This article provides estimates of the potential anti-poverty impacts of eight proposals presented in this double issue of RSF. Using the 2016 Annual Social and Economic Supplement to the Current Population Survey and the Census Bureau and Bureau of Labor Statistics’ Supplemental Poverty Measure, we first discuss the simulation approach taken for each proposal and then provide a consistent set of poverty estimates across proposals that include reductions in the poverty and deep poverty rates and the poverty gap; demographic differences; and net direct government costs. Anti-poverty impacts are largest for the most costly proposals, but less costly and more targeted proposals still have substantial potential impacts for key subgroups.

Keywords

poverty, microsimulation, policy

The articles in this double issue each answer the call and challenge laid out by the Russell Sage Foundation to “showcase a collection of innovative and specific policy proposals intended to reduce poverty in the short- and/or long-term or improve economic well-being.” Authors were given wide latitude to think outside the box in crafting proposals, and the results are striking in both their diversity and creativity. Given this diversity, it is potentially difficult to compare the proposals to assess how they differ in their impacts, costs, and scope. This article is designed to provide an admittedly crude first attempt to make such comparisons. Using a common dataset, the Census Bureau’s Current Population Survey, and measurement tool, the Supplemental Poverty Measure (SPM), this article provides, for eight of the policy proposals in this double issue, a set of estimates to consistently capture the impacts of each on poverty, deep poverty, and the poverty gap; demographic differences; and total direct costs to the government.

data and measures

To simulate the proposals, it is important to use a common starting dataset. Doing so provides a basis for having comparable estimates [End Page 167] across articles such that differences are derived from our simulation methods and not inconsistencies in the underlying data. We use the 2016 Annual Social and Economic Supplement to the Current Population Survey (CPS) as our starting dataset. This dataset is sometimes referred to as the March CPS.

The March CPS is collected every year (primarily in March), and collects detailed income information from a representative sample of households across the United States. Importantly for our purposes, it collects detailed information on multiple sources of income for each person residing in sampled households. The 2016 March CPS is the most current sample available as of this writing. The income data collected references the prior calendar year, in this case 2015. The March CPS is the dataset used by the Census Bureau to calculate annual levels and trends in American incomes and poverty rates, making it an ideal choice to serve as the basis of simulations of the effects of a variety of alternative policy scenarios on incomes and poverty. We use the IPUMS CPS 2015 national dataset (Flood et al. 2016).1 Note that the dataset we use contains only records of the income as reported by sample members, and thus will contain the substantial underreporting of many types of income that has been documented elsewhere (Meyer, Mok, and Sullivan 2009). This has the potential to affect some of the anti-poverty impacts and costs simulated.

To analyze the anti-poverty impacts of each proposal, we use the SPM of the U.S. Census and the Bureau of Labor Statistics, which is widely considered to be an improved measure of income poverty that builds upon the official measure of poverty (Fox et al. 2015; Wimer et al. 2016). The advantages of the SPM and problems with official statistics have been catalogued more extensively in this double issue and elsewhere so we do not go into extensive detail here (see Citro and Michael 1995; Blank 2008; Short 2011; Renwick and Fox 2016). In brief, however, the SPM accounts for resources that come after taxes, such as tax credits like the Earned Income Tax Credit (EITC) or Child Tax Credit (CTC), as well as income tax and payroll tax liabilities that reduce the amount of available income families have to pay for basic needs. The SPM also includes resources in the form of in-kind benefits or near-cash resources that are not accounted for under the official poverty measure. Because many of the proposals in this double issue put forward changes to tax and in-kind programs, as well as cash assistance programs and earnings, the SPM provides a useful measure to assess changes relative to the status quo.

Using the SPM, we are able to provide estimates of simulated proposals’ impacts on the poverty rate of the total U.S. population; the poverty rate of affected proposed program recipients, the deep poverty rates (that is, those falling below 50 percent of the poverty threshold), the reduction in the poverty gap (that is, the total amount of money necessary to lift everyone above the poverty line) for program recipients, the poverty reduction effect of proposed policies for different demographic groups, and the total direct governmental costs of each proposed policy relative to the status quo. To understand how the proposals articulated in the double issue would potentially alter official statistics, we also calculate some measures of poverty reduction under the official measure of poverty. We deliberately include a broad spectrum of metrics of poverty impact in order to capture different types of effects that policy proposals may have at the level of the total population and for direct policy beneficiaries. The change in the poverty rate, SPM and official, is presented as the most traditional measure, but we also include the poverty gap metrics to show changes in the depth of poverty and policies’ reach in terms of the number of individuals affected and differential effects for key demographic groups. Last, we show the average total benefit that recipients of each proposal would newly receive. This is defined at the family (SPM unit) level. That is, if a family had three children and each child received $2,000 in new resources, each child would have $6,000 new dollars coming into their family, and this would be the amount of the benefit that we show in our tables.

Despite the limitations to each of these estimates, which we discuss in the concluding [End Page 168] discussion, we argue here that using the March CPS and the SPM provides a common starting point for providing a consistent set of poverty estimates. We hope that these estimates will, in turn, help readers understand the trade-offs in considering a very diverse and differentially targeted set of anti-poverty proposals.

the simulated proposals

In this section, we outline the proposals that we simulated in our analyses and the approach we took for each. We did not simulate seven of the proposals, including those in which the policy was not designed to alter total family resources, for which the microdata provided no clear method for simulating an anti-poverty impact, or for which the assumptions required to do so were deemed too great. The eight proposals we did simulate were explicitly geared toward income poverty reduction and provided parameters that allowed for a consistent, plausible, and credible set of estimates on poverty reduction and cost. We briefly outline each proposal and its simulation. The proposals and simulations are detailed in the order they appear in this double issue. A summary of each proposal is also featured in table 1. Unless otherwise specified, we do not model macroeconomic or behavioral changes that may result from the proposals or affect their outcomes.

