Congress’s Policy Support Agencies

How Congress makes sense of the world was the focus of a House Modernization Committee hearing today that honed in on the operations of three support agencies: the Government Accountability Office, the Congressional Research Service, and the Congressional Budget Office. It was one of the most insightful hearings of the 117th Congress.

Congress created its support agencies to provide the Legislative branch with information and analysis unsullied by Executive branch bias. Congress began funding its own expert policy support agencies because funding can shape an organization’s behavior and responsiveness, including the perspectives and focus of its staff.

Today’s hearing had the agency heads as witnesses: Gene Dodaro for GAO, Mary Mazanec for CRS, and Phillip Swagel for CBO; it also had a second panel of experts who provided critical insight into those agencies: Zach Graves on GAO, Wendy Ginsberg on CRS, and Philip Joyce on CBO. While the agency heads were not placed on the same panel with the experts, they were asked to respond to the experts’ recommendations that had through happenstance been released ahead of time, which provided a useful starting point. (This makes me wonder whether a best practice for committees might be to require and publish written testimony a week in advance and request that oral testimony incorporate responses to the testimony from others.)

This was a modernization hearing, not an oversight hearing, so the focus was on the direction the policy support agencies are heading and not specific details on where and why they are falling short. Oversight work generally falls upon the Legislative branch Appropriations Committees that have jurisdiction over all three agencies and, to varying degrees, the authorizing committees: House Admin and Senate Rules for CRS; House Oversight and Senate HSGAC for GAO; and the Budget Committees for CBO.

The Big Picture

The Modernization Committee is focused on actions that it can recommend to strengthen the capabilities of Congress. The big question: are the agency missions and practices aligned with the needs of those they support in Congress?

Drawing on the testimony from all the witnesses and having attended dozens of hearings on funding and oversight over the last decade, the following four issues raised at the hearing highlight the challenges facing these agencies: (1) Funding, (2) Legislative branch agency access to Executive branch-held information; (3) Congressional and public access to Legislative branch agency information; (4) Workforce management.


Funding Legislative branch policy support agencies at an appropriate level is essential for them to function at a high level, but over the decades we have seen quite large funding cuts. As Zach Graves highlighted in his testimony, the co-location of funding for GAO, CRS, and CBO inside the Legislative Branch Appropriations Bill means they compete for funding with other entities (the Capitol Police and the Architect of the Capitol) that have obtained the lion’s share of new funding over recent decades. Congress has been holding down the overall level of funding for the Legislative branch vis-a-vis other appropriations bills for political reasons, and with respect to funding inside the Leg Branch Approps Bill, security and infrastructure almost always takes priority over policymaking. Moreover, because members of the Legislative Branch Appropriations Subcommittee are short-termers — they hope to move up to more “important” appropriations subcommittees — they often do not possess a long-term perspective or insight into prior funding decisions. The consequence is funding insecurity for Legislative branch policy support agencies.

Funding insecurity also creates pressure for the agencies to be risk averse in their analysis and operations. In a highly political environment, you do not want to irritate the people who sign your paychecks by issuing a report that reaches politically controversial conclusions, even if what you say is accurate and well-founded. The object lesson of Newt Gingrich’s defunding of Congress’s science and technology agency, the OTA, for the purposes of showing that he was committed to decreasing the size of government and punishing it for saying unpopular scientific truths, is one lesson learned particularly by CRS. (Other agencies had been considered for defunding and all suffered huge cuts.) As a consequence of this existential threat, CRS watered down its analysis and created an insular, inwardly-focused culture that is risk averse and often cloaks even uncontested facts in a bureaucratic version of #bothsides. CRS’s Dr. Susan Thaul made this point in her 2019 testimony that identified “increasing caution among senior staff,” and Kevin Kosar made a similar point in his January 2015 article Why I Quit the Congressional Research Service. This shif to “neutrality” was put into effect in this 2004 memo.

One remedy is to significantly increase funding for the legislative support agencies and to get them out of the annual appropriations vise. Graves suggests pegging GAO’s funding levels to a percentage of overall federal discretionary spending, which is an eminently reasonable approach. GAO largely conducts oversight work, and oversight should scale to the size of government. (It also would save significant money, as GAO returns more than $100 for every $1 spent on the agency.) CBO and CRS’s funding probably should scale as well. Moreover, if funding for these agencies is made more automatic, whether through the discretionary appropriations process or even through mandatory spending, that can help lessen some of the incentives for agencies to be risk-averse because it becomes harder for unhappy politicians to extract retribution.


