Socioeconomic correlates of the lottery rollover effect in Toronto, Canada

Abstract

Background: In a lottery with a progressive jackpot, the rollover effect refers to an increase in revenue or engagement with an accumulating jackpot size. Using a geospatial dataset of lottery ticket sales aggregated by postcode, we test two corollaries of the rollover effect. First, how does the rollover effect change in neighborhoods with higher or lower socioeconomic status? Second, how do fluctuations in progressive-lottery sales affect the consumption of fixed-prize lottery tickets in the same neighborhoods, in line with economic notions of ‘substitution’ versus ‘complementarity’? Methods: We used time-series data on ticket sales from 2012-2015 from 3 progressive-prize lotteries (Lotto 649, Lotto Max, and Lottario) in Toronto, Canada, aggregated for 95 forward-sortation area (FSA) postcodes. Our regression models controlled for other cyclical fluctuations including day of week, month of year, and common paydays. Results: Jackpot size positively predicted lottery ticket sales in all models, with a large effect size. There was a significant interaction between jackpot size and neighborhood SES, such that lottery sales in higher SES neighborhoods were more sensitive to jackpot size, although the effect sizes were small. Sales of fixed-prize lotteries were positively related to sales of progressive-prize lotteries, supporting complementarity. We observed a significant interaction between SES and progressive-prize sales, in which fixed-prize sales were more affected by progressive-prize sales in higher SES neighborhoods. Conclusion: Both the effect of larger jackpots on ticket sales, and the effects of progressive-prize sales on a second lottery type, are attenuated within more disadvantaged (i.e. lower SES) neighborhoods.