Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/open-contracting/cardinal-rs/llms.txt

Use this file to discover all available pages before exploring further.

The ratio of disqualified bids to submitted bids is a high outlier per buyer, procuring entity, or tenderer.
See also: R035 - All except winning bid disqualified

​
Methodology

This indicator calculates ratios and uses statistical outlier detection to flag suspicious patterns.

​
For Buyers and Procuring Entities

For each buyer, the ratio is calculated as: numberOfBidsDisqualifiedByBuyernumberOfBidsSubmittedToBuyer\frac{\text{numberOfBidsDisqualifiedByBuyer}}{\text{numberOfBidsSubmittedToBuyer}} A buyer is flagged if its ratio is greater than or equal to the upper fence of: Q3+1.5(IQR)Q_3 + 1.5(IQR) where Q3Q_3 is the third quartile and IQRIQR is the interquartile range for the set of ratios. The same methodology applies to procuring entities.

​
For Tenderers

For each tenderer, the ratio is calculated as: numberOfBidsDisqualifiedForTenderernumberOfBidsSubmittedByTenderer\frac{\text{numberOfBidsDisqualifiedForTenderer}}{\text{numberOfBidsSubmittedByTenderer}} A tenderer is flagged using the same upper fence calculation.
Example for a buyer or procuring entity25% (Q1Q_1) of buyers in Atlantis disqualify at most 2% of bids received. 75% (Q3Q_3) of buyers disqualify at most 22% of bids received. This yields an upper fence of 52%.The State Water Corporation disqualified 59% of bids received. This is more than the upper fence of 52%. Therefore, the State Water Corporation is flagged.
Example for a tenderer25% (Q1Q_1) of bidders in Atlantis have no disqualified bids. 75% (Q3Q_3) of bidders have at most 24% disqualified bids. This yields an upper fence of 60%.73% of the bids submitted by MisconductCorp were disqualified. This is higher than the upper fence of 60%. Therefore, MisconductCorp is flagged.
Why is this a red flag?A corrupt buyer or procuring entity can award pre-determined bidders by disqualifying other bidders’ bids.A colluding bidder can mimic competition by submitting deliberately unqualified bids.
Based on “Excluding qualified bidders” in Guide to Combating Corruption & Fraud in Infrastructure Development Projects and Behind the Scenes of ProZorro: Does Ukrainian business trust public procurement?.

​
Output

The indicator’s value is the ratio, as a decimal (between 0.0 and 1.0). If the --map command-line flag is set, the Maps key contains:
​
ocid_buyer_r038
map
The buyer for each ocid in which at least one bid is disqualified.
​
ocid_procuringentity_r038
map
The procuring entity for each ocid in which at least one bid is disqualified.

​
Configuration

All configuration is optional.
​
R038.threshold
float
Override the calculated upper fence with a fixed threshold. If not set, the upper fence is calculated using Q3+1.5(IQR)Q_3 + 1.5(IQR).
​
R038.minimum_submitted_bids
integer
default:2
Minimum number of submitted bids for a tenderer to be included in the analysis. Tenderers with fewer bids are excluded.
​
R038.minimum_contracting_processes
integer
default:2
Minimum number of contracting processes for a buyer or procuring entity to be included in the analysis.
[R038]
threshold = 0.5  # optional: override calculated upper fence
minimum_submitted_bids = 2  # default
minimum_contracting_processes = 2  # default
Put all properties under one [R038] section in your settings file.

​
Implementation Details

The indicator (from src/indicators/r038.rs):
  • Tracks disqualified and submitted bid counts for buyers, procuring entities, and tenderers
  • Calculates ratios as fractions (numerator/denominator)
  • Uses statistical analysis (quartiles and IQR) to determine outliers
  • Only flags entities meeting minimum threshold requirements
  • Skips flagging if 75% of cases have no disqualified bids (upper fence = 0)
  • Stores metadata about Q1, Q3, and upper fence calculations

​
Assumptions

This indicator assumes that /buyer/id, /tender/procuringEntity/id and /bids/details[]/tenderers[]/id are stable across contracting processes.
Is this assumption not true for your data? Please contribute to GitHub issue #32.

​
Demonstration

Input 
// Example OCDS releases showing excessive disqualification pattern
Output 
ocdscardinal indicators --settings docs/examples/settings.ini --no-meta docs/examples/R/038.jsonl

{"Buyer":{"B":{"R038":0.5}}}

Build docs developers (and LLMs) love

Get started for freeTalk to us