All three policy agencies have problems obtaining information and data from the Executive branch. I was shocked to hear the CRS Director testify that the agency files FOIA requests for information. That is highly concerning for a number of reasons, including that Congress has said repeatedly that agencies should coordinate with it, FOIA is slow, and FOIA has many exemptions that should not apply to congressional requests. Congress has a right to the information it needs to conduct its job and Executive branch agencies should be responsive and forthcoming in providing information to Congress’s agencies, just as they should be to Congress itself.

A legislatively-ratified MOU template may help to resolve some of these problems that arise from bureaucratic haggling over the terms of access. We suggest a statutorily-created framework for legislative branch agencies to request and obtain information from the Executive branch — a form MOU — subject to negotiation only over the scope of what is provided. This could significantly expedite the process of reaching agreements by reducing the areas of negotiation and providing assurances to all parties. This framework could be applied to many Legislative branch support agency data requests.

The Executive branch is particularly reluctant to provide information to GAO regarding matters from the Intelligence Community, making it difficult to look into waste, fraud, and abuse. Kel McClanahan lays out the issue in great detail in his testimony to appropriators, where he suggests that the Congress put into law legislative language previously passed by the House that would ensure the IC is responsive.


Products generated by Legislative branch support agencies should be timely, contain the right level of detail, and anticipate the needs of Congress. In addition, they should be findable by congressional staff. In other words, agency work must be user focused and accessible. But, as we saw in the hearing, agencies may inadvertently create disincentives so that products are shaped to meet the agency’s bureaucratic needs instead of its intended users. And risk aversion — the fear of saying an unwelcome truth — incentivizes bureaucratic controls that decrease the quality of reports.

The committee should call for a study to look at how congressional staff actually consume information. As Vice Chair Timmons pointed out, the way he finds CRS reports is by going to Google and typing in “CRS,” which is also what CRS employees do. For this reason, CRS’s use of its internal website to gauge interest and utility is fatally flawed. (So too is its policy of only publishing its reports to the public as PDF files.)

Many excellent agency products are invisible because they are published online in the wrong format and thus not SEO-friendly or are not available where people actually look for information. For example, the #1 source for information is Wikipedia eagerly incorporates CRS reports — but CRS would never host a Wikipedia edit-a-thons to increase the incorporation of their material on that website. How are agencies working to push their reports into the information ecosystem? Are we measuring, like the United Kingdom does, when reports are incorporated in committee reports, congressional testimony, oversight hearing, legislation, and elsewhere? These metrics should be included in each agency’s annual report and budget justifications.

A culture of transparency is equally important. While GAO and CBO have a culture of making their information publicly available, so that at least it is searchable on the web, CRS has just the opposite inclination and has fought for decades against public release. Agencies should be pushing their current and historical products onto the open web. The public release of new/updated products means that they are more likely to be reported on by the press, get into the hands of constituents, and make their way to experts and academics. In other words, they go to the sources that staff rely upon to make sense of the world. Why not take advantage of these information networks to get official resources into the hands of staff? A policy of proactive disclosure, especially around all non-confidential CRS products, as Dr. Ginsburg mentioned, would be most welcome. Even agencies that do this well, such as GAO, could do more.

Dr. Ginsburg rightly raised the issue of a newsletter, which would help make policy support agency products more digestible. Imagine a weekly newsletter aimed at congressional staff that work on an issue. A newsletter on a narrow policy topic, authored by a legislative support agencies, could include summaries and links to relevant CRS, GAO, CBO, and IG reports, a list of upcoming hearings and markups & what’s anticipated at those proceedings, a summary of hearings/markups from the previous week, identifying relevant Dear Colleague letters, clips of relevant news stories, forthcoming agency regulations, and ongoing litigation. That would be a tremendously powerful tool. This is the heart of a user-centric approach. And if publicly released, the newsletter would help drive additional debate, research, and analysis.

Where should a newsletter live? Perhaps inside the new absorptive entity inside Congress that Controller Dodaro suggested, which can play a synthesizing and collating role. It could also be inside CRS, which has the capabilities appropriate for this role but not may not have the appetite.


Support agencies have a number of challenges with respect to recruiting, retaining, and supporting their staff, and making sure that they come from diverse backgrounds.

By way of example, CRS’s American Law Division hemorrhaged staff at 19% a year for 2016, 2017, and 2018, a rate three-times higher than that at CRS. This severely depleted the number of experienced attorneys, but it did not come to light until the House Administration Committee held a hearing on this topic. Similar issues were identified with a lack of diversity among CRS’s senior leadership. Have these problems persisted? No one knows.

