Means-Tested Transfers, Asset Limits, and Universal Basic Income

Inequality and Transfers: How the Means Are Distributed

To understand how means-testing for different programs affects households, what the main financial characteristics of those subject to testing are, and how inequality plays a role in this context, we start by documenting how households’ means are distributed along the income distribution. To do so, we use data obtained from the 2018 panel of the SIPP. The SIPP is a nationally representative survey administered by the US Census Bureau and focuses on transfer programs, which makes it particularly well-suited for our analysis. Furthermore, the SIPP has detailed data on wealth, with a granular breakdown of households’ portfolios.⁸

Our approach throughout the analysis is as follows: (i) we show data for annual values at the household level, that is, within a household potentially shared by multiple people; (ii) we order our data by households’ total income and allocate these data into the quantiles of its distribution; (iii) at each bin of this distribution, we categorize households into recipients and nonrecipients.

Figure 1 shows the distribution of total household wealth as defined by net worth in our data. Household wealth is distributed along the quintiles of the income distribution. That is, the first quintile, in which households have an annual income below $23,700, represents the bottom 20 percent of the income distribution, the second quintile, in which households have annual income between $23,701 and $46,500, represents households that are in the bottom 21 percent through 40 percent of the income distribution, and so on up to 100 percent. It is important to note that these income quintiles reflect all household sizes and locations, and these could consist of, for example, a family of three living in a small town in the southern United States or a family of five or more living in New York City. At every quintile, the average accumulated wealth of households that are benefit recipients is substantially lower than that of households that are nonrecipients. This fact suggests that nonrecipient households with the same level of household income as benefit recipient households are better able to build wealth in the longer term.

We observe that at the bottom 20 percent of the household income distribution, nonrecipients have nine times the wealth of recipients, whose estimated average net worth is $13,528. This value more than doubles at the second quintile, with the average net worth of recipients being $33,932, which is 7 times smaller than the average for nonrecipients. The gap narrows as household income increases, but it remains large. In the third quintile, for example, where household income is between $46,501 and $76,400, this gap remains significant: households that do not receive benefits have 3.3 times the average wealth of recipient households. Furthermore, almost all recipients—approximately 91 percent of them—are, as expected, in the bottom three quintiles of the income distribution. Of those recipients in the bottom three quintiles, the vast majority (88 percent) have household incomes below $65,000.⁹

We now turn to depicting how transfers are distributed across the income distribution and which programs are most relevant for the households in our sample. Figure 2 shows the total annual transfers received per household as a percentage of the average total annual income of recipients of the decile in which that household is located. ¹⁰ One of the advantages of the SIPP is that we can directly compute the annual dollar value received by households of all the programs analyzed excepting the EITC because the EITC is a tax credit; however, the survey also provides enough information to allow us to overcome the EITC’s absence and estimate the annual dollar value per household. ¹¹

We can observe in Figure 2 that the lowest-income households receive, on average, more than 60 percent of their income from transfer benefits and that this reliance upon transfer programs decreases as income increases, particularly beyond median income levels (about $60,000 annually). The graph also highlights the sharp relevance of both SNAP and SSI as the main sources of transfer income at all deciles but especially at the bottom 30 percent (below $34,600 annually), thus emphasizing their role as the main programs targeted to poverty alleviation inside the social safety net available to households. We can also see the diminished role that TANF plays relative to the other transfer programs, something that is consistent with the historical decline in relative federal spending on TANF.¹² Finally, another important fact is that the EITC increases its relative importance as we move upward in income distribution from the bottom to the second and third deciles, emphasizing its role as a wage subsidy that by itself is more relevant for working lower- and middle-income households.

