DACA Lawsuit Part 2: The Baseline Problem in Disparate Impact Cases
by Michael Dorf
In my post last Friday, I discussed one of the claims raised in the lawsuit against the Trump administration brought by state Attorneys General to challenge the rescission of DACA--that it violates equal protection because it is motivated by racial animus. I noted that the challenge to DACA is structurally similar to the challenge to the travel ban but stronger in one way and weaker in two ways. It's (1) stronger because the claimants are in the U.S. already but (2)weaker in that discrimination on the basis of nationality is not, strictly speaking, the same thing as discrimination on the basis of national origin, and, (3) by contrast with the Travel Ban, which has an obvious and obviously intended disparate impact on Muslims, the DACA rescission applies to all undocumented immigrants, regardless of their country of origin.
During the panel discussion on immigration and executive action (video available here), Ilya Somin pointed to another difference that makes the DACA challenge weaker than the Travel Ban challenge: Whereas there is a straight line from Trump's "total and complete shutdown of Muslims entering the United States" to the Travel Ban, the path from Trump's anti-Mexican statements to DACA rescission is less clear; thus, attributing DACA rescission to unconstitutional bias is not as easy as attributing the Travel Ban to unconstitutional bias. As I noted during the discussion, I agree with Prof. Somin's assessment. I don't think that's fatal to the case against DACA rescission, however (and perhaps neither does he). Where there is a prima facie case of illicit motive, the burden shifts to the government to prove that the action (here DACA rescission) would have occurred even absent the illicit motive. Perhaps the government will be unable to make that case persuasively.
Meanwhile, a commenter on my prior post made more or less the same point as Prof. Somin. The commenter also called into question my analysis of Palmer v Thompson, the case in which the SCOTUS held that a Mississippi municipality's decision to close its public swimming pools rather than desegregate them was not racially discriminatory because doing so affected everyone equally. I wrote in the post that in light of the more recent disparate impact cases, if there had been proof of a disparate impact on African Americans in Palmer, the plaintiffs would have won. The commenter was dubious. Wasn't it obvious, he asked, that the closing of the public pools would have a disparate impact on African Americans? And if so, doesn't that mean that the subsequent disparate impact cases implicitly overrule Palmer (or that Palmer is wrong)? Good questions. I'll see if I can make any headway on them.
To my mind, there are two sorts of questions here. One, which does not interest me much (and which I lack the ability to answer in any event) is what the majority of justices who decided Palmer would have thought or actually did think about the disparate impact question. Maybe they thought that the decision to close the pools had a disparate impact on African Americans and didn't care. Or maybe the justices didn't think about disparate impact as such. Palmer was decided five years before Washington v. Davis--which held that the disparate impact of a race-neutral policy was not constitutionally objectionable absent evidence that the disparate impact was the product of intentional race discrimination. Just three years before Palmer, the Court--in a free speech case, United States v. O'Brien--said this: "It is a familiar principle of constitutional law that this Court will not strike down an otherwise constitutional statute on the basis of an alleged illicit legislative motive." Still operating under that principle, the Court in Palmer might well have thought that disparate impact was simply irrelevant.
That brings me to the second kind of question posed by the commenter: Setting aside what the Palmer Court itself had in mind, can Palmer be reconciled with modern disparate impact cases? I think that it can be, if Palmer is taken to stand for the following proposition: A facially neutral law or policy will be subject to heightened equal protection scrutiny if and only if it both was motivated by an illicit motive (such as racial subordination) and it disproportionately burdens people on that basis.
Now, understanding Palmer as standing for the foregoing proposition only makes sense if we also understand Palmer as tacitly assuming that the closing of the swimming pools affected persons of all races equally. It probably did not have such an equal impact, but perhaps the Palmer Court itself wasn't aware of that fact. In any event, as I've said, I'm not interested in reconstructing the psychology of the justices of the Palmer Court.
What am I interested in? Chiefly in understanding the baseline against which one measures a disparate impact. I can explain why this is a puzzle using a hypothetical example.
Suppose that a state university admits students strictly based on a combination of high school grade and standardized test scores. Suppose further that the university's board of governors is concerned that using these criteria leads to the admission of "too many" Asian Americans and "too few" white students. Accordingly, the board of governors switches to a lottery in which everyone in the state with or on track to receive a high school degree and at least a B average has an equal chance of admission. Although the lottery is formally race-neutral, let us stipulate that this program has an illicit purpose. Adoption of the lottery cannot be justified as a form of affirmative action for white applicants; it's simply illicit. But does adoption of the lottery have a disparate impact on Asian American applicants?
