David Kane on After SFFA v. Harvard

Yeah, so you emphasize in the paper three sets of predictions.

One, most narrowly made by the economist David Card, who tried to estimate or simulate the effect of taking out race as a criterion for admission, what effect that would have on the black share of Harvard's entering class.

And then more broadly, a larger group of selective colleges.

So let's focus on David Card's prediction first.

What did he say would happen if Harvard was not allowed to use race?

So David Card's headline quote was that if race were not used in admissions, then the entering class of 2015 would have gone from 14% Black to 6% Black.

So a decrease of more than 50%.

Now, there are a lot of technical details associated with the specific numbers of that prediction, but the broader claim of a greater than 50% drop in black enrollment is the central claim that CARD and other colleges and universities made.

So for the... economists in the audience.

I looked at the CARD study and he does a multivariate regression, gets the coefficient on black, and then says, I can simulate the effect of removing racial, of figuring that into, of Harvard being allowed to use that as a plus factor.

by dropping that from regression and then seeing what percent of the class is going to be black.

And right, he says that it would be more than 50% drop in the number of black students holding class size constant and holding other things constant.

So he later says, well, what if you increased the share of the amount of weight put on lower socioeconomic status, which is correlated with black? as a regressor.

And yes, you could offset some of it, but it would have other effects.

Yes, and I always don't think that the core of the issue is all the technical decisions that Card made, because obviously Card, a Nobel Prize winner, is pretty good at these sorts of regressions, although one can quibble about it.

The important issue is the greater than 50% drop prediction.

The goal of the paper is to collect Card's prediction and the predictions of other people as a whole, and then compare those predictions to what actually happened.

Okay,

so

what actually happened at Harvard?

So what actually happened at Harvard is tough to see because either on purpose or by accident, Harvard and Card presented things in a way that makes it very hard to see what happens.

Because on the one hand, we have this headline number, 14% to 6%.

And then you look at the percentage of Black students in Harvard's admitted classes in 2024 and 2025 after the SFFA decision went down, and it's around 6%.

And so to that, you sort of say like, hey, here we go.

Harvard was told not to use race.

Yeah, so you emphasize in the paper three sets of predictions.

One, most narrowly made by the economist David Card, who tried to estimate or simulate the effect of taking out race as a criterion for admission, what effect that would have on the black share of Harvard's entering class.

And then more broadly, a larger group of selective colleges.

So let's focus on David Card's prediction first.

What did he say would happen if Harvard was not allowed to use race?

So David Card's headline quote was that if race were not used in admissions, then the entering class of 2015 would have gone from 14% Black to 6% Black.

So a decrease of more than 50%.

Now, there are a lot of technical details associated with the specific numbers of that prediction, but the broader claim of a greater than 50% drop in black enrollment is the central claim that CARD and other colleges and universities made.

So for the... economists in the audience.

I looked at the CARD study and he does a multivariate regression, gets the coefficient on black, and then says, I can simulate the effect of removing racial, of figuring that into, of Harvard being allowed to use that as a plus factor.

by dropping that from regression and then seeing what percent of the class is going to be black.

And right, he says that it would be more than 50% drop in the number of black students holding class size constant and holding other things constant.

So he later says, well, what if you increased the share of the amount of weight put on lower socioeconomic status, which is correlated with black? as a regressor.

And yes, you could offset some of it, but it would have other effects.

Yes, and I always don't think that the core of the issue is all the technical decisions that Card made, because obviously Card, a Nobel Prize winner, is pretty good at these sorts of regressions, although one can quibble about it.

The important issue is the greater than 50% drop prediction.

The goal of the paper is to collect Card's prediction and the predictions of other people as a whole, and then compare those predictions to what actually happened.

Okay,

so

what actually happened at Harvard?

So what actually happened at Harvard is tough to see because either on purpose or by accident, Harvard and Card presented things in a way that makes it very hard to see what happens.

Because on the one hand, we have this headline number, 14% to 6%.

And then you look at the percentage of Black students in Harvard's admitted classes in 2024 and 2025 after the SFFA decision went down, and it's around 6%.

And so to that, you sort of say like, hey, here we go.

Harvard was told not to use race.