Most policymaking occurs in federal agencies, rather than Congress, and interest groups know that’s where the action is. That’s led many to fear that agencies are captured by regulated industries and can’t make good policy. But Rachel Augustine Potter finds that agencies strategically propose complicated rules and design short rulemaking periods when they are facing interest group pressure, avoiding pressure from lobbyists and other branches of government. Maraam Dwidar finds that agencies do change their rules in response to the interest group comments, but especially if they come from coalitions representing organizationally diverse constituents. Big data text analysis now helps scholars see bureaucrats in action.

Studies:Bending the Rules,” and “Diverse Lobbying Coalitions and Influence in Notice-and-Comment Rulemaking.”

Interviews: Rachel Augustine Potter, University of Virginia; Maraam Dwidar, University of Texas

Transcript

Grossmann: This week on the Science of Politics, how agencies respond to lobbying to make good policy. For the Niskanen Center, I’m Matt Grossmann. Most policy-making occurs in federal agencies rather than Congress, and interest groups know that’s where the action is. That’s led many to fear that agencies are captured by regulated industries and can’t make good policy. But new evidence indicates that they can insulate themselves from outside pressure and respond to useful comments from diverse coalitions. Today, I talk to Rachel Augustine Potter of the University of Virginia about her new University of Chicago Press book, Bending the Rules.

Grossmann: She finds that agencies strategically propose complicated rules and design short rulemakings when they are facing interest group pressure avoiding scrutiny from lobbyists and other branches of government. I also talked to Maraam Dwidar are of the University of Texas about her paper and dissertation, Diverse Lobbying Coalitions and Influence in Notice-and-Comment Rulemaking. She finds that agencies do change their rules in response to interest group comments, especially if they come from coalitions representing organizationally diverse constituents. They both use new big data text analysis methods to analyze many rulemaking processes across American government. Potter finds that agencies use power to strategically insulate themselves.

Potter: This is a book about bureaucratic policy-making. There’s a lot of policy-making that happens in the bureaucracy from the fuel standards for the cars we drive, whether plan B is sold at the pharmacy or AT the counter or behind the counter. Lots of things that are happening, and we have a sense that bureaucrats might be really influential and affect public policy. I think that’s a really well shared sense right now in this current political environment. But the question is we don’t know how are bureaucrats influencing policy. So I wanted to think about where bureaucratic power comes from and how that influence might be exercised.

Potter: And the thing I focus on in the book is procedures. So if we think about procedures, there are procedures everywhere in bureaucracy from the contracting process to how we hire federal employees, and to actually the way new public policy is created by agencies, which is called the notice-and-comment rulemaking process. And if you look at these procedures, there’s actually a tremendous amount of variation in how they play out across cases. And so what I do in the book is I look at the notice-and-comment rulemaking process, and I look at these variation and procedures. And I argue that agencies are using this discretion that they have in procedures to actually advance in a bureaucratic agenda. I call this procedural politicking, which is this strategic way of using procedures to advance a bureaucratic agenda.

Grossmann: And Dwidar finds that organizationally diverse coalitions change policy.

Dwidar: This paper is about interest group influence and rulemaking by federal agencies in the United States. And in particular, I ask whether interest groups are more influential in lobbying federal agencies when they work in organizationally diverse or bi-partisan coalitions? So what I mean by organizationally diverse is, for example, a business like Sony working with an advocacy group like the ACLU. And by bi-partisan I mean for instance a group like the ACLU with clear democratic leanings working with a group with clear Republican leanings like the NRA. And what I find is that organizationally diverse coalitions do in fact appear to experience significantly greater lobbying influence with really striking effect sizes.

Dwidar: And in contrast, I don’t observe this relationship with bi-partisan coalitions or coalitions containing business interests or greater resources. So in terms of implications, I think these findings both support existing research on the role of businesses and money in lobbying, which is comforting. And presents some evidence that agencies appear to favor policy proposals that represent diverse constituencies and interests. So if we’re worried about unelected agents like bureaucrats making policy decisions, this is some evidence that they prioritize wide ranging informed and representative policy ideas, which is a very good thing.

