The Effects of Tax-Time Interventions on Savings Rate and Amount: A Systematic Review and Meta-Analysis

Abstract

Purpose: This systematic review examined the effects of tax-time saving interventions that promote saving with tax refunds from relevant experimental or quasi-experimental studies of interventions aimed toward low- and moderate-income adults delivered when filing U.S. income taxes. Method: A systematic review process was used to search for published and unpublished studies from sources through September 2021. Two reviewers screened studies, extracted data and assessed risk of bias, and effects on savings rate and amount were synthesized using robust variance estimation. Results: This review included 14 unique studies. Five studies reporting 13 effect sizes for savings amount found a small, statistically significant effect (d = 0.06, 95% CI [0.05, 0.08]). Nine studies reporting 35 effect sizes found no statistically significant effect for savings rate (LOR = 4.11, 95% CI [0.42, 40.44]). Discussion: Results suggest some evidence that tax-time savings can be a relatively simple method for increasing the amount low- to moderate-income adults save.

Keywords

tax refund VITA saving emergency saving

Encouraging and facilitating saving for a variety of purposes is a focus of many interventions aimed at building financial well-being (Birkenmaier, Maynard et al., 2022). Saving has been shown to be critical to both short-term financial wellbeing to cover income shortfalls, such as during periods of unemployment or smaller paychecks, and long-term financial well-being to build wealth and fund retirement (Despard, Friedline et al., 2020). However, many U.S. households fall short of possessing adequate amounts of both short-term and long-term savings. The amount of personal savings has fallen in recent years, to an average of $42,000, with higher income households having higher levels of saving (Board of Governors of the Federal Reserve System, 2020). While the source of the savings, such as tax refunds, gifts, and earned income, is unknown, the average U.S. saving rate in 2022 was 4–5% of income (Federal Reserve Bank of St. Louis, 2023). About 40% of U.S. households do not save at all (Board of Governors of the Federal Reserve System, 2020). Most U.S. households have less short-term savings than recommended by financial planners (i.e., enough to cover expenses for at least 3 months in the absence of income) (FINRA, 2019), and half of all adults nearing retirement have no personal retirement funds (King, 2022).

One intervention to assist lower-income households with savings occurs with the income tax filing process. The annual tax filing process results in millions of dollars in tax refunds (i.e., money sent to the taxpayer) for the majority of filing households due to overpayment or tax credits (Internal Revenue Service, 2021). Tax refunds provide an opportunity for accumulating savings because they are funds that are outside the usual flow of income for households. These infusions of income are often the largest sum of money that households have access to within a year because they come as a lump sum (Black & Schreur, 2014). Given that the sums are outside of typical income, the refunds have the potential to be used for a variety of purposes that are atypical for the household (Beverly et al., 2000; Despard et al., 2015; Grinstein-Weiss, Comer et al., 2015; Mendenhall et al., 2012). Households often have plans for these funds, such as saving money, catching up on past due bills, paying down short- and long-term debt, pre-paying for some expenses, and/or making major purchases that were not possible with their regular flow of income (Beverly et al., 2000; Despard et al., 2015; Mendenhall et al., 2012; Sykes et al., 2015). Most households receiving the Earned Income Tax Credit (EITC), a tax credit for lower-income households based on their earned income, use their refund to pay bills and debt (Mendenhall et al., 2012). Some households spend their refunds to avoid material hardship, such as purchasing basic necessities (e.g., food) (Kondratjeva et al., 2022). Households with children often spend a portion of their refunds on their children's well-being, such as for school supplies, clothing, and vacations (Despard et al., 2015). Few EITC recipients save refund funds for long-term purposes, such as for retirement or education (Beverly et al., 2000; Despard et al., 2015; Mendenhall et al., 2012).

To capitalize on the possibility of capturing tax refund funds for short- and long-term saving, tax-time saving is being widely promoted across the country (CFPB, 2015; Tufano et al., 2005). A variety of interventions mostly targeted toward low- and moderate-income households have been designed and implemented. The goal of these interventions is to increase savings and/or investment, and thereby increase financial security and well-being. The details of the interventions vary, but in general, they provide education about the importance of savings and/or investments, and incentivize, facilitate, and/or encourage participants to save or invest either a portion or all of their refund. While systematic reviews have been conducted on related financial interventions (e.g., Birkenmaier, Maynard et al., 2022), to date, no systematic review of tax-refund interventions has been conducted to provide evidence of their effectiveness. To evaluate the effectiveness of these interventions on savings and/or investment, this article will fill in the gap by synthesizing effects of tax time interventions reported between 2011 and 2021.

Background

Despite scarce funds, research has demonstrated the ability of low- and moderate-income households to save money (Loibl et al., 2018; Mills et al., 2019; Richards & Thyer, 2011). The institutional theory of saving points to environmental influences on individual behavior as an important determinant of saving among low-income households. Beverly and Sherraden (1999), building on institutional theory, posit that saving behavior is shaped by the processes that institutions provide for saving. Later work by Sherraden et al. (2010) identified seven institutional dimensions of savings: access, security, incentives, information, facilitation, expectations, and limits. Saving interventions, described next, implement the framework of institutional theory of saving by providing institutional structures and demonstrate various institutional dimensions.

Saving Interventions

Social work researchers and practitioners have taken a prominent role in testing interventions that encourage or incentivize short- and long-term savings for lower-income households. For example, social work researchers have taken the lead on two internationally known savings interventions. The first is matched savings accounts (i.e., Individual Development Accounts [IDAs]), which provide a financial incentive to save. IDAs have demonstrated that low-income households that are provided institutional supports can save (Han & Sherraden, 2009; Loibl et al., 2018; Schreiner & Sherraden, 2007). Furthermore, IDA research has shown that participant families, who have income constraints, can save without experiencing material hardship (Mills et al., 2019). Social workers have also launched a similar intervention focused on saving, called Child Development Accounts (CDA). These accounts provide special investment accounts during childhood whose proceeds are to be used for the child's lifelong assets, such as post-secondary education. While varying in design and saving platforms (e.g., college savings account), many of the CDA programs provide the initial investments in an account to lower barriers to savings as a part of the intervention and encourage participants to directly save for their child's education, which can also be matched if income is eligible (Clancy et al., 2016). Other bodies of saving-related interventions have focused on employer-provided and individual retirement savings plans (e.g., Collins & Urban, 2016; Duflo et al., 2006; Goda et al., 2012; Lusardi et al., 2009), and financial education, counseling and coaching (Fernandes et al., 2014; Kaiser & Menkhoff, 2020; Miller et al., 2015; Richards & Thyer, 2011; Theodos et al., 2018).

While individual studies on many of these interventions have found positive saving effects, recent systematic reviews on IDA programs, CDA programs, and retirement saving interventions have found mostly mixed or weak effects on saving amounts or were inconclusive for several legitimate reasons (Birkenmaier et al., 2021a; 2021b; Birkenmaier, Kim et al., 2022). Individual studies of financial coaching (e.g., Theodos et al., 2018) and at least one systematic review of financial education (e.g., Kaiser et al., 2022) have found positive effects on saving amounts. However, other reviews, including systematic reviews, have concluded that financial education, counseling, and coaching has no or small effects on later behavior, including short-term saving (Birkenmaier, Maynard et al., 2022; Fernandes et al., 2014; Kaiser & Menkhoff, 2020; Miller et al., 2015; Richards & Thyer, 2011).

Tax-Time Saving Interventions

The tax-time saving interventions are a specific type of saving intervention centered around the tax filing process. Lower-income Americans have the same tax filing process options as all Americans, apart from also being eligible to utilize tax clinics operated by nonprofit organizations that will assist in the process with qualified volunteers. These clinics, called the Volunteer Income Tax Assistance (VITA) and Tax Counseling for the Elderly (TCE) sites, offer free basic tax return preparation by IRS-certified volunteers. The VITA sites assist people who earn $58,000 a year or less, people with disabilities, and people with limited English-speaking abilities, while the TCE sites serve lower-income older adults (IRS, 2022a). When people file their tax return and are due a refund, they have four choices: (1) direct deposit of the refund into (an existing) bank account or on a prepaid card; (2) receive a paper check through the mail; (3) split the refund into a maximum of three separate accounts, including savings accounts; or (4) purchase up to $5,000 in U.S. Series I Savings Bonds with all or a portion of their refund, while sending any remaining funds to their bank account or receiving them as a paper check (IRS, 2022b). The clustering of lower-income Americans at these sites, as well as at for-profit tax return companies that offer same-day filing and refunds for a fee, have offered unique access to suitable populations for interventions. More recently, tax filing software has also been utilized as the platform to offer interventions geared toward eligible lower-income households (e.g., Despard, Grinstein-Weiss et al., 2018).

