Hello everyone,
For my dissertation I want to measure the impact of CEO and CFO compensation on payout policy. One of the measures I use for payout policy is cash dividends scaled by total assets. This value is either positive when the firm pays out dividends or 0 when there is no dividend payout. In payout literature it is common to use a Tobit regression in this case. I don't understand why since the zero values seem to be natural zeros and not stem from censoring. Wouldn't it be more logical to use OLS in this case?