199. How trade economists busted corruption at the port

Episode Summary

Executive Summary: The episode examines customs corruption in Madagascar, where tariffs fund a large share of government revenue. Using unusually detailed customs data, World Bank economists identified suspicious inspector-broker pairings enabled by an IT administrator who manipulated supposedly random assignments. Their findings helped trigger investigations and reforms, but the scheme evolved, showing both the power and limits of data-driven anti-corruption efforts.

Main Topics: Why tariffs matter in Madagascar (Priority: 5/5): Madagascar depends heavily on import tariffs because it imports most goods and has limited capacity to collect other taxes, making border revenue critical for public services. How customs clearance works (Priority: 4/5): The episode explains the customs process: brokers file declarations, a risk model flags shipments, inspectors review goods, and taxes are assessed and collected. The corruption mechanism (Priority: 5/5): Corruption occurred through manipulated inspector assignment: an IT administrator overrode randomization so colluding brokers and inspectors could work together and understate taxes. Econometric detection using administrative data (Priority: 5/5): Researchers compared expected random inspector-broker pairings with actual pairings, identifying excess interactions and then showing these pairs had higher fraud risk and preferential treatment. Policy response and adaptive evasion (Priority: 4/5): Madagascar customs acted on the findings by investigating staff and outsourcing assignment randomness, but corrupt actors adapted by withholding some declarations from the third party. Lessons for customs reform (Priority: 5/5): The episode emphasizes that technology alone cannot eliminate corruption; oversight, incentives, and credible sanctions are necessary, especially in revenue-dependent developing countries.

Key Arguments: Tariff revenue is essential in low-income countries like Madagascar because it finances social programs and constitutes a large share of total tax revenue. Customs fraud is easier when inspectors, brokers, and IT staff can collude in small, high-stakes environments with weak enforcement. Random assignment rules in customs can be exploited if IT administrators can override the system. Administrative data can reveal corruption patterns even when direct evidence of bribes is unavailable. Corrupt pairings were associated with higher fraud risk, faster clearance, and lower tariff collection. Anti-corruption reforms must anticipate adaptation by corrupt actors; fixing one loophole may simply shift the scheme elsewhere. Trade facilitation and revenue collection can conflict: speeding clearance may reduce tax enforcement if safeguards are weak.

Data Points: Tariff revenue share of total tax revenue in Madagascar: 50% - Tariffs at the border account for half of Madagascar’s government 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 Tamatina/Tomasina port: about 16 per year - The main port had very few inspectors relative to its importance. Port share of national tax revenue: more than one-third - Tomasina alone generated over a third of Madagascar’s total tax revenue. Broker-inspector universe: about 46 brokers and 16 inspectors - Illustrates the small, interconnected environment at the port. Inspectors surveyed who said unethical behavior is punished: 6% - Shows low perceived risk of sanctions. Inspector annual salary: about $10,000 - High by Madagascar standards, yet still insufficient to deter bribery for some officials. Potential inspector income gains from corruption: double or triple annual salary - Indicates the lucrative incentives from collusion and bribe splitting. Suspicious broker-inspector pairings detected: about 10% of all import declarations - Excess interaction occurred across the study period. Inspectors involved in suspicious pairings: 10 of 16 - Majority of inspectors were implicated in the suspicious assignment patterns. Brokers involved in suspicious pairings: 14 of about 45 - A substantial share of brokers were linked to the suspicious pairings. Average tariff revenue loss from suspicious declarations: 26% higher absent corruption - Declared taxes would have been materially higher without the scheme. Overall tax collection impact in Tomasina: 3% higher absent corruption - Estimated aggregate revenue gain if the initial corruption scheme had not existed. Third-party randomization coverage after reform: 93% - Most declarations were routed through an independent company after the first scheme was exposed. Declarations withheld from third party: 7% - These remained in customs and were used in a renewed manipulation scheme.

Pivotal Quotes: "Tariff revenue is super important." — Ana Fernandez: Explaining why border taxation is central to Madagascar’s public finances. "This IT system is going to indicate that particular shipment should go through a physical inspection. It's called the red channel." — Ana Fernandez: Describing how the customs risk-assessment and inspection process works. "IT solutions are not a panacea in the fight against corruption." — Ana Fernandez: Summarizing the main policy lesson after the corruption schemes evolved despite reforms.

Implications: The episode shows that customs corruption can be uncovered with data, but reforms must be monitored continuously because corrupt networks adapt quickly. For developing countries, protecting tariff revenue is vital to state capacity and service delivery.