Race and Regulation Podcast

Daniel E. Ho is the William Benjamin Scott and Luna M. Scott Professor of Law at Stanford Law School, Professor of Political Science, Senior Fellow at the Stanford Institute for Economic Policy Research, Associate Director of the Stanford Institute for Human-Centered Artificial Intelligence, and Director of the Regulation, Evaluation, and Governance Lab (RegLab).

Ho serves as an appointee to the National Artificial Intelligence Advisory Commission (NAIAC), advising the White House on artificial intelligence, and is a Public Member of the Administrative Conference of the United States (ACUS).

Professor Ho’s scholarship centers on quantitative empirical legal studies, with a substantive focus on administrative law and regulatory policy, antidiscrimination law, and courts. With the RegLab, his work has developed high-demonstration projects of data science in public policy, through partnerships with a range of government agencies, including the Internal Revenue Service, the Environmental Protection Agency, the Santa Clara County Public Health Department, and Seattle and King County Public Health. The collaboration with Santa Clara County was awarded the Innovative Practice Gold Award for “the highest level of program innovation” to serve “community during the COVID-19 pandemic” by the National Association of County and City Health Officials.

He is also a Faculty Fellow at the Center for Advanced Study in the Behavioral Sciences, Faculty Affiliate at the Woods Institute for the Environment, and Faculty Affiliate at the Wilson Sheehan Lab for Economic Opportunities.

Professor Ho received his J.D. from Yale Law School and his Ph.D. in political science from Harvard University, and he clerked for Judge Stephen F. Williams on the U.S. Court of Appeals, District of Columbia Circuit. His research has appeared in a wide range of journals such as the Stanford Law Review, the Yale Law Journal, the N.Y.U. Law Review, PNAS, BMJ, Nature Medicine, JAMA Health Forum, the American Journal of Public Health, Nature Sustainability, the Journal of the American Statistical Association, the American Statistician, ACM FAccT, the Journal of Law, Economics, and Organization, the American Economic Journal: Economic Policy, the American Law and Economics Review, Quarterly Journal of Political Science, Political Analysis, the Journal of Legal Studies, and the Journal of Empirical Legal Studies. He has also written for popular outlets, such as the New York Times, the Boston Review, and The Hill, and his work has been featured in many news outlets, including the New York Times, the Wall Street Journal, Forbes, Atlantic, Slate, and Scientific American.

Professor Ho is the recipient of numerous awards, including the John Bingham Hurlbut Award for Excellence in Teaching at Stanford Law School (2010), the Carole Hafner Award for the best paper at the International Conference on Artificial Intelligence and Law (2021), the Best Empirical Paper Prize from the American Law and Economics Review, the Warren Miller prize for the best paper published in Political Analysis (2008), and the Pi Sigma Alpha award for the best paper delivered at the Midwest Political Science Association meeting (2004).

He served as President for the Society of Empirical Legal Studies (2011-12), as co-editor of the Journal of Law, Economics, and Organization (2013-16), and is on the Faculty Advisory Board for Stanford Impact Labs, the Board for the Society for Empirical Legal Studies (SELS), and the Faculty Advisory Council for Women in Data Science (WiDS).

To watch the video of the full lecture on which this podcast episode is based, visit PPR’s YouTube channel.

• Engstrom, David Freeman, Daniel E. Ho, and Cristina Isabel Ceballos, “Disparate Limbo: How Administrative Law Erased Antidiscrimination,” Yale Law Journal 131, no. 2 (2021): 370-474.
• Coston, Amanda, Neel Guha, Derek Ouyang, Lisa Lu, Alexandra Chouldechova, and Daniel E. Ho. “Leveraging Administrative Data for Bias Audits: Assessing Disparate Coverage with Mobility Data for COVID-19 Policy.” In Conference on Fairness, Accountability, and Transparency (FAccT ’21), March 3–10, 2021, Virtual Event, 173-184. New York: ACM, 2021.
• Engstrom, David Freeman, Daniel E. Ho, Catherine M. Sharkey, and Mariano-Florentino Cuéllar, Government by Algorithm: Artificial Intelligence in Federal Administrative Agencies. Washington, DC: Administrative Conference of the United States, 2020.

Music: Joy Ike’s “The Fall Song”

Dan Ho: We are living in moment of racial reckoning, and it is absolutely the case that people hold quite divergent perspectives on the role of race, though I think a lot of people are also reflecting on their understanding of the role of race.

Cary Coglianese: That’s Dan Ho, an expert on administrative law, delivering a lecture organized by the Penn Program on Regulation at the University of Pennsylvania. I’m Cary Coglianese, the director of the Penn Program on Regulation and a professor at the University of Pennsylvania. Welcome to our podcast, Race and Regulation. 

