In the Nobel-winning economist George Akerlof’s article "The Market for Lemons," he notes that used car buyers will rationally offer less for a car than the dealer would expect, believing that the dealer has more information on the vehicle than the buyer, and assuming that negative information is unlikely to be conveyed.
Of course, this also affects the offer for really good used cars. So, the dealer might solve this problem by offering certified used cars with an extended warranty. Or the buyer might leave the used car market and buy a new car (if he or she can afford one).
Now, when a dean (or chair) proposes to a universitywide committee or the president to award a faculty member tenure, presumably the dean knows more about the faculty member than the university committee or the provost. In effect the dean is a used-car salesman. The letter-writers might have been subtly chosen, the weaknesses might not be revealed. What economists (Kahneman and Tversky) call the "endowment effect" may also apply, where we value what we have in our possession more than if it were offered for sale by someone else. In some institutions, the tenure advocate may also fear that the slot could be eliminated or replaced with a non-tenure-track line if the person is not approved.
The university committee and the provost would like to believe that the dean tells them all, that the letter-writers are forthcoming, that all are committed to the whole university. But human nature being what it is, the provost might want to be sure the university is not getting stuck with a lemon. You can sell that used car that turns out to be a lemon, perhaps for much less than you paid for it. But tenure is for 30 years or so. On the other hand, when really strong candidates come up, both the dean and the provost want to know that.
Moreover, there is evidence  that deans can count but do not read. That is, there are studies that suggest that numbers of articles are a better predictor of deans' decisions than the quality of the work. So even an honest dean may not be reliable.
Faculty tell their chairs and deans their future plans. Given what we know about what people accomplish on their sabbaticals, it is reasonable to suppose that faculty overpromise. The book is almost done, when in fact there are a few draft chapters and outlines at best for the rest. The articles will go out in the next month. The grant should be coming down the line in three months.
Now some professors actually deliver. They do just what they said what they would do -- in grant applications, in annual reports, in personal statements at promotion time. But many faculty overpromise, at best, or they actually are deceptive. Their good intentions dominate their capacity to say where they are in projects.
Used-car dealers (and deans, chairs, faculty) can avoid some of these problems by building a reputation for honesty and reliability, and for taking charge of their mistakes. At the same time, provosts, deans, and chairs must be willing to hear mixed messages and not immediately say no to every candidate who isn’t perfect in every way, and be willing to consider extraordinary cases.
Now, deans have limited appointments, usually for 5 to 10 years. But tenure for the “lemon” faculty member is lifetime. Their colleagues are stuck with them, and if they are not forthcoming with negative opinions at tenure time, at least they suffer the consequences of their acts. But deans do not.
Of course, the university could have a devil’s advocate, providing the strongest negative case for a tenure decision, in effect balancing the dean. It would be much as if you had your mechanic go over the car before you buy it. But, still, it may make sense for deans to face clawbacks for their lack of transparency. It’s hard to know what these might be unless the dean is still in office. That dean might find that its recommendations are less effective or they might not be reappointed. Since a dean may now be a provost elsewhere, they may well be bamboozled by the deans under them. Still, reputational damage is a strong disincentive, much as rescinding of the doctorate might be a strong disincentive to plagiarism.
As for faculty who overpromise, salary raises can be delayed until the promised material arrives. If the faculty member is a tenured associate professor, promotion can be delayed as well. Those who are promising a big book year after year have to be given deadlines.
Now, people could go buy a new car, avoiding the used car market. Universities could get a new assistant professor whenever there was any uncertainty about the person coming up for tenure (saying no to all but the most obviously in good working order -- making the tenure track rather more tenuous).
Or they could make tenure decisions further down the line, say at year 10, the car having proven itself as reliable after 50,000 miles, so to speak. Or they could only hire more senior scholars, for whom information is much more widely shared. The new assistant professor and the two-book scholar are in effect the new cars of the academic marketplace: in one case you don't know if you have a Yugo or a Prius, and in the other you don't know if the well-tested vehicle will continue to perform at a high level and when it will begin to break down (and this does depend on the model).
You might have a repair shop that takes in lemons and makes them good. You bought a lemon, but rather than getting rid of it, you bring it to the repair shop and they do the best they can without breaking your bank. Provosts should have active programs of faculty improvement, where the lemons are in fact repaired and set on a more fruitful path -- although the mechanics here have to be very sophisticated in their people skills and their technical skills.
Martin H. Krieger is professor of planning at the Sol Price School of Public Policy at the University of Southern California. His new book, The Scholar’s Survival Manual: A Road Map for Students, Faculty, and Administrators,  has just been published by Indiana University Press. His blog is here.