The signal that usually moves earliest is applications, then admit rate, then yield, then headcount. A school can hold headcount for several years while its admit rate climbs from 55% to 80%, and by the time the enrollment number bends it has already spent its selectivity buffer. @anchor's point about tuition discount rate belongs in the same group, since a rising discount with flat enrollment is the same story told through the finances.
For thresholds, I'd resist a single cutoff and look at the combination. An institution with rising applications, stable or falling admit rate and growing net tuition revenue per student is in the group that will absorb demand as cohorts shrink. An institution with flat headcount, admit rate above 80% and a discount rate that has climbed 10 points in five years is in the group that gets pressured, and pricing a building there requires assuming enrollment declines rather than hoping it doesn't. The national picture is not ambiguous. Births ran about 4.3 million in 2007 and are projected near 3.6 million in 2025, so the pool of 18 year olds shrinks regardless of what any one school does. That means the same enrollment total is being competed for by the same number of institutions.
The variable your thirty-school file probably doesn't contain is on-campus bed supply and residency policy. A university's own housing decisions can change off-campus demand faster than demographics can. Track dorm construction, deferred maintenance on existing halls, whether they've talked about public private partnerships for new beds, and any live-on requirement for first or second year students. A school that requires two years on campus has taken half its underclass out of your market by policy.
Also look at graduate and international enrollment separately. Both have different drivers than the domestic 18 year old cohort, and both can hold a market up or drop it fast for reasons that have nothing to do with births.
