However, if a revenue estimate is ques­tioned, is the ensuing debate a good thing or a bad thing? For many years, legislators, the press, and the public have been accustomed to accepting rev­enue estimates as not-to-be-questioned "givens." As a consequence, some serious errors and bad judgments have doubtless been obscured from view. Why is it inappropriate, at last, to provide a basis for questioning the process?

There will certainly be cases in which the tax debate is distracted by a dispute over estimating procedures. In some cases, we may be confronted with several differing revenue estimates, each from a different source. But that is healthy, not bad.

As a result of honest public debate, adjust­ments will be made in the estimating process, procedures will become accepted and far better understood, and consensus will emerge. What will have disappeared is pervasive suspicion about the validity of the numbers, the possible biases of the estimators, and the rationality of the estimating process. Bloodying one's nose in an occasional dispute over numbers is a small price to pay for long-term certainty about the integrity of the revenue-estimating process. As public con­fidence in the quality of revenue estimates grows, the debate can again focus on the merits of the tax policy issues at hand.

Argument #8: Loss of Mystique

Thirty or more years ago, revenue estimates had very nearly the authority of utterances from the ancient oracle at Delphi, at least to outsiders.[93] Like the utterances of the oracle, estimates resulted from a mysterious and secretive process known only to a few. Estimates could seldom be chal­lenged, and few dared to try.

Today, much of this mystique has been lost, partly as a result of the spread of revenue-estimat­ing capability in the private sector and slowly growing public knowledge about the nuts and bolts of the estimating process. Opponents of transparency fear a further loss of authority if even more information is revealed about the Treasury and JCT estimating processes. For example, at an Airlie House conference in 1988, David Brockway, who was then Chief of Staff of the Joint Commit­tee, told the conferees:

[W]e generally attempt to avoid laying out what our estimating model is, and how we've reached our conclusions, so that people can't come back and challenge those aspects of the estimate that lead to results they don't like….[94]

The short answer to these concerns is that, in a democracy, authority should be earned rather than based on one's position or on privileged access to information. In the case of revenue estimating, this means that the authority of a revenue estimate should derive from the reasonableness of the assumptions on which an estimate is based, the comprehensiveness and recentness of the data used, and the clarity of the reasoning applied to produce the estimate.

Regrettably, however, lack of transparency in the Treasury and JCT estimating processes makes it impossible at present for those agencies to rely on these legitimate sources of authority. Meanwhile, the deference traditionally accorded to their pro­nouncements continues to erode.

III. Lessons from Others

Lessons from the States

At the state level, revenue and cost estimates play a central, crucial role in the legislative pro­cess. That role is even more important than at the federal level because substantially all states have a balanced budget requirement of some sort. With­out accurate estimates of likely revenues and expenditures, it is not possible to know whether balance has been achieved. In effect, revenue fore­casts establish the allowable parameters of annual state budgets.

Given these facts, "The federal government has a lot to learn [about revenue estimating] from the states," according to James Francis, Director of Tax Research for the Florida Department of Revenue and a member of the Joint Committee on Taxa­tion's 1989 Advisory Board.[95] "We have the ulti­mate fiscal constraint-a balanced budget with no fudging," Francis said. "Thus we have been facing the problem of revenue-driven tax legislation for a long time."

Because almost all states require balanced bud­gets, it is natural for disputes to arise over the accuracy of state-level revenue estimates and the reliability of the process that produces them. Some states have found themselves embroiled in budget disputes that are rooted in numerical duels over estimating assumptions and methods. These dis­putes have sometimes slowed and disrupted the legislative process. In contrast, when state legisla­tors and executive branch officials adopt similar estimating methods and "use the same numbers," that fact "greatly aids the legislative process."[96]

Consensus Revenue Estimating. To deal with disputes over revenue estimates, about half the states[97] have adopted what is known as "consen­sus revenue estimating." The consensus estimating process involves a concerted effort, prior to a legis­lative session, to resolve disputes over the likely revenue and expenditure figures that will be included in an upcoming state budget.[98] The Council of State Governments describes the pro­cess in the following words:

By far the most interesting trend in recent years is the movement by states away from forecasting dominated by the executive or legislative branch to a method known as "consensus estimating."… Consensus revenue estimating involves participants from the legislative and executive branch, as well as representatives of business and academia. Through discussion and analysis, these members must come to unanimous agreement on the estimates for the coming fiscal year. While no forecasting method is ever 100 percent accurate, studies show that on average a consensus forecast will be more accurate than any of the individual forecasts that compose the consensus.[99]

In contrast, within its areas of responsibility, the Congressional Budget Office seems to be doing its best to promote a transparent forecasting regime. Perhaps more could be done, but the agency has made a good start along the road to full transparency.

Calls for greater transparency on the part of Treasury and Joint Committee revenue estimators have been made for a long time, but without much result.[122] Clearly, something must be done to con­vince these two agencies that it is time to address the transparency problem seriously and to devise responses that will work.