Cash Child Allowance (#1)

Luke Shaefer, Sophie Collyer, Greg Duncan, Kathryn Edin, Irwin Garfinkel, David Harris, Timothy Smeeding, Jane Waldfogel, Christopher Wimer, and Hiro Yoshikawa’s article proposes a universal monthly child allowance (2018). Under it, all families would receive a child allowance of $250 per month, per child for all children under age eighteen. The allowance would be taxed at the marginal tax rate of the parent(s) or guardian(s). The proposal also calls for the elimination of the Child Tax Credit, the Additional Child Tax Credit (ACTC), and the tax exemption for dependent children. To simulate this proposal, we provided all families with children under the age of eighteen an allowance of $250 per month per child ($3,000 per year). The allowance value was then “taxed back” at the parent’s marginal tax rate. We then added the net value of the child allowance for each SPM unit to their SPM unit’s resources. To account for the elimination of the CTC and ACTC, the total value of the CTC and ACTC received by each SPM unit, as reported in CPS data, was subtracted from its resources. To account for the elimination of the tax exemption for dependent children, we calculated the amount of additional federal tax that tax filers who received this tax exemption would owe if the exemption was eliminated. We determined this amount by constructing tax units using the CPS data, identifying the number of dependents under the age of eighteen that each filer claimed, calculating the value of each filer’s child exemption using the Internal Revenue Service (IRS) guidelines, and taxing their exemption at their marginal tax rate.2 The 2015 exemption for dependent children reported by the IRS was $4,000 per dependent child for all filers with adjusted gross incomes (AGI) below the 2015 exemption phase-out thresholds. The exemption decreased by 2 percent for each $2,500 in AGI a filer claimed in excess of their exemption phase-out threshold.3 The marginal [End Page 169]

Table 1. Summary of Simulated Proposals
Click for larger view
View full resolution
Table 1.

Summary of Simulated Proposals

[End Page 170]

tax rate values we used came from the CPS and were calculated by the Census Bureau using their tax model and the income values reported in the CPS. We then subtracted the total amount of tax owed on the child exemption by each SPM unit from their unit’s resources. Post-reform poverty was then measured using the SPM unit’s altered total resources and the SPM thresholds.

To evaluate the reform’s effect on poverty measured using the official measure, we added the total value of the unit’s pre-tax child allowance to their resources. These resources were not adjusted in any other way because official resources are measured before taxes.

Cash Child Allowance (#2)

Marianne Bitler, Annie Hines, and Marianne Page’s proposal (2018) guarantees a child allowance that would increase the cash resources of families with children. The allowance, set at $2,000 per year for all citizen children would, in part, be financed by eliminating the CTC, the ACTC, the tax exemption for dependent children, and the child portion of the EITC, that is, the part of the EITC calculated based on the number of dependent children that recipients claim. We took a similar approach to that taken with the Shaefer and colleagues child allowance proposal to simulate this proposal. To begin, we calculated each unit’s total allowance value based on the number of citizen children under the age of eighteen in the unit and the proposal’s allowance value of $2,000 per citizen child. The total allowance value for each SPM unit was then added to their total SPM resources. We then calculated the value of the each unit’s CTC, ACTC, and tax owed on the tax exemption for dependent children using the same method outlined in the explanation of the Shaefer and colleagues simulation. To simulate the suggested changes to the EITC, we excluded children under eighteen from the number of dependents that tax filers claimed and then recalculated the value of their EITC using the National Bureau of Economic Research’s TAXSIM tax calculator (for a description, see Feenberg and Coutts 1993). We then subtracted their original EITC value from their SPM unit’s resources and replaced it with the adjusted EITC value.

We determined the post-reform poverty rate using the SPM resources that took into account each unit’s child allowance, the elimination of the CTC, the ACTC, the tax owed on the unit’s child exemption(s), and the adjusted EITC. To measure the benefit’s impact on poverty under the official measure, the child allowance was added to total resources used under the official measure. Because the official poverty measure is measured pre-tax, we did not update the resources based on the proposed changes to the CTC, ACTC, the child exemption, or the EITC.

Minimum Benefit Plan for the Elderly Poor

Pamela Herd, Melissa Favreault, Madonna Harrington Meyer, and Timothy Smeeding propose a Social Security–based minimum benefit plan (MBP) that safeguards Social Security recipients against falling into poverty (2018). Under the plan, elderly Social Security recipients living in poverty receive additional benefits that bring their resources up to the level of the official poverty threshold for an elderly single individual or couple.

To measure the poverty effects of this proposal, we identified eligible recipients, defined as individuals age sixty-five and over that received Social Security payments and had an individual or joint income (if married) below the official elderly poverty line ($11,367 for individuals, $14,326 for couples in 2015). The income of other individuals living with the potential recipient(s) was not included when evaluating eligibility status. The value of the benefit was calculated as the difference between the official poverty line and their individual or joint income. For example, if a single recipient’s total income was $10,000, they would receive an additional benefit of $1,367; if a couple’s joint income was $10,000, together they would receive a benefit totaling $4,326.

