Work Requirements Have a February Problem
Among other issues, sometimes you work fewer days because there are fewer workdays.
There are lots of issues with making people work a certain number of hours each month to receive some social benefit. Here is a very small one, in two simple observations:
A shorter month has fewer days than a longer month.
People tend to work fewer hours when there are fewer hours during which to work.
We can see this issue in earnings data. The following chart shows data from the Survey of Income and Program Participation (SIPP) on average earnings among workers in each calendar month from January 2019 to December 2023.1 Each year, earnings are at their lowest in February.
As you likely know, the One Big Beautiful Bill Act (OBBBA) is set to impose work requirements for Medicaid in January 2027 and expand them for the Supplemental Nutrition Assistance Program (SNAP) next month. The details of those work requirements will vary by state, but in general, able-bodied adults without dependents (ABAWDs) aged 18 to 64 will have to work at least 80 hours per month to maintain benefits, or otherwise participate in 80 hours of approved activities, such as community service.
Given the predictable drop in earnings and work hours every February, it seems plausible that, if the requirements are measured at the calendar month level, some small number of people could end up losing access to SNAP or Medicaid simply because it is February. In particular, you can imagine this happening to people working part-time jobs with volatile schedules.2
One way OBBBA will allow you to prove that you meet the work requirements for Medicaid is by showing that you have earned an amount greater than or equal to the federal hourly minimum wage multiplied by 80. The minimum wage is currently $7.25, so that amount is currently $580. The chart below shows people earning less than $580 as a percentage of all people with earnings greater than $0 in each month of 2023. By a small margin, the percentage of workers with earnings below that threshold is highest in February. The same relationship holds for any given income threshold — I show it also for workers earning less than $1000.
Showing your work
I’m not a lawyer, but I do not think that OBBBA’s statutory language will necessarily require states to use calendar months to verify Medicaid beneficiaries’ work hours. States that have previously implemented or sought to implement Medicaid work requirements under section 1115 waivers have done so in a variety of ways, and the same is true of SNAP work requirements.
Some states’ programs have used or do use calendar months for work hours requirements:
New Hampshire’s brief effort to implement Medicaid work requirements specified that “[b]eneficiaries without an exemption must participate for at least 100 hours per calendar month in one or more qualifying activities.”
Iowa’s SNAP work hours reporting and verification standards for ABAWDs specify that the “number of hours actually worked in a calendar month is the determining factor for meeting the ABAWD work requirement.” The state even says to “[u]se caution if using pay stubs to verify hours of work, because pay periods often overlap calendar months and do not reflect the number of hours actually worked in the month in which the person received the pay.”
But other states use alternative periods that simulate months:
Georgia is the only state that currently has work requirements for Medicaid in place. It requires applicants to “provide documentation for verification that they meet the hours and activities threshold for the most recent four weeks available,” as well as specifying that “[t]he submitted four weeks must be within the eight weeks prior to the application date.”
Kansas’s verification rules for its SNAP work requirements for ABAWDs specify that individuals “must provide verification of the most recent 30 days of pay.”
For states that wish to minimize the burden of reporting work hours, it is probably best to take a both/and approach. States should look at what has been done in the past and allow individuals to verify their earnings or work hours through any of the common methods. The most relevant definition of “month” for an individual trying to prove that they meet the work requirements will vary depending on their pay schedule. Someone who is paid at the end of each week may prefer to provide their earnings for the previous four full weeks, while someone who is paid monthly would likely find it easiest to provide their work hours on the basis of calendar months.
Also relevant here is the frequency with which states reverify an individual’s compliance with work requirements, as well as the length of the lookback period they use during redeterminations. States must reverify individuals’ work statuses at least every six months using a lookback period of at least one month, but they are allowed reverify more frequently and use a longer lookback period. Simply put, a state that requires individuals to submit their hours every month will make life more difficult for beneficiaries (and for the employees of the state Medicaid office) than a state that requires them to do so once every six months. States that reverify beneficiaries’ compliance using a three-month lookback period will make life more difficult than states using a one-month lookback period.
The Centers for Medicare and Medicaid Services (CMS) must issue a rule guiding states’ implementation of Medicaid work requirements by June 2026, and it should tell states to ease the costs of complying with the requirements by using a flexible definition of “month” in verifying work hours in both the initial eligibility determination and reverification processes. CMS should also ensure that states have access to already-existing sources of income data that can help to automatically recertify individuals. It should do all of this sooner rather than later, as states will likely have to get started with the implementation process well before June 2026.
Some bigger issues
Admittedly, the February problem as specified is quite minor. But it is related to larger issues with work requirements. Many people — especially those working low-income jobs — have unstable hours that are set by their employer. These people will be penalized for dropping below 80 work hours in a given month, even if they wanted to work more than that, which they often do. People also experience unemployment fairly regularly, and making them participate in community service to maintain their health insurance when they could be looking for a new job doesn’t seem all that likely to encourage work.
In the following chart, I look at ABAWDs between the ages of 19 and 64 who were on Medicaid for all twelve months of 2023 and had an average monthly income of at least $580 over the course of the year. Roughly a quarter of them earned less than $580 in at least one month. Most of those individuals had some earnings in at least nine months out of the year,3 but presumably, many would lose Medicaid coverage under OBBBA for failing to meet work requirements.
There simply isn’t a significant population of long-term unemployed ABAWDs living large on food stamps and public health insurance. Using the SIPP, I estimate that in December 2023, ABAWDs who did not work at any point during the year made up less than five percent of Medicaid recipients. Other estimates using different surveys have put the number around three percent.4
Accordingly, there is little evidence that work requirements boost employment — the primary stated goal of the policy. The main effect, as we have observed when states have tried to implement Medicaid work requirements, is actually to remove people from benefit programs because they don’t know that they have to consistently report their work to the government. This is why work requirements are often derisively called work reporting requirements or, even more derisively, paperwork requirements.
Even if we could implement work requirements perfectly, I would personally still think that they are bad. But there are seemingly many people who disagree with me, and to them I would simply say that, as with any regulation, work requirements don’t administer themselves. In practice, they mostly serve as a cost-inefficient way to kick people off of social programs regardless of their actual work situation.
I use earnings rather than hours because the SIPP doesn’t allow you to directly observe hours worked in each calendar month. In fact, the SIPP is the only survey I am aware of where you actually can observe earnings this way. Most data has some sort of standardized period that it treats as a month but does not perfectly correspond with a calendar month, because it is not actually desirable for most uses of economic data to capture the economic effect of February being a shorter month.
Additionally, I exclude bonus income because it substantially boosts average earnings in March and December (and to a lesser extent February and April) in a manner that isn’t reflective of most workers’ situations, since large bonuses are concentrated at the top of the income distribution. I find this pretty interesting; chart provided below in case you do as well.
I highly recommend The Hamilton Project’s post on how work requirements affect people in jobs with volatile hours.
The chart below breaks down the months with >$0 earnings among that quarter of ABAWDs Medicaid recipients who had average monthly earnings >$580 but earned <$580 in at least one month.
Also relevant is that the average length of unemployment spells has continually increased. Many of the ABAWDs on Medicaid who have not worked recently are not in that situation by choice. The long-term unemployed need healthcare too!