Grossmann: They’re both revising our knowledge. Potter says past studies have emphasized too much constraint, but there’s real power to make policy in the bureaucracy.

Potter: There’s been two camps. One camp is the sort of conveyor belt camp, which thinks about, this is just sort of a way that agencies fill up the details of laws passed by Congress. So this is a mundane administrative thing that nobody should care about. So that’s an older view of their process. A more recent view of dating from the 1980s is this idea that notice-and-comment rulemaking is a very constrained process. It’s a process where Congress and the president and the courts have come in, and they have set up a very strong structural process. And because of that, agencies are really hemmed in. And the agencies don’t have a lot of discretion, and the choices they make are more likely to align with those of their overseers in Congress, in the courts, and in the white house.

Potter: And I think I pushed back on both of these ideas. First, rules are not mundane, some of them are, I should say, but not all of them. So in the book, I talk about what we might think about as a boring rule, and it’s a rule by the Office of Personnel Management that changes the reimbursement rate for employees’ uniforms from 400 to $800. That’s a big price increase, but I think most of us don’t care about that from a public policy perspective. So there’s a big chunk of rulemaking that does stuff like that that we just don’t think about, and I don’t think should think about in the broader sense. The things we should be thinking about and where I say rulemaking is an important political activity are things like when the department of education came in during the Obama administration and issued rules that required for-profit colleges to meet new benchmark standards.

Potter: And the intent of these rules was really to crack down on some malpractice and fraud in for-profit schools and in the money they were getting from the federal government. So this is traditionally an area we would think Congress would pass a law. But instead what we saw is agencies wrote a rule. We see that again in the most important climate change actions we’ve seen at the federal level, especially during the Obama years were written by the Environmental Protection Agency, not by Congress. And those are the same roles that we see the Trump administration very actively trying to roll back because they are seen as an important public policy move.

Potter: So I think I pushed back on this idea that these are rote administrative activities because they’re clearly important policy-making activity. And then in thinking about the procedure constraints agencies, I think it certainly constrains them more than if they had no procedures. But the fact that agencies run the administrative process, that they’re given the discretion to do so, gives them a source of power. And if we think bureaucrats have preferences, and I think they do, then we see this power being used as a way to influence the policy that’s created.

Grossmann: Dwidar’s study follows up on findings of interest group influence to specify who wins and loses.

Dwidar: So the prior research has found both that interest groups are influential in policymaking and that businesses and moneyed interests are not any more effective than your non rich, non business counterpart groups. I think the other primary school of thought that I run up against is the work that argues that comments might not matter very much because agencies take comments into consideration prior to issuing a proposed rule because it’s very arduous for agencies to develop these rules, and it’s arduous for them to change them even after they’ve gone through the notice-and-comment period and they have the opportunity to change them. Also, agencies aren’t legally required to make changes to their proposed rules. So I think if anything, I’m running up against the literature that says that comments don’t matter. Obviously, I think that they do.

Grossmann: Potter’s project started from her real-world experience in the bureaucracy.

Potter: After college, got a master’s in public policy and was lucky enough to land a job at the Office of Management and Budget in Washington DC. So I was working in a small part of this organization called the Office of Information and Regulatory Affairs or OIRA, which plays a really important role in the book. And my role was as a desk officer. So I was given a handful of agencies and charged with overseeing their rulemaking activity. And when I got hired, I knew nothing about notice-and-comment, and I had a really steep learning curve. I was interacting with high level people all throughout the executive branch, and I was learning a lot. But the biggest thing I was learning and the steepest part of my learning curve came from dealing with the people themselves. These were very professional people, and they were very strategic.

Potter: So it was a game. I was the overseer, and the people I was overseeing were really smart about doing things, giving me things at different times or how they would frame an issue to me. And I had to be on my toes all the time. So when I left that job to go do research, what I found was that the academic view of rulemaking and bureaucratic politics more generally didn’t have a sense of this game. Of the sense that it wasn’t all a top down process, that the bureaucrats on the other end were really smart and understood the procedures and the way that game was played and knew how to play their side of the game too. And so what I wanted to do with this book was bring that real-world perspective into the academic discussion of rulemaking and bureaucratic politics as well.

