Tax incentive and household saving strategy: A regression discontinuity approach to catch-up contributions

ABSTRACT

This study examined how the catch-up provision—a tax-deferral incentive—impacts total retirement saving and household saving strategies. Using the Survey of Consumer Finance from 2004 to 2016, this study found that the catch-up provision does not have a positive impact on total retirement saving in the majority of cases; however, this provision was found to positively impact contributions to defined contribution plans among middle-class households whose income ranges were between $50,00 and $100,000. The study also found that household saving and consumption habits, which include self-control factors, as well as employer-match programs, play an important role in increasing savings for retirement. The results of this study have implications for future work on a broad spectrum of social policies and programs that promote financial security in retirement.

KEYWORDS:

• Retirement savings
• Tax-deferred incentive
• Financial security

Highlights

• No significant impact of the catch-up provision is found in a majority of cases.

• This provision raises contributions with income levels between $50,000–$100,000.

• Demographic, saving behavior, and financial resources impact retirement savings.

• A key implication is that tax-incentive is not enough to boost retirement saving.

• Additional programs to promote saving and employer’s match contributions are needed.

Notes

¹ Sharp RD: This term is used when the probability of an observation receiving the treatment is either zero (0) or one (1) which differ from fuzzy RD that probability of receiving treatment is between 0 and 1 (Imbens & Lemieux, 2008, pp. 618–619). In this study, observations who receive the treatment (catch-up provision) is coded as 1 while the control group or the group that does not receive the treatment (not be eligible for catch-up provision) is coded as zero.

² Running variable or assignment variable is the unit scoring on one side of the cutoff is assigned to one condition and those on the other side to another (Cook et al., 2002, p. 208). In this paper, age is the running variable, a group whose age is equal 50 and above is assigned in the treatment group (eligible to make catch-up contribution); a group whose age is lower than 50 (between 40 and 49) is assigned in the control group (not eligible to make catch-up contribution).

³ Besides the results of linear models that are presented in this table, we run additional quadratic model with and without constraints for robustness check. The age-squared in the quadratic models does not appear as significant and the coefficients of other variables similar to the results presented in the linear models above. The consistency of estimated results of total savings across different types of models demonstrates the robustness of the regression discontinuity design.