# Data audit

• data/panel.csv SHA-256 968410b993df4d696194ba82b35a030d9c4f5f69e3c2de2c16c7b304d55ce5a8 (matches bundle.json)
• 2000 rows x 3 cols: firm_id int64, x float64, y float64 (dimensionless; synthetic). No dates.
• Missing: 0 in all columns. Non-finite: 0. Exact duplicate rows: 0.
• firm_id 0..49: 50 firms, exactly 40 obs each (balanced).
• x: mean 0.161, sd 1.325, range [-4.34, 4.99]; y: mean 1.573, sd 1.782, range [-3.50, 7.20]. No apparent anomalies.
• Between-firm share of variance: x ≈ 0.45, OLS residual ≈ 0.53 -> clustering will matter materially.
• No availability-rule violations (single permitted input).