Means-Tested Transfer Recipients: Resource Distribution and Portfolio Composition

The potential critical role of asset-testing and the compressed distribution of household wealth depicted in Figure 1 warrants taking a deeper look into households’ balance sheets. In particular, we want to understand how assets that are subject to means-testing in the transfers programs are distributed. Since we have so far shown wealth as total household net worth in our data, we start by restricting our measure of wealth into a narrower category and creating a variable that, following the usual program nomenclature, we call “resources.” For this variable, we select assets that are broadly subject to means-testing by the different programs studied such as the amount of cash in checking and savings accounts, retirement funds, vehicles, real estate that is not a primary residence, and so on.¹³

In Figure 3, we show how such resources are distributed across the income distribution. For a clearer view, we show only the bottom half of the distribution, and we subdivide it into deciles. We also add lines for the highest asset level tested for SNAP, the highest asset level for TANF, and a resource threshold below which households are generally considered to be in “asset poverty.”¹⁴

We find that, once again, households that are receiving means-tested transfers exhibit, at all deciles, a significantly lower average level of resources than those of nonrecipient households. At the smallest difference, households not receiving benefits have 4.7 times the dollar value of resources than those households that do receive benefits. Furthermore, we can see that at the lowest decile, transfer recipients have an average level of resources of slightly less than $4,000. These households are within asset poverty and hold resources below the cutoff to qualify for TANF but above the cutoff to qualify for SNAP. Entering the second decile, on average, benefit-recipient resources hover slightly above $7,000, escaping asset poverty but also exiting eligibility for most transfer programs.

Figure 4 offers a breakdown of the portfolio composition of households’ assets into all categories of our data except for primary residence and debt. Maintaining our focus on households for which testing constraints have a higher probability of being binding, we restrict our attention to the bottom 40% of the income distribution. We find that recipients allocate a larger part of their assets into categories that are less liquid and not as frequently tested by programs as cash. In particular, the allocation of a larger share of recipients’ wealth into retirement accounts or business equity can be related to the less-strict testing of such asset types in the rules of transfer programs. For instance, there is substantial diversity in the way states test and exempt vehicle values in both SNAP and TANF, a fact widely studied and documented by the literature (Gehr, 2018).

The Role of Asset-Testing

While there may be compelling reasons for a transfer system to be means-tested, such as efficiently distributing taxpayer-funded assistance to those most in need, researchers have identified potential reasons to believe that such tests are related to the differences in wealth within the individual quintiles we have observed in our data. Overall, studies have found a positive but not necessarily strong relationship between limits on liquid assets and low-income household assets; for some cases, as limits increase, so, too, do recipients’ liquid savings. Chen and Lermann (2005) provide an excellent review of the early empirical literature that finds that asset limits have an overall effect of lowering the net worth of potentially eligible low-income individuals but with different sizes and significance of those effects depending on the program that is analyzed.

More recent analyses have identified other mechanisms through which asset limits impact households’ choices beyond the limited impact on families’ savings levels. For instance, relaxing or eliminating limits does not increase program enrollment or participation (PEW Charitable Trusts, 2017). In states that relaxed SNAP asset limits through broad-based categorical eligibility (BBCE) policies, benefit-recipient households became more likely to hold bank accounts (and it became more likely for those accounts to be funded) and less likely to cycle on and off benefit programs. Also, without BBCE policies, 16 percent of households eligible to receive SNAP benefits would become ineligible because of the federal asset test (Ratcliffe, McKernan, Wheaton, and Kalish, 2016).¹⁵

The literature has historically found more robust evidence of the effects of asset-testing for the case of vehicle limits. When these asset limits are relaxed or eliminated, there were associated increases in vehicle ownership and the number of vehicles owned, but not of vehicle equity for an individual or household (Hurst and Ziliak, 2006; Sullivan, 2006; Pirog et al., 2017). Overall, the evidence suggests that strict asset limits may offer a disincentive to low-income households from establishing a more stable financial position. Instability in financial position poses hardships for many families and makes it more likely that they remain on benefit programs for extended periods of time.

Universal Basic Income: An Alternative to Asset Means-Testing?

