Top universities already have plenty of brand recognition and get way more applicants than what they can accept. The monetary value of any additional marketing is probably very close to zero for them.
Every once in a while people call into question the value of higher education and ask whether MOOCs mark the end of traditional universities. I find it ironic that when top universities flex their muscles and start asking those people to pay up for classes - the same classes that you otherwise would have had to go through admissions and tuition to get into - they're being called money grabbers.
The judgment on exercises that is helpful to the students in traditional universities is a factor proportional to the size of courses. On the one hand there is automatic testing, that's feasible in basic matters. Take Khan Academy for example. A test that anyone can do with advanced knowledge is a simple application.
The motivation to really learn, that some get from deadlines, is learned and it's less direct than an actual want to understand a topic. In that sense, while a job doesn't need to become a passion, the drive to really dive deep into a topic comes as part of a hobby and so the tax payer would probably not want to afford that. But that is a different topic and I'm biased against taxes.
In any case, costs could probably be reduced so much per student, that it needn't be the driving factor, far from the tuition that is payed now.
You can apply that reasoning to any large organization: once it's known/famous for its excellence, an overwhelming majority of their marketing investments have small return. Still, they continue to invest in marketing because they know that this excellence has to be maintained. That's exactly what they should use MOOCs for - and they initially did