Some Guiding Principles

It is important to keep costs and benefits in mind when devising solutions to transparency problems. For example, it is clear that publishing in-depth supporting analyses for thousands of rev­enue estimates would not be worth the cost. Few people would use most of them. But publishing detailed analyses for proposals that are important for revenue or other reasons makes a great deal of sense. A careful weighing of costs and benefits is therefore needed when deciding which revenue estimates require in-depth supporting analysis.

In the end, however, asking about the costs and benefits of transparency is a little like asking about the costs and benefits of democracy in contrast to other forms of government. In both cases, the evaluation involves weighing intangible values against tangible costs. Cost-benefit analysis can help us come up with sensible solutions that work, but in the end, we cannot quantify the intangible values at stake.

When devising solutions to revenue-estimating transparency problems, it is also important to avoid proposals that create an "us versus them" reaction on the part of Treasury and the Joint Committee or that reinforce the existing siege mentality that too often affects the revenue-esti­mating staffs.[123] Both agencies need to recognize that greater transparency for the revenue-estimat­ing process has benefits for them and for the pub­lic, not just drawbacks, and need to know that their concerns and staffing requirements will be taken into account when resolving transparency problems.

It is also important to make it clear that calls for greater transparency do not constitute an attack on Treasury and JCT revenue estimators. These indi­viduals are professionals who do their best under sometimes difficult circumstances. The work they produce can make a significant contribution to the tax policy debate if it is made more broadly avail­able to the public and if the estimators themselves are encouraged to play a larger, more visible role in the tax policy community. Enhanced transparency can help to show that, far from being second-class citizens, revenue estimators are among the best of public-finance economists.

In general, the details of effective, lasting solu­tions to transparency questions can best be devised by the affected agencies themselves. They know their procedures better than any outsider possibly can. They are also better positioned to conduct experiments to find out what solutions work or do not work. And it is they who must request the additional staff necessary to implement any effective solution. The impetus for change probably has to come from outside in the form of a policy directive mandating greater transparency, but the detailed nuts-and-bolts implementation can be handled best by the agencies themselves.

Proposed solutions also need to address the dis­trust of one another's motives that currently char­acterizes relations between government estimating staffs and private estimators in accounting firms, think tanks, and academia. In the view of one gov­ernment estimator, "No one in the private sector seriously thinks that Joint Committee or Treasury estimates are 'cooked.' What these guys really want is the ability to estimate their own proposals-not to improve JCT or Treasury estimates." Conversely, many private estimators are concerned about the secrecy surrounding official estimates, which they suspect masks inadequate analysis or questionable assumptions.

Dealing with this distrust will require more than mere changes in agency procedures. An ongoing, meaningful, public-private dialog is needed. This in turn will require a lessening of the isolation that separates revenue estimators from the rest of the tax policy community. It will also require real informational reciprocity in pub­lic-private discussions of revenue estimates. Both sides need to share information and calculations if trust is to be established; transparency is not a one-way street.

Finally, we need to be willing to accept incre­mental solutions to transparency problems. Refer­ring to the 11 transparency components listed earlier in this paper,[124] a leading revenue estima­tor in private practice cautioned, "Perfection is the enemy of the good. If we moved Treasury and the JCT to take the first five of these [component] steps, and included descriptions of the assump­tions and data in steps 6 and 7, we would achieve 98 percent of the goal."

Is Outside Intervention Needed?

Most of the proposals set forth below could be implemented promptly by the officials in charge of the Treasury and Joint Tax Committee revenue estimating staffs. Aside from the restrictions on disclosure of tax return data, there are no stat­utes or regulations that bar Treasury and the Joint Committee from becoming more open about their revenue-estimating assumptions and procedures.

Despite these facts, however, finding a solu­tion to the transparency problem cannot simply be left to Treasury and the Joint Committee. Those two agencies have had decades to come up with solutions since calls for greater transpar­ency first arose. They have very largely failed to do so. Clearly, outside policy direction is now needed.

Why have Treasury and the Joint Committee failed to resolve the transparency issues discussed in this paper? The short answer is that transpar­ency does not serve their short-term interests. All the internal pressures and institutional barriers at both agencies are against openness. As a Treasury official recently put it, "transparency doesn't give any return to the Assistant Secretary [for Tax Pol­icy]." So, he concluded, "nothing is going to hap­pen on transparency without a mandate."

For example, greater transparency would inev­itably delay the production of estimates because extra time would be needed to assure that the estimates and the underlying documentation could withstand public scrutiny. This would aggravate complaints about delays in producing estimates. Further, transparency would make it easier for outside parties to criticize revenue esti­mates, and staff time would be required to respond to those criticisms. The managers of the Treasury and JCT estimating staffs have no desire to pile these new burdens on the shoulders of already overworked staffers. It is therefore unlikely that either Treasury or the JCT will enhance the transparency of their revenue-esti­mating practices on their own.

