Sondra Beverly (University of Kansas), Peter Tufano (Harvard Business School) & Daniel Schneider (Harvard Business School) have posted Splitting Tax Refunds and Building Savings: An Empirical Test on SSRN.  Here is the abstract:

Families are more likely to save if they can commit to savings before funds are in-hand (and subject to spending temptations). For low- and moderate-income U.S. families, an important savings opportunity arises annually, during income tax season. We study a group of low-income individuals in Tulsa, Oklahoma, who, at the time of tax filing, were encouraged to save parts of their federal refunds. Those who agreed directed a portion of their refund to a savings account, and arranged to have the rest sent to them in the form of a check. Eligible individuals could also open low-cost savings accounts. We document the demand for these services, the characteristics of those who sought to participate, the savings goals of those who participated, the immediate savings generated by the program, and the disposition of savings a few months after receipt. This pilot study suggests that there may be demand among low-income families for a refund-splitting program that supports emergency needs as well as asset building, especially if a basic savings product is available to all at the time of tax filing.