Agencies should —

  • Gather information about the workforce on a regular basis, including the total number of employees with breakdowns by pay level, turnover, diversity, division, etc. Highly granular data should be used to identify emerging problems inside agency components. High level information should be published annually to support external review. The unions should be routinely kept in the loop.
  • Identify best practices for hiring expert staff and report on whether those practices are in place. Hiring practices should be continually reviewed to see the extent to which they result in hiring staff that are a good fit for the agency and result in a diverse workforce.
  • Identify best practices for individuals who manage staff and identify the extent to which the agency provides ongoing training to them. These practices should be continually reviewed and measured against employee satisfaction levels and production.
  • Regularly conduct employee satisfaction surveys and report highly detailed information to the public, including trends over time. These surveys should be run independent of agency leadership.
  • Maximize employee benefits to the extent possible, including low cost benefits like providing full time telework availability. Additional consideration should be given around childcare services.
  • Provide on-the-job training to employees and regular opportunities to interact with their peers inside and outside the agency. It is good for staff to be renowned in their fields.
  • Have an office of diversity and inclusion that is empowered to advocate on these issues. (It is our understanding that CRS does not have such an office.)
  • Increase the number of staff detailees so that employees can better understand the needs and rhythms of Congressional offices from the inside-out.

Connecting congressional staff and support agency staff is challenging. Congress does not maintain a list of each personal, committee, and leadership office staffer and the issues on which they work. Nor is there a comprehensive directory across all the support agencies of staff and the issues on which they work. We believe one should be created.

Stray Ideas

I have a few additional ideas that don’t fit neatly into the conceptual framework above.

First, the hearing did not cover all of Congress’s support agencies. Outside the scope of the hearing, for example, were the Office of Congressional Workplace Rights, the Government Publishing Office, and others. I think that GPO is often overlooked in its role providing policy support to Congress through the centralization and publication of official records and provision of detailees to support committee work. The hearing did not address other entities that provide policy support either, such as the Library of Congress with access to legislative information on, and the Law Library of Congress with its foreign law reports. This isn’t problematic — you can’t focus on everything — but it’s worth noting.

Second, Comptroller Dodaro’s statements opposing providing its science and technology component, the STAA, a separate budgetary line item and separate congressional budget justifications were unpersuasive. He argued that creating a separate entity would result in bureaucratic duplication and that Congress hadn’t liked the old OTA model. The OTA model was replicated around the world, continues to be quite effective, and it’s worth remembering that OTA’s defunding had to do with the politics of the 90s and not the merits. But more importantly, there are reasons to be concerned that GAO’s bureaucracy could undermine the nimbleness necessary for a science and technology component, especially with regard to applying GAO’s congressional protocols to the STAA, so some separation may be in GAO’s interest. I respect his views, but am unconvinced.

Third, and on the prior point, I think more consideration must be given to how the agencies decide what research projects to undertake. GAO’s congressional protocols make sense for its oversight work but not necessarily for its science and technology work. How do we balance the products that are intended for a general congressional audience versus those for a narrow committee audience? Those that have bipartisan interest versus partisan interest? Those that are initiatives by Congress and those initiated by the agency in anticipation of the Congress’s needs? I play some of this out in this report on building a modern technology assessment office.

Fourth, the public release of announcements of committee activities more than a week in advance would make it easier to synchronize research products with upcoming hearings and mark-ups.

Finally, how much of this work should be done by the support agencies and how much by congressional committees? I often think congressional committee staff should probably be several multiples larger and include a number of non-partisan, permanent staff that act as the institutional memory. They also should be paid much better, and perhaps put on an equivalent scale for the support agencies. Additionally, in theory, House and Senate committees with overlapping jurisdiction should collaborate more often. In practice, however, cross-chamber committees are unlikely to collaborate and the incentives around cutting pay for political appointees means that committees are unlikely to be a strong source of institutional memory. So, by default, the obligation to understand the big picture falls to the support agencies, which are more insulated from the political winds and can pay staff more. However, they also have their own institutional incentives and bureaucracies and are limited in how much policy-making they can prompt. In theory, this is where civil society steps in, to fill the gap, but as someone who often plays that role, it is untenable for most issue areas.


Legislative support agencies should be designed to meet the needs of Congress. Their products should be designed for how Congress accesses, absorbs, and utilizes information. The Legislative branch policy support agencies were created at different times to meet varying needs. The Congress has evolved, but some agency processes remain driven by their bureaucratic structures and internal incentives. Others are driven by fear. Now is the time to reconsider the structure and rebuild it around meeting the needs of Congress as a user of information.

Today’s hearing is exactly the kind of congressional proceeding where expert support from the policy support agencies would have been helpful. A timely report that explored the evolution of the support agencies, proposals for reform over the decades, ongoing problems and concerns raised in prior committee proceedings, and the experiences of overseas equivalents would have shed valuable light on the proceedings. That, in a nutshell, is the challenge and the opportunity.

It is timely that the House Modernization Committee took a close look at CRS, CBO, and GAO. We hope Congress will take the time to take another look.