The Initial Effects of EMV Migration on Chargebacks in the United States

To reduce counterfeit fraud in the card-present (CP) environment, the United States started migrating to EMV chip technology in the mid-2010s. Since October 2015, merchants have been liable for counterfeit fraud committed using EMV cards if the merchants have not adopted EMV chip-readable terminals. Merchants are held liable through chargebacks. This study examines the initial effects of the EMV liability shift on fraud chargeback and merchant loss rates using data from merchant processors and PIN debit networks. Combined with gross fraud rates—overall fraud rates regardless of who incurs fraud losses—estimated in other studies, the results of our study suggest that merchants have faced a significantly higher share of fraud losses since the shift; however, this spike will decline if merchants continue to adopt EMV. Merchant fraud loss rates for signature-based transactions in the CP channel increased sixfold, but the rates significantly vary between magnetic stripe (mag-stripe) and chip-to-chip transactions. The rates for mag-stripe transactions are over 9 basis points in value for all merchants combined, and vary across merchant categories, while the rates for chip-to-chip transactions are very low, around 0.02 basis points, across all merchant categories. Because the gross fraud rates of mag-stripe transactions did not increase in the same period, our results suggest that the higher merchant fraud loss rates for mag-stripe transactions after the liability shift is mainly due to the liability shift. Compared to signature-based transactions, fraud chargeback rates of PIN debit transactions in the CP channel are much lower. Our results suggest that both EMV and PIN are effective in reducing merchant fraud loss rates. However, we need detailed gross fraud rates to examine how effective EMV and PIN are in reducing fraud itself in the CP channel. Our results for card-not present (CNP) fraud chargeback and merchant loss rates are mixed. Both rates increased for some merchant categories, but the rates for all merchants combined actually declined in our data. This decline is likely due to the underrepresentation of signature-based CNP transactions in our data. The gross fraud rates for CNP transactions actually increase d in the same period and merchants are generally liable for CNP fraud