Most Workers in Low-Wage Labor Market Work Substantial Hours, in Volatile Jobs
July 24, 2018
State and federal policymakers are considering, or have already imposed, work requirements that would take away SNAP (formerly food stamp) benefits or Medicaid coverage from people who are not working or engaged in work-related activities for a required number of hours each month. A large share of the people who would face work requirements are employed, but they also experience high levels of job displacement and unemployment, and their wages have grown little, as our analysis of the low-wage labor market shows. An understanding of the labor market that SNAP and Medicaid beneficiaries face makes clear that it will be difficult for many individuals and families to meet proposed work requirements. Depending on the details of the work requirements, it also may be quite administratively burdensome to monitor them. An understanding of the labor market that SNAP and Medicaid beneficiaries face makes clear that it will be difficult for many individuals and families to meet proposed work requirements.
When an individual cannot work, there are not enough jobs available, or the jobs that are available pay too little, programs like SNAP and Medicaid help ensure that people have adequate access to food and medical care. SNAP also helps the broader economy by providing a counter-cyclical stimulus that brings a quicker economic recovery in times of recession.
Given the current policy debate about imposing rigid work requirements in order to receive food or medical insurance assistance, it is important to understand the characteristics of those who participate in assistance programs, and the nature of the labor market facing those individuals who are most likely to need assistance from such programs. Proposals for work requirements in SNAP, Medicaid, and other assistance programs vary in terms of who is covered, how many hours of work are required, and the economic conditions under which the requirements may be temporarily waived. This report does not aim to predict the impact of any specific policy proposal, but rather to describe employment outcomes for those individuals around whom current policy discussions about work requirements are largely centered: non-disabled adults who do not have young children (under the age of 6) at home.
Using Current Population Survey data, we have analyzed the labor market facing individuals who are most likely to need access to SNAP or Medicaid benefits, including those with low education levels, particularly women, and workers in particular occupations. (For more on our methodology, see the box below and the Technical Appendix.)
As Figure 1 shows, in 2017, SNAP or Medicaid beneficiaries who are not currently engaged in substantial work are disproportionately likely to have an education of only a high school degree or less, and are likelier to be women with a high school degree or less. A substantial share also report a disability: almost a quarter (23.5 percent) of these SNAP or Medicaid participants who do not engage in substantial work report a work-limiting disability (not shown in the chart), even though this group is limited to those not receiving disability assistance.
Notably, SNAP or Medicaid beneficiaries report different occupations from those with middle-class incomes. Our approach is to analyze the labor market outcomes for people who have similar measurable characteristics and report similar occupations as SNAP and Medicaid participants.
Our analysis shows that most people with low education levels are working, and most of those are working a substantial amount. However, their earnings are low. Their employment prospects from one year to the next are volatile. Among those who work a substantial amount in one year, the next year is likely to be worse, as they revert to the mean outcome among workers with low levels of education. A closer look at the types of occupations disproportionately reported by SNAP or Medicaid beneficiaries shows that workers in these occupations face lower wages, less wage growth, and more volatility, measured in a variety of ways, than those who report “middle-class” occupations.
In particular, we find:
- Even though the majority of adults with low levels of education worked a substantial amount, their annual incomes from wages were low and grew almost not at all between 2002 and 2017. In 2017, 9.5 percent of women with a high school degree or less who worked at least 30 hours per week for at least 20 weeks per year lived in households where total income from wages was below the poverty line.
- Following the same workers from one year to the next, we find that working substantially in one year is no guarantee that one will accumulate work experience and skills that lead to increased work or wages in the next year.
- The most common occupations reported by SNAP or Medicaid recipients include nursing aides, orderlies, and attendants; cashiers; cooks; and retail sales clerks.
- Real median annual incomes from wages for people in occupations common among SNAP or Medicaid recipients did not grow between 2002 and 2017, after adjusting for inflation. Total hours worked were lower in 2017 than 2002.