This section proposes a few modest steps that should help to move Treasury and the Joint Com­mittee along the road toward greater openness for their revenue-estimating activities. If these propos­als fail to produce results, more vigorous interven­tion may be needed, but it is too early to assume failure. To the contrary, the appropriate current starting point is to assume goodwill and an intelli­gent response from both Treasury and the Joint Committee.

A Few Modest Proposals

The transparency suggestions set forth below fall into two categories: "process" proposals and "substance" proposals. Process proposals relate to the rules and customs that govern the preparation of revenue estimates. Substance proposals relate to the actual content of estimates and the underly­ing data. Both types of suggestion deserve careful consideration.

Process Proposals

Enact Statutory Disclosure Mandates. The disclosure mandates in the Congressional Budget Act[125] and the Unfunded Mandates Reform Act[126] have played a crucial role in encouraging disclosure of the underpinnings of the cost estimates prepared by the Congressional Budget Office. A similar stat­utory mandate[127] addressed to Treasury and mod­eled on the CBO disclosure mandates should be added to the Internal Revenue Code.[128] A separate but substantially identical mandate should be added to the provisions of the code that relate to the Joint Committee on Taxation.[129]

These mandates would make clear to both Trea­sury and the Joint Committee that "disclosure is part of the job" of preparing revenue estimates and that publication of a mere revenue number with­out any explanation or rational underpinning rep­resents incomplete work. These mandates would also provide a basis for requests from both Trea­sury and the Joint Committee for the additional staff and funding needed to make transparency a reality.

At the same time, because these mandates would speak in general terms, Treasury and the Joint Committee would remain free within rather broad limits, as the CBO has been, to experiment with possible solutions to transparency issues.

Statutory mandates are important because they help to change the mindset of those who are sub­ject to the mandate. At present, JCT estimators can regard their job as "done" when they send an esti­mate to the Congressman requesting it, and their counterparts at Treasury can feel the same way when they have provided Treasury or White House officials with revenue gain or loss figures. Understandable though those attitudes may be, however, these mindsets leave the information needs of the public out of account.

Mere exhortation is unlikely to change these existing mindsets; the habits of thinking that block disclosure are simply too ingrained.[130] That is why a congressional command is needed to make it clear that the job of preparing a revenue estimate is not completed until the public is adequately informed.

The idea of using a statutory mandate to encourage revenue-estimating transparency is not new. Starting in 1988, Senator Robert W. Kasten, Jr. (R-WI), began to champion his Tax Policy Free­dom of Information and Sunshine Act.[131] Among other things, that legislation would have required the Joint Committee on Taxation to report on "the economic assumptions and methodology that underlie" its revenue estimates. It would also have required preparation by the Joint Committee of "technical explanations setting forth the economic data, assumptions, and methodology [underlying a revenue estimate] in sufficient detail to permit replications of the results by nongovernmental analysts…."[132]

A similar, more recent call for greater revenue-estimating transparency came on May 22, 2003, from Senators Max Baucus (D-MT) and Kent Con­rad (D-ND), who outlined "five areas of concern" regarding the revenue-estimating process in an open letter to the Joint Committee on Taxation.[133] "We are concerned with the lack of transparency in the revenue estimating process," the Senators said. In addition, they called for enhanced quality controls over the estimating process, greater time­liness in the production of estimates, and better information about the relationship between reve­nue yields and IRS enforcement efforts.

Publish Assumptions. The assumptions on which a revenue estimate is based-especially the behavioral assumptions-can have an enormous impact on the resulting estimate. Recall, for exam­ple, the furor that arose when, in 1990, Treasury and the Joint Committee made somewhat different assumptions about the likely taxpayer response to a change in capital gains tax rates.[134] As one govern­ment revenue estimator has put it, "The real differ­ences in revenue estimates arise from the behavioral assumptions, which have a large effect on some esti­mates and little or no effect on many others."

The assumptions that underlie revenue esti­mates should be released precisely because they are so important in understanding and evaluating the estimate. Indeed, publishing the assumptions on which estimates are based is probably the single most important step that should be taken to enhance the transparency of the federal revenue-estimating process. Publication of assumptions will not, in itself, resolve disputes about the valid­ity of the assumptions, but it is a necessary first step to talking about and eventually resolving those disputes.

Releasing estimators' assumptions along with their estimates is a step that could be taken volun­tarily, and almost immediately, by Treasury and the Joint Committee. A change in internal directions and procedures is all that is needed. Indeed, the Joint Committee made just such a pledge in 1995.[135] The existence of that pledge shows that release of the Joint Committee's estimating assump­tions is feasible and is not barred by statute or rule. The situation at Treasury appears to be similar.[136]

Releasing assumptions would not involve ques­tions relating to the confidentiality of tax returns and return information, except possibly in a few very rare cases. In those rare instances, redaction could be used to safeguard taxpayer data.

