Association for Public Policy Analysis & Management

The Earned Income Tax Credit (EITC) is a critical income support for working families. An extensive literature documents the program’s positive effects on work, educational attainment, health, and poverty reduction (Eissa and Hoynes 2004; Dahl and Lochner 2012; Evans and Garthwaite 2014; Hoynes, Miller and Simon 2015; Fox 2017; Hoynes and Patel 2018; Bastian and Michelmore 2018; Manoli and Turner 2018).

However, not all eligible families receive the credit (Nichols and Rothstein 2015). Upwards of half of the lowest-income eligible households may fail to claim it (Blumenthal, Erard and Ho 2005). Maximizing take-up of the EITC is a priority for many key stakeholders, but to date, most outreach efforts pursued by public agencies and community organizations have relied on broad awareness-building strategies and evidence about the effectiveness of these efforts remains limited.

This paper reports on the results of three large-scale field experiments (conducted alongside two state agencies and a non-profit partner over two tax seasons) in which we test the effect of targeted, personalized outreach – informed by insights from behavioral science – to over one million potentially eligible Californians. We varied the content and design of our communications in each experiment in order to separately consider the importance of learning costs, compliance cost, and psychological costs as explanations for incomplete take-up.

The three experiments are briefly summarized below:

• The state’s tax agency sent letters to over 140,000 households (excluding a control group). These letters were varied so that we could compare the impact of (1) including more information about the value of the credit versus personalized information about the subject’s nearest tax assistance site; and (2) more or less formal designs.
• The state’s social services agency sent personalized text messages to over 60,000 households (excluding a control group) enrolled in SNAP and, using data on earnings and household composition, deemed likely eligible for the EITC. These text messages were varied so that we could compare the effect of providing basic information about the credit versus personalized estimates of a household’s EITC amounts.
• A non-profit partner sent text messages to over 1.2 million potentially eligible individuals, letters to 20,000 households, and advertised to over 100,000 households on social media (all excluding control groups). Text messages and letters were varied to test the relative impact of basic information about the EITC, non-monetary behavior motivations, and information about how to access assistance with filing.

This paper draws on the literature on the impact of administrative burdens (Herd and Moynihan 2019). The design also explicitly extends recent research by Notowidigdo and Finkelstein (2018), Bhargava and Manoli (2015), Bettinger et al (2009) on the effect of targeted nudges to increase take-up of public programs.

We will measure our outreach’s impact by analyzing deidentified state tax records shared by the California’s tax agency, the Franchise Tax Board. We expect to have initial results prepared by early Fall 2019.