Learnings Part 2: Outreach

2.1 Messengers
2.1.1 The Program Office as the Primary Source of Information
2.1.2 Employers
2.1.3 Other Sources

2.2. Using Data to Track Progress and Target Outreach

Especially when PFML programs are new, administrators must pursue explicit outreach strategies to ensure that potential beneficiaries know about the programs and how to access them—and that businesses in the state understand their new obligations, and the rights of their employees. The quality and design of these efforts have significant bearing on the success of the programs.

2.1.1 The Program Office as the Primary Source of Information

The program office should ultimately own communications about its own program. This is not to say that the government should go it alone on outreach; other non-government actors—including workplaces, health and medical offices, and advocacy organizations—play important roles in getting the word out. But program administrators should create clear and versatile enough materials that they remain the core—and, perhaps more importantly, a trusted—source of information.

This means that outreach and communications have to be an explicit focus of the program office—and, in this sense, New Jersey under the current administration is a role model, with a highly empowered outreach team reporting directly to agency leadership.

But this empowered outreach team is a relatively recent development. New Jersey’s experience in previous years also demonstrates what happens when the program office does not take charge of communications, and responsibility falls to outside actors. Because the previous administration made such little effort to publicize and communicate about the program, it had become common for other players to create and disseminate their own program materials. During the course of our research in New Jersey, we came across written materials on the PFML program created by a coalition of outside advocates and service providers, by well-intentioned HR representatives, even by security staff in the NJDOL lobby in Trenton—all inspired by perceived deficiencies in the official materials.

These players were doing vital work to fill the void left by the program office, and are meaningfully to thank for the level of success the program had achieved. But even when these materials are accurate, they unfortunately and subtly help perpetuate a downward spiral. Beneficiaries rely ever more on informal sources of information, such as peer groups and Facebook pages, where the program office has even less control over the information circulating.

“It was quite confusing. . . I got most information from other moms that had kids.” – Interviewee #8.

As the process continues, the information can become increasingly less reliable, especially if there are changes in program requirements. Moreover, outside actors with less control can perpetuate a sense that the program is too complicated to understand, since—intentionally or not—their rhetoric can be seen as blaming excessive complexity on the program, rather than fully owning and addressing it.¹

Ultimately, potential beneficiaries we spoke to faced an avalanche of information that they do not know how to parse, and do not know which parts of it to trust.

“Nobody knows anything, which includes my HR person. I’m getting conflicting information from all these forums, and you can’t get through to the state. I got through once, when I was four months pregnant, and it’s been impossible… I don’t know if they [the state] don’t want people to know about it, but there’s nothing helpful except for the facebook forum [a group run by advocates that assists with applications]… Like for this, you need a ‘maternity leave pay for dummies’ so people know what to do, when to do, and how to do it.” – Interviewee #1

Notably, New Jersey’s experience in this regard was not unique. Early paid leave states generally did not allocate sustained funds for outreach and education, and this has hampered awareness. Previous research by the National Partnership for Women & Families and others has found that “low program awareness is common,” and is often a key barrier that keeps eligible workers with care responsibilities from using PFML, particularly among workers with lower wages.

It bears repeating that the fault here does not lie with the outside actors, whose herculean efforts kept New Jersey’s PFML program afloat when the previous administration essentially declined to do so. Rather, the experience simply underscores that communicating clearly and proactively must be seen as a core part of the PFML program—one that administrators neglect, outsource, or leave to others at their own peril.

Recommendations:

• Take outreach and communication seriously as a core responsibility of the program. A devoted outreach and communication team should be a central division within the leave administering agency, with the team’s manager a member of leadership.
• If other actors begin circulating entirely separate publicity materials, this should be taken as a warning sign that the program office may need to communicate more clearly. Those shadow materials may also provide useful clues about what the formal materials are lacking.