To measure the benefit’s impact on poverty under the SPM, the benefit was added to the total resources of each recipient’s SPM units. Each unit’s total resources with the benefit were measured against their SPM threshold to determine whether it was below the SPM poverty line post-benefit. We followed the same steps using the resources and thresholds used [End Page 171] for official poverty statistics to measure the post-reform poverty under the official measure.

Note that the MBP benefit effectively brings poor recipients’ resources up to the official poverty threshold and therefore “out of poverty” if they live on their own or only with their spouse, according to the official measure.4 However, for recipients living in family units with other members besides a spouse, total family resources under the official poverty measure include the resources of all family members and the corresponding official poverty threshold for the family is larger than the threshold for a single person or couple. Thus, for these poor seniors, the MBP may not increase total family resources enough to lift the family over the official poverty threshold. SPM thresholds are often higher than the official poverty thresholds, both overall and in different geographic areas of the country, as described earlier, meaning that the MBP similarly may not increase family resources above the poverty threshold when calculating post-reform poverty statistics under the SPM.

Guaranteed Child Support Payment

Maria Cancian and Daniel R. Meyer’s proposal addresses child poverty by way of a government-provided guaranteed child support payment of $150 per month, per child (2018). Under their plan, all single custodial parents, including those that do not receive child support, are entitled to the guarantee. To simulate this proposal, we first increased the child support received by families with child support income less than the guarantee, as reported in the CPS, to $150 per month, per child. For example, if a parent in the March CPS reported $1,200 in child support income over the calendar year, our simulation would assume they already received $100 per month and we would add $600 to his or her annual income in the form of this newly proposed benefit.

We then assume that some single custodial parents not currently receiving any child support income would also benefit from the guaranteed payment. Following Cancian and Meyer, we assume that 12 percent of these parents would enroll in the program, so we randomly selected 12 percent of this group and allotted them $150 per month per child.

The additional child support that families received was then added to their SPM unit’s resources and measured against the SPM thresholds to determine the post-reform poverty rate. The child support was also added to the unit’s resources used in the official measure and compared to official thresholds.5

Additional Supplemental Nutrition Assistance Program Benefit Payments

Craig Gundersen, Brent Kreider, and John Pepper’s proposal (2018) provides all households receiving Supplemental Nutrition Assistance Program (SNAP) benefits with an additional benefit of $41.62 per week to lower household food insecurity (an important correlate of poverty). This is based on the fact that SNAP recipients report a monetary shortfall from their SNAP allotments relative to the money necessary to meet their routine food expenditures, and based on their analyses, the authors find that a little over $40 per week would be enough to meet most recipients’ budget shortfalls. To simulate the effect of this policy, we identified all SPM units with at least one SNAP recipient and added $2,164.24 ($41.62 × 52 weeks) to their total SPM unit resources (the value of the additional SNAP subsidy for one year). We then recalculated the poverty rate and associated metrics after the additional SNAP resources were included. We did not make any changes to resources when calculating poverty under the official measure because SNAP benefits are not included in the definition of total resources under this measure. [End Page 172]

Renter’s Tax Credit

Sara Kimberlin, Laura Tach, and Christopher Wimer propose a refundable renter’s tax credit, designed to address high housing cost burdens among low-income renters as housing costs are a primary driver of poverty under the SPM (2018). The credit is designed to reduce a tax filer’s rental housing cost burden to 40 percent of after-tax cash income, a conservative afford-ability level given that the U.S. Department of Housing and Urban Development (HUD) identifies a 30 percent or less housing cost burden as affordable, and 50 percent is considered severely housing burdened. Caps are applied to the amount of rent that can be claimed in order to target the credit to renters with the most need, and so tax filers cannot claim rent paid that exceeds an assigned fair market rent (FMR) value (determined by the population-weighted mean HUD FMR across metropolitan or non-metropolitan areas for the tax filer’s state). In addition, under the proposal, filers cannot claim rent paid that exceeds 80 percent of after-tax cash income. Tax filers already receiving a housing subsidy cannot claim the credit.

To simulate this policy, we used the methods proposed by the authors to calculate credit amounts. First, tax units were constructed in the CPS data, and total after-tax cash income for all tax unit members was calculated for the tax units living in rental housing using 2015 income and federal income tax or credit amounts from CPS data. Gross rent paid was imputed for each renter household via a regression model using coefficients derived from the American Community Survey (ACS) data, where rent is reported directly, and prorated to the tax unit. To calculate the caps on claimable rent paid, each tax unit was assigned an FMR cap based on the population-weighted mean two-bedroom HUD FMR across metropolitan or nonmetropolitan areas for the tax filer’s state. The FMR cap was adjusted for number of bedrooms by applying a multiplier, with the number of bedrooms for each tax unit determined by the number of tax dependents. The claimable share of rent paid was identified as the lowest of actual rent paid, the assigned FMR cap, or 80 percent of after-tax cash income. The renter’s tax credit was then calculated as the difference between capped rent paid and 40 percent of after-tax cash income. Tax filers reported in CPS data as housing subsidy recipients were assigned no credit. The credit amount was summed across tax units to the level of the SPM family unit. We then added the renter’s tax credit to family resources to calculate a revised SPM poverty status that was used to calculate the post-reform SPM poverty rate. There were no changes to official poverty post-reform because tax credits are not included in the definition of total resources under the official measure.

