Higher-Ed A/G Research Extension A public IPEDS study of institutional closure risk in a validated higher-ed subdomain.

ED Financial Responsibility Baseline

Purpose. The U.S. Department of Education publishes an annual financial-responsibility composite score for institutions filing audited financial statements. The composite is the obvious regulator-side comparator for any institutional-distress signal. This track tests how the higher-ed A/G geometry compares.

Source path

The federal data was fetched via the Urban Institute Education Data Portal API (educationdata.urban.org), which mirrors federal U.S. Department of Education / Federal Student Aid releases under public-domain terms. The mirror exposes the financial-responsibility composite at the path /api/v1/college-university/fsa/financial-responsibility/{year}/ with a direct unitid identifier, removing the need for an OPEID-to-UNITID crosswalk.

Coverage in the public mirror used here: 2006–2016. Records returned per fetched year (2014 / 2015 / 2016): 3,396 / 3,709 / 3,627. Data dictionary fields include unitid, year, opeid, inst_name_fsa, and the headline financial_resp_score.

Findings

Head-to-head on the locked test cohort is not runnable. The locked test cohort years are 2017–2018; the public mirror used here stops at 2016. We therefore cannot run an apples-to-apples test on the locked cohort. What we can do is run a descriptive comparison on the train cohort (2014–2016).

On the FASB × Sector 2 train cohort (2014–2016, 4,495 institution-years with composite scores joined; 84 closure-positive rows; 92 % join rate):

PredictorAUROC (train cohort)95 % CI (analytic)
Federal composite, sign-flipped0.747[0.686, 0.808]
Negated net-position cushion0.643[0.579, 0.708]
Negated cash operating margin0.631[0.566, 0.695]
Negated low six-year graduation rate0.788[0.718, 0.859]

The federal composite is a real closure-relevant signal on this descriptive train cohort. Within the same train cohort, the simple low six-year graduation rate baseline has a slightly higher point estimate.

What is and is not claimed

What would close the gap

A future iteration with direct access to federal Financial Responsibility data covering the locked test cohort years (2017–2018 or later) would let us compute joint-weak and the federal composite on the same held-out cohort and report a paired difference, in the same way the cluster-bootstrap track reports the paired difference vs single-variable baselines.

← Back to Evidence