A Few Rotten Apples : How a Few External Fraud Events Drive Most Operational Losses in a Leading Indian Government-Owned Bank

Using novel, proprietary data on operational losses from a large government-owned bank in India, we provide evidence that only 3% of operational risk events (particularly external frauds) account for more than 80% of aggregate operational losses. Operational losses are power-law distributed and exhibit steep increases in their tail operational value-at-risk. We also show that extreme operational losses in a given year are driven mostly by the previous year’s NPA (non-performing asset) level, with a one standard deviation rise in the former associated with about 0.1% rise in extreme operational losses