During COVID-19, Congress vastly
expanded the breadth and generosity of the nation’s unemployment insurance (UI)
programs. Doing so under emergency conditions negatively impacted
implementation, leading to a host of challenges. If we’re to avoid a replay of these
challenges in future crises, we need to be clear on what mistakes were made as
well as the inadequacies of the underlying system during periods of normal
operations. This will mean taking the necessary steps to form and strengthen
the underlying UI program as well as its crisis response mechanisms.

To explore these questions, in late October, I was joined at an AEI event by Amy Simon, of Simon Advisory, who led implementation of the emergency pandemic UI programs and authored a paper on UI performance and reform, as well as AEI experts Matt Weidinger and Mason Bishop. You can watch the full event here.

It was a vibrant conversation with many audience questions that we were unable to address during the program. Below you will find responses written responses to some of the audience questions that we did not get to during the event.

Research shows that more people would have filed for unemployment insurance during the pandemic had the process been more accessible. How do we balance the issue of unemployment fraud with the issue of not getting support to those who really need it? What are the alternative options to self-certification?

Matt Weidinger: In sum, we can and must be able to do both.

Amy Simon: I don’t think fraud protection needs to be the antithesis of getting support to eligible unemployed workers. In fact, I’d argue they are different sides of the same coin. With better fraud structures in place, states would have been able to more quickly cut through the noisy high volumes of fake applications and get benefits to eligible individuals more smoothly. Fighting fraud is one very important part of making sure that eligible applicants can access unemployment insurance, but there are, of course, other facets to the accessibility conversation.

In terms of alternatives to self-certification, I would simply note two things. First, a clarification: The self-certification standard mostly refers to Pandemic Unemployment Assistance (PUA) applicants choosing which COVID-19-related category applies to them; basic evidence of non-wage income could have (and should have) been requested by the state as part of the PUA eligibility determination process. States likely improved on that front over the course of the pandemic, but certainly at the start, states were often accepting very thin evidence of 1099 employment and issuing effectively the minimum PUA benefit amount to everyone. See UIPL 16-20, Change 1.

Second, Disaster Unemployment
Assistance (DUA) is also designed for similarly difficult situations; DUA requires
that applicants demonstrate they have lost a primary source of income, prorates
the benefit amount when the person is not employed full time, and requires
evidence of employment within 21 days. These safeguards were not carried into the
original incarnation of PUA; those or similar policy choices could have
provided more balance to the income portion of the eligibility determination.

Third, one of the challenges to
the entire PUA concept was that state agencies don’t typically have access to
the tax records (state or federal) to validate contract, gig, self-, or other
non-wage earnings. They have very well-established methods of verifying wage
data because that has been the heart of the program. I think pushing state
agencies to find or develop alternate and better sources of earnings
verification will be an important part of the UI conversation going forward.

How does handling fraud look different during normal times versus during times of crisis? What were the instances of unemployment insurance fraud like pre-COVID-19?

Matt Weidinger: Pre-COVID-19, a significant share of fraud involved individuals who continued to collect benefits after returning to work, and the overall volume and share of benefits lost to fraud was significantly lower. During the pandemic, the surge in claims, as well as the unprecedented increase in benefits and expansion in eligibility, created a richer target for criminals bent on defrauding the system. Overwhelmed states could not keep up; some intentionally let their guard down, for example by suspending the matching of state caseloads against prisoner databases. The system needs to be better prepared to address both types of fraud in the years ahead.

Amy Simon: Yes, fraud as commonly defined in the pandemic era (synthetic identities, imposter claims, altering bank account information) was not really a factor. The three largest culprits in overpayments have consistently been separation issues, benefit year earnings issues, and work search issues.

Is it possible for the Department of Labor (DOL) to establish standard practices/technology (ID Verification, for example) or minimum standards in order to qualify for federal funds?

Mason Bishop: It is important to remember that the UI program is a federal-state program whereby states tax employers and pay benefits, while the federal government provides annual allotments to pay for the administration of state-based UI programs. Historically, UI was created at the state level and in the 1930s adopted nationally.

This
history and state-level operationalization of the program is critical for
understanding the long-standing federal practice of providing basic regulatory
requirements, such as being “able and available” to work while also primarily
providing guidance for the operations of UI. However, when a crisis such as the
Great Recession or COVID-19 comes along, to address immediate and catastrophic
unemployment, Congress passes a 100 percent federally-funded benefits program and
then administers it through the states. The federal government has no existing
infrastructure of its own to pay and manage a UI benefits program.