- Among those in common SNAP or Medicaid occupations, unemployment rates are higher; the share stably employed from one year to the next is lower; the share that lost or left a job in the last three years due to plant or company closing/moved, position abolished, insufficient work, or other similar reason is higher; and job tenure is lower than those in middle-class occupations.
Together, these results suggest that it will be difficult for individuals who work and participate in benefits programs to meet proposed work requirements in the private sector alone. Although employment levels are high among many of these types of workers, employment volatility is also quite high. Much of this volatility reflects characteristics of these types of occupations and is not necessarily due to decisions made by the workers. Since wage growth also has been stagnant in these jobs, an increasing share of families have combined work with participation in SNAP and/or Medicaid to make ends meet.
We first used Current Population Survey (CPS)* data to investigate characteristics of those most likely to be affected by work requirements: non-disabled SNAP and Medicaid beneficiaries, without young children, who do not exhibit substantial labor market attachment. For these purposes, we define “substantial work” as usually working at least 30 hours per week for at least 20 weeks in the previous year, and “non-disabled” as not currently receiving Social Security or Supplemental Security Income for reason of disability.
We next examined labor market outcomes among individuals who belong to these groups. Specifically, we examined labor market outcomes for everyone, for those with a high school degree or less education, for women with a high school degree or less, and for those with a self-reported work-limiting disability. We find that there are high levels of employment among these groups, and among workers the majority are employed at least 30 hours per week, 20 weeks per year, or both. Nonetheless, for these workers, wages are low and have been stagnant over time. Even among those who are substantially employed, 1 out of every 10 does not earn enough to lift their family above the poverty line. Earnings among these groups are highly variable from year to year.
We then switched to examining outcomes by occupation group. We considered the top ten occupations reported by SNAP or Medicaid recipients and the top ten occupations among those that had annual incomes from wages at the median or above, and considered labor market outcomes among all individuals who reported these as their occupations.
People who are employed at the time of the CPS survey are asked to report their occupation during the preceding week. If respondents held more than one job, they are to report the occupation in which they worked the most hours. People who are unemployed or not in the labor force are asked to report their most recent occupation.
We find that occupations typically reported by SNAP or Medicaid recipients are low paid, have exhibited little or no recent wage growth, and are volatile across a number of metrics.
Throughout the analysis, we used data from the 2002-2017 Current Population Surveys. We limited the sample to 18- to 64-year-old U.S. citizens, who are not active-duty military, and who do not have children under the age of 6 at home. We further excluded anyone who reported receiving Social Security or Supplemental Security Income for reasons of disability. In most cases, we used the Annual Social Economic Supplement of the CPS (ASEC) for the analysis. When we compared changes from year to year, we matched individuals’ records from one year to the next. Further, in some cases, we used data from the Job Tenure and Occupational Mobility Supplement, and from the Displaced Worker Supplement of the CPS available in January every other year, matched to the ASEC for that year.
*IPUMS-CPS, University of Minnesota, www.ipums.org.
Many of Those Likely to Be Subject to New Work Requirements Are Working, But for Low Wages
In this section, we show labor market outcomes among non-disabled adults without young children at home, including three subgroups that are particularly likely to receive SNAP or Medicaid benefits: those with a high school diploma or less, women with a high school diploma or less, and those who report a work-limiting disability but are not receiving disability benefits.
As the figures below show, many of the individuals who are particularly likely to receive SNAP or Medicaid benefits are working, usually work 30 or more hours per week, and are employed at least 20 weeks per year. However, even among those who work substantial hours and weeks, jobs are unstable and they have not experienced wage growth. Median wages for these workers are low and did not rise much between 2002 and 2017. When we match individuals across two years of data (2016-2017), those who had substantial work in one year were likely to experience drops in their income, hours, and wages in the next year.