Requiring publication of assumptions would impose an additional workload on the estimators at both Treasury and the Joint Committee because neither agency currently requires their estimators to document all the estimates they produce. Today, in many cases, the assumptions that underpin a revenue estimate either are not written down at all or take the form of brief jottings for the file.

Further, even in those cases in which assump­tions are formally documented internally, the doc­umentation is often less detailed than would be required if the assumptions were open to public scrutiny. Thus, even when internal documentation is available, estimators' workloads would probably increase modestly if their assumptions were released to the public.

Another problem involves the source and nature of the assumptions currently employed in making revenue estimates. Sometimes, estimators rely on private-sector data sources such as trade associa­tions that would not want their assistance to be revealed when a revenue estimate is released. Therefore, release of estimating assumptions might constrict Treasury and Joint Committee access to data, at least to some degree.

Further, all estimates involve some degree of per­sonal judgment by the estimator. If publication of estimators' assumptions were required, many esti­mators would probably be reluctant to list "My best guess" as the assumption underpinning an estimate.

This reluctance would probably push estimators to base their assumptions on more than guesses to the extent that this is possible. However, especially when estimates must be produced under severe time pressure, it will often be necessary to base estimates on assumptions that are little more than educated guesses. Even in those cases, however, the assumptions should be published. In most cases, it is more important to know that an esti­mating assumption is based on a guess than to know that it is based on rock-hard data.

Educated guessing is required in all professional work, whether the work involves law, medicine, engineering, or revenue estimating. However, even educated guessing should have some basis, such as discussions with industry experts or consideration of analogous behavioral effects. When good pro­fessionals are required to guess, they normally make that fact clear, together with the basis for their guess, so that judges, patients, clients, and others can evaluate the information provided. The same standard applies to revenue estimators.

Explore User Certifications. When govern­ment agencies wish to spur private research that involves confidential data, they often make use of data protection plans and "user certifications" to preserve confidentiality while allowing research to go forward. Subject to similar safeguards, both the Treasury Department and the Joint Committee should consider providing access to IRS statistics of income files to researchers in the private sector, including the major accounting firms, universities and nonprofit institutions, and others. This would provide a substantial spur to in-depth research on the revenue-estimating process.

The typical user certification procedure involves the signing of data release agreements by the researcher and the relevant government agency.[137] These agreements require preparation of data pro­tection plans by researchers for agency approval. Bonding, encryption procedures, and firewall certi­fications can also be required. Surprise inspections can be used to enforce the restrictions. In the case of tax data, researchers should be made subject to the same civil and criminal penalties[138] that apply to IRS personnel who have access to tax data.

Development of release procedures such as this would enable Treasury and the Joint Committee to commission private research on the revenue-esti­mating process. Private contractors could also assist in the task of model building and testing to facilitate and support the work of Treasury and Joint Committee estimators. Private contractors have sometimes assisted in this way in the past,[139] and adoption of user certification procedures could facilitate expanded use of their services in the future.

There is a precedent for the use of contract-based estimating data release procedures such as those suggested above. When the Joint Committee on Taxation established its first private-sector, rev­enue-estimating advisory board in 1989, it required all board members to sign "contracts strictly binding them to observe the confidentiality of JCT taxpayer data."[140] The procedures sug­gested here are simply a modern elaboration of that earlier approach.

Establish an IRS Research Data Center. Many federal agencies find themselves under the dual obligation of protecting personal identities and information while at the same time releasing infor­mation and statistics to the public. Treasury, the IRS, and the JCT are not alone in facing these con­flicting demands. However, other agencies, notably the Bureau of the Census,[141] Department of Health and Human Services,[142] Department of Educa­tion,[143] and Bureau of Labor Statistics,[144] have done better than tax-related agencies in devising creative solutions to their information dilemmas.

Chief among these responses has been the Research Data Center approach. A Research Data Center (RDC) is a way to grant private researchers access to confidential statistical data while carefully limiting the risk of any disclosure of confidential information.[145] RDCs allow licensed researchers to make use of confidential files (either on site or over secure electronic lines) within a closely restricted area of study using specialized equipment and extreme security. By using an RDC, an agency can encourage research without breaching legal and ethical restrictions on data disclosure.

The RDC approach offers a way to resolve the problems caused by the increasing difficulty of dis­closure-proofing the IRS public-use file. It also provides a means of allowing sophisticated private researchers to gain access to needed data, such as data relating to larger corporations and multi-year longitudinal data, without jeopardizing taxpayer privacy or confidentiality.

The RDC concept was pioneered by the Census Bureau starting in 1994 through its Center for Eco­nomic Studies. The success of the Bureau's pilot projects led to the establishment of regional RDCs at several major universities. That success also led other agencies, including the National Center for Health Statistics and the Agency for Health Care Research and Quality, to establish RDCs.