2.1.2 Employers

Many workers rely on their employers as the primary, if not exclusive, source of information about PFML programs, and inaccurate or incomplete information from employers can easily prevent potential beneficiaries from using PFML at all. If program administrators were able to prioritize just one avenue of outreach partnership, it would be employers.² Across all of our interviews in New Jersey, potential beneficiaries expected their employers to be reliable sources of information on the PFML program:

“I wish my HR person gave me a packet. She gave me no direction at all. Why even have an HR person?” – Interviewee #1, who resorted to other methods of communication only after HR was unhelpful.

Interviewee #16, on whether she knew about the program pre-pregnancy: “Nope, and honestly I didn’t expect to. I expected my employer to know.”

Previous research on state PFML programs supports this point as well. According to a brief by the National Partnership for Women & Families and the Main Street Alliance, “employers who embrace paid leave programs are among the most valuable people to educate employees about the programs and help them apply.” And yet, employers are generally found to be insufficiently informed about PFML programs. For instance, one survey of New Jersey small businesses found 40 percent did not know if FLI covered their employees, even though all NJ employers (save the federal government) must pay in to FLI and are required to post information about it.

Employers are central not only because potential beneficiaries expect to learn about work-related benefits from their employers, but also because of the central role of job protections in any decision to take leave. As discussed above (Learning 1.1), according to our user interviews, many beneficiaries’ use of PFML is completely contingent on the confidence that they can return to their job afterwards. And while the government has an important role to play in enforcing workers’ rights to job protection, in practice, job protections are frequently only as strong as employers understand them to be. Many workers have neither the capital nor the resources to bring civil rights complaints against their employers if their leave is denied. For them, according to our interviews, the limits of the job-protected leave they can take are dictated by their HR departments, and they will not take leave if their employer does not provide accurate and actionable information about its availability—regardless of what their doctors, their friends, or local community organizations tell them.

While our interviews with employers and employer associations were far from conclusive, our research suggested that program administrators may frequently find willing partners in this work—employers who are trying to do the right thing, and would like it to be just a little bit easier to gather and disseminate accurate information about TDI/FLI. One HR manager, who processes several dozen New Jersey leaves annually, talked about the lengths she goes to in order to help her employees access leave:

“We try to go back to the employees to ensure the whole form is correct before it goes to the state. We’re trying to help employees get their money and their benefits, and it is a frustrating process.” – Employer #9

The same manager reported the difficulty of identifying reliable program information: “If there’s a webinar we try and tap into it. We get a lot of emails, we google a lot. We are members of [the Employers Association of New Jersey, which provides training for members], which is extremely helpful.” Another senior HR coordinator said she would appreciate more direct support from the state around the PFML program:

“I’d love for them to bring back some courses; like a seminar, we can go and they’d give a presentation, they go over each different scenario.” – Employer #5

A third HR manager,³ who manages a predominantly Spanish-speaking workforce, spoke of regularly filling out both the employer⁴ and employee portions of the PFML applications, given language barriers. Surely, the sample of employers willing to talk to our team were biased towards cooperation, and other employers may be more recalcitrant; but administrators could make significant strides by starting with the employers that do embrace PFML and want to make sure it works well, and ensuring they have what they need.

Of course, on the other hand, there are also employers who implicitly or explicitly discourage their employees from taking leave—and most employees are not in a position to fight back individually. We heard this from several interviewees:

“All of the weird stuff was after I came back. Four weeks into my leave I started getting emails I was going to have to come back, even though I’d told everyone I’d be out for another couple months. It was nerve-wracking because apparently my date wasn’t good enough… I feel like I got put on the mom track.” – Interviewee #16

It may be too much to expect that such employers will suddenly become devoted and proactive messengers on behalf of PFML programs. But communicating clearly and concisely about the law to these employers is important for another reason: More concerted messaging to employers about what the law is, how they should be disseminating useful information, and what job protections workers are entitled to, is likely the fastest way to significantly change this discriminatory behavior.⁵

In disseminating information to employers, administrators should keep in mind the unique challenges facing employers that are based primarily in states with a different or no public paid leave policy. We heard from several employers and beneficiaries about HR departments struggling to keep straight the varying rules in different states, and about the steep learning curve of reading up on the New Jersey law just for a handful of employees in that state:

“The company doesn’t know a lot about rules and laws that other states have, and this is pretty new.” – Interviewee #1, whose company was headquartered in Pennsylvania, and had only a small footprint in New Jersey

As more and more states enact PFML programs, keeping track of the wide range of program details will add more complexity for multi-state employers. Navigating such complexity is no deal breaker for businesses well accustomed to the vagaries of operating in multiple states; but program administrators should keep these employers in mind and try to present their rules as clearly and transparently as possible. Meanwhile, in the long run, a national PFML program would go a long way toward getting all businesses on the same page, even if some states continue to offer more generous programs than the national baseline.

A systematic employer outreach program could include:

• Toolkit of materials for employers that have been user tested with employers and available on a specific page on the program website.
• Such materials could include recorded webinars or videos, if user testing suggests this would be valuable.
• Trainings delivered via business associations such as the Chamber of Commerce, business advocacy groups like the Main Street Alliance, or other employer or HR associations.

(TDI/FLI has implemented several of these since the sprint. A toolkit of employer materials went live in fall 2020, and the team did several presentations to HR associations in the state. The pandemic, however, served as a significant impediment to further outreach; according to the outreach team, COVID-related programs are currently the only ones employers want to hear about.)

In addition to testing materials directly with employers, administrators may consider best practices around employer messaging developed from other research. A case study by CLASP and Family Values @ Work in Washington State found that the employers responded best to messages around “need” (describing why employees would need to take leave) and “connection” (as the report puts it: “supporting employees when they need it most connects us to each other as human beings and creates a bond of loyalty and a sense of belonging within a company”).

Even though wage replacement comes through the state program for most employees, the pathway to PFML for getting job-protected paid leave approved runs directly through employers. While other sources of information (discussed below) may help workers learn and stand up for their rights, there is no denying the important, affirmative role that employers play. It is incumbent on administrators to ensure that all employers—and especially those that may not understand the program well—have what they need to provide workers with the right information, not only about the TDI/FLI program but about job protections as well.

2.1.3 Other Sources

A full-fledged outreach program may also pursue a variety of other avenues. Some major opportunities include medical professionals, community organizations, and other state programs—e.g. Medicaid or the Special Supplemental Nutrition Program for Women, Infants, and Children (WIC).

Healthcare providers are a robust route of outreach in New Jersey, especially for maternity leave, where doctors routinely tell their pregnant patients about it and encourage them to apply; several interviewees mentioned hearing about PFML and learning how to apply for it from their doctor. Such outreach indeed makes particular sense for TDI, where doctors are involved in certifying the beneficiary's own disability—and particularly for pregnancy, when a doctor has time to raise the issue well in advance. It may be slightly harder for FLI, when the medical provider may not have as much access to the relevant caregiver, at least in advance.

Meanwhile, community leaders and local organizations have played an especially large role publicizing FLI in New Jersey, especially given the limited outreach program under the Christie administration. That said, working through advocacy groups comes with challenges. Community organizations only have the reach they have, and ensuring comprehensive coverage via a patchwork of such organizations is challenging at best.

Working through these and other third-party messengers is important—though it is no replacement for creating clear public-facing materials and working through employers with access to every eligible worker in the state.

While the picture is somewhat more complicated for other types of leave, parental leave has a convenient feature: It is relatively easy to determine who is eligible for it, and who is not receiving it. Government records and even publicly available Census data record with reasonable accuracy the number of births per year, with geographic and sociodemographic detail. Administrators can use this data to determine what portion of their prospective beneficiaries they are reaching, within which populations, and how to better target outreach to potential beneficiaries.

When the team began its research in New Jersey, we asked the administrators this question: What portion of new parents in New Jersey use the state’s PFML program? Despite hundreds of reports the program receives on a weekly and monthly basis from the NJDOL analytics department, administrators did not know; the reports, designed and programmed decades ago, did not directly answer this question. They referred us to a New Jersey Policy Perspective report estimating 12 percent of new parents in the state take paid leave (referring only to FLI bonding leave, with no reference to TDI birth and recovery leave available only to birthing parents).