In 2017, 81 percent of adults in our sample reported having worked at least one week in the previous year. Among those with high school degrees or less, 74 percent reported working during the prior year. Among women who have a high school education or less, 69 percent reported at least some work.
In 2017 among non-workers in our sample, 28 percent reported not working because they were a student, while 23 percent reported that they had caregiving responsibilities and 22 percent reported that they retired early. Another 17 percent reported they were not working due to an illness or disability, and 6 percent reported they could not find work.
The overwhelming majority of working non-disabled 18- to 64-year-olds without young children performed substantial work, as Figure 2 shows. Some 95 percent of all workers were employed at least 20 weeks in 2017, while 94 percent of those with a high school diploma or less were. Among all workers, 87 percent reported that they usually worked at least 30 hours per week; 81 percent of those with a high school diploma or less did. The share working both at least 20 weeks and at least 30 hours per week — a level we term “substantial” work — was 85 percent, 84 percent, and 78 percent, respectively, for all, for those with a high school degree or less education, and for women with a high school degree or less.
Although those with low levels of education are working, and the majority of those are working a substantial amount, real wages have remained low. Figure 3 shows real median annual income from wages and salaries in 2002 and 2017 for those who worked at least 20 weeks per year and usually worked at least 30 hours per week. While overall median income from wages and salaries increased somewhat over this period — from $39,700 to $42,300 — it has been flat or declining for workers with low levels of education. Among all workers with a high school diploma or less engaged in substantial work, inflation-adjusted wages were $31,000 in 2017, down slightly from $31,700 in 2002. Among women with low education, median wages were $27,300 in 2017 — nearly the same as in 2002. Earnings differences over time and across groups are not primarily driven by differences in labor market participation levels, as this figure includes only those who report working at least 30 usual hours per week and 20 weeks per year.
Figure 4 shows the percent of the individuals in our sample who in 2017 worked at least 20 weeks in the past year and for at least 30 hours per week, but nonetheless had household-level income from wages and salary that was below the poverty line. (This sums wages and salary across household members and compares it to the relevant income-to-poverty threshold for that year.)
Overall, about 5.4 percent of these workers lived in households where those who were working did not earn enough to lift the household out of poverty in 2017. Those shares are substantially higher for those workers with a high school diploma or less and women with a high school diploma or less. And among Medicaid and SNAP recipients with substantial work, 18.3 percent and 28.6 percent, respectively, remained below the poverty line despite their wages and salaries (not shown in graph).
Some observers may believe that once a worker gets a job that allows them to be highly attached to the labor market — usually working 30 hours or more per week for at least 20 weeks per year — it will serve as a toehold on a ladder toward upward mobility and subsequent years of wage growth and job stability. However, when we match workers across two adjacent years, the data indicate that median income from wages after a “good” employment year is lower in the subsequent year. The chart shows these calculations for 2016 to 2017, but in all year pairs from 2002 to 2017, median income fell for workers with a high school degree or less education, including women with a high school degree or less, when we restrict the sample to those who were highly attached in the first year.
This does not mean that a good year causes a bad year. For example, we similarly find that when we limit the sample to those who have good employment outcomes in the second year in which we observe them, it is typically preceded by a worse year. These numbers suggest that these workers face a relatively volatile labor market, and one year’s success does not translate into continued growth in the next year. Programs like SNAP and Medicaid provide some insurance for workers who lose jobs or can only find limited work.
As the bottom panel of Figure 5 shows, in 2016 median earnings were $32,700 among those with a high school diploma or less who worked at least 30 hours per week for at least 20 weeks that year. Between 2016 and 2017, the median income for workers in this group fell by more than $2,100. In fact, the median income fell from 2016 to 2017 across all subgroups shown in the figure. We see a similar pattern in 2009-2010, as the top panel of the figure shows. Further, we find a similar pattern regardless of what pair of adjacent years between 2002 and 2017 we use for these calculations (not shown).
