Finance research suggests that firms cater to investors’ time-varying preference for low-priced stocks by managing nominal share prices. We show that the existing empirical tests of such catering use highly persistent data that may lead to finding significant relations between variables that are actually independent (spurious regression bias). We revisit the catering hypothesis applying five estimation techniques that reduce the effects of data persistence. Adjusted for persistence, the data offer little evidence that stock splits respond to catering incentives. There is some, although mixed, evidence that catering incentives affect the firms’ choice of the post-split prices and the IPO prices.