Increasing Federal Minimum Wage to $12.00 per Hour

Jennifer Romich and Heather Hill’s proposal raises the federal minimum wage to $12.00 per hour from the current $7.25 per hour (2018). In accounting for the costs and benefits of their proposal, they assume that the higher minimum wage will trigger some unemployment as employers eliminate some jobs given higher labor costs. They propose increasing employer tax credits available through the Work Opportunity Tax Credit by $1 billion (doubling the current allocation) in order to reduce the expected disemployment rate to 6.5 percent of workers earning less than the new minimum wage pre-reform. They also consider changes in payroll tax revenues, income tax revenues and expenditures, and SNAP outlays in accounting for the policy’s total government costs and savings.

To simulate the effect of this policy, we identified all workers earning less than $12 per hour, as calculated from annual wage income, weeks worked, and usual hours worked per week. We randomly assigned 6.5 percent of these workers to become unemployed with revised wage income of $0. The workers remaining were assigned new earnings by multiplying their weeks worked and usual hours per week by the new minimum wage of $12 per hour. We then calculated changes in these workers’ income and payroll taxes by constructing tax units in the CPS sample and recalculating their federal income tax liabilities and credits (including the EITC), and federal payroll taxes (FICA) based on new earnings for tax units that included workers with increased or decreased wages. These tax adjustments were calculated using [End Page 173] the NBER’s TAXSIM tax simulator (as described earlier). Finally, we simulated changes in SNAP benefits expected to result from implementation of the policy. For workers with increased wages who reported receiving SNAP, we followed the USDA’s SNAP benefit calculation formula and reduced annual SNAP benefits by $0.24 for every $1.00 in increased earnings, including the earned income deduction (for the formula, see CBPP 2017).6 For disemployed workers who reported receiving SNAP, we followed the same formula to increase annual SNAP benefits by $0.24 for every $1 in lost earnings. We also assumed that 85 percent of disemployed workers who did not report receiving SNAP and were not initially income-eligible for SNAP but became eligible due to lost earnings would newly enroll in SNAP. We assigned new SNAP participation randomly within this group and assigned new annual SNAP benefits based on the SNAP benefits calculation formula. These changes in SNAP resources were modeled explicitly given they are a key focus of Romich and Hill’s proposal.

After these simulations, we incorporated these changes in family resources from earnings, income tax liabilities and credits, payroll taxes, and SNAP into a revised calculation of family resources to calculate post-reform SPM poverty rates. The family resources were adjusted based only on changes in earnings to measure the post-reform poverty rate under the official measure. To calculate the net government cost for the policy, we summed the aggregate net changes in income tax revenues and expenditures, payroll tax revenues, SNAP outlays, and increased Work Opportunity Tax Credit (WOTC) expenditures. As we show, this procedure produced a net negative government cost (net government savings) associated with the policy. This is because increased tax revenues and decreased SNAP payments from workers with increased earnings exceeded increased tax expenditures for the WOTC and SNAP outlays for disemployed workers.

Guaranteed Minimum Wage Employment Program

Indivar Dutta-Gupta, Kali Grant, Julie Kersick, Dan Bloom, and Ajay Chaudry address the struggle many Americans face when not employed full-time by providing a framework for a national subsidized employment program (2018). The authors argue that subsidizing employment is a cost-effective way to improve the labor market outcomes and well-being of disadvantaged workers and to reduce poverty in the short and long run. The program they propose guarantees a thirty-five-hour per week job at participants’ state minimum wage for nine months with the opportunity to reapply for continued participation. Simulating the poverty effects of the program required defining the universe of eligible participants in the CPS data, randomly selecting program participants, estimating their length of program participation, and recalculating their total income with the addition of their new wages from the program.

To be eligible for the employment program, individuals had to be either: long-term unemployed (unemployed for sixteen or more weeks), working involuntarily part time for economic reasons, or marginally attached to the labor force (for definitions, see BLS 2016). In the CPS, marginally attached workers are defined as persons available for work who looked for work in the past twelve months, but not in the four weeks immediately preceding the survey. We identified the group of potential program participants based on self-reported length of unemployment, reasons for working part time, and the interest in securing a job among individuals that were marginally attached to the labor force.7 Working under the assumption that 20 percent of potential participants would enroll in the program, we randomly selected 20 percent of the universe of potential participants. After random selection, each “new” participant was then randomly assigned a number of weeks of participation. We assumed that the [End Page 174] length of program participation would be normally distributed and have a mean of twenty-six weeks (approximately six months). Participants’ potential length of participation was randomly assigned based on this distribution.

For participants who were unemployed or marginally attached to the labor force, we added the number of weeks of participation in the program to their total weeks actually worked for the year.8 The maximum number of weeks a participant who was long-term unemployed could work after their program weeks were added to their actual weeks worked was thirty-six weeks; for those marginally attached to the labor force, the maximum was forty-eight weeks.9 For these groups of participants, if the new total weeks when program weeks were included exceeded these maximums, we reduced the number of weeks of program participation so that their weeks worked for the year totaled thirty-six for the long-term unemployed and forty-eight for the marginally attached workers. We then calculated the additional income from the program using their state’s minimum wage rate and the number of weeks participants were employed by the program (that is, wage × 35 hours × program weeks) and added it to their annual earnings.10

A similar process was followed for those employed part-time for economic reasons. However, the maximum number of weeks in the year that participants from this group could be employed when program weeks were included with their actual weeks worked was fifty-two. When annual weeks worked for this group exceeded fifty-two (including program weeks), we had two approaches. First, if the weekly salary at the part-time job was greater than the weekly salary in the program, we reduced the number of weeks the participant spent in the program, Alternatively, if the weekly salary at the part-time job was less than the program’s weekly salary, we reduced the number of weeks the participant spent at the part-time job and replaced it with weeks they participated in the program. The implicit assumption here is that if the part-time job pays less than the new full-time minimum wage job, the worker would rather be in the full-time job (that is, the program). Conversely, if the part-time job pays more than the new full-time minimum wage job, the worker would rather be in the part-time job for as long as they were recorded as having that job in the CPS. Individual annual earnings were updated according to the number of weeks involuntary part-time workers spent in the program.