The income loss is explained in part by a decline in usual hours worked or weeks worked per year. For example, in the most recent data, highly attached workers with a high school degree or less usually worked a mean of 42.4 hours per week in 2016, but that fell by 1.2 hours per week in 2017. The same group worked a mean of 49.9 weeks per year in 2016, but in 2017 that fell by 3.4 weeks per year.
People With Disabilities
In our sample, almost a quarter of SNAP or Medicaid recipients who do not engage in substantial work report that they have a work-limiting disability. The sample already excludes anyone who reports receiving Social Security or SSI benefits due to disability, suggesting that a substantial portion of those who would likely be subjected to work requirements report a work-limiting disability. Among this group, 44 percent worked in the previous year. Among these workers, 85 percent reported working at least 20 weeks over the year, 79 percent reported usually working at least 30 hours per week, and 70 percent reported working both at least 20 weeks and at least 30 hours per week. Eleven percent of those who worked substantially, despite their work-limiting disability, lived in households where the workers did not earn enough to raise their household out of poverty. The year-to-year income fluctuations are particularly large for these workers.
SNAP and Medicaid Beneficiaries Work in Lower-Wage, More Volatile Jobs
While the preceding tables and figures show labor market outcomes for individuals with different education levels and characteristics, this section examines the broader job market faced by those who are likely to participate in benefit programs.
Those who received SNAP or Medicaid are concentrated in certain occupations, and these occupations are different from occupations commonly reported by people with above-median incomes. Our analysis finds that among all people in the types of occupations that SNAP and Medicaid recipients often report, income and income growth are lower and employment is more volatile than among people in above-median income occupations. These patterns suggest that many willing workers would struggle to meet proposed work requirements.
By focusing on outcomes by occupation, one can see patterns in working and wages that are linked to the type of job, rather than to the type of worker. Figure 6 shows the ten most common occupations listed by people in our sample who report receiving SNAP or Medicaid benefits in the last year. About one-third of benefit recipients who listed any occupation reported that they fell into these ten occupations. For example, 5.6 percent of those who received benefits in the last year reported their occupations were in health fields such as nursing aides, orderlies, and medical attendants, and 5.4 percent were cashiers.
While the top ten list for those who receive benefits includes janitors, truck drivers, sales clerks, waitstaff, among others, the top ten list for those with middle-class occupations is topped by managers and administrators, and includes registered nurses, school teachers, accountants and auditors, and computer systems analysts, as Figure 7 shows.
Only two occupations appear on both “top ten” lists: truck drivers and supervisors of sales jobs. This pair of figures shows that workers who would be subject to proposed work requirements in SNAP and Medicaid report substantially different occupations from those who earn higher incomes.
Next, we show that wages and work patterns also differ markedly across these groups of occupations.
Figure 8 shows median wages in inflation-adjusted 2017 dollars among all people reporting these occupations, using wage data averaged across 2002-2017. The data are not restricted to those who received SNAP or Medicaid, nor to those who report working. While above-median, or middle-class, occupations had median earnings of $47,600, the top occupations among those who receive SNAP or Medicaid had median earnings of $22,000.
Individual occupations vary in their median earnings. Only two (drivers and sales supervisors) of the top ten occupations common among SNAP or Medicaid beneficiaries had median wages above $25,000 per year; these are the two occupations that also appear on the top ten list among middle-class occupations. Cashiers, waiters, and housekeepers — encompassing 10 percent of people who received SNAP or Medicaid and reported an occupation — had median wages below $15,000 per year.
Figure 9 calculates inflation-adjusted median annual income from wages and salary (in 2017 dollars) in 2002 and 2017, for all people in the two sets of top ten occupation groups: those common among SNAP or Medicaid beneficiaries and those common among middle-class wage income earners. Wages are substantially lower among top occupations of SNAP or Medicaid beneficiaries — $23,800 in 2017, compared with $50,000 for the top ten middle-class occupations.
