Vendor Concentration.

What “flagged” means here

Glassbox surfaces correlations and observations, not findings of misconduct. Severity is a calibrated read on how strongly the dataset matches the pattern definition — a starting point for review, not a verdict.

• ● Observation

  The match meets the pattern threshold but the signal is the lower band. Normal review queue.

• ● Attention

  The match exceeds the routine threshold. Worth a closer look — pull the source rows, walk the citation.

• ● Flag

  The match is in the highest band of the calibrated scale. Recommend: prioritise for active review.

Severity is computed from the detection signal · per (department, program) over the rolling 36 months to the corpus as-of date · program total ≥ $10M · ≥ 3 recipients · HHI = Σ recipient share² × 100² ≥ p90 of program HHIs within the department AND ≥ 1500 floor · bands: ≥5000 extreme, ≥2500 highly, ≥1500 moderately concentrated · signal = HHI

Recommended next steps

Once a match is identified, the dataset suggests these review actions. Glassbox does not prescribe outcomes — these are calibrated options for funders, auditors, and program officers, framed as correlations to investigate.

1. 01
   Decompose the HHI into volume vs. unit-price drivers.

   Concentration can reflect a few large contracts (volume) or many small ones (incumbency). Each implies different responses — the former may be one-off necessity, the latter suggests competitive design issues.

2. 02
   Compare the department's HHI against peer departments.

   Some functions are inherently concentrated (specialty equipment, niche expertise). Compare with peer departments delivering similar mandates to identify outliers.

3. 03
   Evaluate whether competitive procurement design needs review.

   If concentration persists across multiple FYs, the procurement strategy itself may merit redesign — open tenders, longer notice periods, or explicit competition mandates.

4. 04
   Maintain a watchlist of departments above the 2500 HHI threshold.

   Annual re-scoring lets the funder track whether concentration is stable, increasing, or decreasing — and whether interventions are working.

These are calibrated suggestions, not directives. The dataset shows the pattern; the funder decides the response.