As part of our research, we set out to answer this question. (Note that this research looked only at usage of the state PFML program, not private temporary disability insurance plans that the state permits employers to offer in lieu of state TDI, and as such may somewhat understate the total portion of workers who take paid medical leave.) We derived the numerator from NJDOL’s table of all program users over the last ten years, and the denominator from American Community Survey (ACS) data from the Census. The following topline findings are specific to New Jersey, but they are offered as an illustration of the types of learnings that administrators may find from similar analyses. Many of these findings raise further research questions that should be explored with qualitative research akin to the interviews we did during the sprint.⁶

• Use of PFML by birthing parents was quite high overall; 51 percent of eligible mothers⁷ use PFML to some degree.
• However this high overall rate masked important and actionable discrepancies: although most birthing parents are eligible for both TDI (for birth and recovery) and FLI (for bonding), only about one-third of the eligible mothers used both programs. About a third used only TDI, and a third used only FLI. This analysis reveals an important and counterintuitive point: NJDOL can improve program coverage not just by reaching new populations, but by better communicating with existing beneficiaries.
  • TDI-only beneficiaries among this group skewed significantly lower-income than overall beneficiaries, at the time of our analysis in fall 2019. One likely possibility is that the wage replacement rate at the time (66 percent) was insufficient to maximize use among lower-income workers, who could not afford to take more than six weeks off at such a rate. Advocates expect that full leave usage by low-income workers will have increased with the wage replacement rate rising to 85 percent in July 2020, and future research should investigate this hypothesis. It is also important to explore, though, whether there was insufficient messaging around FLI bonding leave for low-income birthing parents, whether those parents felt stronger pressures to return to work sooner, or whether the process of how to apply to and transition from one program to the other was particularly opaque for some populations.
  • FLI-only beneficiaries in this group may not have been eligible for TDI,⁸ or may not have been aware that TDI is available for pregnancy and birth.
• Use of paternity leave was very low; only 8 percent of eligible fathers use FLI.⁹ Like with all other state paid leave programs, there remains much work to be done on engaging fathers.
• There was generally no clear geographic pattern to PFML usage throughout the state.¹⁰
• Lower-income people had lower usage: lower-income zip codes had lower overall usage and lower-income applications were more likely to be rejected. To a degree, this might be an expected outcome of policy design—many low-income workers have insufficient income to qualify¹¹—but it also may be that lower-income people are struggling to navigate the red tape of the program.
• Zip codes with higher rates of graduate degrees had lower usage. There are several explanations. First, higher earners are more likely to have access to private, employer-provided paid parental leave. (Though just 21 percent of the civilian workforce has access to paid family leave, high earners are more than seven times more likely than the lowest earners to have access to employer-provided paid leave.) Second, wage replacement caps may also make the state PFML program less attractive to high earners, though this dynamic may have been ameliorated with the program updates.¹² Third, they may work in demanding industries/professions or at companies which strongly discourage their employees from taking paid leave for caregiving. Because education level is not collected on TDI/FLI applications, this analysis was only done at the zip code level.
• Zip codes with higher rates of non-citizens had lower usage—and the correlation between usage and citizenship is much stronger than the correlation between usage and immigration status, or English-speaking status. More direct engagement with immigrant communities would be needed to better understand the barriers facing them.¹³
• Contrary to the hypotheses of program staff,¹⁴ after controlling for income, we did not find any correlation between program use and race at the zip code level.