Median incomes among top occupations of SNAP or Medicaid beneficiaries fell slightly from 2002 to 2017 in inflation-adjusted dollars, similar to the patterns shown in Figure 3 for workers with low levels of education. The lack of income growth among these occupations suggests that stagnation reflects characteristics of the jobs available to workers with low levels of education. Over the same time period, middle-class occupations’ incomes rose. The median income from wages and salary for the jobs commonly occupied by SNAP or Medicaid beneficiaries are lower, in part, because these jobs afford less work, and less stable work, as shown below.
Only two of the top ten occupations held by SNAP or Medicaid beneficiaries experienced statistically meaningful real median income growth from 2002 to 2017: cooks and waiters/waitresses. Despite having experienced some earnings growth over time, these occupations are nonetheless low paying. For many of the occupations that are common among beneficiaries, the number of hours worked per year fell between 2002 and 2017. The exceptions are that cooks, waiters/waitresses, and housekeepers reported more hours worked in 2017, and the former two groups saw income gains that coincided with these increases in hours.
Even if one takes the beneficiaries themselves out of these calculations, the picture remains virtually the same. In other words, these patterns do not simply reflect that SNAP or Medicaid beneficiaries must be earning low wages in order to qualify for benefits. Median income from wages among all people in the types of occupations that are common among beneficiaries are substantially lower than those in the middle-class jobs.
Figure 10 shows the unemployment rate for the two groups of occupations in 2017, as well as the top ten occupations individually for SNAP or Medicaid beneficiaries. The definition of “unemployed” is not having a job (during the reference week) and both looking for and able to work. As described above, reported occupation among those who are unemployed is their most recent occupation. People in occupations common among SNAP or Medicaid beneficiaries face high unemployment rates. In 2017, their unemployment rate was 5.3 percent — over twice as high as the unemployment rate for typical middle-class occupations.
Many individual occupations common among SNAP or Medicaid participants have even higher unemployment rates. Three of the top occupations had unemployment rates of 7.0 percent or higher in 2017.
Figure 11 shows two measures of the stability of employment in the different occupation groups. The right bar in each set shows the share of people in these occupations who were employed two years in a row during the reference week in both 2016 and 2017. (Work activity in the reference week is used to calculate official unemployment statistics.) Only 8.9 percent of people in middle-class jobs were not consistently employed in the reference week across two years; for people in jobs commonly held by SNAP or Medicaid beneficiaries, that figure was nearly twice as large, at 16.9 percent. Individual occupations have a range of employment stability; for example, only 63.3 percent of cashiers were employed two years in a row.
The bars on the left show the share of workers who participated in substantial work two years in a row, measured by reporting that they usually worked at least 30 hours per week and for at least 20 weeks per year across both years. While 84.9 percent of those in middle-class occupations had substantial work across both years, only 64.3 percent of those in occupations common among SNAP or Medicaid beneficiaries had substantial work in both years. Again, there are marked differences across occupations, with fewer than half of individuals being substantially employed in two years among waiters/waitresses, housekeepers, and cashiers. Employment instability in occupations common among benefit recipients was even more severe in the aftermath of the Great Recession.
High levels of unemployment, job turnover and instability, and hours volatility in occupations commonly held by SNAP or Medicaid beneficiaries will make it challenging for participants to meet proposed work requirements and will make monitoring whether they meet them difficult.
Every other year the Displaced Worker Supplement of the Current Population Survey (CPS) asks those age 20 and older if they have lost or left a job in the last three years because their plant or company closed or moved, their position was abolished, they had insufficient work, or for another similar reason. Figure 12 shows the share of people, aged 20-64, who say they lost or left a job in the prior three years for these reasons. This does not include workers who voluntarily left a job, for example to take another job. Nearly 10 percent of those with common occupations among those who would be subject to proposed work requirements reported being displaced from a job in the three years before 2016, compared to 6.3 percent with common middle-class occupations.