For all program participants, federal taxes and credits, state taxes and credits, and FICA payroll taxes were recalculated based on their adjusted earnings using the TAXSIM program described earlier. To measure the program’s impact on poverty under the SPM, participants’ reported annual earnings were subtracted from their SPM unit’s resources and replaced with their annual earnings including those from the program. Any additional taxes owed by and credits earned by participants were also accounted for in their unit’s total resources. Total SPM resources with these alterations were measured against the SPM thresholds to determine the post-reform poverty rates.

To measure the benefit’s impact on poverty under the official measure, the annual earnings of participants, including those from the program, were added to their unit’s total resources used under the official measure. Total resources were not adjusted based on additional taxes owed and credits received by participants because these resources are measured pre-tax and are not included in resources under the official poverty measure. [End Page 175]

results

Table 2 presents our topline simulation results for the eight articles included in the cross-proposal simulation. The first panel provides the total likely effects of each policy on the two groups: the total U.S. population, and all recipients of new resources provided by the policy. The second is an important population to examine given that the policies vary tremendously in how targeted they are at broad versus specific populations. Considering first the total U.S. population, we see that the policy proposals with the largest total impact on the poverty rate are those from Shaefer and his colleagues’ universal child allowance and Romich and Hill’s increased minimum wage. This makes sense because, aside from Bitler, Hines, and Page’s child allowance proposal, resources from these simulations reach the most people, at 158.5 million and 99.7 million, respectively. These proposals also provide some of the largest simulated benefits, as illustrated in the bottom row of table 2. The eight proposals range quite a bit in the average benefit that recipients would see, from approximately $1,000 in Bitler, Hines, and Page to more than $5,000 in Romich and Hill.

Not surprisingly, these proposals with the largest overall impact on the poverty rate also come with some of the largest price tags. The Bitler, Hines, and Page proposal, though also touching many recipients, has a much lower net cost than the Shaefer and colleagues proposal because it is more concerned with containing direct costs of the proposal than the Shaefer et al. proposal. Interestingly, the Romich and Hill proposal is the only proposal that shows a net savings in direct costs, though our estimates of direct costs are only the costs to government programs, and with a minimum wage proposal like Romich and Hill’s, other stakeholders (in this case businesses) would potentially bear a larger direct cost. It is important that many of the authors’ individual articles consider how these costs could be paid for. Mechanisms for financing these costs could, of course, alter the topline results presented here.

As noted earlier, however, the proposals vary widely in the size of the population targeted. By examining effects on recipients of the proposal, we generally see much larger effects. Take Cancian and Meyer’s proposal of a guaranteed child support benefit, for instance. Although it barely budges the total population poverty rate, the proposal is targeted specifically to custodial parents with low child support income and a subset of custodial parents without a child support agreement. For this latter group, the proposal reduces poverty by 22.5 percent. We find that the Dutta-Gupta and colleagues proposal to provide guaranteed jobs to discouraged, marginally attached, and involuntary part-time workers would reduce poverty overall negligibly, but by more than 40 percent among recipients. Again, some of the largest effects are found among those proposals that reach the most people or cost the most money. But we still find fairly substantial effects even among less costly programs.

The second panel of table 2 shows analogous results to the first panel, but with deep SPM poverty rates. The results are largely similar, so we do not dwell on them here. But it is worth noting that the Herd and colleagues MBP for seniors has a large effect on its targeted recipients, reducing poverty among (officially) poor elderly singles and couples by about two-thirds (from 22.5 percent to 7.3 percent). Alternatively, Romich and Hill’s minimum wage proposal has less of an effect on deep poverty among recipients than it does on total poverty, most likely because many workers affected originally earned enough to avoid deep poverty while not earning enough to place them above the poverty line.

The third panel reports an alternative statistic, the percent reduction in the poverty gap among recipients of resources stemming from each policy proposal. The largest effects are found among the Dutta-Gupta and colleagues job guarantee and the Herd and colleagues MBP for the elderly, but all of the proposals achieve sizable reductions in the poverty gap among their intended recipients.

We also show, in panel 4, the simulated effects of each proposal on official poverty rates. Of note here is that some proposals, particularly the Herd and colleagues MBP plan for (officially) poor seniors show much larger anti-poverty effects under the official measure than under the SPM. For the MBP plan, this makes sense, as eligibility for the plan is based on being [End Page 176] counted as officially in poverty, and the value of the benefit is explicitly tied to the depth of that poverty below the official poverty line. The Bitler, Hines, and Page child allowance proposal also shows a more sizable decline in official versus supplemental poverty. In this case, the difference stems from the fact that the proposed Cash for Kids resources come in the form of cash, savings coming mainly from tax credits not counted in the official measure but counted in the SPM. Two proposals, Gundersen, Kreider, and Pepper’s SNAP supplement and Kimberlin, Tach, and Wimer’s renter’s credit, are shown to have no effects on the official poverty rate because they concern forms of income that are not included as resources under the official poverty measure.

Table 2. Poverty Effects Across Proposals
Click for larger view
View full resolution
Table 2.

Poverty Effects Across Proposals

Tables 3 through 6 focus exclusively on poverty reductions using the SPM and the official measure. These tables, however, show the anti-poverty effects of each proposal by four different demographic characteristics: race-ethnicity, gender, age, and family structure. Each table [End Page 177] shows effects for the total population within each group.