Amy Simon: The department has published some guidelines for identity management (see UIPL 4-21 here). What the department does not have is the capacity to create and/or deploy technical teams who can evaluate if states are following the guidance, and/or the ability to impose penalties for not doing so. The states do not necessarily always have the proportionate resources (funding or staff) and may view something like a federal technical audit team as a violation of the existing partnership structure. State agencies are supposed to affirm to DOL via their annual State Quality Service Plan. There are some items related to fraud prevention or technology usage (see UIPL 24-21 here), but again, I am not sure that non-affirmation results in different outcomes for any given state. On the flip side, as mentioned previously, states are also not necessarily rewarded for technical excellence or technical fraud prevention.

Unemployment trust fund solvency is also a major problem in many states, both now and prior to the pandemic. How do you think the federal government can better incentivize states to maintain adequate levels of solvency in their trust funds in anticipation of the next economic crisis?

Matt Weidinger: There have been a number of proposals that seek to address this issue, such as offering preferential terms to states that achieve a target level of solvency. However, states view such questions differently. Some states like Texas both strive to maintain low trust fund balances (preferring instead that funds circulate in the economy instead of remaining idle in government trust funds) in advance of a recession and resort to private instead of federal borrowing if needed in the wake of a downturn.

For the record, solvency prior to the pandemic was improving rapidly and in most states met or exceeded federal norms for weathering a typical recession. See the following AEI report.

Given the problems with administrative funding, why not devolve the Federal Unemployment Tax Act tax collection and administrative funding responsibility to the states?

Mason Bishop: The George W. Bush administration proposed a set of UI improvements that reformed administrative funding and looked to devolve funding to the states, but Congress did not act on the proposal.

Matt Weidinger: The current administrative financing system is complicated, creates winners and losers among states in terms of what they get back versus the taxes paid in, and requires some states to levy their own taxes to effectively administer the system. However, fixing it will require a series of compromises through the legislative process.

The history of UI advisory councils has been that they are heavily weighted in membership toward nationalizing the system and increasing coverage and employer tax burden. How do the councils rely on existing policy groups for evaluation?

Matt Weidinger: The Advisory Council on Unemployment Compensation recommendations in the 1990s tended to support bigger benefits and higher taxes. Most “expert” panels in DC tend to do the same, and are overrepresented by interests that support more and bigger government “solutions.”

Amy Simon: Excellent point. I don’t think the UI Advisory Council is currently active so I can’t speak to how they do or do not rely on existing policy groups. I think there are many competing voices for a small number of stakeholders and decision-makers, and data-driven researchers should be louder in those discussions.

Can you be more concrete about what you mean by modernizing UI services? How do we balance technological advancement of the system with the system serving persons who, for whatever reason, are not technologically literate or do not have access to technology, broadband, or smart devices?

Mason Bishop: Modernization can be thought of in three broad areas: (1) Policies, (2) Service Delivery Practices, and (3) Technology. UI should be “modernized” in all three areas. First, at the policy level, we see that today’s UI system is not aligned with today’s economy and work arrangements. The PUA fiasco is just one example of a system that has not evolved to recognize the growing segment of independent practitioners, freelance workers, and “gig” workers. Service Delivery modernization would restructure and realign programs to ensure that, for instance, UI claimants, who primarily apply for, and receive, UI benefits would receive immediate access to employment and training services, and the agencies providing those services would have access to UI claimant data.

Amy Simon: In light of Mason’s definition, I was using modernization as an overarching term for lane 3: the types of systems, structures of those systems, and methods of deploying those systems that reflect more recent technology innovations. In my mind, the desired outcomes of UI system modernization, regardless of approach, are more reliable, more secure, more scalable and resilient, and more user-experience-centric systems.

I
agree that the unemployment insurance world will always need to maintain a
variety of platforms to communicate with end-users. However, modernized systems
should present fewer barriers to diverse populations with a range of
technical literacy skills or internet connectivity. One basic example: Many
state websites were not mobile-friendly, even though much traffic has shifted
toward mobile devices. That limitation is not always a pure reflection of the
underlying system design, but simply a small example that out-of-date technology
can quickly turn into a barrier for claimants who are mobile-first.

Regarding gig workers, what are the benefits and drawbacks of including gig workers in the UI system? One of the issues is gig workers work for different employers or move between them. Should there be a pooled benefits method for gig workers and independent contractors that is different from UI? What would this look like in an ideal world?

Amy Simon: First, I think there are a lot of nuances in definitions. Gig work can be an entire spectrum encompassing everything from occasional to full-time work for one or various employers. Relatedly, the regular UI system has internal boundaries with wage thresholds or labor market attachment time, which would not necessarily translate to gig work. Definitions aside, I think there are more fundamental questions to tackle before figuring out the mechanics of how gig workers might be best covered. The fundamental question is, how do we create a sustainable, defined, reliable, accurate safety net of partial income replacement for workers who lose employment by no fault of their own? That brings up multiple angles: funding and taxes, benefit, coverage, and sustainability.