The share of those who were displaced from a job in the last three years was higher for both groups in 2010, following the Great Recession, than during 2016, but still substantially higher for those in occupations common among SNAP or Medicaid beneficiaries than for those in middle-class occupations.
Cashiers, cooks, janitors, waiters/waitresses, and laborers have high levels of job displacement both in strong economic times and during recessions. Drivers, janitors, laborers, and wait staff had higher-than-average job displacement in the aftermath of the Great Recession. These patterns highlight not only the instability of employment among the types of occupations common among SNAP or Medicaid beneficiaries, but also the need to recognize that these workers experience additional instability during economic downturns.
Another measure of the volatility of work in these occupations is a worker’s job tenure length — that is, how long the individual has been with a particular employer. Every other year, the Job Tenure and Occupational Mobility Supplement of the CPS asks about length of time with current employer. Our measure of tenure can be lower either because an individual leaves an existing firm for a different one (voluntarily or not), or because the firm ceases to exist, or because the individual is not currently working. Whatever the cause, those who report occupations disproportionately held by SNAP or Medicaid beneficiaries have lower tenure than those in middle-class occupations.
As Figure 13 shows, 87.5 percent of people with middle-class occupations had at least one year of tenure with their current employer in 2016, compared to 76.9 percent of those in occupations commonly held by SNAP or Medicaid beneficiaries. For these people, only 40.1 percent had five or more years of tenure with their employer — a rate that is 19.5 percentage points lower than the share of people with middle-class occupations. The patterns are broadly similar when we account for a person’s age, though the differences in the share of people with five or more years of tenure are smaller when age is factored in.
Tenure varies across individual occupations. Some occupations that have high levels of one- and two-year tenure, such as nursing assistants and cooks, drop off substantially in the share with five years of tenure. Other occupations such as waiters/waitresses and cashiers have fewer than half of people in the same job for two years of tenure. Again, the instability of these occupations suggests that meeting and monitoring proposed work requirements will be difficult for SNAP and Medicaid participants and administrators.
The data used for this report come from the Current Population Survey (CPS), available from the Integrated Public Use Microdata Samples. Most of the analyses are using the Annual Social and Economic Supplements (ASEC) of the Current Population Survey from 2002 to 2017. Some analyses link individuals from one year to the next. We were able to successfully link 45 percent of possible matches from the 2016 ASEC to the 2017 ASEC. Our analysis sample includes U.S. citizens who are 18-64 years old, not active-duty military, who do not receive Social Security or Supplemental Security Income (SSI) for reasons of disability, and who do not have a child under age 6 in the household. Additional analyses use the Displaced Worker Supplements and the Job Tenure and Occupational Mobility Supplements, available every other year. In order to exclude those who receive Social Security or SSI for reasons of disability from the sample, we match respondents in these supplements to their responses in the Annual Social and Economic Supplement, as the latter contains information on benefit receipt. We linked 91 percent of possible January supplement respondents to the ASEC.
Figure 1. Most SNAP or Medicaid Participants Without Substantial Work Have Lower Education Levels
Data come from the CPS ASEC (2017). In our sample in 2017, 14.6 percent of individuals reported receiving SNAP or Medicaid in the previous year. Of these SNAP or Medicaid participants, 62 percent reported working at least one week in the previous year and 44 percent reported substantial work, which we define as usually working at least 30 hours per week for 20 or more weeks. Figure 1 shows the gender and education breakdown of the SNAP or Medicaid participants who did not report substantial work.
Figure 2. Most Workers Were Highly Attached to the Labor Market
Data come from the CPS ASEC (2017). The share working at least one week in the previous year is defined as a dummy variable where 1 indicates that the respondent said they worked one or more weeks in the last year. It is missing for those who did not give a valid answer to how many weeks they worked.