Table 3. Poverty Effects by Race-Ethnicity
Click for larger view
View full resolution
Table 3.

Poverty Effects by Race-Ethnicity

We begin with race-ethnicity in table 3. The first panel shows the anti-poverty effects of each proposal on the poverty rate of the white, non-Hispanic population. As with the total population the largest anti-poverty impacts are seen for the larger proposals of Romich and Hill and Schaefer and colleagues. Similar results prevail for the official measure, though of course those based on tax or in-kind benefits by definition show no impacts here. The final column shows the poverty gap reduction but focused only on recipients. Some of the smaller and more narrowly focused proposals show more sizable anti-poverty effects. For example, Dutta-Gupta colleagues’ guaranteed job proposal reduces the poverty gap among its recipients fairly dramatically (65.3 percent) and Herd and colleagues’ MBP cuts the poverty gap by over half among its intended recipients.

The second panel of table 3 shows the results for black non-Hispanics. The results are largely consistent with those for white non-Hispanics, [End Page 178] but are systematically larger, which likely reflects the fact that this group has higher poverty rates in general. The results for Hispanics are again broadly similar despite a few key exceptions. First, both Romich and Hill’s minimum wage proposal and Shaefer and colleagues’ universal child allowance show somewhat larger effects on poverty rates for Hispanics than for black non-Hispanics, even though both groups begin the exercise showing fairly similar poverty rates. This is likely because both are designed to be universal, including no restriction by citizenship or immigration status. In contrast, the Hispanic poverty rate under the Bitler, Hines, and Page child allowance proposal would rise a bit, likely because of the fact that the allowance they propose is restricted to U.S. citizens, while some noncitizens receive resources from programs that would be replaced or eliminated.

Table 4. Poverty Effects by Gender
Click for larger view
View full resolution
Table 4.

Poverty Effects by Gender

Table 4 repeats the analysis but by gender. The top panel shows the results for males and the bottom for females. What is notable about table 4 is how few major differences there are in the effects of each proposal by gender. As is well known, women in general have higher poverty rates to start than their male counterparts do. But the proposals with the largest effects are very similar by gender and even the magnitudes of the anti-poverty impacts and reduction in the poverty gap are broadly similar.

Table 5 presents the results by age group, children under eighteen, adults eighteen to sixty-four, and adults sixty-five and older. Looking first at child poverty, the Shaefer and colleagues proposal again has the largest effects, which makes sense given that it presents a universal child allowance that would go to every child in the United States. Proposals focused on other groups, such as the MBP and guaranteed jobs programs, show less impact on child poverty, as expected. For adults eighteen through sixty-four, we see less dramatic reductions [End Page 179] than when we focus on children. For the poverty gap analysis, though, the Dutta-Gupta and colleagues proposal achieves substantial and large reductions in the poverty gap among the adults that benefit from the proposal. For adults sixty-five and older, the only program that explicitly targets this group (the MBP) of course shows the largest impact in absolute magnitude. But other proposals also make a bit of difference, including the renter’s credit and the SNAP supplement.

Table 5. Poverty Effects by Age Group
Click for larger view
View full resolution
Table 5.

Poverty Effects by Age Group

Last, in table 6, we examine differences by family structure. Everyone in a family was coded as belonging to a single, cohabiting, or married family by first looking for anyone married within a unit, secondarily by looking for anyone cohabiting with a romantic partner, and lastly by finding those where no one was married or cohabiting within a unit. Looking first at people in single families, we see that the Shaefer and colleagues proposal has the largest effect on both the poverty rate and the Dutta-Gupta and colleagues proposal results in the largest poverty gap reduction for direct beneficiaries. [End Page 180] This is also true for those in married or cohabiting families. The Romich and Hill proposal generally comes next for all groups in the magnitude of the anti-poverty rate reduction. Those in single families have the highest starting poverty rates and those in married families have the lowest, so the absolute magnitude of the drop in poverty tends to be somewhat larger in the single group, though this is not universally the case, as sometimes the drops are quite large in cohabiting families. Note that the poverty gap reduction for recipients is substantial across virtually all the policies once restricted to the targeted recipients.

Table 6. Poverty Effects by Family Structure
Click for larger view
View full resolution
Table 6.

Poverty Effects by Family Structure

Finally, although some policies are explicitly targeted to the demographic subgroups we examined (such as child allowances targeted to children), other policies are targeted to more specific demographic groups that are not captured by the subgroup categories we examined—such as families eligible for child support (Cancian and Meyer) or renters with high rent burdens (Kimberlin, Tach, and Wimer), and we do not compare how the full set of policies [End Page 181] would affect these more specific demographic groups.

A final note of comparison is that the proposals vary substantially in the initial poverty rate and deep poverty rate of their beneficiaries. Some proposals are designed to be universal, and most are targeted based on criteria not directly linked to poverty status, so for some proposals as much as 83.9 percent of beneficiaries are already above the poverty line before receiving any benefits from the new policy. This means that a larger share of total policy benefits and costs are directed to individuals who are already nonpoor than with policies more narrowly targeted to individuals in poverty, potentially reducing the efficiency of anti-poverty impact. Other policy goals can be met by serving nonpoor beneficiaries, however.

discussion

This article attempts to provide a consistently estimated set of simulations using the most recently available data for as many articles in the double issue as possible. In total, we conducted simulations for eight proposals, providing anti-poverty impact estimates for both the SPM and the official measures, and for poverty rates, deep poverty rates, and the overall poverty gap. One key finding is that the policies simulated vary tremendously in their scope and reach, populations targeted, and cost. Thus we also made it a point of the analyses to look at anti-poverty metrics for the specific population of “recipients” that would receive new resources from the proposed policy. Finally, we considered and presented differences by four key demographic characteristics: race-ethnicity, gender, age, and family structure.