While intervention details vary, in general, these interventions have provided information about the importance of short-term savings and/or safe investments via software or through face-to-face interaction during the filing process, followed by facilitation of and opportunities to save and invest that are safe, convenient, and affordable. Some interventions also involve sweepstakes and lotteries to incentive tax filers to save a portion or all of their refund (e.g., AARP Foundation, 2022).

Interventions to encourage savings with tax refunds, or “tax time savings interventions,” have the potential to impact thousands of low- to moderate-income adults to help increase their savings and promote better financial security. Since tax-time savings interventions are quite short in duration (happen at the time participants file their taxes) and are relatively low cost, tax-time interventions can be an efficient method of helping low- and moderate-income adults save money. While there have been a number of both small and very large-scale studies of tax-time intervention savings, the results of individual studies have provided a mixed view of the effects of tax-time savings on various outcomes. For example, The Refund 2 Assets program (R2A) promoted saving part of the refund to a savings account to tax filers at VITA sites. The intervention offered only low-cost savings account without any financial incentive. Study authors found no significant effect of the program on the average amount of refund saved 3–5 months later (Beverly et al., 2006). Using a different strategy to promote saving, the Extra Credit Savings Program required VITA site participants to deposit their entire tax refund into a savings account with the possibility of a 10% match if some funds remained at the end of the calendar year. After 3 months, the median account balance was less than 5% of the original refund amount (Beverly et al., 2000). Other studies, including those contained in this study, have reported significant savings results in the short- and longer-term (e.g., Azurdia & Freedman, 2016; Grinstein-Weiss et al., 2017; Key et al., 2015). In addition, interventions that have focused solely on investment (i.e., purchase of savings bond) rather than savings have found no effect (Bronchetti et al., 2013).

Tax-time interventions may have the potential to increase savings among low- and moderate-income adults; however, relying on individual studies to provide an answer to whether tax-time savings are effective in improving saving and other relevant outcomes does not provide the full picture of the evidence. While there have been numerous studies assessing effects of tax-time interventions, there has not been any synthesis of tax-time intervention to examine the full body of evidence of effects across studies to date. Synthesizing effects across studies can provide a more robust and complete understanding of the evidence and more precise and less biased estimates of effects of the intervention (Lipsey & Wilson, 2001).

Purpose of and Rationale for the Study

The aim of this study is to review and synthesize the scientific evidence for tax-time saving interventions on savings and other relevant outcomes reported by study authors. To achieve this aim, we asked the following research questions: (1) What are the effects of tax-time interventions with low- and moderate-income participants in the United States on (a) savings amount and/or (b) savings rate compared to participants that do not receive the intervention? (2) What are the effects of tax-time interventions with low- and moderate-income participants in the United States on other relevant outcomes compared to those that do not receive the intervention? As a systematic review, this study will provide the highest level of evidence on the effectiveness of tax-time interventions aimed at promoting savings. Finding can inform policy and practice related to saving in general, and tax-time saving initiatives more specifically.

Method

Systematic review methodology was used for all aspects of the search, selection, and coding of studies. Meta-analysis was planned to quantitatively synthesize effects of interventions. Campbell Collaboration procedures and guidelines for systematic review and meta-analytic methods were used (see www.campbellcollaboration.org), including developing an a priori protocol, conducting a rigorous search across multiple sources and including a search for gray literature, using two coders for screening and data extraction, and conducting an assessment of the risk of bias for each study. The protocol, screening and data extraction form for this review are available from the first author. This study did not involve human subjects, thus institutional review board review was not required. The authors did not receive funding for the conduct of this review and the authors have no conflicts of interest to declare.

Inclusion and Exclusion Criteria

Intervention and Participants

To be eligible for this review, the intervention must have been an intervention aimed toward low- and moderate-income participants delivered when filing income taxes with the aim of increasing savings and/or investment with a tax refund. To meet the criteria for delivering a tax-time savings or investment intervention, the intervention delivered information and/or encouragement about saving and/or investing a portion or the entire tax refund. The information or encouragement could have occurred prior to or while the tax filing was occurring, and could have been delivered in a face-to-face, printed, online (email or within tax filing software), or video format. The information and/or encouragement could have been about a variety of ways that tax filers can save and invest, such as splitting the refund between bank accounts or purchasing a savings bond. The information and/or encouragement could have been delivered using behavioral economic concepts that shape and influence choice, and could have involved a financial incentive, such as a savings/investment match. Studies that used multi-component interventions were eligible as long as a component was encouragement or incentive to save a portion or all of the tax refund.

Studies must have been conducted in an Organization for Economic Co-operation and Development (OECD) member country. Non-OECD countries were excluded for several reasons. First, this limitation assisted in maintaining a reasonable scope to the project and could produce findings relevant to a large population in the U.S. and other developed countries. Second, the study focused on tax filing initiated by individuals within developed financial systems, a feature not shared by financial systems in all developing countries.

Types of Outcome Measures

Studies must have measured at least one of the following primary outcomes related to the tax refund: saving amount for any reason (e.g., emergency, retirement), saving rate, purchase of an investment (e.g., saving bond), splitting refund into savings, saving all or a portion of the refund in a checking or savings account, percent of refund saved, refund saved post-tax filing period, and debt repayment. If studies included one of the primary outcomes, we also extracted data on secondary outcomes, such as use of alternative financial services, presence or amount of unsecured debt, debt amount repaid, and experience of material hardship.

Measurement of above outcomes could have been conducted using standardized or unstandardized instruments, and self- or other-reported or researcher-administered measures. Thus, the reviewers did not exclude measures based on the type of measure, but planned to pool effects based on type of measure used (e.g., survey measures pooled with survey measures). In the planned meta-analysis, for a study to be included, study authors must have reported enough information to calculate an effect size. If sufficient information to calculate an effect size was not provided, every effort was made to contact study authors and request the necessary information.

Types of Study Designs

To mitigate threats to internal validity, studies must have used a prospective randomized controlled trial (RCT) or quasi-experimental research design (QED) with parallel cohorts. Studies using single-group pre-posttest design, or single subject design, or historical comparisons were excluded. For RCT and QED studies, waitlist control, no treatment, treatment-as-usual and alternative treatment groups were considered acceptable comparison groups.

Duration of Follow-up

The reviewers included measurement points at post-intervention and all follow-up time points. We planned to synthesize studies that reported similar follow-up time points (e.g., immediately after filing, 6 months, 16–21 months) if there were more than two studies that reported sufficient data. All settings and modes (e.g., face-to-face, online) were eligible for inclusion.

Literature Search and Procedures

We searched for and retrieved both published and unpublished studies through a comprehensive search that included 12 electronic databases, Google and Google Scholar, conference proceedings, and organizational and government websites (e.g., Consumer Financial Protection Bureau, Federal Reserve Banks) (Kugley et al., 2017). The reference lists from included studies were harvested for potential studies. We conducted forward citation searching using Google Scholar to search for studies citing the included studies.

The search terms used in the databases, with necessary modifications, were: (tax-time OR “tax time” OR “refund” OR “tax”) AND (saving OR savings OR “field experiment” OR “refund experiment” OR “refund intervention”), AND (evaluation OR intervention OR treatment OR outcome OR program OR trial OR experiment OR “control group” OR “trial” OR quasi-experiment OR “quasi experiment” OR random*). For example, in ABI/INFORM, the search terms (tax-time OR “tax time” OR “refund” OR “tax”) AND (saving OR savings OR “field experiment” OR “refund experiment” OR “refund intervention”), AND evaluation in any field. We limited the search to publication years ranging from January 2011, when studies about tax-time saving interventions began to be published/reported through September 2021, as we began our search in October 2021.

Data Screening and Analysis

Between October and December 2021, one member of the review team conducted the initial search in all sources. The search results were saved in an electronic format using the reference manager software Endnotes. At this stage, two review team members examined titles and abstracts and discarded results that were obviously ineligible (nonempirical report, book review, editorial, non-OECD country, etc.). For those that were not obviously ineligible, the reviewers retrieved the reports, saved them in an electronic file, removed duplicates, and documented the bibliographic information, source, and date retrieved. The data was then uploaded into the online platform Covidence Systematic Review Software (Health Innovation, Melbourne, Australia, www.covidence.org). Multiple reports of individual studies were collated. Two review team members independently screened each of the reports for eligibility using a screening instrument (available from the authors). The same two reviewers compared the coding and identified all discrepancies. The two reviewers discussed and resolved all discrepancies related to study eligibility through consensus.