Grossmann: And Dwidar started with a search for the ways to see interest group influence.

Dwidar: I come from a research group that likes to collect really big amounts of data, the Policy Agendas Project. And I ended up studying rulemaking because I was trying to figure out how to collect a lot of data on interest group coalitions outside of the survey context. And so we know that interests groups coalesce, but we don’t have records of this. There’s no active repository for all coalitions in lobbying. So I ended up starting rulemaking because public comments, which are a part of the rulemaking process are a record of lobbying by interest groups. And when interest groups lobby with others, they co-sign these comments. So I ended up studying rulemaking because of the data opportunity.

Grossmann: Rulemaking is a critical process that receives less attention than it should, Potter explains how it provides a lot of opportunity for agency strategy.

Potter: The rulemaking process is a process by which federal agencies create binding law. So once agencies have gone through this process, the body of law that is created carries the same force and weight as a law passed by Congress. The rulemaking process starts with an agency having an idea that they need to write a rule. That idea might come from Congress telling them they have to do it, or it might come from their own programmatic oversight and realizing that some change needs to be made. The agency then drafts a proposal, which is called a notice of proposed rulemaking. And that proposal is circulated within the agency and sometimes reviewed by OIRA. After OIRA has signed off on the rule, the rule is published in the federal register. That’s when we all learn for the first time what the agency is thinking about doing.

Potter: From there, there’s a public comment period where anyone can go to the website, regulations.gov and submit a public comment or the agency might do even more than that. They might hold hearings, they might go on a road show and tell the public about what they’re thinking about doing. After that period is over, the agency reads all the comments and deliberates over the comments they got back. Sometimes they get zero comments, sometimes they might get over a million. And they think about them. They’re not required to take them, but they are required to respond in the final rule and to explain why they did what they did. That final rule is then possibly again reviewed by OIRA and then published in the federal register. That’s when we all learn what the actual new law is. And from there, the rule takes effect, and it’s law. And there are thousands of these things done every year by federal agencies.

Grossmann: And Dwidar says it makes room for interest group influence.

Dwidar: So the notice-and-comment process has three stages. The first is that after bills are signed into law, their component parts are sent to federal agencies for implementation by way of crafting rules that regulate their application and enforcement. Agencies are then required to make these draft rules public for a notice-and-comment period where interested parties, including private citizens, interest groups, businesses and more can submit their recommendations for shaping the final rule. And then once this period comes to a close, the agencies review all the comments they’ve received and then issue final rules. So I think the primary thing that folks tend to react to is they’ll say, “Well, how do you know agencies actually read these public comments?” So we know that they read them first because they’re required by law to read them, but also because they are required to issue an agency response to comments in every final rule they issue.

Dwidar: So at the top of every final rule, usually it’s tens and tens and sometimes hundreds of pages of agency responses to every comment they’ve received detailing what the comment’s recommendation was and why they did or didn’t implement it. So we also have evidence that they take the comments pretty seriously. They also have some strong incentives from the courts to take the comments seriously because they don’t want to be sued by the groups that they didn’t take seriously. And courts have historically urged these agencies to seriously consider and implement the recommendations by interest groups in their comments. So they have a lot of incentives to read them, and we have a lot of evidence that they do in fact read them.

Grossmann: Potter says there are a lot of actors that can push back on agencies, courts, Congress, interest groups, and the White House.

Potter: They show that agencies do get overturned by all of these different branches. So all three branches actively oversee agency rulemaking. And what agencies learn from that is not that they have a better chance with Congress and the courts. What they learn from that, I argue is that everybody can do this, anybody can overturn us. And it’s our responsibility as the agency to gauge where the risks are with each of our individual rulemaking projects and to use the procedural politicking that I identify to insulate rules from this type of oversight if we think it’s going to affect a rule.