In this series, we are talking about the most fundamental responsibility of every society: ensuring equal justice, and dignity and respect, to all people. Advancing racial justice calls for all of us to understand better the racial dimensions of regulatory systems and institutions.

We’re glad you can join us as we hear from Dan Ho, a professor at the Stanford Law School. His remarks draw from research he conducted with Cristina Ceballos and David Freeman Engstrom, both at Stanford. That research was published in an article in the Yale Law Journal entitled, “Disparate Limbo: How Administrative Law Erased Antidiscrimination.”

Professor Ho and his colleagues start with a truth about all public policies and laws: they affect some people differently than others. Professor Ho next points to standard principles of administrative law—that is, the law that governs the work of administrative agencies, whether it’s the Internal Revenue Service, the Federal Communications Commission, the Environmental Protection Agency, or many, many more departments and offices in the executive branch of government. Under standard principles of administrative law, these agencies must justify the disparate effects their actions will have. But administrative law doesn’t call for agencies to justify disparate effects based on race. Those are treated as a matter of civil rights law. They are not part of administrative law system’s requirement for reasoned decision-making under what is known as the “arbitrary and capricious” standard of the Administrative Procedure Act. As Professor Ho explains:

DH: It is blackletter administrative law that the reviewing court shall hold unlawful and set aside agency actions, action findings, and conclusions found to be arbitrary and capricious, and claims for disparate effects are entertained seriously across the administrative law canon. 

To give you some examples, when the Forest Service issues a particular rule for park visitors, differential effects between kayakers and jet boaters are considered in an arbitrary and capricious type challenge. We’ve had a Supreme Court case that contemplated seriously the differential effects on types of business owners—small broadcasters, or large broadcasters. And similarly, when the Navy has a kind of rule about the use of sonar, administrative law takes seriously claims for differential effects on small dolphin pods versus large dolphin pods. 

But when it comes to differential effects across race, administrative law is, in a sense, curiously silent. So one example comes from a claim by a public interest group challenging the use of methyl bromide as pesticides, and the case did not recognize the claim for differential effects between Latinx schoolchildren and white schoolchildren that were differentially impacted in California. Nor in a case called Garcia v. Vilsack, where there was a claim about the failure to investigate discrimination and the kind of differential effects were minority farmers’ claims recognized under administrative law. How is it that we have arrived at this curious status quo? 

CC: To understand why courts say that the Administrative Procedure Act—or the APA—requires agencies to justify the differential effects of their policies between, say, kayakers and jet boaters, but does not require them to explain differential effects between different racial communities, Professor Ho looks back to the way the courts have construed the effects of the civil rights laws that Congress passed in the 1960s.

DH: Under the Civil Rights Act of 1964, Title VI protects individuals from discrimination in federal programs or activities. Section 601 says, “No person in the United States shall, on the ground of race, color, or national origin, be excluded from participation in any program or activity receiving federal financial assistance.” Section 602 is the agency mechanism to potentially cut off federal funds when there is a violation of 601 by a recipient of federal financial assistance.

CC: In addition to providing that federal agencies can cut off their funding of state or local entities that discriminate, the Civil Rights Act also provided opportunities for individuals discriminated by these federally funded entities to go to court. The Supreme Court in 1979, in the case of Cannon v. University of Chicago, held that Section 601 of the Civil Rights Act contains an implied right of action allowing aggrieved individuals to go to court. And the statute itself, in Section 603, provides an avenue to go to court to challenge federal agencies’ decisions about federal funding.  Professor Ho quotes a key passage from Section 603:

DH: In the case of action that isn’t otherwise subject to judicial review, terminating or refusing to grant or to continue financial assistance, any person aggrieved may obtain judicial review of such action in accordance with the APA. 

CC: The APA generally allows anyone negatively affected by any federal agency action to challenge that action in federal court. Under the APA, the courts can review agency actions for their constitutionality, their compliance with statutes, and their adherence to any required agency procedures. And the APA also allows for a catchall opportunity for a court to strike down an agency action as being “arbitrary and capricious.” To win an argument that an agency has been arbitrary and capricious, a challenger has to show that the agency failed to provide adequate reasons for its actions. If the action had disproportionate effects on different types of people or different types of businesses, the agency would normally need to provide reasons for those differences.

The APA had been on the books for nearly twenty years before Congress passed the Civil Rights Act. Its passage would come to affect the abilities of individuals discriminated based on race to rely on the APA. You see, a key avenue for plaintiffs suing under the APA would be to force agencies to have to explain any decisions they make to allow programs they fund to continue to discriminate. Because this would be indefensible, it would effectively compel federal agencies to do more to combat racial discrimination at the state and local levels of government. But this avenue was shut off by the courts.

They did so by focusing on Section 704 of the APA, which, as Professor Ho explains, states as follows.