In our search, we found multiple studies that used unique samples of nearly identical interventions (e.g., repeated the same/similar study in different years using different samples). Each unique sample was considered a separate study. We also found duplicate reports of the primary studies, as well as summary reports that spanned the findings of multiple studies. We designated these as secondary reports and extracted data from all reports that were relevant to a particular study/primary report.

Data Extraction and Management

The three review authors participated in data extraction and management; with at least two of the three review authors independently extracted data from all included studies using a structured data extraction form (available from the authors). The review authors compared coding, identified and discussed discrepancies, and resolved them through consensus. The data extraction form included items related to bibliographic information and source descriptors, methods and procedures, context, nature, and implementation of the intervention, sample characteristics, and outcome data needed to calculate effect sizes.

Assessment of Risk of Bias in Included Studies

The three review authors participated in risk of bias assessment using the Cochrane Collaboration's risk of bias tool (Higgins et al., 2011), with at least two authors coding each study independently. The review authors assessed risk of bias for each of the seven following domains: sequence generation, allocation concealment, blinding of participants and personnel for all outcomes, blinding of outcome assessors for all outcomes, incomplete outcome data for all outcomes, selective outcome reporting, and other potential sources of bias (i.e., researcher allegiance, funding source). Each study was coded as “low,” “high,” or “unclear” risk of bias on each of the domains. Following independent coding by at least two review authors, discrepancies were resolved through consensus. We anticipated that most studies included in this review would be at high risk of bias in terms of allocation and blinding; thus, we did not plan to restrict analyses based on risk of bias in any domain.

Synthesis Procedures and Statistical Analysis

The review authors conducted descriptive analyses on variables of interest from all included studies to provide information regarding: (1) publication type; (2) study design; (3) sample size; (4) participant age; (5) participant race; (6) participant gender; (7) participant income; (8) refund amount; (9) whether banked at baseline; (10) intervention mode (e.g., online); (11) intervention goal; (12) intervention incentives; (13) federal refund amount; and (14) outcomes regarding saving and other financial outcomes.

Following descriptive analysis, we examined the data to calculate effect sizes and prepare for the meta-analysis. For continuous outcomes, we used author reported means and standard deviations to calculate the effect size. For dichotomous outcomes (e.g., in the form of raw numbers or percentages), effect sizes were computed as odds ratio, where the odds refer to the odds of saving compared to not saving for an individual that participated in the intervention relative to participants in the control group. We provide a narrative summary and tables describing the study characteristics and effect sizes, calculated using the Practical Meta-Analysis Effect Size Calculator (Wilson, n.d.) or results as reported by the primary author when we could not calculate an effect size, for each of the outcomes of interest. For outcomes used in meta-analysis (savings rate and savings amount), effect size data were exported into R and effects were estimated using the metafor package (Viechtbauer, 2010). Effect sizes could not be calculated for some studies due to authors not reporting sufficient information (most commonly, standard deviations). Review authors contacted five primary study authors for missing outcome data, of which one author provided the data needed for two studies.

We planned to conduct meta-analysis for all outcomes for which there was sufficient data to calculate effect sizes in at least two studies reporting the same outcomes at similar time points. After extracting all outcome data, we had sufficient data to quantitatively synthesize effects for the two primary outcomes of saving amount and saving rate, which were measured immediately after the intervention. Saving amount was reported in all reports as a continuous outcome on the same scale (U.S. dollars saved), and thus we used both the standardized mean difference and the mean difference effect size. Savings amount included savings for any type of purpose, including emergency savings and retirement savings. Saving rate was reported as the proportion of people who saved or did not save money in any fashion, such as depositing money into an account or purchasing savings bond; thus, we used the log-odds ratio effect estimate (log-odds ratio was converted back into odds ratio for the forest plot for easier interpretation). Because multiple effects were reported in the same studies of the same outcome (e.g., multiple treatment conditions against the same control condition), all meta-analysis were run in R using the robumeta package (Fisher et al., 2017) to conduct robust variance estimation (RVE) with small-sample corrections. RVE accounts for the nonindependent sampling errors due to inclusion of multiple effect sizes from the same study. The model used is a correlated effects model with the assumption that the correlation among the effect sizes within studies is 0.8. We additionally performed sensitivity analyses using different assumptions of the correlation among effect sizes. Results were robust to differing assumptions of correlations among effect sizes. Heterogeneity was assessed using I².

Results

Search Results

A total of 5,434 titles were located, including six studies found on websites. The forward citation search on Google Scholar and reference list search did not find any additional studies from the results of our search of the databases, websites, Google Scholar and Google. After removing studies based on relevance, 95 full-text titles were screened in Covidence, and after removal of ineligible reports, 38 titles met criteria for inclusion in the review. After identifying duplicate and summary reports, 14 studies reported on in 25 reports were included. See Figure 1 for flow diagram (figure generated with Haddaway et al., 2022) and the reference list for a full list of included reports.

Characteristics of Included Studies

As seen in Table 1, of the 14 studies, 11 were randomized control trials and three were QEDs. The majority of studies did not specify the mean age, the percent male, the predominant race of the subjects, or whether they were banked at baseline. The largest percentage of the studies have between 5,301 and 285,000 subjects in the study, and low-moderate income participants. Most of the studies did not offer savings incentives, but of those that did, the predominant incentive was matched savings. The most frequent intervention mode was via tax-filing software (43%). The average federal refund amount of the subjects was $1,000–$2,000. The predominant intervention goal was saving all or portion of their refund or purchasing a U.S. savings bond (43%), followed by saving into a bank account (29%). The most frequent outcomes were saving rate (71%) and amount (79%).

Table 1.

Characteristics of Included Studies Across Studies (k = 14, Primary Study Only).

Study design	k	%	Sample size	k	%
Randomized control trial	11	79	200–400	3	21
Quasi-experimental study	3	21	401–5,300	3	21
			5,301–285,000	4	29
Mean age of beneficiaries			285,001–846,000	3	21
34–35	3	21	Unspecified	1	7
36–37	2	14
38–39	2	14	Income
Not specified	7	57	Low	1	7
			Low–moderate	11	79
Percent male			All incomes	1	7
22–27	3	21	Unknown	1	7
28–40	3	21
Not specified	8	57	Banked at baseline
			75–82%	3	21
Predominant subject race			83–100%	2	14
White	3	21	“Mixed”	1	7
African-American	4	29	Not specified	8	57
Not specified	7	57
			Saving incentives
Intervention mode			Matched savings	3	21
VITA site	5	36	10% discount card	1	7
Commercial tax preparer	2	14	$5	1	7
Smartphone app	1	7
Software	6	43	Intervention goal
			Save all/portion of refund or	6	43
Average federal refund amount			purchase U.S. savings bond
$1,000–$2,000	5	36	Save into bank account	4	29
$2,001–$3,000	3	21	U.S. savings bond only	1	7
$3,001 and higher	2	14	Pay down debt and save	1	7
Not specified	4	29	Save on a prepaid card	1	7
Financial outcomes			Purchase certificate of deposit (CD), U.S. savings bond, or IRA savings	1	7
Debt amount	2	14
Debt payment	1	7	Saving/investment outcomes
Use of alternative financial service)	2	14	Savings rate* (have savings)	10	71
Refund saved 6 months post	1	7	Savings amount*	11	79
			Intention to save (retirement)	1	7

*Author-reported outcomes, not necessarily included in the meta-analysis.

Risk of Bias

As seen in Figure 2 (generated using robvis tool, Mcguinness & Higgins, 2021), the risk of bias varied across studies. Most of the studies were RCTs, and the remainder of the studies were QEDs with comparison groups. Because the included studies did not have preregistered protocols, it is difficult to assess reporting bias for incomplete outcome data for all outcomes or selective outcome reporting. Across the studies, the risk of performance bias is high or unclear, as none of the included studies reported that they employed blinding of participants or personnel. There is generally a lack of clarity related to the study authors’ role in the interventions or potential bias originating from study fundings, thus the potential bias related to researcher allegiance or funding is unclear or high. The risk of selection bias (from sequence generation and allocation concealment) and detection bias (blinding of outcome assessment) is mostly low for the outcomes. Generally, the studies were conducted with randomly selected samples or convenience samples, and large sample sizes, and thus findings were generalizable. Study authors recognized study limitations and recommended that research on intervention effectiveness continues, and that rigorous research be conducted on the tax-time interventions. The full Risk of Bias assessment by study with rationale for each item can be requested from the first author.