Grossmann: She sought to measure the many ways the text of a rule can be used strategically.

Potter: In the book, I wanted to get a sense of how agencies write rulemaking proposals in a strategic way and how this can be studied systematically. So I thought about a number of different tools that might be used by agencies in writing in this strategic way. And I thought about framing or how an agency talks about an issue, how an agency analyzes that issue. So what kinds of assumptions are used in the regulatory impact analysis? And then I thought about how the text is actually written. The first two of those tools, framing and analytical assumptions are very context dependent. So you have to think about a specific issue and how it’s framed in light of the politics of that issue. But when we think about how a text is written, that’s one thing that we can study systematically across a large set of rules.

Potter: So that’s exactly what I do. And the argument I make is that agencies try to write less accessible rules when they fear that political overseers might be more likely to be closely scrutinizing them and possibly overturning them. So what they’re doing here is raising the cost of oversight. It takes more time and requires greater expertise to engage with a rule that is several hundred pages long and that is written in a technical way. So what I do is I operationalize this concept of inaccessibility by looking at the number of words or the length of an agency rule as well as the readability of that rule. What I find is that when interest groups are monitoring agencies closely and courts in particular are likely to oversee agencies or hear their cases, I find that agencies do indeed write less accessible rules on both of these metrics.

Grossmann: Potter found that interest group opposition interacts with oversight to stimulate agency action.

Potter: Interest groups provide an important monitoring role. So they are the ones who are waiting for the federal register to be printed hot off the presses and see what agencies are doing. Very rarely do the other three branches have the time, the resources, the staff or sometimes even the expertise to have that level of detailed monitoring. So what I find across several of the empirical measures I look at in the book is that when interest groups are likely to be opposed to what agencies are doing, those same interest groups are more likely to activate political overseers. And we see agencies responding to that by using procedural politicking in the ways that I expect.

Grossmann: She flushed out the mechanisms through a case study of the Pizza Lobby.

Potter: Writing the menu labeling chapter was the most fun part of the book. I got to write about the Pizza Lobby and do all kinds of fun interviews. And the reason I did this case or I wanted to include this case in the book was because I talk about a number of strategies, procedural politicking strategies relating to writing, consultation, and timing. And I look for empirical evidence of these tactics being employed, and I find it. But I wanted to show that these aren’t just tactics that they actually effect outcomes, that they actually affect the policies that agencies make. And so what I do with the menu labeling chapter is dive really deep into this one case where the Food and Drug Administration was implementing a part of the Affordable Care Act that required restaurant menu labels to contain calorie information.

Potter: And this was really a huge challenge for the FDA. They hadn’t really ever done anything like this before, and they had pretty strong beliefs about it. If you talk to people at the FDA, they felt this is a good idea, that this was really going to tackle obesity and that it fit with a lot of public health viewpoints held at the agency. And so I looked at how the agency managed the procedures in that case. And then I looked at the actual evolution of the policy contained in the rule over the approximately six years it took to implement that rule. So I’m able to sort of connect the dots from procedures and tactics to the actual outcome in that rule. To me, that was a really powerful way to try to make the case that procedures do matter a lot for outcomes and for the policies that we as citizens experience from bureaucratic agencies.

Grossmann: Instead of sampling from agencies, Dwidar sampled from the groups attempting to influence them.

Dwidar: I looked at about 350 rules proposed by about 90 federal agencies, and these rules represented most major policy topics present in American politics. So about 20 of the 21 major topics in one of the most common coding schemes we use in American policy are present in the data. There was also a mix of executive branch and independent agencies in the data, so about 40% of the former and 60% of the latter. And qualitatively, I can tell you that there was a pretty wide range in terms of the size of agencies’ presence. So I think it’s a pretty healthy mix of rules in terms of policy and agencies in terms of characteristics. As you mentioned, I ended up with these rules by sampling on the interest group rather than the rules or the agencies. And I did this because I started off interested in the interest groups in particular. I weighted the sample by policy topics so that I could identify a set of groups active across all major policy topics on the national agenda.