Access to an RDC by a researcher is premised on compliance with a number of highly restrictive conditions. These include submission of a research proposal for approval; execution of a written agreement covering the work to be done, the data to be used, and the types of output allowed; limi­tations on the type of analysis allowed, including use of any outside linkable data brought in by the researcher; use of dedicated computers; review of output; inspection of material removed from the site; and in-person oversight by agency staff.

Adoption of the RDC approach by the IRS would allow tax-related private research to go for­ward in ways that are not now possible. The approach also offers a partial solution to the increasing challenge of disclosure-proofing the IRS public-use file and would permit controlled release of large-firm corporate tax data to research­ers even if inclusion of large-firm data in the IRS public-use file is deemed impossible.

The RDC approach would also permit Trea­sury, the IRS, and the JCT to subcontract to pri­vate agencies some of the tax-related research projects that at present can be performed only by agency staff due to existing restrictions on access to tax data.

Imitate and Enhance CBO Publication Prac­tices. From a theoretical point of view, it might make sense to assign the tax estimators who are now part of the Joint Committee staff to the Con­gressional Budget Office;[146] but from an historical and political point of view, a merger of this sort seems highly unlikely. The Joint Committee has opposed the idea.[147] Moreover, such a move would require agreement by the chairs of the Ways and Means, Finance, and Budget Committees, and that is highly unlikely since this move would alter the power relationships between these individuals.

Nevertheless, it would be possible to gain the transparency benefits of such a merger if the Joint Committee were to imitate in toto the CBO's publi­cation practices. This expansion of JCT disclosure should apply to all revenue estimates prepared by the JCT staff, not just the estimates with respect to chairman's marks, reported legislation, House- and Senate-passed bills, conference reports, and tax expenditures, all of which are now made available on the Web by the JCT.[148]

Expanded JCT disclosure of estimates would, of course, require a change in the JCT's custom of permitting most revenue-estimate disclosure deci­sions to be made by the individual Members of Congress who requested the estimate.[149]

On the basis of similar reasoning, the revenue-estimating staff at Treasury should publish all of its revenue estimates within a reasonable time after the estimates have been completed. Treasury's tax expenditure forecasts are already made public in the Administration's annual budget.[150] Expanded publication of revenue estimates would, of course, require enlargement of the existing Treasury Office of Tax Analysis Web site.[151]

These suggestions are not as far-fetched as they may seem because the CBO already publishes on the CBO Web site selected tax estimates supplied by the JCT.[152] However, what is currently pub­lished provides little additional information aside from the JCT estimate itself, with only a few accompanying details. The scope of CBO publica­tion of JCT tax estimates needs to be enhanced.[153]

A possible objection is that release of the thou­sands of revenue estimates done each year by the JCT and Treasury is unlikely to be useful to the public. As a general matter, this is obviously true, just as (in a different context) the current practice of releasing thousands of IRS letter rulings each year is not useful to the general public.

The alternative to general release is to require publication of only "important" revenue estimates, but who is to judge importance? Leaving this judg­ment to agency administrators is not the right answer. Every revenue estimate, just like every let­ter ruling, is of interest to someone. Otherwise, the estimate would not exist. For that reason, every revenue estimate should be made available to the public, even though the number of interested indi­viduals may not be large, just as the CBO makes every one of its cost estimates publicly available.

Consider Peer Review for Selected Estimates. Independent, objective peer review is a very effec­tive way to improve the quality of a study's meth­ods, results, and findings. Peer review is commonly used to assure the quality of articles submitted for publication in scholarly journals.[154] Increasingly, it is also being applied in other areas, such as the federal regulatory process.[155]

Selected revenue estimates published by Trea­sury, the JCT, and the CBO should be subjected to independent peer review. Given the estimating workloads of these three agencies, it would be impractical to subject every one of their estimates, or even the bulk of their estimates, to review; but it would be highly useful to peer review at least some of their estimates. The suggestions made during those reviews would benefit the entire esti­mating process, not just the estimates that are actually subjected to review.

Thought needs to be given to the criteria that might be used to select estimates for peer review. One criterion might be the uniqueness or novelty of the estimating challenge, because peer review is especially needed in cases in which estimating procedures are still in a somewhat experimental stage. Another criterion might be the frequency with which estimates in a particular area are chal­lenged by third parties, because in those cases, peer review might help to resolve or at least clar­ify the sources of dispute. Perhaps priority ought to go to estimates relating to newly enacted tax provisions, with the goal of standardizing the applicable estimating techniques. In any event, some minimum number of peer reviews should be done each year: for example, not less than five per agency per year.