Effects of Tax Time Savings on Saving Amount and Saving Rate

Of the 14 studies included in this review, several outcomes were reported at various time points; however, not all studies reported sufficient data to calculate an effect size or reported outcomes at various time points, thus could not all be included in a meta-analysis. We had a sufficient number of studies to conduct a meta-analysis for saving amount and saving rate at the immediate posttest timepoint. Multiple studies reported on both saving amount and saving rate so were included in each of the meta-analyses. See Table 2 for a list of study outcomes reported by study, including author reported outcomes when effects sizes could not be calculated.

Table 2.

Summary of Effects Sizes and Author-Reported Outcomes.

Study/program name¹	Study authors	Effect size outcome type	Effect size outcome type²	Author-reported outcomes type	Author-reported outcomes or effect sizes
Bronchetti	Bronchetti et al., 2013	*Saving rate	d = 0; 95% CI [−0.46, 0.46]
Johnson	Johnson & Wang-Ly, 2020	Debt repayment (immediate)	d = 0.06; 95% CI [0.04, 0.08]
Knoll	Knoll et al., 2019	Saving rate—encouragement to save (immediate) Saving rate—incentive to save (immediate)	d = 0.13; 95% CI [0.02, 0.25] d = 0.26; 95% CI [0.15, 0.37]	Saving amount (immediate)	Study authors reported that the amount was greater for the control group. The incentive group saved the least. There was no statistically significant difference between the control and the encouragement group, but the incentive group was statistically smaller than the other two groups.
Palmer	Palmer et al., 2016	Saving rate (10% club + video) (immediate) Saving rate (10% club) (immediate) Saving rate (video only) (immediate) Saving amount (10% club + video) (immediate) Saving amount (10% club) (immediate) Saving amount (video only) (immediate)	d = −0.14; 95% CI [−0.06, 0.31] d = −0.1; 95% CI [−0.56, 0.36] d = 0.61; 95% CI [0.14, 1.09] d = −0.28; 95% CI [−0.59, 0.02] d = −0.02; 95% CI [−0.42, 0.38] d = 0.52; 95% CI [0.05, 0.99]
Refund to Savings 2012 Data
	Grinstein-Weiss et al., 2017	Saving rate (immediate) Saving amount (immediate)	d = 0.19; 95% CI [0.15, 0.23] d = 0.04; 95% CI [0.02, 0.06]
Refund to Savings 2013 Data
	Despard, Taylor et al., 2018	Material hardship (6 mo) Health care hardship (6 mo.)	d = 0.03; 95% CI [−0.06, 0.12] d = 0.03; 95% CI [−0.06; 0.12]
	Despard, Grinstein-Weiss et al., 2018	Used Credit Alternative Financial Services (6 mo). Used Transaction Alternative Financial Services (6 mo.)	d = 0.01; 95% CI [−0.11, 0.13] d = −0.02; 95% CI [−0.13; 0.09]
	Despard et al., 2017	*Savings rate (6 mo.)	d = 0.6; 95% CI [0.4; 0.8]
	Grinstein-Weiss, Russell et al., 2014			Savings amount (immediate)	Study authors reported the following saving amounts: Control—$197 Anchor 25%—$216 Anchor 50%—$214 Emergency 25%—$224 Emergency 50%—$221 Future 25%—$213 Family 25%—$198
	Grinstein-Weiss, Perantie et al., 2015	*Savings rate (immediate)	d = 0.07; 95% CI [0.05; 0.08]
	Roll, Grinstein-Weiss et al., 2019	Saving rate (immediate) P1-generic 25% Saving rate (immediate) P1-generic 50% Saving rate (immediate) P1-emergency 25% Saving rate (immediate) P1-emergency 50% Saving rate (immediate) P1-future 25% Saving rate (immediate) P1-future 50% Saving rate (immediate) P2-generic 50% Saving rate (immediate) P2-generic 75% Saving rate (immediate) P2-emergency 50% Saving rate (immediate) P2-emergency 75% Saving rate (immediate) P2-future 75% Saving rate (immediate) P2-family 75% Saving rate (immediate) P3-emergency $100 Saving rate (immediate) P3-emergency $250 Saving rate (immediate) P3-emergency 25% Saving rate (immediate) P3-future $100 Saving rate (immediate) P3-future $250 Saving rate (immediate) P3-future 25%	d = 0.12; 95% CI [0.09, 0.15] d = 0.07; 95% CI [0.04, 0.1] d = 0.11; 95% CI [0.08, 0.14] d = 0.10; 95% CI [0.07, 0.13] d = 0.10; 95% CI [0.07, 0.13] d = 0.05; 95% CI [0.02, 0.08] d = 0.11; 95% CI [0.07, 0.15] d = 0.10; 95% CI [0.06, 0.14] d = 0.11; 95% CI [0.08, 0.15] d = 0.11; 95% CI [0.08, 0.15] d = 0.09; 95% CI [0.05, 0.13] d = 0.08; 95% CI [0.04, 0.12] d = 0.00; 95% CI [−0.04, 0.05] d = 0.02; 95% CI [−0.02, 0.07] d = 0.02; 95% CI [−0.02, 0.07] d = 0.05; 95% CI [0.01, 0.09] d = 0.06; 95% CI [0.02, 0.11] d = −0.01; 95% CI [−0.06, 0.00]
Refund to Savings 2015 Data
	Grinstein-Weiss et al., 2017	Savings rate—emergency savings (immediate) Savings rate—interactive goals (immediate) Savings rate—interactive retirement (immediate) Savings amount—Emergency savings (immediate) Savings amount—interactive goals (immediate) Savings amount—interactive retirement (immediate)	d = 0.28; 95% CI [0.27, 0.3] d = 0.25; 95% CI [0.23, 0.26] d = 0.24; 95% CI [0.22, 0.25] d = 0.08; 95% CI [0.08, 0.1] d = 0.07; 95% CI [0.07, 0.08] d = 0.07; 95% CI [0.07; 0.08]
	Despard et al., 2020 EITC recipient subgroup	Deposit to savings (saving amount) (immediate) Savings amount (control vs. emergency savings) (immediate) Savings amount (control vs. interactive goal) (immediate) *Savings amount (control vs. interactive retirement) (immediate)	d = 0.23; 95% CI [0.21, 0.25] d = 0.1; 95% CI [0.09, 0.11] d = 0.08; 95% CI [0.07, 0.09] d = 0.08; 95% CI [0.07; 0.09]
	Roll, Grinstein-Weiss et al., 2019 Subsample with savings accounts			Any refund saved (6 mo. post-tx) Percent of refund saved (6 mo. post-tx)	Study authors reported the following outcomes: precautionary savings—0.017 (0.019) Interactive goal—0.016 (0.019) Interactive retirement—0.001 (0.019) Precautionary savings—1.028 (1.493) Interactive goal—1.312 (1.471) Interactive retirement—0.991 (1.498)
Refund to Savings 2016 Data
	Roll et al., 2018	Saving rate (control vs. choice arch) (immediate) Saving rate (control vs. ES) (immediate) *Saving rate (control vs. IE) (immediate)	d = 0.23; 95% CI [0.01, 0.24] d = 0.24; 95% CI [0.22, 0.26] d = 0.21; 95% CI [0.19; 0.23]
	Roll, Bufe et al., 2020	Saving amount (choice arch) (immediate) Saving amount (choice arch and passive emergency) (immediate) *Saving amount (choice architecture and interactive emergency) (immediate)	d = 0.6; 95% CI [0.05, 0.07] d = 0.6; 95% CI [0.05, 0.07] d = 0.6; 95% CI [0.05; 0.08]
	Despard, Grinstein-Weiss et al., 2020 EITC recipient subgroup	Saving rate (immediate) Saving amount (control vs. choice arch) Saving amount (control vs. ES) Saving amount (control vs. IE)	d = 0.19; 95% CI [0.16, 0.21] d = 0.06; 95% CI [0.04, 0.08] d = 0.07; 95% CI [0.06, 0.09] d = 0.08; 95% CI [0.06, 0.09]
Refund to Savings 2016 myRA
	Davidson et al., 2018	Click to deposit to myRa—easy email (immediate) Click to deposit to myRa—impossible email (immediate) Click to deposit to myRa—Gibber refund email (immediate) Click to deposit to myRa—combined email (immediate)	d = 1.09, 95% CI [1.18, 0.99] d = −1.5; 95% CI [−1.62, −1.38] d = 0.25; 95% CI [0.17, 0.32] d = 0.88; 95% CI [0.79, 0.97]
Refund to Savings 2017 Data
	Kondratjeva et al., 2020	Debt amt	d = −0.03.; 95% CI [−0.13, 0.07]
	Roll, Grinstein-Weiss et al., 2020	Savings rate condition 2 (immediate) Savings rate condition 3 (immediate) Savings rate condition 4 (immediate) Saving amount condition 2 (immediate) Savings amount condition 3 (immediate) Savings amount condition 4 (immediate)	d = 0.28; 95% CI [0.27, 0.29] d = 0.22; 95% CI [0.21, 0.23] d = 0.22; 95% CI [0.21, 0.24] d = 0.08; 95% CI [0.08, 0.09] d = 0.05; 95% CI [0.04, 0.06] d = 0.5; 95% CI [0.048, 0.063]
$aveNYC 2009
	Key et al., 2015			Saving amount (6–11 mo. post tx)	Study authors reported effect size of 0.27.
$aveNYC 2010
	Tucker et al., 2014			Saving amount (15–21 mo. post tx)	Study authors report that the tx saving amount mean was $1,375, and the control group mean was $940.
SaveUSA
	Azurdia et al., 2014 Azurdia & Freedman, 2016	Debt amount (18 mo.) Debt Amount (42 mo.) Have nonretirement savings (18 mo.) Have nonretirement savings (42 mo.) Have retirement savings (18 mo.) Have retirement savings (42 mo.) Saving amount (nonretirement) (18 mo.) Saving amount (nonretirement) (42 mo.) Saving amount (retirement) (18 mo.) Saving amount (retirement) (42 mo.) Used high-interest credit (18 mo.) Used high-interest credit (42 mo.)	d = −0.03; 95% CI [−0.14, 0.08] d = −0.03; 95% CI [−0.08, 0.15] d = 0.23; 95% CI [0.09, 0.37] d = 0.23; 95% CI [0.09, 0.38] d = 0.12.; 95% CI [−0.02, 0.25] d = 0.03; 95% CI [−0.1, 0.17] d = −0.08; 95% CI [−0.19, 0.08] d = −0.13; 95% CI [0.02, 0.24] d = −0.08; 95% CI [−0.19, 0.08] d = −0.02; 95% CI [−0.13, 0.09] d = −0.02; 95% CI [−0.17, 0.13] d = −0.1; 95% CI [−0.04; 0.24]
Tufano	Tufano, 2011	*Saving rate (immediate)	d = 1.28; 95% CI [0.94, 1.61]	Saving amount (immediate)	Study authors report that treatment group members saved an average of $28.21, compared to a control group mean of $12.95.