Dwidar: So this sample is not intended to represent all agencies, all policy topics, all groups equally. The sample is intended to represent the policy landscape as it pertains to lobbying. So the rules varied quite a bit in terms of their complexity and salience. And in terms of policy topic, the thing that stood out to me the most is that by and large, the topic that was the most present in these rules was health, the major topic for health, which makes a lot of sense because at the time of the data, which was 2005 to 2015, our infrastructure regarding health policy was growing and there was a lot more health policy activity generally in American politics. So one other interesting thing about the sample as well is that I think it was nonprofit organizations and advocacy groups are the most active coalition builders, whereas unions and businesses are the least active, which is really interesting to me because we know that they’re very active in lobbying, just not in lobbying and coalitions apparently.

Grossmann: She measured copy language from the group comments to the final rules.

Dwidar: I measured the influence of coalitions over final rules using textual similarity between public comments that were submitted by coalitions and the corresponding final rules, which represented the policy output the coalitions were trying to influence. So the primary reason that I developed this particular measure of influence is because public comments often recommend really precise changes to the language of final rules. And agencies often lift this language verbatim and adopt it into the final rule. So by measuring the degree to which the comments and the final rules share language, of course, controlling for a lot of things, we can get a sense of the degree to which the comment influenced the final rule. So to develop this measure, I first pre-processed these documents really heavily. So I removed all non-substantive content and conjugations from the documents. So things like headings, subheadings, footnotes and notes, appendices, the stems of all words, all stop words, just everything that is pretty conventional to do when it comes to pre-processing and text analysis.

Dwidar: I then used a plagiarism detection software to identify perfectly matching phrases between comment and final rule documents using a set of standard comparison rules. So in combination, the pre-processing and the comparison rules are detecting a really precise range of common language. But of course, they might also be detecting boiler plate legal language or quotations from the proposed rule that ended up in the comment state and became a part of the final rule. So what I was concerned about in particular was language from the proposed rule quoted in the text of the public comment ending up in the similarity measure. So as a robustness check on this in particular, the one big thing I did was some additional hand pre-processing on all the comments submitted in one year of the data, and that was about 225. Where I went in and removed all quotations from the proposed rule, from the comment, and then redevelop the similarity measure for the data in that year to determine whether the removal of this language would substantially change this measure.

Dwidar: And what I found ultimately is that it does not, so the correlation coefficient for the documents that I had pre-processed in this manner additionally, and those that I did not in this particular year was about 92%. So I’m not really worried about quotations from the proposed rule in the comments ending up in the final rule sort of messing with the results.

Grossmann: That’s different from the hand coding used in prior studies to reach subtle interest group influence.

Dwidar: Folks have attempted to assess the similarities between the comments and the final rule in a lot of different ways, including Susan Yackee and others who have sort of gone in in a very laborious and incredibly impressive way. Read the comments and identified what they were asking for and done this amazing, incredible coding. And in that way, they’ve identified influence or attempted to identify influence. I think that this approach is unique and I think a step forward in the sense that it’s giving us a very, very fine grained measure. So it is not just our assessment, our informed assessment, but also there’s a lot going on in these comments and in these final rules that we might not be able to detect ourselves.

Dwidar: The language is incredibly technical and incredibly detailed. And the rules that I developed to identify that shared language is very attuned to those very slight changes. I think these really slight changes are really important here because can hinge on very, very small changes in language. So you might think, well, five extra words is not really that many words. Well, five extra words might be the clause that outlaws religious discrimination in this particular provision of this particular regulation. Very small changes have really serious consequences. And I think this measure allows us to make sure we’re getting a sense of these very small changes.

Grossmann: Dwidar found that organizational and partisan diversity are both widespread but only organizational diversity helps.