Music: Joy Ike’s “Home”

DH: “Agency action for which there is no other adequate remedy in a court is subject to judicial review.”

CC: This provision says that the APA will provide a backstop, if there is no other adequate way for individuals to obtain judicial review of agency action.  But the flipside is that, if there is an “adequate remedy” elsewhere, the courts won’t let anyone sue under the APA. The courts focused on this “adequate remedy” language in Section 704 and concluded that the private rights of action allowed under the Civil Rights Act effectively barred racial claims from being raised under the APA.

DH: To sketch out the way in which the D.C. Circuit, in particular, fleshes this out is that as a backdrop we have the 1979 Cannon case that is widely seen to provide for a private cause action for disparate impact.

Under that backdrop, when the D.C. Circuit first heard a case that involved the Office of Revenue Sharing in claims of discrimination in block grants to the states, when the challenge was made against the Office of Revenue Sharing, the D.C. Circuit in an en banc decision, relied particularly on APA Section 704, stating that there was an alternative adequate remedy, and that was under Cannon to pursue case-by-case private actions against the recipients of block grants.

What we then get is another en banc decision in the D.C. Circuit, the Women’s Equity case, which involved federally funded educational institutions and invoked the similar kind of move to infer that because of Cannonand the implied private cause of action, there is an alternative adequate remedy. Therefore, the claim cannot be heard under the APA.

CC: The court of appeals in the Women’s Equity case held that the Civil Rights Act provided a “adequate remedy,” so the APA was no longer available as a basis for a lawsuit. The Women’s Equity decision came down in 1990, and the opinion was written by Ruth Bader Ginsburg, who was still at that point in her life a judge on the U.S. Court of Appeals for the D.C. Circuit.  At that time, individuals affected by racial discrimination could sue directly to enforce Section 601 of the Civil Rights Act any time they were adversely affected by a program that relied on federal funds. They could prevail even if they were unable to show that the officials running the state or local program had intended to discriminate. It was enough just to show the existence of a disparate impact—that is, different outcomes between different racial groups. But eleven years later, in a 2001 case called Alexander v. Sandoval, the Supreme Court took away this option. The Court still said plaintiffs could go to court to enforce the antidiscrimination rule in Section 601 of the Civil Rights Act. But the only way going forward that they could win was to show proof of an intent to discriminate—a much harder case to make.

DH: What is particularly curious is that Women’s Equity looms large in this particular space when individual litigants raise both discrimination and APA claims, even after the Supreme Court reverses course and in Alexander v. Sandoval, finds that there is no private cause of action for disparate impact claims, reversing what had been the kind of consensus under Cannon since 1979. And nonetheless, Women’s Equity continues to get cited in this case law as a way to channel discrimination claims away from garden variety APA suits, leading to what we document and call a form of disparate limbo. Civil rights plaintiffs sit in disparate limbo, unable to make out the stringent intent showings, but simultaneously barred from mounting claims invoking the APA’s baseline guarantees of non-arbitrariness.

CC: Legally speaking, this limbo means that individuals subjected to racial discrimination today have almost the worst of all possible options available in court. They can no longer raise disparate impact arguments under the Civil Rights Act. That is, they have to show intent to discriminate and can’t rely on just showing that one racial group suffers worse than others. But they also can’t proceed under the Administrative Procedure Act, or APA, to force the government to explain why they tolerate disparate impacts on the basis of race. After all, they can still force the government to have to explain disparate impacts between kayakers and jet boaters.

So, what is to be done? Professor Ho suggests three changes to the law.

DH: First, it would be wise for federal courts to cabin the scope of Women’s Equity in really two ways. The first is that APA Section 704 should allow for APA claims for differential effects that can no longer be brought per Sandoval under Title VI. And the particular concern animating the en banc D.C. Circuit in Women’s Equityreally stemmed from the colossal nature of the case. The worry was about federal supervision of sub-federal actors and bringing in all sorts of different kinds of schools. Women’s Equity should not apply to cases where there is a direct challenge to the federal agency and where this “overseeing the overseer” concern is not really at issue.

Second, we do think it’s worth taking seriously this analysis of what “adequacy” means under APA in Section 704. One case that was not cited in the Women’s Equity decision is the Bowen v. Massachusetts Supreme Court case that really interrogated the adequacy of the alternative claim much more seriously. And another canon of administrative law, Abbott Labs, of course, painted 704 as being a provision that should be interpreted to broaden access to the courts, not to narrow access to the courts.

Third, we do think that some of the stark contrasts between the kinds of differential effects that are recognized by the courts under the APA, like those relating to the National Environmental Policy Act or the Regulatory Flexibility Act in terms of different kinds of small business interests. Those actually are the kinds of moves that could be brought into administrative law essentially to mandate similar kinds of disparate impact assessments that would, as a result, lead for a more rigorous assessment of these kinds of claims for differential effects.