Notes: 1. If the intervention was not a named intervention, the first author's last name is used to designate the study name.

2. Effect sizes included in the meta-analysis are noted with an *.

To examine the effects of tax-time interventions on savings amount, five studies contributed 13 effect sizes to the meta-analysis. The results of the meta-analytic model, as seen in Figure 3, indicate that those participating in the tax time intervention saved more money than those in the control group (g = 0.06, SE = 0.01, p = .003, 95% CI [0.04, 0.09]; I² = 91.83). In terms of dollars, the intervention group saved an average of $52.30 (SE = 11.30, 95% CI [$16.6, $88.1]; I² = 96.98). The degrees of freedom for the correlated effect model were less than 4 for both analyses indicating that the results of the RVE analysis may be unreliable. Because of the small number of studies, no other models exploring effect size heterogeneity were possible.

To examine the effects of tax-time interventions on saving rate, nine studies contributed 35 outcomes to the meta-analysis. As seen in Figure 4, the results of the meta-analytic model indicated no statistically significant difference in savings rate between the intervention and comparison groups (LOR = 0.35, SE = 0.15, p = .6, 95% CI [−0.27, 0.72]; I² = 98.76), and in terms of odds ratios, the mean OR is 1.41 (SE = 1.17, 95% CI [0.97, 2.05]). As in the prior analysis, the degrees of freedom for the RVE model were smaller than 4 (2.25) indicating potential bias in the estimates.

Five of the nine studies reporting both saving amount and rate outcomes emanated from the Refund to Savings (R2S) studies (2012, 2013, 2015, 2016, and 2017), in which the intervention was provided through online tax filing software. The studies tested different messages and message combinations built directly into the software related to savings, investments, and/or finances to encourage filers to deposit a portion or all of their tax return into savings, which they could do immediately while filing. Each of the studies gathered data from samples that participated in the studies during a different year. Three of the four remaining studies that reported both saving amount and saving rate outcomes delivered a face-to-face or video intervention that included incentives for saving the refund. Knoll et al. (2019) delivered the intervention in a face-to-face format at a commercial tax preparation company. They encouraged their low-income consumers who use prepaid cards to save a portion of their tax refund on the prepaid card. Treatment group participants in one arm were sent an email with encouragement to save a portion of their refund on a separate fund from other funds within their prepaid card, and the second treatment group was sent an email with an incentive to save ($5) of their refund on the separate prepaid card fund. In the SaveUSA intervention delivered at VITA sites, treatment group participants were encouraged to deposit a portion of their tax refund into a special savings account and pledged to save between $200 and $1,000 of their deposit for about a year to receive a 50% match. Palmer et al. (2016) used a video-based solution-focused brief coaching intervention and an incentive-based savings promotion intervention during tax filing at a VITA site. There were three treatment groups to which participants were assigned. The first invited participants to save 10% of their potential tax refund in a restricted savings vehicle (CD, IRA or savings bond) and receive a 10% discount card. The second invited participants to view a solution-focused brief counseling video and complete a worksheet. The third involved pledging to save 10%, watching the video, and completing a worksheet. The remaining study reporting both outcomes used education, rather than messaging or incentives, to encourage refund saving. Tufano (2011) used a commercial tax preparation setting to provide treatment group members with education about saving and investment. Respondents were provided an opportunity to purchase a U.S. Saving Bond with their tax refund funds.

Three studies reported only savings amounts or saving rates. Bronchetti et al. (2013) provided saving rate outcomes from an intervention that occurred while interacting face-to-face with VITA volunteers. After visual exposure to posters and flyers about savings bonds, treatment group members were offered the opportunity to opt-out (rather than opt-in) from 10% of their tax refund directed by default to a U.S. savings bond. Two similar studies, $aveNYC 2009 and 2010, reported on treatment participants who completed their tax returns at VITA sites. Treatment group members were offered a $aveNYC account into which a portion of their tax refund of their choosing was deposited. If the original amount was retained in the account for one year, a 50% match was offered at the end of the year (in 2009, up to $250 was offered for at least $100 retained savings, and in 2010, up to $500 for at least $200 retained savings).

Effects of Tax Time Saving on Other Outcomes

Studies reported a variety of other outcomes that were unable to be grouped together for the meta-analysis due to outcomes being conceptually different and/or time-frames not being similar. Johnson and Wang-Ly (2020) reported debt repayment outcomes on an intervention that used a bank smart phone app to provide messaging, reminding, and payment tools for debt repayment during tax refund season in Australia. The R2S 2016 myRa study (related to the R2S studies mentioned earlier) provided treatment group participants with the ability to indicate their intention to deposit refund monies into a myRA account, a publicly available retirement savings vehicle available at the time of the study. Studies previously discussed also reported the following outcomes: material hardship (R2S 2013), healthcare hardship (R2S 2013), use of credit and transaction alternative (nonbank) services (R2S 2013; SaveUSA), any refund and percent of refund saved 6 months postintervention (R2S, 2015), and debt amount (R2S 2017; SaveUSA).

Discussion and Application to Practice

The purpose of this systematic review and meta-analysis was to determine the state of the evidence on the effectiveness of tax-time interventions on low- and moderate-income population to increase their financial well-being through increased savings and investment. This review included a total of 14 studies that reported on a range of types of tax-time interventions. The predominant intervention mode was the use of prompts and messaging (quick description) using tax filing software at the time of tax filing. The predominant intervention goal was saving all or portion of refund or purchasing a U.S. savings bond, followed by saving into a bank account. The most frequent outcomes reported were saving rate and amount.

Overall, the evidence on the effects of tax-time interventions for saving amounts and saving rate is relatively weak. The results of the meta-analytic model indicate that those participating in the tax time intervention saved more money than those in the control group, saving an average of about $52 more. However, the results of the meta-analytic model indicated no statistically significant difference in savings rate between the intervention and comparison groups.