As one would expect, the larger, costlier proposals tended to show the largest aggregate impacts. In most cases, these would be governmental costs, with the exception of Romich and Hill’s minimum wage plan, which would direct the new costs to employers. It is worth emphasizing two points on costs. First, we model only what we call the direct costs of what the authors put in their primary proposal. Many authors present alternative policy scenarios, some of which would have very different cost structures than those presented. In addition, some authors focus explicitly on containing costs whereas others discuss the numerous ways one might finance the direct costs presented here. Those choices about how to pay for these direct costs could obviously shape the poverty estimates presented here. The results for the Bitler, Hines, and Page child allowance proposal are instructive on this point. These authors go to great lengths to figure out how to contain costs and pay for their proposal. As a result, the proposal often shows smaller anti-poverty impacts than other proposals that have not fully grappled with the same cost challenge. Fuller consideration of how these costs would be paid for would allow us to make more exact comparisons of the relative pros and cons of various approaches and how these affect the distribution of resources across the total population.

Our approach and resulting estimates have several limitations. First, as noted earlier, our underlying dataset is not corrected for the underreporting of income or benefits. Our estimates may therefore not be fully accurate, though this bias would likely still evince similar patterns across proposals if incomes and benefits were fully reported. Second, our simulations generally do not model behavioral responses, changes in the macro-economy that might result from a proposed policy and thus alter poverty rates in unexpected ways, or cross-program interactions. This is not always the case because in some instances the authors explicitly built such factors and assumptions into their proposal. Third, many of the policy proposals simulated here would provide new resources not only to the poor but also to the near poor and in some cases even the well off. Our focus on the anti-poverty effects and reductions in the poverty gap do not account for these important impacts. Many of the limitations are discussed in more detail in the individual articles in this volume.

Christopher Wimer

Christopher Wimer is co-director of the Center on Poverty and Social Policy at Columbia University.

Sophie Collyer

Sophie Collyer is a research analyst at the Center on Poverty and Social Policy at Columbia University.

Sara Kimberlin

Sara Kimberlin is an affiliate at the Stanford Center on Poverty and Inequality and a senior policy analyst at the California Budget & Policy Center.

Direct correspondence to: Christopher Wimer at cw2727@columbia.edu, 1255 Amsterdam Ave., New York, NY 10028.

references

Bitler, Marianne P., Annie Laurie Hines, and Marianne Page. 2018. “Cash for Kids.” RSF: The Russell Sage Foundation Journal of the Social Sciences 4(2): 43–73. DOI: 10.7758/RSF.2018.4.2.03.
Blank, Rebecca M. 2008. “Presidential Address: How [End Page 182] to Improve Poverty Measurement in the United States.” Journal of Policy Analysis and Management 27(2): 233–54.
Bureau of Labor Statistics (BLS). 2016. “Glossary.” Washington: U.S. Department of Labor. Last modified June 7, 2016. Accessed October 19, 2017. https://www.bls.gov/bls/glossary.htm#.
Burnett, Jennifer. 2014. “Minimum Wage Watch 2015.” Washington, D.C.: The Council of State Governments. Accessed October 19, 2017. http://knowledgecenter.csg.org/kc/content/minimum-wage-watch-2015-0.
Cancian, Maria, and Daniel R. Meyer. 2018. “Reforming Policy for Single-Parent Families to Reduce Child Poverty.” RSF: The Russell Sage Foundation Journal of the Social Sciences 4(2): 91–112. DOI: 10.7758/RSF.2018.4.2.05.
Center on Budget and Policy Priorities (CBPP). 2017. “A Quick Guide to SNAP Eligibility and Benefits.” Washington, D.C.: CBPP. Accessed October 18, 2017. https://www.cbpp.org/research/a-quick-guide-to-snap-eligibility-and-benefits.
Citro, Constance F., and Robert T. Michael. 1995. “Measuring Poverty: A New Approach.” Washington, D.C.: National Academies Press. Accessed October 18, 2017. http://www.nap.edu/catalog/4759.html.
Dutta-Gupta, Indivar, Kali Grant, Julie Kerksick, Dan Bloom, and Ajay Chaudry. 2018. “Working to Reduce Poverty: A National Subsidized Employment Proposal.” RSF: The Russell Sage Foundation Journal of the Social Sciences 4(3): 64–83. DOI: 10.7758/RSF.2018.4.3.04.
Feenberg, Daniel, and Elisabeth Coutts. 1993. “An Introduction to the TAXSIM Model.” Journal of Policy Analysis and Management 12(1): 189–94.
Flood, Sarah, Miriam King, Steven Ruggles, and J. Robert Warren. 2016. Integrated Public Use Microdata Series, Current Population Survey: Version 4.0. [Data file]. DOI: 10.18128/D030.V4.0.
Fox, Liana, Christopher Wimer, Irwin Garfinkel, Neeraj Kaushal, and Jane Waldfogel. 2015. “Waging War on Poverty: Poverty Trends Using a Historical Supplemental Poverty Measure.” Journal of Policy Analysis and Management 34(3): 567–92.
Gundersen, Craig, Brent Kreider, and John V. Pepper. 2018. “Reconstructing the Supplemental Nutrition Assistance Program to More Effectively Alleviate Food Insecurity in the United States.” RSF: The Russell Sage Foundation Journal of the Social Sciences 4(2): 113–30. DOI: 10.7758/RSF.2018.4.2.06.
Herd, Pamela, Melissa Favreault, Madonna Harrington Meyer, and Timothy M. Smeeding. 2018. “A Targeted Minimum Benefit Plan: A New Proposal to Reduce Poverty Among Older Social Security Recipients.” RSF: The Russell Sage Foundation Journal of the Social Sciences 4(2): 74–90. DOI: 10.7758/RSF.2018.4.2.04.
Kimberlin, Sara, Laura Tach, and Christopher Wimer. 2018. “A Renter’s Tax Credit to Curtail the Affordable Housing Crisis.” RSF: The Russell Sage Foundation Journal of the Social Sciences 4(2): 131–60. DOI: 10.7758/RSF.2018.4.2.07.
Meyer, Bruce D., Wallace K. Mok, and James X. Sullivan. 2009. “The Under-Reporting of Transfers in Household Surveys: Its Nature and Consequences.” NBER working paper no. w15181. Cambridge, Mass.: National Bureau of Economic Research.
Renwick, Trudi, and Liana Fox. 2016. “The Supplemental Poverty Measure: 2015.” Current Population Reports, series P60, no. 258. Washington: U.S. Census Bureau.
Romich, Jennifer, and Heather D. Hill. 2018. “Coupling a Federal Minimum Wage Hike with Public Investments to Make Work Pay and Reduce Poverty.” RSF: The Russell Sage Foundation Journal of the Social Sciences 4(3): 22–43. DOI: 10.7758/RSF.2018.4.3.02.
Shaefer, H. Luke, Sophie Collyer, Greg Duncan, Kathryn Edin, Irwin Garfinkel, David Harris, Timothy M. Smeeding, Jane Waldfogel, Christopher Wimer, and Hirokazu Yoshikawa. 2018. “A Universal Child Allowance: A Plan to Reduce Poverty and Income Instability Among Children in the United States.” RSF: The Russell Sage Foundation Journal of the Social Sciences 4(2): 22–42. DOI: 10.7758/RSF.2018.4.2.02.
Short, Kathleen. 2011. “The Research Supplemental Poverty Measure: 2010.” Current Population Reports, series P60, no. 241. Washington: U.S. Census Bureau.
Wimer, Christopher, Liana Fox, Irwin Garfinkel, Neeraj Kaushal, and Jane Waldfogel. 2016. “Progress on Poverty? New Estimates of Historical Trends Using an Anchored Supplemental Poverty Measure.” Demography 53(4): 1207–18. [End Page 183]

