A Letter to Google

I sent the following to 'scholar-support@google.com'. If you agree, you might send them an email yourself.

---

Since Google has changed the world in so many good ways, I thought you might be open to suggestions for new projects.

The traditional method of academic publishing is way too expensive and way too slow. Worst of all it's exclusive, presenting high cost barriers to those who want to access the information (I should know--one of my jobs is to administer a college library). The fact that much research is produced by public funds but then sequestered by for-profit publishing houses, is not just inefficient--it's an insult to the general population.

My suggestion: follow the lead of pre-print sites (see http://xxx.lanl.gov/ for example), many of which exist for specific disciplines. Create a software framework that can encompass that idea the same way that a 'blog' is now pretty much universally recognized. I imagine that components would include cool ways to post feedback, cross-reference, and provide permanent links to research papers. If one has to log into the system, you could even create cool ratings of papers if you wanted, by measuring not just the number of hits on a page, but how important those eyeballs are by examining how that person's profile corresponds to everyone else's (the kind of thing Google excels at).

A standard, reliable implementation with a central index has the potential to revolutionize research by making it truly a democratic marketplace of ideas.

Marginal Discount Rates

For a private institution, setting tuition is a difficult decision. Sooner or later the conversation gets around to the number on everyone's mind: how much of the new money do we get to keep?

Suppose tuition is $10,000 and our discount rate is 40%. That means we nominally charge each student 10K, but in actuality 4K of that is institutional aid, or funny money, which is excluded from reported revenue in the annual audit. Essentially, it's a way to customize the price per student according to: 1. how much the school wants them to enroll, and 2. how much money they have. If policies are altruistic, the institution may award institutional aid to students with low incomes. A future post on that subject...

Suppose after meeting for hours we decide to raise tuition $1,000 to 11K. How much do we get to keep? We might assume that since our discount rate is 40%, we can keep 60% of the tuition hike, or $600 per student. Wrong!

Huh?
The original discount rate is just the ratio Rold = A/T, where A is institutional aid awarded, and T is tuition. So if Suzy gets $2,000 institutional aid toward the 10K tuition, that's 20% off. Simple. But there's no reason to assume that when tuition increases by 10% that average awards will also increase by 10%, which works out to

Rnew = [A*(1.10)]/[T*(1.10)] = A/T = Rold, our implicit assumption.

This troublesome enthymeme can easily creep in to an unsophisticated analysis. Let's see what it implies.

Follow the money
Typically these calculations are done for new students. It may be assumed that returning students generally do not have their aid packages adjusted for the new tuition rate. Our assumption that we should discount new tuition revenue by the old 40% rate leaves us with the conclusion that these new students will pay $600 more for tuition in real dollars. We can enforce this with financial aid policies, but we may not want to if it means sacrificing recruiting targets. If your institution has to turn away lots of qualified applicants this probably won't be a problem. But if you are trying to balance an aid budget against enrollment targets, you may very well end up with less net revenue.

The cost of a Coke
If the soda company wants to raise revenue, it's not as simple as just raising the cost of a can of the stuff from .55 to .60. Since revenue is (price)*(number sold), we have to consider the fact that we may lose more revenue through lost customers than we gain through increased prices. Trying to figure out the right balance is the holy grail of setting tuition, and the topic of a future post. The bottom line is, if we have no reason to believe that students can or will pay more tuition, we shouldn't assume that they will (and hence keep the discount rate the same).

We may try to strike a happy medium between higher prices and fewer enrollees by allowing some flexibility in the awarding policies. Set the goals higher, but be willing to back off if admitted students can't or won't pay the sticker price. Rather like buying a car. In this way we can let the market decide how much an education at The College is worth. After the dust settles from Fall enrollment, we can then calculate the real discount rate.

Calculating Marginal Discount
Suppose, after raising tuition 10% to 11K, our new enrollees averaged 5K in institutional aid. Even with tuition leveraging methods, you can't completely predict what kind of applicants you're getting--rich, poor, talented, not--so complete control of your destiny is unattainable. To compute the marginal discount, calculate

R' = (additional aid)/(additional tuition).

Simple, right? In our example, the additional aid given to incoming students averaged 1K, while additional tuition was set at 1K also. The marginal discount rate is 100%! This means than any additional tuition revenue we were counting on from the new class did not materialize. If you look at historical marginal discount rates, you may be able to get a sense of your price point in the market. Bear in mind that any mathematical operation where you subtract and divide is very sensitive to inputs, so bad data can really screw up the numbers. Even random fluctuations from year to year can mess it up. Look for a broad pattern. We can do more sophisticated things, but that's another post.

Summing up
If your R' is less than your R, then tuition is priced lower than the market "thinks" it should be--your institution is a bargain! If R' creeps up to around 100% (or it can be even greater), you can interpret that as the market pushing back against your increases--you're becoming overpriced. In any event, these ideas can inform the budget discussions when a number for additional tuition revenue is bandied about. A good hedge is to assume that the marginal rate will be 10-20% higher than the current R.

Contradictions in the Academy, Part I

One of the perennial contradictions I run into when helping to plan an annual budget is the conflict between financial aid and admissions. FA is typically given a fixed budget of institutional aid (for a private institution) whereas admissions is give an enrollment target. So FA wants above all not to overspend its budget, while admissions is trying to pack them in whatever the cost.

Good leadership (which we have at The College) goes a long way to alleviate that issue, but a better approach to experiment with is to include the revenue generated by an admit as part of the FA leveraging strategy. Without this big picture outlook, things run fine until near the end of an annual Fall admit process, when money is running low. FA will be pulling back on awards because odds are it has already overspent its budget by 20% or more anticipating that a significant percentage of new students won't show up or will drop after a semester, bringing the budget back into line.

So you can have a situation where good students apply and don't get the scholarship they would have gotten if they had applied earlier. This makes no sense from an academic point of view, and assuming that the scholarships are near the average discount rate (about 42% for The College), it also makes no sense from a financial perspective. The poster children for this are the Spring enrollees, who get significantly less aid from about every source. They have terrible retention stats too. It would seem to make sense to let these students revise (upward) their institutional aid packages the next Fall in the new budget at a normal rate. It would make even more sense to have each student considered on his or her merits with a more individualistic approach to talent and revenue.