One example that Professor Olatunde Johnson has written about in her work that is an exemplar along those lines, comes from the Obama Administration, where the City of Oakland had proposed building an airport connector from the BART, the highspeed rail system, to the Oakland Airport. It was a $500 million infrastructure project, but the Federal Transit Administration initiated a civil rights review because the concern was that the routing of this airport connector would have gone through the heavily African American neighborhood in the East Oakland community, and the fare structure would have fundamentally changed to have actually made it a much less affordable option to get to the airport. It also said that there was insufficient participation by local neighborhood groups in this particular proceeding. And under that mandated equity analysis, ultimately, the $70 million of federal funds that were used for this project were diverted toward other means.

CC: In addition to changing how the courts view the relationship between civil rights laws and the APA, Professor Ho urges a rethinking of how the federal government collects data related to race.

DH: The other intervention that I will highlight that I think is a very affirmative step forward is Executive Order 13985, Biden’s Racial Justice Executive Order, which says that each agency must assess whether its programs and policies perpetuate systemic barriers for people of color. Many federal data sets are not disaggregated by race, ethnicity, gender, disability, income, veteran status, or other key demographic variables. And this Executive Order, if implemented faithfully, would really lead to much more of a kind of equity directive in federal regulatory agencies.

That said, the language right here points to one of the really substantial epistemic challenges here, which is that under the Privacy Act of 1974, there are many agencies that don’t actually have the ability to link their records easily with information about protected attributes. That runs into particular challenges when we think about what is on the horizon in terms of algorithmic bias. Cary has done some leading work on this, and with David Engstrom, Cathy Sharkey, and Tino Cuéllar, we wrote a report about the use of algorithmic decision tools by federal regulatory agencies. One kind of emerging mantra from those coming from within the machine learning community is that there is no fairness without awareness. It is very difficult to understand what the potential impact might be and how to mitigate potential bias without understanding these disparities. And that was the subject of some extensive discussion in the GAO’s Controller Generals Forum, where the GAO was trying to come up with audit standards for algorithmic decision tools, and remains a really big challenge if you simply don’t have access to protected attributes to understand these dimensions.

A lot of the future work that remains to be done is to make these kinds of disparities more visible.

Music: Joy Ike’s “Home”

CC: How might this actually work?

DH: Let me give you one example from the COVID response. This was done in Georgia, where a lot of the times, just with federal regulatory agencies, race and ethnicity information was missing in COVID surveillance data. And they used some imputation techniques based on name and address information of where patients resided and showed that basically once you add those imputations, the absolute racial and ethnic disparity increased 1.3- and 1.6-fold for classified Black and Hispanic persons.

CC: And this is possible to do more extensively, as Professor Ho has shown with students involved in his Regulatory Research Lab at Stanford.

DH: Our team at the RegLab has done some work with Santa Clara County’s Public Health Department. And if you look at the 1997 OMB categories for race, if anything, it looks like Asian Americans are faring the best in terms of most measures of COVID outcomes, case rates, positivity rates, and vaccination rates. But as many of you know, one of the big points of contention within the Asian American community has been to disaggregate and come up with better measurement since the category of Asian Americans is an omnibus category. And it turns out when you do do that, case rates look significantly worse, for instance, for Vietnamese Americans in Santa Clara County than if you are simply aggregating across a large omnibus category of Asian Americans.

Of course, once you have uncovered those kinds of blind spots, that enables you to actually reduce these disparities by understanding the sources. So I do think there is a lot that we can do to uncover these kinds of blind spots. Part of all these discussions is to heighten awareness of the role of the regulatory state in these kinds of dimensions.

CC: Uncover the blind spots, indeed, because we can’t fix problems we don’t see. And we can’t learn to do better if we keep overlooking racial claims in administrative law, thinking there exist adequate remedies under prevailing interpretations of civil rights law.

To move forward, then, would be to do more to study and understand, rather than to overlook or wish away, to open our eyes to the ways that governmental policies and programs can have disparate impacts across different racial communities.

Music: Joy Ike’s “Walk”

CC: Thank you for listening to this episode of Race and Regulation. We hope you have learned more about the role administrative law can play in the quest for racial justice. 

This podcast has been adapted from a lecture delivered by Professor Dan Ho in 2022. He spoke as part of the Penn Program on Regulation’s lecture series on race and regulation, co-sponsored by the Office on Equity and Inclusion at the University of Pennsylvania Carey Law School. 

I’m Cary Coglianese, the director of the Penn Program on Regulation. For more about our program and free public events, visit us at pennreg.org. You can also find other episodes in our Race and Regulation series wherever you get your podcasts.

This podcast was produced by Patty McMahon, with help from Andy Coopersmith, our program’s managing director. Our music is by Philadelphia-based artist, Joy Ike.