Matt Weidinger: I would note that one of the many complexities with such proposals is the question of when individuals return to work and whether they are expected to search for other work while on benefits. For example, in the regular UI system, when a laid-off employee is recalled, he or she must return to work and forego further UI benefits. But when must a gig worker or independent contractor return to work, and who decides that? If the answer is “when the individual decides,” then compared with UI that will tend in the direction of longer durations, resulting in more expensive benefits and thus higher taxes if they are levied on these groups. Among the many questions worth considering are whether the taxes paid by gig workers and independent contractors would be experience rated as UI taxes are, and how would those taxes compare with UI taxes for the average worker?

Amy mentioned that we cannot wait until the next crisis to reconstruct the system. What are the conversations happening now about reconstruction and who are they including? Are they bringing claimants, employers, and other stakeholders to the table? If not how can they?

Matt Weidinger: I described several Wyden bills during my verbal comments, neither of which was included in the current trillion-dollar spending “framework” released this week. While Congress (and especially the committees central to that legislation) are distracted with those matters, such UI questions will have to wait until 2022.

Amy Simon: The conversations are not being held in public, to my knowledge. Certainly, there have been many thoughtful reform proposals from various think tanks or research organizations over the past two years. Many states have been huddling internally about how to address their technology issues, hence the ongoing wave of RFPs for new systems. I also think the department, the relevant White House entities, and Congressional leaders have been thinking about this behind closed doors, hence various investments. This is a particularly difficult conversation to corral because of the wide number of stakeholders and multiple layers of governance. That being said, my hope is that the department will use their institutional position to convene relevant voices on the policy front or even issue a Request for Information on the technology side.

Is it possible for DOL to mandate implementation of certain standards? If not, why? Especially in the case of fraud management?

Amy Simon: Yes, DOL does mandate some program integrity measures (see UIPL 23-20, UIPL 16-21 or UIPL 4-21 here), but I think the larger issue is if and how the department can evaluate state adherence in practice and in real-time to these regulations.

Is there a way to make the administration itself more efficient and reduce administrative burdens and costs? This could free up more money for workers. More efficiency for the sake of helping unemployed workers seems like a necessary step. Where can we be more efficient in administrative costs and how?

Matt Weidinger: Spending on program administration should increase, not decline, in the years ahead. And it is already a fraction of benefit spending, which reached unprecedented levels during the pandemic. If anything, as the conversation around the $600 bonus suggests, administrative inefficiency contributed to that excessively large benefit spending, both in terms of not stemming fraud and abuse and also in terms of preventing lawmakers from deploying more deft policy solutions than paying every claimant a flat $600 on top of their other weekly benefits. We should be able to do better than that and should hold policymakers accountable for doing so.

Is there a future where we could leverage machine learning and other big-data-empowered tech solutions to deliver individualized extended benefits packages in a pinch? If so, is this a route we want to tread down given the sensitivity of the information involved?

Amy Simon: The original (and some of the current) program structure is built around the timing and availability of data sources. The program does not need to be and should not be unnecessarily hindered by those constraints when alternatives, including options mentioned, are more available. In the best case scenario, this could free both policy and technology choices at the state level, and lead to a much more responsive program.

On the privacy and safety front, I will note that unemployment insurance data is under its own set of stricter confidentiality regulations (here) as the risk of such data’s misuse has been well understood. The status quo is more constrained by the well-known obstacles to exchanging some kinds of UI data between employers, states, and benefit programs. For example, all states ask for wage verification data from employers, but they often ask for similar data elements in disparate ways which causes unnecessary friction for multistate employers or third-party agents. In the short term, there should be ways to smooth and share data more efficiently within the current constraints. Meanwhile, piloting approaches to safely and productively explore machine learning or big data is well worth considering.

How do you balance the tension between the idea of giving DOL more intermediate-level tools to use in corralling state labor and workforce agencies toward standardization with the freedom to innovate unhindered in the federalist system?

Amy Simon: The goal of any intermediate-level tools should be to promote operational excellence and service delivery improvement rather than pressuring states into certain policy stances, although I recognize the lines could be blurred in some cases. For instance, DOL should be able to insist that states meet certain technological standards or can defend against a defined set of fraud scenarios.

In order for that to work, the department must have the staff to both set and evaluate against the standards and then have tools to obtain compliance when the states don’t meet them. The department currently has neither. On the flip side, states should be given room to innovate, especially if they are able to demonstrate operational improvements or the ability to better meet program outcomes.

The post Unemployment insurance at a crossroads: Tracing program design during and beyond COVID-19 appeared first on American Enterprise Institute – AEI.