Dwidar: Both of these types of diversity are very, very common independent of each other. So about 77% of coalitions have some level of either partisan or organizational diversity. What I found really interesting is that most coalitions don’t contain both. So I think the correlation coefficient between organizational and partisan diversity is like 0.1. In terms of the type of diversity itself, we rarely observed coalitions with businesses or groups that have very high levels of resources. So these groups lobby just not in these diverse coalitions in particular. Otherwise, there’s just a huge range of the types of coalitions that form in terms of their organizational diversity. So there’s a lot of unions working with advocacy groups, working with business leagues. The categories that I look at for partisan diversity are Democrat leaning, Republican leaning and nonpartisan. I would say that most coalitions are mostly nonpartisan in nature or are nonpartisan and then working with one party and not both of the parties represented.

Grossmann: It helps coalitions due to attention, signaling, and real changes in the content they produce.

Dwidar: They attract a lot of attention, which means they might be more likely to be paid attention to by the agents who are going through their rulemaking process. But also because I think they’re producing better policy information. So they are bridging their differences, which is a very hard thing to do. That signals credibility. And they’re also bringing in a multitude of information and resources together to produce a common policy consensus without the sort of divisive differences that come with partisanship. So I think that’s why we observed the relationship we observed here. Organizational coalitions are producing more informed policy proposals and sending credibility signals to bureaucrats who have strong incentives to go with prioritized good information and save their time because they’re reading thousands and thousands and sometimes millions and millions of comments. Any heuristic will help.

Grossmann: Low salients issues and long comments lead to more influence.

Dwidar: So one of the relationships that I observe is that coalitions are more influential when they’re lobbying on policies and on proposed rules that are less salient to the folks in the policy subsystem. This is a pretty intuitive relationship. It’s telling us that when the scope of conflict is tighter, so this goes back to the work of Schattschneider and others, one less attention is being paid to a policy domain. When there are fewer actors at play, it’s a lot easier to effect change on that particular domain. That may or may not be a good thing, but it is in line with what we should expect in public policymaking.

Dwidar: The other relationship is about the length of public comments and influence. So I observed that when coalitions submit comments that are longer in terms of their overall word count, they are more likely to share influence or to share language with the final rule. I think this is a pretty comforting finding, there’s just more language to be shared and so they’re sharing more language. I think I would be pretty disturbed if I found the opposite, which is the shorter comments were more influential.

Grossmann: And coalitions are better than individual group lobbying.

Dwidar: Coalitions are really effective in lobbying, and they’re effective in lobbying the bureaucracy. I don’t present these results in this paper, but I have replicated this finding using my data. The big takeaway is that if you can manage it, lobbying with others is a lot more effective than lobbying on your own across a lot of different contexts. And apparently lobbying in a diverse coalition is more effective as well.

Grossmann: Groups know this is where the action is, but it’s still hard to influence agencies.

Dwidar: We have a lot of evidence that most lobbying targets a federal agency. I don’t know that most coalitional lobbying targets a federal agency, but interest groups are really, really active in agency rulemaking. What I think is really interesting about this, and I think a common misconception is that agency rulemaking is cheap and that’s why interest groups lobby federal agencies. I don’t think that’s true at all. I think lobbying agencies and rulemaking is incredibly costly. Effective public comments are very detailed and very technical and often require interest groups to hire law firms to crack the appropriate language or hire independent research firms to go out and collect data and conduct studies. So I think rulemaking is interesting in the sense that it’s both more accessible as a lobbying avenue, but also probably just as arduous as lobbying a member of Congress if not more.

Grossmann: Dwidar says her research fits well with Potter’s findings.

Dwidar: I’m convinced by, and totally amenable to her argument and her findings. I think my findings run in parallel to hers first because even if agencies don’t often change their proposed rules because they have employed a number of tactics to make it such that they don’t have to, they still do on occasion. And so at the very least, I think my findings speak to the conditions under which these groups are effective on these particular occasions. Sort of more generally, I think there are always actors in policy subsystems that are paying attention to what is happening. There might not be many, but I think of them as interest group entrepreneurs who are paying attention to rulemaking activity on a particular policy area. Once they see that there’s an opportunity to comment or get involved if they haven’t been involved previously, they rally others who also have interest in this area. So I think regardless of some of the tactics that agencies can use, there are always interest group actors paying attention because the likelihood of their constituents depends on it.