It is important to note that the full value of tax-time savings interventions may be difficult to ascertain. While the finding that those who participated in the tax-time interventions saved on average $52 more than the control group, the amount of money may not seem practically significant to warrant excitement over this statistically significant finding. However, Tucker et al. (2014) point out that the value of tax-time savings interventions may not be detected by short- or long-term effect on savings volume. Any short-term savings from tax refunds could be used for emergency purposes, whether shortly after receiving the funds or months later, fulfilling the precautionary purpose of the savings. Withdrawing savings from tax refunds for these purposes could help to fulfill a major purpose of the savings: to avoid material hardship and/or use of predatory short-term consumer loan products, such as payday or auto-title lending, during periods of negative income volatility (Beverly et al., 2000; Grinstein-Weiss et al., 2016). Beyond facilitation of saving a portion or all of a refund, tax-time saving interventions may also have other positive effects, such as influencing participants to spend more slowly and carefully, and resist spending temptations, and/or progress toward financial goals (Beverly et al., 2000), as well as avoiding material hardship (Grinstein-Weiss et al., 2016). Therefore, evaluation of tax time interventions based solely on saving amounts and rate offers an incomplete, albeit valuable, picture of the value of the interventions. Future research in this area that measures these other outcomes may assist in generating new evidence about the value of these interventions. This area of intervention research could also benefit from qualitative studies examining the meaning, purpose, and value (monetary and nonmonetary) of savings. Even though many would perceive $52 as a relatively small amount of money, it could be meaningful to those who participate. For example, Williams and Beal (2020) found that maintaining a balance of $100 or more in an account was associated with increased household financial well-being for lower-income Americans. While cost data for the interventions discussed here are not available, there is a possibility that some of these interventions (e.g., the R2S intervention embedded in software) hold the possibility of reaching a large number of people with relatively low cost; thus the cost-benefit may be favorable, even with small amounts of money.

The quality and risk of bias varies across studies. Some potential bias was difficult to assess (income outcome data and selective outcome reporting) because the included studies did not have preregistered protocols. There is a general absence of information related to the blinding of participants or personnel, as well as other indicators of bias, such as allegiance and funding. The included studies have important strengths that should not be overlooked. Most of the studies are RCTs (k = 11) with large sample sizes, and therefore their findings are generalizable and well powered. Many of the studies at post-test had low attrition. While most of the studies used an RCT design, several studies had important methodological weaknesses. For instance, most of the studies did not use a manualized intervention or examine fidelity. We did not find protocols for the included studies, few of the included studies reported using blinding, and the majority did not report on allocation concealment nor the researchers’ role in intervention development or implementation. Several of the studies experienced high attrition in the control and treatment groups. None of the groups assessed whether the respondents had savings prior to the intervention. Future research could improve upon the prior study designs to improve internal validity, as well as improve upon the outcomes measured and the time points at which outcomes are measured to be more consistent and therefore more comparable for synthesizing across studies.

This review is not without its limitations, and therefore the findings must be interpreted in light of the study's limitations. While we made every attempt to search for published and unpublished studies, studies could possibly have been missed and not be representative of the tax-time interventions being used. The included studies presented various risks of bias and thus caution must be used when interpreting findings as study quality varied. Moreover, a considerable proportion of included studies were derived from the R2S initiative, with repeating authors across the studies. Also, it should be noted that there was variation among the specific elements and strategies used in the tax-time intervention studies, with some studies testing multiple strategies within one study. Thus, it is not clear which elements or strategies are sufficient or necessary to be effective when implementing a savings intervention at the time of tax filing. Also, for both meta-analyses, the degrees of freedom for the RVE model were smaller than 4 (2.25), indicating potential bias in the estimates. Therefore, future systematic review and meta-analysis are needed with additional studies. There is also an overall lack of longer-term findings. Some studies did assess outcomes at later time points, but these were fraught with some challenges, including high attrition, and few studies measuring the same outcomes as the same time points, thus could not be synthesized. It is important to examine longer-term effects, as well as repeated exposure year over year to this intervention. If participants had the opportunity to save something from tax savings every year, the savings could add up over time, or at least become a more meaningful and consistent vehicle for savings than a one-time opportunity.

As a systematic review and meta-analysis, this study provides a rigorous review of the evidence on the effectiveness of tax-time interventions aimed at promoting savings. The limited evidence on the effectiveness of tax-time saving interventions is an important finding for policy and practice related to saving in general, and tax-time saving initiatives more specifically. Social work practitioners implementing this intervention should examine the utility of providing these interventions, given the small amount of savings generated. Social work practitioners and policy advocates who wish to advance these tax-time interventions should support the generation of additional evidence. Research on this topic should use RCT with large sample sizes, with preregistered protocols and clearly meeting the criteria of low bias on all elements of the Campbell Collaborative risk of bias elements.

These findings have relevance for social work practitioners focused on assisting historically oppressed and minoritized populations to gain emergency saving, long-term savings, and wealth. A wide range of interventions aimed at these goals are designed to increase people's financial capability: in particular, those whose financial access and information is limited. In addition to tax-time interventions, financial capability interventions include financial education, counseling, and coaching, matched savings accounts, children's savings accounts, and others (Birkenmaier, Maynard et al., 2022). These interventions all seek to provide financial information and the tools needed to build financial well-being through inclusive and equitable institutional support. Thus far, evidence is still needed that the financial capability interventions are effective. Practitioners working in this area can collaborate with researchers to design high-quality experimental research on financial capability interventions, such as tax-time saving interventions, either independently or in combination, to gather more definitive evidence on these interventions.

Practitioners can also use these findings to advance new policy initiatives. For example, the SECURE Act of 2022, passed in December 2022, will allow employers to automatically enroll workers in an emergency savings account alongside their retirement plan, up to $2,500 (Bernard, 2022). While this is a step forward that will assist some families, this policy change is optional for employers, and thus, some employees will not have access. Additionally, low income workers, even if they do have access to this savings vehicle, may not see the benefit or feel that they can spare any of their paycheck to go to savings. Access to a savings vehicle may be necessary to help low-income workers save money, but not sufficient. Practitioners can connect clients to these opportunities, encourage advocacy with employers to make this opportunity available, and help identify and manage barriers to savings. In the long run, more experimentation with tax time and other interventions to assist families in generating savings to avoid financial hardship is needed.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Julie Birkenmaier

Supplemental Material

Supplemental material for this article is available online.

References

AARP Foundation (2022). SaveYourRefund. https://saveyourrefund.aarpfoundation.org/partners/.

^ Azurdia

Freedman

S. R.

(2016). Encouraging nonretirement savings at tax time: Final impact findings from the SaveUSA evaluation. MDRC. https://www.mdrc.org/publication/encouraging-nonretirement-savings-tax-time

^ Azurdia

Freedman

S. R.

Hamilton

Schultz

C. B.

(2014). Encouraging low-and moderate-income tax filers to save: Implementation and interim impact findings from the SaveUSA evaluation. MDRC. https://www.mdrc.org/publication/encouraging-low-and-moderate-income-tax-filers-save

Bernard

T. S.

(2022). New spending bill makes it easier for Americans saving for retirement. New York Times.

Beverly

Schneider

Tufano

(2006). Splitting tax refunds and building savings: An empirical test. Tax Policy and the Economy, 20, 111–162. https://www.journals.uchicago.edu/doi/pdf/10.1086/tpe.20.20061906 https://doi.org/10.1086/tpe.20.20061906

Beverly

S. G.

Sherraden

(1999). Institutional determinants of saving: Implications for low-income households and public policy. Journal of Socioeconomics, 28(4), 457–473.

Beverly

S. G.

Tescher

Marzahl

(2000). Linking tax refunds and low-cost bank accounts. Working Paper 00-19. Center for Social Development, Washington University in St Louis. https://openscholarship.wustl.edu/cgi/viewcontent.cgi?article=1274&context=csd_research

Birkenmaier

J. M.

Kim

Maynard

(2022). Financial outcomes of interventions designed to improve financial capability through individual development accounts: A systematic review. Journal of Evidence-Based Social Work, 19(4), 408–439. https://doi.org/10.1080/26408066.2022.2059427

Birkenmaier

J. M.

Kim

Maynard

(2021a). Financial outcomes of interventions designed to improve financial capability through children’s development accounts: A systematic review. Journal of the Society for Social Work and Research. https://doi.org/10.1086/716103

10.