The answer depends on the baseline. Relative to their prior admission rate under the grades-plus-scores approach, the lottery has a disparate impact. But relative to their distribution in the population or the applicant pool, the lottery has no disparate impact. So which measure is correct?
One might think that Title VII case law provides an answer, because the lower courts have decided a fair number of Title VII disparate impact cases. Under Title VII, a plaintiff or class of plaintiffs can prevail in an employment discrimination case by showing disparate impact even without a showing of intent. Yet my admittedly cursory research failed to turn up cases involving the precise question I've raised here. Various cases address a different baseline question: They say that in measuring disparate impact, the relevant baseline is the qualified applicant pool, rather than the general population.
That approach has been criticized by some as failing to account for the fact that the applicant pool is itself a product of discrimination and exists in dynamic relation to employment opportunities. But whatever one thinks of that baseline problem, it is not the baseline problem I've identified here: Whether the baseline is the status quo ante or an idealized result given the applicants. It's possible that an answer can be found in Title VII cases or cases in other statutory areas where disparate impact liability is available. I invite readers with a deeper familiarity of those fields to enlighten me in the comments.
Of course, even if statutory case law does address the baseline question I've raised, that would not be dispositive of the constitutional question, because the constitutional law governing disparate impact differs from the statutory case law. Meanwhile, perhaps constitutional law does provide some guidance after all.
Consider Schuette v. Coalition to Defend Affirmative Action By Any Means Necessary (BAMN). There the Court rejected the argument (which had been accepted by the Sixth Circuit) that a Michigan ballot initiative that banned affirmative action in public universities violated equal protection because it removed a racial issue from the ordinary electoral process. In so doing, Justice Kennedy's plurality opinion distinguished earlier cases that comprised the so-called "political process doctrine." In one such case, Hunter v. Erickson, the Court invalidated an Akron ballot initiative that amended the city charter to disempower the ordinary political process from enacting housing anti-discrimination ordinances and also repealed an existing such ordinance.
In neither BAMN nor Hunter did the Court speak in disparate impact terms, but they nonetheless fit. There is little doubt that both the Hunter Court itself and the BAMN Court saw the Akron ballot initiative as racially discriminatory in purpose. And both Courts had little difficulty in seeing that it also disadvantaged African Americans--even though there is no constitutional right to a fair housing ordinance forbidding private discrimination. Here is how the BAMN Court characterized the holding in Hunter: "Hunter rests on the unremarkable principle that the State may not alter the procedures of government to target racial minorities." Note that the key word is "alter." Whether the impacted group is disadvantaged is measured relative to the status quo ante, not relative to what the government could do in theory.
The BAMN Court found no constitutional violation in Michigan's ballot initiative--but that's not because the Court thought that there was no disparate impact. If the Court thought that, it simply could have said "there's no right to affirmative action so there's no problem with the government taking it away." What the Court actually said, in effect, was that the taking away of affirmative action, while dealing with a question of race, was not based on an illicit racial motive. Whether to have affirmative action programs, the Court said, was a matter for the democratic process.
In light of that approach, it seems reasonably clear that if ending a program that benefits a group is based on racial or similar animus (as in my state university hypothetical example involving anti-Asian American animus), the termination of the benefit suffices to show a disparate impact--even though the affected group is no worse off than it might have been had the program never been adopted.
Prop 8 provides another example. Before the SCOTUS held that there was no appellate standing in the case, the Ninth Circuit ruled that California could not take away same-sex marriage based on animus, even if doing so put same-sex couples in no worse a position than they would have occupied if they had never been granted a right to SSM. (The case was decided in 2012 on the assumption that there was no freestanding constitutional right to SSM.) As I explained at the time, the decision made sense in those terms for more or less the sorts of reasons I have set forth here.
And so, I conclude, if it can be shown that DACA is being terminated because of illicit animus (an admittedly substantial "if"), then the fact that erstwhile DACA recipients would be worse off than before DACA termination suffices to make out an equal protection claim--even though they will be no worse off than they would have been if DACA had never been adopted.
***
Postscript 1: In characterizing DACA recipients as no worse off if DACA is rescinded than if it had never been adopted, I am speaking only with respect to the equal protection claim. The complaint also raises a due process objection to the government's use of information that it only received as a result of people coming forward in reliance on assurances that such information would not be used against them.
Postscript 2: In the foregoing I am assuming that the challenge to the Travel Ban is quite strong. Yesterday's order by Justice Kennedy staying the Ninth Circuit mandate with regard to travel assurances for refugees does not undermine that assumption--as it goes to questions about who may obtain relief rather than the validity of the Travel Ban as such.