Grossmann: And Potter says they jointly highlight outsiders and insider strategy.

Potter: It really highlights how important interest groups are in terms of monitoring agencies. Given the scope of bureaucracy today, it is nearly impossible for the small number of staff we have in OMB or on The Hill or even in the court system to oversee the gargantuan nature of the administrative state. So that monitoring is really conducted by interest groups. And so I think this is a really important strand of the literature. And what I think these arguments contribute is they give a sense of which groups are powerful and which groups agencies listen to. So I think I kind of black box that in my study. You don’t read my book and get a sense of, “Oh, this interest group is really the kind that makes a difference.” So I think that pairs really well with digging into the weeds of this broader argument that I make.

Grossmann: But Potter sees more opportunities for influence beyond the text of rules.

Potter: And one thing I think my study does is look and say, well, if we look across these rules, not all of the opportunities for comment are the same. And what I mean is that agencies often do more or less consultations. They’re actually manipulating the opportunity for groups to participate. And what I argue is that agencies think about what groups are going to say because groups do have the ear of Congress, of OIRA and are the ones who initiate lawsuits. And so they’re very powerful. And so the agencies take this potential feedback into account and offer more consultation if they think that feedback might be helpful to them and might shore up their case, and less feedback in instances where they think that the responses they get might be harmful to them. So they’re actually manipulating this.

Potter: I also look at the timing of agency rules. So we only think about agency rules at certain points in their life cycle. So when an agency is sitting in their DC office working on some rule, which they’re probably doing right now, we’re not thinking about that rule. We’re thinking about that rule when we hear about it because it came out in the federal register or when it’s open for public comment or when we, the regulated public have to comply with it. So there’s certain points in the lifecycle of an agency’s role where we pay attention to it.

Potter: And what I argue is that agencies manipulate that. They know that it might be better to sit on this for a little bit, right now is not a good time to release a new policy on X because people are not interested in X or people are really worked up about X. And so they’ll wait, and they’ll wait until the political environment is more supportive of that policy. And so I find a pretty strong timing effects where agencies think about how specific audiences, by which I mean those in Congress, the presidency or the White House may think about a rule and release the rule at times when the political environment is more favorable.

Grossmann: They both see some potential for change under Trump. “A lot of rules have changed,” Potter says, “but we don’t know if that will change behavior on the ground.”

Potter: I’d cover this 20-year period between 1995 and 2014 where the rules of the game are pretty set for agencies in terms of rulemaking, not too much major change in that timeframe. In the past few years, there have been, as you alluded to, many, many changes. And I think we’re still trying to unpack what changes are important and what they mean. So the Congressional Review Act, which was used once in my 20-year period study was used 16 times since 2016. There’s been much more public attention given to the rulemaking process. Trump has an executive order that says for every one rule in, two have to be taken off the books.

Potter: And OIRA recently extended its review to cover independent agencies, which is something that has been talked about for decades and finally has happened. And so I think we’re looking at a whole new set of procedures. Rules of the game have changed. And what I think that means for agencies is not that the era of procedural politicking is over, but instead, how procedural politicking is done is likely to change. So agencies need to adapt strategies to the current political environment. And I think that might suggest that in a decade from now, there should be a bending the rules part two.

Grossmann: Dwidar says agencies respond to the administration, so their culture might be changing.

Dwidar: So some administrations definitely pay a lot more attention to rulemaking activity than others. Some are more concerned with rolling back regulations than advancing the regulatory states. Agencies are definitely influenced by the executive and aspects of their political and policy environment. For instance, agencies might rely much less on outside expertise under an administration that is trying to roll back regulations or dampen the regulatory state, which might mean that rules are just not as informed and also that we might observe more judicial review of agency choices, which is something I think we’ve seen quite a bit of recently under the Trump administration.