Figure 3. Real Wages Have Been Stagnant for Workers With Low Levels of Education
Data come from the CPS ASEC (2002 and 2017). Income from wages and salary is reported in 2017 dollars and adjusted for inflation using the Personal Consumption Expenditure Price Index (PCEPI) deflator, obtained from https://fred.stlouisfed.org/series/PCEPI. Nominal income from wages and salary was multiplied by the PCEPI value from January of each year divided by the PCEPI value for January 2017 to get real income from wages and salary in 2017 dollars.
Figure 4. Workers With Substantial Work and Lower Education Levels Are at Risk of Poverty
Data come from the CPS ASEC (2017). Household income from wages and salary were calculated by adding the income from wages of all household members in the ASEC, including those out of sample (non-citizens, all ages, those who received SSI or Social Security for disability, active military, and those with children under age 6). Household income from wages and salary was then compared to the Census’s official poverty line (for the appropriate household size) to determine poverty status. This means that all income from wages and salary, whether from out-of-sample or in-sample people in a household, was counted towards determining the household poverty status of those in the sample.
Figure 5. Labor Market Earnings in One Year Not Guaranteed in Subsequent Year
Data come from the CPS ASEC (2016-2017), where individuals’ responses were matched with to their responses in the subsequent year. Sample was limited to those who could be matched across years in the ASEC. Income from wages and salary is reported in 2017 dollars and adjusted for inflation using the PCEPI deflator (as described under Figure 3).
Each year of the ASEC was matched to the subsequent year using the method suggested by Madrian and Lefgren. Matches were checked and omitted for incorrect matches of race, age, and sex. In 2016-2017, we successfully matched 45.1 percent of possible matches, for a total of 17,133 people successfully matched across these years.
Figure 6. Most Common Jobs Among Working SNAP or Medicaid Participants Include Health Care Workers, Cashiers, and Cooks
Data come from the CPS ASEC (2002-2017). For employed people, the occupation reported is the occupation that they work in for the largest number of hours per week. For people who are unemployed or not in the labor force, the occupation is reported as the most recent occupation held within the last five years. Occupation is not reported for those who have not worked in the last five years.
Figure 7. Most Common Occupations Among Workers with Above-Median Wages are Managers, Teachers, and Salespersons
Data come from the CPS ASEC (2002-2017). See description of occupation under Figure 6. We calculate the real median income from wages and salary for 2002 to 2017 (including zeros). If an individual reports a real income from wages and salary above that threshold, then that individual is used to calculate the top ten occupations reported by people who earn above-median incomes from wages and salary (among those who report an occupation). We then examine labor market outcomes among all those who report one of these ten occupations.
Figure 8. Many Jobs Commonly Held by Working SNAP or Medicaid Beneficiaries Have Low Wages
Data come from the CPS ASEC (2002-2017). Income from wages is reported in 2017 dollars and adjusted for inflation using the PCEPI deflator (as described under Figure 3). Median real incomes from wages are calculated across all people in the sample who report each given occupation, whether they indicated working in the previous year or not.
Figure 9. Income from Occupations Common Among SNAP or Medicaid Participants Has Stagnated Over Past 15 Years
Data come from the CPS ASEC (2002, 2017). Income from wages is reported in 2017 dollars and adjusted for inflation using the PCEPI deflator (as described under Figure 3). Median real incomes from wages are calculated across all people in the sample who report each given occupation, whether they indicated working in the previous year or not.
Figure 10. Unemployment Rates Higher in SNAP or Medicaid Beneficiaries’ Typical Jobs
Data come from the CPS ASEC (2017). Unemployment rates are calculated across all people in the sample who report each given occupation. Occupations were reported for all people who had worked in the last five years, and people who were currently unemployed would be reporting their most recent occupation, not the occupations they are searching for. The reference week is the week prior to the week in which the person participated in the CPS interview.