The effects of means-testing on households’ behavior can also have broader consequences for macroeconomic inequality and households’ welfare, as shown in our documentation of the disparities between recipients and nonrecipients along the income distribution. Evidence of these disparities raises questions as to whether alternative designs of the transfers system that depart from the means-testing paradigm, such as a UBI, could improve overall economic welfare. An important component of the distributional effects of means-testing is the role of precautionary savings, that is, savings for an emergency. Precautionary savings can be discouraged by the presence of transfers focused on the bottom of the income distribution. In this case, asset-testing can add an extra effect in a household’s consumption and savings decisions to frontload the former by diminishing the latter because households are confronted with the potential loss of benefits prompted by crossing testing thresholds (Hubbard, Skinner, and Zeldes, 1995). For example, in the case of Medicaid’s asset test, Pashchenko and Porapakkarm (2016) find that asset tests that differ depending on the work status of households, with stricter limits to nonworkers, can generate fewer distortionary effects.

In a structural model of the economy, Wellschmied (2021) shows that removing all types of asset means-testing in transfer programs decreases by half the share of households that have low wealth and become jobless. The elimination of means-testing leads low-income households to accumulate more savings, an action which improves a household’s financial stability and reduces the potential vulnerability of the nonemployed. Such elimination of means-testing in transfer programs is in similar spirit to a UBI. According to the theoretical model, this reform increases the overall welfare of households. But the reform also means that in the absence of a test, a fixed budget for government spending on transfers decreases the dollar value of transfer per household as transfers are paid to more families. This issue is especially relevant when one expands the analysis to depend on the family composition of households. In another theoretical model of the economy, Ortigueira and Siassi (2022) find that the current system has large distortionary effects on the decisions of working households with children. The authors assess that the main distortions do not stem from the EITC, given the nonbinding nature of the investment income limit, but mainly from SNAP and TANF, further finding that asset-testing thresholds will induce single parents that are dependent on such benefit programs to allocate less money toward savings. Accordingly, these households accumulate less wealth, making them more financially vulnerable.

The idea of reforming the current system of income transfers toward a UBI program is tackled in a recent working paper by Luduvice (2021).¹⁶ Using a structural model of the economy, the author conducts counterfactual exercises in which the income security system, such as that in place in the United States, is replaced by a UBI that is funded by a federal tax on consumer spending. In this analysis, a nationwide budget-neutral reform toward a UBI leads to increased savings by households. But assuming that total spending on transfers is held fixed from the original set of programs to the move to the UBI, the effective transfer per household declines because the same total amount of money is distributed across more households. In turn, the new program provides less insurance to lower-income households than the old program. As a result, under this treatment, the UBI actually reduces the welfare of poorer households rather than making them better off. If, instead, the budget for transfer payments is increased so that the UBI transfer amounts to $12,000 per year—a value commonly referred to in the public policy debate—the result is the converse: Poorer households are made better off by the UBI. However, the federal tax rate on consumption needed to sustain the high level of transfer needs to be substantially increased. The increased tax rate and the high transfer combined with less need to save for a rainy day and less need to work decrease hours worked, labor force participation, and aggregate capital, thus yielding a drop in output. Despite that, the redistribution effect and the consumption floor generated by the UBI improves overall household welfare in the model, especially for poorer households.

Conclusion

In this Commentary, we have documented that means-tested program recipients have a smaller average net worth than nonbeneficiaries at every quintile of the income distribution. We also observe that transfers are substantially more relevant at the bottom of the income distribution, and we identify which of the programs studied plays a more pronounced role of providing economic cushion for benefit recipients. We find that resources subject to testing are unevenly distributed between recipients and nonrecipients and that participants at the very bottom of the income distribution are in asset poverty, with assets surrounding testing thresholds. Finally, we observe that transfer beneficiaries have a relatively larger share of their portfolios allocated toward assets less prone to being tested when compared to nonrecipients. Recent research suggests that the elimination of testing limits, such as in policies similar to a UBI, could present a welfare-improving alternative to the current system, though not without economic trade-offs that broadly impact macroeconomic aggregates and transfer levels.