In my post last Friday, I discussed one of the claims raised in the lawsuit against the Trump administration brought by state Attorneys General to challenge the rescission of DACA--that it violates equal protection because it is motivated by racial animus. I noted that the challenge to DACA is structurally similar to the challenge to the travel ban but stronger in one way and weaker in two ways. It's (1) stronger because the claimants are in the U.S. already but (2)weaker in that discrimination on the basis of nationality is not, strictly speaking, the same thing as discrimination on the basis of national origin, and, (3) by contrast with the Travel Ban, which has an obvious and obviously intended disparate impact on Muslims, the DACA rescission applies to all undocumented immigrants, regardless of their country of origin.
During the panel discussion on immigration and executive action (video available here), Ilya Somin pointed to another difference that makes the DACA challenge weaker than the Travel Ban challenge: Whereas there is a straight line from Trump's "total and complete shutdown of Muslims entering the United States" to the Travel Ban, the path from Trump's anti-Mexican statements to DACA rescission is less clear; thus, attributing DACA rescission to unconstitutional bias is not as easy as attributing the Travel Ban to unconstitutional bias. As I noted during the discussion, I agree with Prof. Somin's assessment. I don't think that's fatal to the case against DACA rescission, however (and perhaps neither does he). Where there is a prima facie case of illicit motive, the burden shifts to the government to prove that the action (here DACA rescission) would have occurred even absent the illicit motive. Perhaps the government will be unable to make that case persuasively.
Meanwhile, a commenter on my prior post made more or less the same point as Prof. Somin. The commenter also called into question my analysis of Palmer v Thompson, the case in which the SCOTUS held that a Mississippi municipality's decision to close its public swimming pools rather than desegregate them was not racially discriminatory because doing so affected everyone equally. I wrote in the post that in light of the more recent disparate impact cases, if there had been proof of a disparate impact on African Americans in Palmer, the plaintiffs would have won. The commenter was dubious. Wasn't it obvious, he asked, that the closing of the public pools would have a disparate impact on African Americans? And if so, doesn't that mean that the subsequent disparate impact cases implicitly overrule Palmer (or that Palmer is wrong)? Good questions. I'll see if I can make any headway on them.
To my mind, there are two sorts of questions here. One, which does not interest me much (and which I lack the ability to answer in any event) is what the majority of justices who decided Palmer would have thought or actually did think about the disparate impact question. Maybe they thought that the decision to close the pools had a disparate impact on African Americans and didn't care. Or maybe the justices didn't think about disparate impact as such. Palmer was decided five years before Washington v. Davis--which held that the disparate impact of a race-neutral policy was not constitutionally objectionable absent evidence that the disparate impact was the product of intentional race discrimination. Just three years before Palmer, the Court--in a free speech case, United States v. O'Brien--said this: "It is a familiar principle of constitutional law that this Court will not strike down an otherwise constitutional statute on the basis of an alleged illicit legislative motive." Still operating under that principle, the Court in Palmer might well have thought that disparate impact was simply irrelevant.
That brings me to the second kind of question posed by the commenter: Setting aside what the Palmer Court itself had in mind, can Palmer be reconciled with modern disparate impact cases? I think that it can be, if Palmer is taken to stand for the following proposition: A facially neutral law or policy will be subject to heightened equal protection scrutiny if and only if it both was motivated by an illicit motive (such as racial subordination) and it disproportionately burdens people on that basis.
Now, understanding Palmer as standing for the foregoing proposition only makes sense if we also understand Palmer as tacitly assuming that the closing of the swimming pools affected persons of all races equally. It probably did not have such an equal impact, but perhaps the Palmer Court itself wasn't aware of that fact. In any event, as I've said, I'm not interested in reconstructing the psychology of the justices of the Palmer Court.
What am I interested in? Chiefly in understanding the baseline against which one measures a disparate impact. I can explain why this is a puzzle using a hypothetical example.
Suppose that a state university admits students strictly based on a combination of high school grade and standardized test scores. Suppose further that the university's board of governors is concerned that using these criteria leads to the admission of "too many" Asian Americans and "too few" white students. Accordingly, the board of governors switches to a lottery in which everyone in the state with or on track to receive a high school degree and at least a B average has an equal chance of admission. Although the lottery is formally race-neutral, let us stipulate that this program has an illicit purpose. Adoption of the lottery cannot be justified as a form of affirmative action for white applicants; it's simply illicit. But does adoption of the lottery have a disparate impact on Asian American applicants?