Figure 11. Individuals Are Less Stably Employed in SNAP or Medicaid Participants’ Typical Occupations
Data come from the CPS ASEC (2016-2017). See notes on ASEC-to-ASEC merge under Figure 5. Employed in the reference week in both years indicates that the person was employed, whether at work or not, during the week prior to their completion of the CPS ASEC survey in both years. Employed substantially in both years indicates that the respondent reported working both 30 or more usual hours per week and 20 or more weeks per year in both years. Both of these measures of employment are defined as dummy variables where 1 indicates that the person was employed or substantially employed in both 2016 and 2017. This is only calculated across people who could be matched from the 2016 ASEC to the 2017 ASEC. Individuals are assigned to the occupation they reported in 2016.
Figure 12. Job Displacement Is Higher in Typical Occupations of SNAP or Medicaid Participants
Data come from the CPS ASEC (2010, 2016) and the CPS Displaced Worker Supplement (2010, 2016). Data from the CPS Displaced Worker Supplement is available for every other year and is taken in January. We used data from 2010 and 2016 in order to evaluate the most recent year, and to look at a year following the Great Recession.
In order to match the Displaced Worker Supplement with the ASEC, we first merged the January data with the March data using IPUMS CPS identifiers (cpsidp), and then merged that data set to the ASEC using IPUMS CPS identifiers (marbasecidp). Incorrect matches were dropped based on race, sex, and birthplace. 46.3 percent of Displaced Worker Supplement respondents were successfully matched to the ASEC (92.7 percent of possible matches). Only those who matched are included in the calculations.
The variable for “displaced” is a dummy variable equal to 1 if the person answers “yes” to the following question (years for 2016): “During the last 3 calendar years, that is, January 2013 through December 2015, did (name/you) lose a job, or leave one because: (your/his/her) plant or company closed or moved, (your/his/her) position or shift was abolished, insufficient work or another similar reason?”
This was only asked of people 20 years or older, and so the calculations in the figure only include people 20-64 years old, as well as only including people who were eligible and interviewed for the Displaced Worker Supplement.
Figure 13. Job Tenure Lower in Typical Occupations of SNAP or Medicaid Participants
Data come from the CPS ASEC (2016) and the CPS Job Tenure and Occupational Mobility Supplement (2016). Data from the Job Tenure and Occupational Mobility Supplement are available every other year and are taken in January.
In order to match the Job Tenure and Occupational Mobility Supplement with the ASEC, we first merged the January data with the March data using IPUMS CPS identifiers (cpsidp), and then merged that data set with the ASEC using IPUMS CPS identifiers (marbasecidp). Incorrect matches were dropped base on race, sex, and birthplace. 46.6 percent of Job Tenure and Occupational Mobility Supplement respondents were successfully matched to the ASEC (93.2 percent of possible matches). Only those who matched are included in the calculations.
Job tenure for each category is coded as a dummy variable where 1 indicates that the person reported working for their current employer one or more years, two or more years, or five or more years, and 0 indicates that they have worked for less than one year, two years, or five years. The job tenure variable is only asked of those who are working, so we have additionally coded those who are not working as a 0.
 Kristin F. Butcher is the Marshall I. Goldman Professor of Economics at Wellesley College. Diane Whitmore Schanzenbach is the Margaret Walker Alexander Professor of Education and Social Policy and Director of the Institute for Policy Research at Northwestern University. The opinions and conclusions expressed in this report are solely those of the authors. Views expressed in this report do not necessarily reflect the views of the Center on Budget and Policy Priorities. We thank Abigail Pitts for excellent research assistance.
 We code individuals as displaced if they report losing or leaving a job because the plant or company closed or moved, their position or shift was abolished, or for other similar reasons. “Other similar reason” includes those who lost or left their job due to seasonality, self-operated business failure, for example.
 Brigitte C. Madrian and Lars John Lefgren, “An Approach To Longitudinally Matching Population Survey (CPS) Respondents,” Journal of Economic and Social Measurement, 2000, Vol. 26 (1), 31-62.