# Data audit (P1)

| file | sha256 | rows (excl. header) | cols |
|---|---|---|---|
| data/endowments.csv | fc130a2f54c750a5834ec7314060d91f7984769c7c006ad04b98582a016ba74b | 2 | 3 (agent, good1, good2) |
| data/parameters.csv | 717807e75337fbf4aff8e4d74e0fc0b83b799f78fa689172be0d501cb3e43983 | 4 | 3 (parameter, value, description) |

- endowments: A=(8.0, 2.0), B=(4.0, 8.0). Units: units of each good. No missing values, no duplicate agents, all strictly positive.
- parameters: alpha_A=0.4, alpha_B=0.6 (both in (0,1)); aggregate_endowment_good1=14.0, aggregate_endowment_good2=10.0.
- No dates (theoretical exercise).

## Anomaly
- **aggregate_endowment_good1 = 14.0 in the summary sheet, but the sum of individual endowments of good 1 is 8+4 = 12.0.**
  aggregate_endowment_good2 = 10.0 equals 2+8 = 10.0 (consistent).
  In a pure exchange economy the aggregate endowment is by definition the sum of individual endowments; TASK.md calls
  endowments.csv the primary data. Clearing good 1 against 14 would require 2 units nobody owns and, by Walras' law,
  would leave the good-2 market not clearing.
  **Decision:** use the individual-endowment sums (12, 10); report the discrepancy in data_issues; report, as a
  diagnostic only, the price that would result from clearing good 1 against 14 and the induced good-2 imbalance.
- No rows excluded (exclusions.csv records a no-op step).