The answer depends on the baseline. Relative to their prior admission rate under the grades-plus-scores approach, the lottery has a disparate impact. But relative to their distribution in the population or the applicant pool, the lottery has no disparate impact. So which measure is correct?
One might think that Title VII case law provides an answer, because the lower courts have decided a fair number of Title VII disparate impact cases. Under Title VII, a plaintiff or class of plaintiffs can prevail in an employment discrimination case by showing disparate impact even without a showing of intent. Yet my admittedly cursory research failed to turn up cases involving the precise question I've raised here. Various cases address a different baseline question: They say that in measuring disparate impact, the relevant baseline is the qualified applicant pool, rather than the general population.
That approach has been criticized by some as failing to account for the fact that the applicant pool is itself a product of discrimination and exists in dynamic relation to employment opportunities. But whatever one thinks of that baseline problem, it is not the baseline problem I've identified here: Whether the baseline is the status quo ante or an idealized result given the applicants. It's possible that an answer can be found in Title VII cases or cases in other statutory areas where disparate impact liability is available. I invite readers with a deeper familiarity of those fields to enlighten me in the comments.
Of course, even if statutory case law does address the baseline question I've raised, that would not be dispositive of the constitutional question, because the constitutional law governing disparate impact differs from the statutory case law. Meanwhile, perhaps constitutional law does provide some guidance after all.
Consider Schuette v. Coalition to Defend Affirmative Action By Any Means Necessary (BAMN). There the Court rejected the argument (which had been accepted by the Sixth Circuit) that a Michigan ballot initiative that banned affirmative action in public universities violated equal protection because it removed a racial issue from the ordinary electoral process. In so doing, Justice Kennedy's plurality opinion distinguished earlier cases that comprised the so-called "political process doctrine." In one such case, Hunter v. Erickson, the Court invalidated an Akron ballot initiative that amended the city charter to disempower the ordinary political process from enacting housing anti-discrimination ordinances and also repealed an existing such ordinance.
In neither BAMN nor Hunter did the Court speak in disparate impact terms, but they nonetheless fit. There is little doubt that both the Hunter Court itself and the BAMN Court saw the Akron ballot initiative as racially discriminatory in purpose. And both Courts had little difficulty in seeing that it also disadvantaged African Americans--even though there is no constitutional right to a fair housing ordinance forbidding private discrimination. Here is how the BAMN Court characterized the holding in Hunter: "Hunter rests on the unremarkable principle that the State may not alter the procedures of government to target racial minorities." Note that the key word is "alter." Whether the impacted group is disadvantaged is measured relative to the status quo ante, not relative to what the government could do in theory.
The BAMN Court found no constitutional violation in Michigan's ballot initiative--but that's not because the Court thought that there was no disparate impact. If the Court thought that, it simply could have said "there's no right to affirmative action so there's no problem with the government taking it away." What the Court actually said, in effect, was that the taking away of affirmative action, while dealing with a question of race, was not based on an illicit racial motive. Whether to have affirmative action programs, the Court said, was a matter for the democratic process.
In light of that approach, it seems reasonably clear that if ending a program that benefits a group is based on racial or similar animus (as in my state university hypothetical example involving anti-Asian American animus), the termination of the benefit suffices to show a disparate impact--even though the affected group is no worse off than it might have been had the program never been adopted.
Prop 8 provides another example. Before the SCOTUS held that there was no appellate standing in the case, the Ninth Circuit ruled that California could not take away same-sex marriage based on animus, even if doing so put same-sex couples in no worse a position than they would have occupied if they had never been granted a right to SSM. (The case was decided in 2012 on the assumption that there was no freestanding constitutional right to SSM.) As I explained at the time, the decision made sense in those terms for more or less the sorts of reasons I have set forth here.
And so, I conclude, if it can be shown that DACA is being terminated because of illicit animus (an admittedly substantial "if"), then the fact that erstwhile DACA recipients would be worse off than before DACA termination suffices to make out an equal protection claim--even though they will be no worse off than they would have been if DACA had never been adopted.
***
Postscript 1: In characterizing DACA recipients as no worse off if DACA is rescinded than if it had never been adopted, I am speaking only with respect to the equal protection claim. The complaint also raises a due process objection to the government's use of information that it only received as a result of people coming forward in reliance on assurances that such information would not be used against them.
Postscript 2: In the foregoing I am assuming that the challenge to the Travel Ban is quite strong. Yesterday's order by Justice Kennedy staying the Ninth Circuit mandate with regard to travel assurances for refugees does not undermine that assumption--as it goes to questions about who may obtain relief rather than the validity of the Travel Ban as such.