I agree, but socioeconomic class is already highly correlated with race (and sometimes gender), leading to race being used as a possible signal for class.

I think their point was that passive racial biases lead to class differences that can't be summed up in a hypothetical comparison.