Peer reviewers would necessarily require access to tax return data, and perhaps other confidential information, to evaluate the quality of a revenue estimate. This issue could be dealt with by requir­ing peer reviewers to sign confidentiality agree­ments, as is commonly done in other instances when the government requires outside advice on confidential subjects.[156]

Peer review cannot be allowed to delay submis­sion of revenue estimates to the agencies request­ing them. Legislative time schedules will not permit delay. Consequently, except in rare instances, peer review of revenue estimates will have to be done after estimates have been prepared and submitted to requesters rather than prior to publication, as in the case of scholarly articles. The focus of the review will therefore be on the improvement of future estimates rather than on correction of the estimate at hand.

It is important to assure the independence of the individuals who participate in a peer review of agency revenue estimates. The three agencies that prepare revenue estimates already conduct their own internal reviews before releasing their esti­mates. An important function of independent peer review is to overcome the suspicions to which self-review inevitably gives rise, including fears that self-reviewers might feel under pressure to support their agency's position and to refrain from frank comment. Independent reviewers can also provide a fresh look at old problems and can thus supply insights that might not arise during self-review.

Independent peer reviewers should include individuals from the private sector (subject to safe­guards similar to those outlined above in connec­tion with user mandates and research data centers). The Government Accountability Office, the Congressional Budget Office, the Congres­sional Research Service, and the National Academy of Sciences could also provide peer reviewers for revenue estimates.

When independent peer review of selected reve­nue estimates occurs, there should be an opportu­nity for public discussion and participation. At a minimum, the reviewers should present a public report regarding their findings, and the private sector should be given an opportunity to ask ques­tions about that report.

Substance Proposals

Publish Estimating Memorandums. Publish­ing revenue-estimating assumptions would be a good first step toward greater transparency. How­ever, a more comprehensive and satisfactory solu­tion would involve publication of more extensive revenue-estimating technical memorandums, including assumptions. At a minimum, technical memorandums should be prepared and published for every significant estimate.

These technical estimating memorandums should be modeled, in the case of routine esti­mates, on the fiscal notes prepared by the revenue departments of states such as Minnesota[157] or the cost estimates published by the Congressional Budget Office.[158] For more important estimates, the technical memorandums should be modeled on those prepared by private-sector estimators for clients and on the Joint Committee's rare but valu­able technical explanations, such as its recent tech­nical explanation regarding dynamic scoring of the President's 2003 tax proposals.[159]

The information in technical memorandums should include a description of the rules of thumb used by estimators to adjust the current-law base­lines provided by the CBO or the Office of Manage­ment and Budget. For example, the memorandums should describe how the JCT and Treasury divide estimated current-law receipts by income class and how the estimators move from adjusted gross income to imputed income figures, such as the value of owner-occupied housing. Knowing the rules of thumb used to resolve these nitty-gritty issues can help the interested public understand why a given estimate turned out as it did.

Estimating memorandums would also be an appropriate place to describe the techniques used by Treasury and Joint Committee estimators to extrapolate current estimating data into periods five or 10 years in the future. These extrapolations are needed when revenue estimates cover decade-long future periods, as required by current budget rules, but the extrapolation techniques used by Treasury and the JCT are largely unknown.

Estimating memorandums would also be an appropriate vehicle for providing basic sensitivity analysis with respect to the point figures normally provided in revenue estimates. A leading private-sector estimator states:

I would argue that some simple sensitivity analysis around [revenue-estimating] results would be much more helpful than detailed information about the data sources, assumptions, and modeling procedures that underlie an estimate. Let's figure out a way to provide some meaningful information about the potential range of results around a point estimate. In addition, I'd like to know about the estimator's relative comfort level with an estimate, on a one to five scale.[160]

Why are technical memorandums outlining this information not routinely prepared and released by Treasury and the Joint Committee, at least for signif­icant revenue estimates? Personnel constraints pro­vide one answer: More staff would be required to write the required memorandums. Another answer doubtless involves political considerations: Making more information public about a tax proposal might provide the proposal's opponents with ammunition to mount a critique.[161]

Neither of these considerations constitutes a decisive objection to the proposal to accompany Treasury and Joint Committee estimates with tech­nical memorandums. A desire to suppress infor­mation to gain political advantage, while perhaps understandable, is inappropriate in a democracy. Responsible civil servants and their superiors have a duty to request the staffing necessary to serve the public properly.

Preparation and release of revenue-estimating technical memorandums might actually be a boon to morale on both the Treasury and Joint Commit­tee estimating staffs. At present, staff members have no vehicle that enables them to explain or justify their conclusions. This means that they can easily be criticized without having a chance to defend their work. Release of technical memoran­dums would provide staff members with a means of defending themselves and their conclusions.

As a first step, both Treasury and the Joint Com­mittee should begin to accompany all significant revenue estimates with a technical explanation that is released to the public. At least at the outset, it should be left to them to decide what constitutes "significance." In some cases, revenue impact alone might determine an estimate's significance. In other cases (such as the Joint Committee's release of its first dynamic scoring), the novelty of the estimating approach might lend significance to an estimate.

Once the Treasury and Joint Committee staffs grow accustomed to preparing and releasing tech­nical memorandums in significant cases, they can consider whether abbreviated memorandums or fiscal notes would be useful to them and to the public in additional cases, or perhaps in all cases.

