We present an empirical test for the addictiveness of lottery gambling that exploits an exogenous shock to local market consumption of lottery gambling. It uses the sale of a winning jackpot ticket in a zip code as an instrument for present consumption and tests for a causal relationship between present and future consumption. This test estimates the time path of persistence nonparametrically. Data from the Texas State Lottery suggests that after 6 months, roughly half of the initial increase in lottery consumption is maintained. After 18 months, roughly 40 percent of the initial shock persists, though estimates become less precise.