Dwidar: As far as what I can speak to empirically using my data, it’s about an 11-year period between 2005 and 2015, which includes both Republican and Democratic administrations. So I think the trends that I observe in the paper can speak to the behavior of both parties right across a wide range of contexts. I’m not sure if I’m comfortable speculating about the Trump administration in particular, but I do think the executive bear’s substantial influence over agency decisions and preferences and culture. And so I expect that we’ll probably see some pretty wacky things in terms of rulemaking in the future.

Grossmann: Both researchers mostly see their findings as good news. Potter says it’s good that bureaucrats have power.

Potter: Should we care that if I’m right that bureaucrats have so much power. On the one hand, bureaucrats, if we look at them as a group, by that I mean both political appointees and civil servants, they are more demographically representative of the American public than members of Congress who are many, many millionaires in that group or than most people that work in the Office of Management and Budget or in the courts. They are also more ideologically in step with the average citizen than members of Congress are. So we know from lots of work in political science that members of Congress are very polarized and that there tends to be leapfrogging among districts or within districts. What I say shows that if we have bureaucrats making policy, we don’t have this leapfrogging and we have a pretty stable group of people who are pretty moderate making decisions. Oh, and by the way, they’re also experts in their respective areas.

Potter: So in a normative sense, I’m not too worried about the policies that might emerge from that given what we know about descriptive representation generally and how that affects choices that leaders might make. Now, the broader question I think is, are we okay with this level of delegation to a non-elected body. And I think given that we’re a country of 320 odd million people and we have a government that does so many things that we have to be a little bit okay with that because we can’t live in a world where Congress could ever take on these responsibilities. Delegation is a fact of modern government. What I’m showing, that’s not necessarily a bad fact to have to live with.

Grossmann: And Dwidar says there is some capture, but there’s also real informed policymaking.

Dwidar: There are some topics, and there are some domains that might be owned by certain industries where we both might perceive there to be dominant influenced by the oil industry and for that influence actually to be true. However, I think at the same time, bureaucrats in particular have really strong incentives to prioritize good policy information. They don’t want to deal with massive public scandal, congressional or presidential critique. They don’t want to be sued. And so I think, by and large, for the most part, they’re trying to produce good public policy. And interest groups for the most part are trying to give them the information they need to produce that good public policy. So I think we’re both often a little bit not.

Grossmann: In her larger project, Dwidar is even finding that underrepresented groups can achieve influence through diverse coalitions.

Dwidar: So my broader dissertation and research agenda is actually about whether coalitional lobbying allows interest groups that represent minorities. So those historically marginalized on the basis of race, class, and gender to more effectively represent their most disadvantaged constituents. And so I asked some of the same questions that I present in this paper, does coalition lobbying enhance your influence? Does diverse coalition lobbying enhance your influence? I’d find both similar results and much, much larger effect sizes, which I find really remarkable. And it’s what I was expecting, and so I’m very excited about it. More broadly, some questions I would like to tackle in the future stemming from this project have to do with the relationship between partisan coalitions, lobbying agencies with historically partisan leanings and whether we tend to see certain constituencies represented more often by certain coalitions. So are we going to see poor black women more likely to be represented when the coalition includes groups that represent women, impoverished groups, and African-Americans, for example. So those are some of the things I’m hoping to get out of the future.

Grossmann: Potter’s next step is to look at agencies’ use of evidence and contracts.

Potter: What I’m currently looking at is the use of procedural politicking in a different venue, which is government contract. And thinking about how agencies apply these same lessons from rulemaking in a different venue. And I’m also really interested in not just how they apply it in contracting, but how they apply it in the use of evidence. So agencies use evidence in lots of different ways. And so I’m thinking about which kinds of evidence can be used where. I think we have a lot of questions left in the rulemaking literature, but I’m tuned to other areas at the moment

Grossmann: There’s a lot more to learn. The Science of Politics is available biweekly from the Niskanen Center. I’m your host, Matt Grossmann. Thanks to Rachel Augustine Potter and Maraam Dwidar for joining me. Please check out, Bending the Rules and Diverse Lobbying Coalitions and Influence in Notice-and-Comment Rulemaking, and then listen in next time.