A gradualist approach along these lines would enable revenue estimators to find out what style of reporting would best serve the public. This would minimize the dangers that one former Treasury estimator sees in required estimating memoran­dums. "Publication of memorandums" he warns, "raises the prospect that disclosure will result in non-meaningful reports that have real costs but produce minimum benefit. Compliance-oriented memos might appear to be addressing the prob­lem, while serving only as window-dressing."

In addition to preparing memorandums on spe­cific estimates, both Treasury and the JCT should publish every three to five years a general descrip­tion and overview outlining the status of their rev­enue-estimating efforts. This overview should include a discussion of any innovations that have been introduced, changes in data sources, alter­ations in estimating techniques, challenges faced, workloads, and problems yet to be resolved. The discussions in the overview should relate to each of the major estimating areas: individual, corpo­rate, payroll, excise, and estate and gift. The goal should be to provide the interested public with a periodic official update regarding the status of Treasury and JCT revenue-estimating efforts.

Enhance the Public-Use File. Private-sector revenue estimating depends on the availability of tax data. For that reason, it is important both to cor­rect the deficiencies in the public-use tax file pre­pared by the IRS Statistics of Income Division and to enhance that file in several ways. Unless these improvements are made, private-sector revenue estimating will not flourish and university research on revenue-estimating topics cannot be undertaken.

Timeliness. First, and most important, the IRS public-use file needs to be produced in a timely way and released on a regular basis. In recent years, that has not been the case. Two-to-three-year release delays have been common.[162] This has seriously hampered the work of private revenue estimators.

As of early 2004, the most recently released IRS public-use file related to the taxable year 1999. Even longer delays have occurred with respect to earlier years.[163] As a consequence, the information in the public-use file is often outdated by the time it becomes available to private estimators.

Much of the delay in releasing public-use files has resulted from IRS concern about the possibil­ity that private-sector operatives could glean indi­vidual taxpayer data from the public-use file. These are serious and legitimate concerns. How­ever, as outlined earlier,[164] the IRS, with the help of private contractors, has developed improved techniques to disclosure-proof the public-use file. As a result, speedier release of the public-use file should now be possible.[165]

Timeliness problems could be somewhat ame­liorated if the Treasury and the JCT made available their extrapolated files, which show their current and future-year data extrapolated from prior-year figures. As one private-sector estimator has put it, "I can easily get the public-use file, but without a careful extrapolation it is not that helpful, since the [tax] law and the underlying demographics change over time." He states flatly that "Treasury and the JCT should make available their extrapo­lated files."

If the public-use file is to be published in a timely way, the release effort needs to be ade­quately staffed. The goal should be to release the public-use file contemporaneously with the statis­tics of income file for the same year.[166]

Scope. The individual public-use file should include basic classifiers, such as information relat­ing to age distribution. In addition, it should be expanded to include multi-year, longitudinal infor­mation in light of the increasing importance of lon­gitudinal files in the revenue-estimating process.

In the case of business returns, a public-use file relating to small and mid-size corporate and non-corporate business returns should also be made available, despite the obstacles outlined earlier,[167] with confidentiality-preserving aggregation of the data. Small and mid-size firms frequently play a disproportionately large role in the formulation of tax policy, and a public-use business file relating to those firms is therefore needed. Obviously, prepa­ration and release of this file would require addi­tional staffing in the IRS Statistics of Income Division.

Disclosure-proofing business return data for larger firms is probably impossible. To be useful, a large-firm business return file would have to iden­tify firms by industry, since many important cor­porate tax rules vary from one industry to another. But if industry identifiers were provided, it would be easy in many cases to pick out individual firms. Hence, at least for larger corporations, the Research Data Center approach described ear­lier[168] is probably the only appropriate way to permit private research on large corporate data to go forward.

Competition as a Driving Force for Change

The ideas suggested in the preceding section may sound innocuous. They are intentionally designed both to intrude as little as possible into the current estimating work of Treasury and the JCT and to leave those agencies largely free to devise their own ways of responding to the need for greater transparency.

Nevertheless, some government experts who have reviewed these suggestions in draft form con­clude, in the words of one of them, that the pre­ceding discussion "grossly underestimates the difficulty of doing what is suggested" and "grossly overestimates the utility of doing so." As to the utility of revenue-estimating transparency, another government critic adds that transparency concerns are not a priority. "Revenue estimating transpar­ency is not in the top 100 issues that concern tax policymakers," he states.

Comments such as these, which are honestly advanced and deeply felt, show how unlikely it is that Treasury or the Joint Committee will voluntar­ily enhance their existing disclosure practices. Outside pressure will almost certainly be needed to produce change.

