Gera Education ROI Index (GEROI) — Methodology

Data source

Step-by-step formula

1. 1

   Download the College Scorecard CSV

   Fetch Most-Recent-Cohorts-Institution_06102026.zip from https://collegescorecard.ed.gov/data/. No API key required. Extract the CSV. The file contains 3,844 rows (one per institution).

2. 2

   Extract and validate required columns

   Parse UGDS, MD_EARN_WNE_P10, GRAD_DEBT_MDN, and NPT4_PUB/NPT4_PRIV. Discard any row where any of these four values is empty, "NULL", "PrivacySuppressed", "NA", or ≤ 0. This left 3,844 valid rows in the June 2026 dataset.

3. 3

   Sort by enrollment and cap to top 300

   Sort valid rows by UGDS (total undergrad enrollment) descending. Take the top 300 as the published cohort. The remaining 3,544 institutions are excluded from individual leaf pages (logged in the data module header).

4. 4

   Compute raw GEROI

   raw_geroi = (MD_EARN_WNE_P10 − GRAD_DEBT_MDN) / NPT4. This ratio is positive when 10-year median earnings exceed cumulative debt. It is negative when debt exceeds earnings — and will still produce a valid GEROI (the normalisation step handles negative raw values).

5. 5

   Normalise to 0–100 (min-max across the top-300 cohort)

   Find min_raw and max_raw across all 300 institutions. GEROI = (raw_geroi − min_raw) / (max_raw − min_raw) × 100, rounded to 1 d.p. The institution with the highest raw ratio (College of the Sequoias, CA, raw=72.1) scores 100; the lowest (University of Arizona Global Campus, AZ, raw=0.13) scores 0.

6. 6

   Generate slugs and publish

   Institution name → lowercase → non-alphanumeric → hyphen → trim to 80 chars. Collision (same slug for two institutions) → append UNITID. Pages live at /us-college-roi/[slug].

Pseudocode

# Input: Most-Recent-Cohorts-Institution_06102026.csv
# Output: GEROI_INSTITUTIONS array (top 300 by enrollment)

rows = []
for each row in csv:
  ugds      = parse_float(row['UGDS'])
  earn10    = parse_float(row['MD_EARN_WNE_P10'])
  debt      = parse_float(row['GRAD_DEBT_MDN'])
  net_price = parse_float(row['NPT4_PUB']) ?? parse_float(row['NPT4_PRIV'])

  if any of [ugds, earn10, debt, net_price] is null or <= 0: skip
  rows.append({ name, city, state, ugds, earn10, debt, net_price })

rows.sort(by=ugds, descending=True)
top300 = rows[:300]

for each r in top300:
  r.raw_geroi = (r.earn10 - r.debt) / r.net_price

min_raw = min(r.raw_geroi for r in top300)
max_raw = max(r.raw_geroi for r in top300)

for each r in top300:
  r.geroi = round((r.raw_geroi - min_raw) / (max_raw - min_raw) * 100, 1)
  r.slug  = slugify(r.name)[:80]

publish top300 as GEROI_INSTITUTIONS

Limitations & caveats

• 10-year window is post-entry, not post-graduation. MD_EARN_WNE_P10 counts 10 years from first enrollment, not degree completion. Students who took longer to graduate will have fewer post-graduation work years in the window.
• Debt is at graduation, not 10 years later. GRAD_DEBT_MDN reflects cumulative debt at the time of leaving, not after 10 years of repayment. GEROI is a return-on-investment context metric, not a net-worth projection.
• Top-300 by enrollment is a size filter, not a quality filter.Smaller specialised institutions that may have high GEROI are excluded because they fall outside the top 300 by UGDS. The enrollment cap ensures data volume is sufficient for the College Scorecard's privacy thresholds.
• PrivacySuppressed values are excluded. Institutions where any of the four required columns is privacy-suppressed (too few students for DoE to publish without risking re-identification) are excluded. This typically affects very small or specialised schools.
• Not financial advice. GEROI is a statistical summary of publicly reported outcomes. It is not a guarantee of individual earnings, a financial recommendation, or an endorsement of any institution. Field of study, location, job market, and individual circumstances all affect actual outcomes.