199. How trade economists busted corruption at the port

Episode Summary

Executive Summary: The episode examines corruption in Madagascar’s customs system, where tariffs fund much of government revenue. Using detailed administrative data, World Bank economists showed that inspector assignments were manipulated to pair brokers with favored inspectors, reducing tax collection and speeding clearance. The analysis helped trigger investigations and reforms, but corrupt actors adapted, showing both the power and limits of IT-based anti-corruption tools.

Main Topics: Madagascar’s dependence on tariff revenue (Priority: 5/5): The episode explains why customs matter so much in Madagascar: tariffs are a major source of tax revenue in a poor country with limited ability to collect other taxes. How customs clearance works (Priority: 5/5): Anna Fernandez outlines the clearance process, including brokers filing declarations, secret risk scoring, and the assignment of inspectors to shipments. Corruption through inspector assignment (Priority: 5/5): The core scheme involved manipulating the supposedly random assignment system so certain brokers and inspectors could repeatedly work together and underpay tariffs. Detecting fraud with administrative data (Priority: 5/5): Researchers compared expected random assignment patterns with actual pairings, then tested whether suspicious pairings showed higher fraud risk and preferential treatment. Reform, retaliation, and adaptation (Priority: 4/5): Madagascar Customs responded with investigations and IT reforms, but corrupt actors later adapted by withholding some declarations from the new randomization system. Broader lessons for customs policy (Priority: 4/5): The discussion highlights that IT can help but is not sufficient; enforcement, incentives, and oversight are necessary, especially where customs must both facilitate trade and collect revenue.

Key Arguments: Tariffs are exceptionally important in low-income countries because they are easier to collect than income or property taxes, making customs a major fiscal pillar. Customs corruption is more likely where a small number of officials control large revenue flows and where oversight and sanctions are weak. Random assignment of inspectors is supposed to reduce collusion, but the IT system itself can be captured and manipulated by insiders. Undervaluation is the most common fraud method because it is harder to detect than misclassifying goods; the episode says 70% of observed fraud involved underreporting value. Administrative data can reveal corruption patterns even without direct evidence of bribes by comparing theoretical random assignment with actual assignment behavior. Suspicious inspector-broker pairings were associated with higher-risk declarations and preferential treatment, indicating that the pairing pattern was not random. The corruption scheme had measurable fiscal costs, raising the importance of better enforcement in countries where customs revenue funds essential public services. Anti-corruption reforms can be partially effective, but corrupt networks may adapt quickly, so reforms need monitoring and repeated updating.

Data Points: Tariff revenue share of total tax revenue in Madagascar: 50% - Tariffs collected at the border account for half of Madagascar’s total tax revenue. Tariff revenue share of total tax revenue in the U.S.: less than 2% - Used as a comparison to show how much more dependent Madagascar is on tariffs. Inspectors at Tomasina port: about 16 per year - Average number of customs inspectors working at Madagascar’s main port. Brokers at Tomasina port: about 46 - Number of brokers interacting with inspectors at the port. Port share of national tax revenue: more than one-third - Tomasina port by itself generates over a third of Madagascar’s total tax revenue. Revenue managed per inspector: 1.3% of total taxes - Each individual inspector effectively controls a meaningful share of national revenue. Inspectors believing non-ethical behavior is punished: 6% - Survey result showing very weak perceived deterrence. Annual salary of a customs inspector: about $10,000 - Inspector pay is high by Madagascar standards, but still small compared with potential gains from corruption. Potential income from cheating: double or triple annual salary - Illustrates the strong financial incentive to collude. Example shipment value: $1 million - Used to illustrate how underreporting can create large illicit gains. Example tariff rate: 10% - On the $1 million example shipment. Potential tariff revenue on example shipment: $100,000 - What the government should collect if value is properly declared. Corruptly reduced payment in example: $50,000 - Illustrates how the broker and inspector can split the difference. Observed fraud due to undervaluation: 70% - Share of detected fraud in Madagascar attributed to underreporting values. Suspicious declarations in study period: about 10% - Portion of import declarations showing excess interaction between certain brokers and inspectors. Inspectors involved in suspicious pairings: 10 of 16 - Number of inspectors linked to suspicious interaction patterns. Brokers involved in suspicious pairings: 14 of about 45 - Number of brokers linked to suspicious interaction patterns. Higher tariff revenue without corruption: 26% higher - Average tariff revenue that suspicious-pair declarations would have generated absent the scheme. Overall tax revenue gain without initial scheme: 3% higher - Estimated increase in taxes collected in Tomasina if the corruption scheme had not existed. Third-party randomization coverage: 93% - After reforms, the outsourced system handled 93% of declarations. Declarations kept in-house after reform: 7% - The share withheld by customs and manipulated through the renewed scheme. Time corrupt scheme lasted after outsourcing: about 4 months - How long the first reform lasted before a new manipulation method emerged.

Pivotal Quotes: "Tariff revenue is super important." — Anna Fernandez: Explaining why customs corruption matters so much in Madagascar and similar developing countries. "The corruption that we uncovered was taking place through the use of the IT system." — Anna Fernandez: Describing the mechanism used to manipulate inspector assignments and enable collusion. "IT solutions are not a panacea in the fight against corruption." — Anna Fernandez: Summarizing the episode’s broader policy lesson about the limits of technology without oversight and enforcement.

Implications: The episode shows that customs corruption can be detected through data, but stopping it requires more than automation. Countries need oversight, credible penalties, and incentive reforms because corrupt insiders can quickly adapt to weakly governed IT systems.