For future research: Two other worthwhile pieces of analysis we aimed to do but were not possible with the data available:

• Program use by industry. New Jersey’s application collects occupation data; but because they ask applicants for it in a free text field, and applicants use a variety of terms, it is nearly impossible to standardize. New Jersey and other states interested in tracking these trends could likely merge industry information from the unemployment system to search for industry trends—or collect more systematized information from applicants on the front end (e.g., the form prompts applicants to select from a searchable list of industries). (On the other hand, if states cannot systematize the data this way, they should strike the question from the application entirely; the program should not put additional barriers in front of applicants unless there is good reason to do so.)
• Program use by race and ethnicity, at an individual level. Our analysis found no trends by race after controlling for income, at the zip code level. But because New Jersey’s application did not collect data on race and ethnicity until mid-2020,¹⁵ we could not perform any racial and ethnic analysis at the individual level.

The conclusion from this exercise is not that other jurisdictions are likely to have these same findings, but that this type of analysis is likely to be instructive, and implementing it on an ongoing basis in the form of a data dashboard is a powerful reality check for outreach efforts. In the simplest case, such a dashboard need not be especially difficult to implement. Here are a few data points that administrators could figure in a matter of days:

• Portion of birthing parents who take parental leave. In New Jersey, our analysis found that 73 percent of birthing parents were eligible for leave.¹⁶ While this fraction will vary by demographics and eligibility rules, it is arguably reasonable that around three quarters of birthing parents will be eligible.¹⁷ The number of births per year is readily discernible from the Census Bureau,¹⁸ and program use by new parents should be relatively easy to derive from program data.
• Portion of non-birthing parents who take leave. In New Jersey, our analysis found that, for 69 percent of births, there was a non-birthing parent eligible for leave.¹⁹ Again, the fraction will vary, but it is reasonable to assume around seven in 10 births will have another parent eligible for leave.
• For jurisdictions with clear geographic disparities, geographic breakdown of program use. As a rough approximation, programs could calculate the portion of usage that occurs in given geographies to the portion of population in those geographies. Some small mismatches are to be expected given that the number of eligible parents is not directly proportional to population, especially if program administrators use very small geographies. But if, say, a given geography has triple the population but half the usage of another, it is fair to conclude outreach efforts have flagged in that area.
• For programs that collect reliable industry data, industry breakdown of program use. As with geography, programs can calculate the portion of usage that occurs in given industries to the portion of the child-bearing age workforce in those industries. The latter data should be readily available to state departments of labor who usually implement these programs. Results would allow program administrators to better target outreach activities, to beneficiaries and employers.
• Income and demographic correlates with applications. (Technically this reflects more on the application pipeline than on the outreach process.) Most programs either collect income and demographic data in their applications, or can derive this from other administrative sources. While programs may expect some differences (lower-income applicants are more likely to be ineligible, which may be correlated with other demographic indicators), they should look for approximate socioeconomic parity between accepted and rejected applications, and investigate any discrepancies that occur.

Straightforward and commonsense checks like these can give administrators critical insights regarding where and how outreach and engagement efforts must be improved. Critically, these are not checks that require data science teams or months (or years) of custom development. A few staff and a calculator—and ideally a statutory mandate—can get these underway in a matter of days.

As noted above, such analysis is especially feasible for parental leave, where the universe of potential beneficiaries is relatively easy to define. That said, while it was outside the scope of this sprint, programs may be able to develop heuristics for other types of leave as well. Perhaps, for example, there may be data on injuries and illnesses that commonly require temporary caregiving.

On the other hand, one note of caution: It is easy to draw erroneous conclusions when such calculations get slightly more complex. For example: In New Jersey, the mean annual salary of a birthing parent approved for TDI in 2018 was $46,121, whereas the average salary in the state was $59,980. From these figures, one might conclude that parental leave is successfully reaching income workers. But the average parent is young, and the average young person tends to earn less than the average older person; and, moreover, birthing parents are usually women, who earn on average less than men. The comparison is misleading. It is easy to stumble into such mismatches.

Recommendations:

• Develop a few key metrics to summarize program usage along key dimensions, perhaps based on some of those outlined above (e.g. overall usage among birthing and non-birthing parents, usage by income bucket, usage by geographic area).
• Start simple and build out over time, rather than sinking large sums into a data product up front.