Birkenmaier

J. M.

Kim

Maynard

(2021b). Financial outcomes of interventions designed to improve retirement savings: A systematic review. Journal of Gerontological Social Work, 64(3), 238–256. https://doi.org/10.1080/01634372.2020.1865496

11.

Birkenmaier

J. M.

Maynard

Kim

(2022). Interventions designed to improve financial capability: A systematic review. (Final report). Campbell Systematic Reviews, 18(1), e1225. https://doi.org/10.1002/cl2.1225

12.

Black

Schreur

(2014). Connecting tax time to financial security. New America Foundation. https://static.newamerica.org/attachments/3820-connecting-tax-time-to-financial-security-2/Connecting_Tax_Time_to_Financial_Security-Final.c9c150fbaddf449281871c572f2169b5.pdf

13.

Board of Governors of the Federal Reserve System. (2023). Changes in U.S. family finances from 2016 to 2019: Evidence from the survey of consumer finances. Federal Reserve Bulletin, 106(5). https://www.federalreserve.gov/publications/files/scf20.pdf

14.

^ Bronchetti

E. T.

Dee

T. S.

Huffman

D. B.

Magenheim

(2013). When a nudge isn't enough: Defaults and saving among low-income tax filers. NBER Working PaperSeries. http://www.nber.org/papers/w16887.

15.

Clancy

M. M.

Beverly

S. G.

Sherraden

Huang

(2016). Testing universal child development accounts: Financial effects in a large social experiment. Social Service Review, 90(4), 683–708. https://doi.org/10.1086/689756

16.

Collins

M. J.

Urban

(2016). The role of information on retirement planning: Evidence from a field study. Economic Inquiry, 54(4), 1860–1872. https://doi.org/10.1111/ecin.12349

17.

Consumer Financial Protection Bureau [CFPB]. (2015). Increasing saving at tax time and promising practices for the field. https://files.consumerfinance.gov/f/201509_cfpb_increasing-saving-at-tax-time-and-promising-practices-for-the-field.pdf .

18.

^ Davidson

Frank-Miller

Roll

S. P.

Grinstein-Weiss

(2018). Promoting savings at tax time: Insights from online and in-person tax preparation services. CSD Research Report No. 18–33. https://openscholarship.wustl.edu/cgi/viewcontent.cgi?article = 1908&context = csd_research.

19.

^ Despard

Grinstein-Weiss

Deruyter

Guo

Oliphant

J. E.

Friedline

(2018). Effects of a randomized tax-time savings intervention on savings account ownership among low-and moderate-income households. Journal of Financial Counseling and Planning, 29(2), 219–233. https://doi.org/10.1891/1052-3073.29.2.219

20.

^ Despard

Grinstein-Weiss

Roll

Hardy

Perantie

Oliphant

(2020). Tax-time saving and the Earned Income Tax Credit: Results from online field and survey experiments. https://doi.org/10.7936/rjqe-g266 https://openscholarship.wustl.edu/spi_research/26/ .

21.

Despard

M. R.

Friedline

Martin-West

(2020). Why do households lack emergency savings? The role of financial capability. Journal of Family and Economic Issues, 41(3), 542–557. https://doi.org/10.1007/s10834-020-09679-8

22.

^Despard

M. R.

Grinstein-Weiss

Ren

Guo

Raghavan

(2017). Effects of a tax-time savings intervention on use of alternative financial services among lower-income households. Journal of Consumer Affairs, 51(2), 355–379. https://doi.org/10.1111/joca.1213

23.

Despard

M. R.

Parantie

D. C.

Oliphant

Grinstein-Weiss

(2015). Do EITC recipients use their tax refunds to get ahead? Evidence from the refund to savings initiative. https://openscholarship.wustl.edu/cgi/viewcontent.cgi? article = 1589&context = csd_research.

24.

^ Despard

M. R.

Taylor

Ren

Russell

Grinstein-Weiss

Raghavan

(2018). Effects of a tax-time savings experiment on material and health care hardship among low-income filers. Journal of Poverty, 22(2), 156–178. https://doi.org/10.1080/10875549.2017.1348431

25.

Duflo

Gale

Liebman

Orszag

Saez

(2006). Saving incentives for low- and middle-income families: Evidence from a field experiment with H&R block. The Quarterly Journal of Economics, 121(4), 1311–1346. https://doi.org/10.1162/qjec.121.4.1311

26.

Federal Reserve Bank of St. Louis. (2023). Personal saving rate. https://fred.stlouisfed.org/series/PSAVERT.

27.

Fernandes

Lynch

J. G.

Netemeyer

R. G.

(2014). Financial literacy, financial education and downstream financial behaviors. Management Science, 60(8), 1861–1883. https://doi.org/10.1287/mnsc.2013.1849

28.

FINRA Investor Education Foundation (2019). Financial capability in the United States 2018. https://www.usfinancialcapability.org/downloads/NFCS_2018_Report_Natl_Findings.pdf .

29.

Fisher

Tipton

Zhipeng

(2017). Robumeta: Robust variance meta-regression. https://CRAN.R-project.org/package = robumeta.

30.

Goda

Manchester

Sojourner

(2012). What will my account really be worth? Experimental evidence on how retirement income projections affect saving. Journal of Public Economics, 119, 80–92. https://doi.org/10.1016/j.jpubeco.2014.08.005

31.

Grinstein-Weiss

Comer

Russell

Key

Perantie

Ariely

(2015). Refund to savings: Creating contingency savings at tax time. In Collins

J. M.

(Ed.), A fragile balance (pp. 87–106). Palgrave Macmillan. https://doi.org/10.1057/9781137482372_6

32.

^ Grinstein-Weiss

Comer

Russell

Key

Perantie

D. C.

Ariely

(2014). Refund to Savings: 2013 evidence of tax-time saving in a national randomized control trial.https://openscholarship.wustl.edu/cgi/viewcontent.cgi?article = 1604&context = csd_research.

33.

^ Grinstein-Weiss

Cryder

Despard

M. R.

Perantie

D. C.

Oliphant

J. E.

Ariely

(2017). The role of choice architecture in promoting saving at tax time: Evidence from a large-scale field experiment. Behavioral Science & Policy, 3(2), 20–38. https://doi.org/10.1353/bsp.2017.0011

34.

Grinstein-Weiss

Despard

Guo

Russell

Key

Raghavan

(2016). Do tax-time savings deposits reduce hardship among low-income filers? A propensity score analysis. Journal of the Society for Social Work and Research, 7(4), 707–728. https://doi.org/10.1086/689357

35.

^ Grinstein-Weiss

Perantie

D. C.

Russell

B. D.

Comer

Taylor

S. H.

Luo

Key

Ariely

(2015). Refund to Savings 2013: Comprehensive report on a large-scale tax-time saving program. CSD Research Report No. 15-06. https://csd.wustl.edu/15-06/.

36.

^ Grinstein-Weiss

Russell

Tucker

Comer

(2014). Lack of emergency savings puts American households at risk. CSD Policy Brief No. 14-13. https://csd.wustl.edu/14-13/10.1353/bsp.2017.0011.

37.

^ Grinstein-Weiss

Russell

B. D.

Gale

W. G.

Key

Ariely

(2017). Behavioral interventions to increase tax-time saving: Evidence from a national randomized trial. Journal of Consumer Affairs, 51(1), 3–26. https://doi.org/10.1111/joca.12114

38.

Haddaway

N. R.

Page

M. J.

Pritchard

C. C.

McGuinness

L. A.

(2022). PRISMA2020: An R package and shiny app for producing PRISMA 2020-compliant flow diagrams, with interactivity for optimised digital transparency and open synthesis. Campbell Systematic Reviews, 18, e1230. https://doi.org/10.1002/cl2.1230. Download citation (.ris)

39.

Han

C. K.

Sherraden

(2009). Do institutions really matter for saving among low-income households? A comparative approach. The Journal of Socio-Economics, 38(3), 475–483. https://doi.org/10.1016/j.socec.2008.07.002

40.

Higgins

J. P. T.

Altman

D. G.

Gotzsche

P. D.

Juni

Moher

Oxman

A. D.

Savovic

Schulz

Weeks

Sterne

J. A. C.

(2011). The Cochrane collaboration’s tool for assessing risk of bias in randomized trials. BMJ, 343, https://doi.org/10.1136/bmj.d5928

41.

Internal Revenue Service [IRS]. (2022a). Free tax return preparation for qualifying taxpayers. https://www.irs.gov/individuals/free-tax-return-preparation-for-qualifying-taxpayers#:∼:text = The%20IRS’s%20Volunteer%20Income%20Tax,operated%20for%20over%2050%20years.