Footnotes

2. In some cases, including the exemption value in a filer’s AGI might increase their marginal tax rate. We did not make this adjustment in this simulation, meaning the cost saving from eliminating the child exemption might be underestimated. We constructed tax units using a modified version of Stata code created by Judith Scott-Clayton, which is available on the TAXSIM website of the National Bureau of Economic Research (http://users.nber.org/~taxsim, accessed November 17, 2017).

3. We used each filer’s AGI and the number of dependents under eighteen they claimed, as reported in the CPS, to calculate their exemption value. In 2015, the exemption phase-out thresholds began at $258,250 and ended at $380,750 for single filers, $390,900 and $432,400 for joint filers, and $284,050 and $406,550 for heads of household. Filers with AGIs greater than the upper bound of the phase-out thresholds did not receive exemptions for dependent children. For a description of the exemption for dependent children and exemption phase-outs for 2015, see the IRS’s “Publication 501(2015): Exemptions, Standard Deduction, and Filing Information” (https://www.irs.gov/pub/irs-prior/p501--2015.pdf, accessed November 17, 2017).

4. The official poverty threshold for couples differs depending on the age of seniors in the couple. For couples where the householder is sixty-four years old or younger and his or her spouse is over sixty-four, the official poverty threshold is $15,871, so under this simulation, these families would not be brought up to the official poverty line with the additional benefit. To bring this couple to the poverty line, their MBP benefit amount would need to be calculated using the poverty line of $15,871.

5. We did not make any adjustments to the unit resources of non-custodial parents responsible for paying child support as we cannot link noncustodial parents to the children for whom they owe support.

6. We ignore other allowable deductions for dependent care, child support, medical expenses, and excess shelter expenses.

7. If an involuntary part-time worker worked for fifty-two weeks and their take-home pay was greater than their potential take-home pay from a thirty-five-hour per week job at their state’s minimum wage, they were excluded from the group of potential participants.

8. The number of weeks they participated was based on the normal distribution detailed above.

9. The long-term unemployed needed to be unemployed for sixteen weeks of the year to qualify for the program. The maximum number of weeks they could work needed to account for their sixteen weeks of unemployment. To qualify as a marginally attached worker, one would need to have not looked for work for four weeks, which also needed to be accounted for in total weeks worked when program weeks were included.

10. Because state minimum wage rates are not reported in the CPS data, we merged a state minimum wage rate dataset onto our CPS dataset (Burnett 2014). The 2015 state-level minimum wages were merged onto the CPS data such that if a worker resided in a higher minimum wage state, that worker would be assigned a higher wage in the simulation.

Additional Information

ISSN
2377-8261
Print ISSN
2377-8253
Pages
167-183
Launched on MUSE
2018-03-09
Open Access
Yes
Back To Top

This website uses cookies to ensure you get the best experience on our website. Without cookies your experience may not be seamless.