Panel Paper: How Does Savings Affect Participation in the Gig Economy? Evidence from a Field Experiment

Thursday, November 7, 2019
Plaza Building: Concourse Level, Plaza Ballroom F (Sheraton Denver Downtown)

*Names in bold indicate Presenter

Sam Bufe1, Stephen Roll1, Olga Kondratjeva1, Bradley Hardy2 and Michal Grinstein-Weiss1, (1)Washington University in St. Louis, (2)American University


Background: Nearly one tenth of Americans earn money through the “gig economy”—the segment of the labor market in which workers find very short-term contractual employment through websites and online applications (Smith, 2016). As the labor market continues to change, there is an increasing need to understand the financial circumstances of this growing segment of the workforce and the factors that create barriers or opportunities for entry into this sector. To that end, this paper uses a savings experiment to identify the effects of tax refund savings on labor supplied to the gig economy for low- and moderate-income (LMI) households. Theory does not clearly point to an expected direction of this effect. On one hand, increased access to liquidity through refund savings may assure some that their necessities will be covered, reducing the incentive to work in the gig economy. Alternatively, access to extra liquidity may give some liquidity-constrained individuals the ability to cover some capital costs of gig work, allowing them to increase participation in this labor market.

Data and Methods: This paper uses data from the two waves of the 2017 Household Financial Survey (HFS). The first wave of the HFS, which was offered to a random subset of TurboTax Freedom Edition LMI tax filers, was administered immediately after tax-filing. In addition to questions about respondents’ financial circumstances and characteristics, the survey included questions about respondents’ participation in the gig economy. Six months after tax-filing, a follow-up survey was administered. Survey responses were merged with individual-level administrative tax data.

When filing their taxes, HFS respondents participated in a randomized controlled trial testing the effects of several low-touch behavioral interventions designed to encourage filers to deposit their refunds into savings vehicles. Savings interventions were embedded directly in the tax preparation software. We conduct both intent-to-treat and treatment-on-treated analyses to examine the effects of this experiment (and, more generally, refund savings) on participation in the gig economy. Due to the heterogeneity in the LMI population, we conduct subsample analyses in addition to analyses on the full sample.

Results: Although we find that refund savings had no effect on gig economy participation for our LMI sample as whole, we find strong heterogeneous effects of refund savings in different subsamples. For students, the interventions reduced the likelihood of working in the gig economy in the six months after tax-filing. For liquidity-constrained non-students, however, the interventions increased the likelihood of working in the gig economy. We also find evidence that these effects are occurring at the extensive margin.