42.

Internal Revenue Service [IRS]. (2022b). Frequently asked questions about splitting federal income tax refunds. https://www.irs.gov/refunds/frequently-asked-questions-about-splitting-federal-income-tax-refunds.

43.

Internal Revenue Service [IRS]. (2021). Data Book. https://www.irs.gov/pub/irs-pdf/p55b.pdf.

44.

^ Johnson

Wang-Ly

(2020). Using tax refunds for debt repayment. A study with Commonwealth Bank of Australia. Financial Health Network. https://finhealthnetwork.org/research/financial-health-solutions-using-tax-refunds-for-debt-repayment/.

45.

Kaiser

Lusardi

Menkhoff

Urban

(2022). Financial education affects financial knowledge and downstream behaviors. Journal of Financial Economics, 145(2), 255–272. https://doi.org/10.1016/j.jfineco.2021.09.022

46.

Kaiser

Menkhoff

(2020). Financial education in schools: A meta-analysis of experimental studies. Economics of Education Review, 78, 1–15. https://doi.org/10.1016/j.econedurev.2019.101930

47.

^ Key

Tucker

J. N.

Grinstein-Weiss

Comer

(2015). Tax-time savings among low-income households in the $aveNYC program. Journal of Consumer Affairs, 49(3), 489–518. https://doi.org/10.1111/joca.12070

48.

King

(2022). Those who married once more likely than others to have retirement savings. U.S. Census. https://www.census.gov/library/stories/2022/01/women-more-likely-than-men-to-have-no-retirement-savings.html

49.

^ Knoll

Cooper

Johnson

Sieminski

Banker

(2019). Planning for tax-time savings: Innovation insights. Consumer Financial Protection Bureau. https://www.consumerfinance.gov/data-research/research-reports/planning-tax-time-savings/

50.

^ Kondratjeva

Roll

S. P.

Bufe

Grinstein-Weiss

(2020). Using financial tips to guide debt repayment: Experimental evidence from low-and moderate-income tax filers. Working paper. Social Policy Institute, Washington University in St. Louis. https://openscholarship.wustl.edu/cgi/viewcontent.cgi?article = 1024&context = spi_research.

51.

Kondratjeva

Roll

S. P.

Despard

Grinstein-Weiss

(2022). The impact of tax refund delays on the experience of hardship among lower-income households. Journal of Consumer Policy, 45, 239–280. https://doi.org/10.1007/s10603-021-09501-4

52.

Kugley

Wade

Thomas

Mahood

Jorgensen

A.-M.

Kammerstron

Sathe

(2017). Searching for studies: A guide to information retrieval for Campbell systematic reviews. The Campbell Collaboration. https://doi.org/10.4073/cmg.2016.1

53.

Lipsey

Wilson

D. B.

(2001). Practical meta-analysis. Sage.

54.

Loibl

Jones

Haisley

(2018). Testing strategies to increase saving in individual development account programs. Journal of Economic Psychology, 66, 45–63. https://doi.org/10.1016/j.joep.2018.04.002

55.

Lusardi

Keller

P. A.

Keller

(2009). Increasing the effectiveness of retirement saving programs for females and low-income employees: A marketing approach. National Endowment for Financial Education. https://www.nefe.org/Portals/0/WhatWeProvide/PrimaryResearch/PDF/DartmouthStudy.pdf

56.

Mcguinness

L. A.

Higgins

J. P. T.

(2021). Risk-of-bias VISualization (robvis): An R package and Shing web app for visualizing risk-of-bias assessments. Research Synthesis Methods, 12(1), 55–61. https://doi.org/10.1002/jrsm.1411

57.

Mendenhall

Edin

Crowley

Sykes

Tach

Kriz

Kling

J. R.

(2012). The role of earned income tax credit in the budgets of low-income households. Social Service Review, 86(3), 367–400. https://doi.org/10.1086/667972

58.

Miller

Reichelstein

Salas

Zia

(2015). Can you help someone become financially capable? A meta-analysis of the literature. The World Bank Research Observer, 30(2), 220–246. https://doi.org/10.1093/wbro/lkv009

59.

Mills

McKernan

S. M.

Ratcliffe

Edelstein

Pergamit

Braga

(2019). First-year impacts on savings and economic well-being from the assets for independence program randomized evaluation. Journal of Consumer Affairs, 53(3), 848–868. https://doi.org/10.1111/joca.12247

60.

^ Palmer

Pichot

Kunovskaya

(2016). Promoting savings at tax time through a video-based solution-focused brief coaching intervention. Journal of Financial Therapy, 7(1), 2. https://doi.org/10.4148/1944-9771.1103

61.

Richards

K. V.

Thyer

B. A.

(2011). Does individual development account participation help the poor? A review. Research on Social Work Practice, 21(3), 348–362. https://doi.org/10.1177/1049731510395609

62.

^ Roll

Bufe

Kondratjeva

Grinstein-Weiss

(2020). Promoting public retirement savings accounts during tax filing: Evidence from a field experiment. Working paper. Social Policy Institute, Washington University in St. Louis. https://socialpolicyinstitute.wustl.edu/promoting-public-retirement-savings-accounts-during-tax-filing/.

63.

^ Roll

Davison

Grinstein-Weiss

Despard

Bufe

(2018). Refund to Savings 2015-2016: Field experiments to promote tax-time saving in low-and moderate-income households. CSD Research Report No. 18-28. Center for Social Development, Washington University. https://openscholarship.wustl.edu/cgi/viewcontent.cgi?article = 1603&context = csd_research.

64.

^ Roll

Grinstein-Weiss

Despard

(2019). Long-term impacts of online tax-time savings interventions: Effects among persistently poor and resource-constrained households. Consumer Interests Annual, 65, 1–7. https://www.consumerinterests.org/assets/docs/CIA/CIA2019/RollCIA19.pdf

65.

^ Roll

Grinstein-Weiss

Gallagher

Cryder

(2020). Can pre-commitment increase savings deposits? Evidence from a tax-time field experiment. Journal of Economic Behavior & Organization, 180, 357–380. https://doi.org/10.1016/j.jebo.2020.10.011

66.

^ Roll

S. P.

Russell

B. D.

Perantie

D. C.

Grinstein-Weiss

(2019). Encouraging tax-time savings with a low-touch, large-scale intervention: Evidence from the refund to savings experiment. Journal of Consumer Affairs, 53(1), 87–125. https://doi.org/10.1111/joca.12194

67.

Schreiner

Sherraden

(2007). Can the poor save? Routledge. https://doi.org/10.4324/9781315081816

68.

Sherraden

M. S.

McBride

A. M.

Beverly

S. G.

(2010). Striving to save: Creating policies for financial security of low-income families. University of Michigan Press.

69.

Sykes

Križ

Edin

Halpern-Meekin

(2015). Dignity and dreams: What the earned income tax credit (EITC) means to low-income families. American Sociological Review, 80(2), 243–267. https://doi.org/10.1177/0003122414551552

70.

Theodos

Stacy

C. P.

Daniels

(2018). Client led coaching: A random assignment evaluation of the impacts of financial coaching programs. Journal of Economic Behavior & Organization, 155, 140–158. https://doi.org/10.1016/j.jebo.2018.08.019

71.

^ Tucker

J. N.

Key

C. C.

Grinstein-Weiss

(2014). The benefits of saving at tax time: Evidence from the $aveNYC evaluation. The Journal of Socio-Economics, 48, 50–61. https://doi.org/10.1016/j.socec.2013.08.011

72.

^ Tufano

(2011). Just keep my money! Supporting tax-time savings with US savings bonds. American Economic Journal: Economic Policy, 3(4), 172–200. https://doi.org/10.1257/pol.3.4.172

73.

Tufano

Schneider

Beverly

S. G.

(2005). Leveraging tax refunds to encourage savings. Brookings. https://www.brookings.edu/wpcontent/uploads/2016/07/08_leveraging_tax_refunds.pdf

74.

Viechtbauer

(2010). Conducting meta-analyses in R with the metafor package. Journal of Statistical Software, 36(3), 1–48. https://www.jstatsoft.org/v36/i03/

75.

Williams

Beal

(2020). Maintaining over $100 in savings linked to increased likelihood of financial stability in lower income households. FINRA and SaverLife. https://about.saverlife.org/research-posts/finra

76.

Wilson

D. B.

(n.d.). Practical meta-analysis effect size calculator [Online calculator]. https://campbellcollaboration.org/research-resources/effect-size-calculator.html.