Congress is a possible source of outside pressure supporting greater transparency. Congress is responsible for mandating the CBO's existing trans­parency practices[169] and for changing the rules of the House of Representatives to require the JCT to produce dynamic scores.[170] Now and again, indi­vidual Members of Congress have argued for greater revenue-estimating transparency,[171] but there is no sign that Congress is currently interested in trans­parency issues.

Private competition is a more likely source of pressure for greater transparency. If properly encouraged,[172] competition could become a driv­ing force for improvement and change in the reve­nue-estimating process.[173]

Recall, for example, the 1989-1990 competi­tion between Treasury and the Joint Committee over the assumptions and techniques to be used to estimate the revenue effects of changes in capital gains taxes. That competition forced both agencies to provide Congress and the public with detailed explanations of the methods they used to arrive at their numbers.[174] Recall, too, the competition between advocates of static and dynamic scoring, which spurred the Joint Committee estimators in 1995 to begin a serious investigation of ways to produce a reliable dynamic score. More recently, there has been dynamic scoring competition between the CBO and the JCT.[175] These various efforts culminated in the 2003 publication of the JCT's first full-scale dynamic score for a tax pro­posal.[176] Competition was an effective spur to action.

From whom might a competitive critique of rev­enue estimates come? Certainly, in the first instance, it might come from some of the major accounting firms and their clients.[177] Would that critique be one-sided and self-interested? In a few instances it might be, at least if clients were allowed to insist on assertion of a predetermined viewpoint; but this is not likely, and even self-interested comments contain gems of wisdom. In general, however, the professionals in accounting and law firms are fully capable of providing high-quality, public-spirited commentary and are will­ing to do so if given the opportunity.

Government agencies other than the JCT and Treasury are another possible source of comment regarding revenue estimates. Interested agencies might include the Congressional Research Service, the Government Accountability Office, and execu­tive branch departments outside Treasury. All have the resources needed to provide thoughtful cri­tiques of the estimating process.

Further, at least six think tanks and research groups have developed an ability to estimate broad-based individual income tax provisions that can be simulated with a tax model using the IRS public-use file. As outlined earlier,[178] these groups span the political spectrum from left to right. If the information available to these groups was enhanced in the ways outlined earlier, they would be poised to make a significant contribution[179] to the debate over the revenue-estimating process. Several universities would also be able to contrib­ute to the debate.

However, these private charitable and educa­tional groups currently operate on a shoestring in most cases. To compete effectively with Treasury and the JCT, they would have to find the money and staff to make modest enlargements in their revenue-estimating staffs. The situation seems made to order for demonstration projects lasting three-five years, funded by private philanthropy.

After the three-five year period, these groups could either convert their revenue-estimating prowess into a self-supporting enterprise[180] or, if the ongoing need for their work had been demon­strated, ask for continued foundation support. In either event, their public-spirited competition, along with similar competition from elsewhere in the private sector, would push Treasury and the Joint Committee to improve their own procedures and would provide examples of how to do that.

V. Conclusion: What Is to Be Done?

Public-private competition can lay the ground­work for changes in revenue-estimating transpar­ency practices, but competition by itself is unlikely to be the catalyst that is needed to get changes underway. In the interim, in the words of one dis­tinguished economist with both Treasury and pri­vate-sector tax experience, "Treasury and JCT will stonewall, with perhaps a dose of arrogance or hubris to boot."

One hopes that prediction will not come true, but it might. If it does, an outside catalyst will be needed to produce action on the transparency issues discussed in this paper.

Congress is the most likely catalyst. It can set the rules, not only for the JCT revenue estimators, but also for those at Treasury. Further, as men­tioned earlier,[181] individual Members of Con­gress, including members of the tax-writing committees, have sometimes expressed their unhappiness with the current lack of revenue-estimating transparency. If a senior member of a tax-writing committee became consistently inter­ested in transparency issues, change would occur sooner rather than later.

Alternatively, a "Blue Ribbon Commission" sponsored by a government agency, nonprofit group, or accounting or law firm might get the ball rolling, especially if it included some of the indi­viduals who lived through the revenue-estimating wars of the 1980s and 1990s. Many of them now occupy senior positions in both government and the private sector, and if the commission included at least some of those individuals, it might have the influence required to produce change.

One thing is sure: Change is needed in the reve­nue-estimating transparency practices at both Treasury and the JCT. The current secrecy aids no one. The agencies themselves, the professions, and the public would all benefit from greater openness. Given these facts, change is desirable and perhaps inevitable, even though it is not now possible to outline the circumstances that will produce it.

Thomas F. Field is the founder of Tax Analysts, a nonprofit publisher located in Arlington, Virginia. This paper is based on interviews with scores of individuals, both in and outside of government. Most asked not to be identified, but the author wishes to thank each of them for their assistance. Their comments helped to correct errors and substantially improved the recom­mendations. Thanks to these comments, this paper has truly become a collective effort by dozens of individu­als who share a common interest in the improvement of the estimating process. Any errors, however, are the author's own.