A path to transforming the nation’s unemployment insurance systems

When digital services fail, like what we saw with the nation’s unemployment insurance (UI) systems during the COVID-19 pandemic, there are real human and social costs that undermine the public’s trust in government. These failures can also serve as catalysts for modernization, similar to what we saw in the wake of the HealthCare.gov rescue and the rise of digital service teams like the U.S. Digital Service, 18F, and Ad Hoc.

With new political leadership and initiatives like the Technology Modernization Fund, the time is ripe to rethink how the nation’s unemployment systems, many of which still rely on manual and paper-based processes, can be improved through new technology and approaches. This effort should include the use of modular and reusable solutions, an idea included in a proposed bill from Senator Ron Wyden, that states can adopt to address common challenges, such as identity verification, eligibility determination, claims tracking, and document management.

However, while new technology and approaches can help states reduce sources of friction that affect service delivery, above all, the federal government should set a clear intent for modernization of the nation’s unemployment systems, with measurable goals that support this aim.

The DOL should set a clear intent and measurable goals that guide the modernization of state UI systems

The Department of Labor (DOL) oversees the nation’s unemployment systems and defines the metrics that measure the success of these programs. In the past, funding for unemployment system modernization has been allocated but has not resulted in measurable improvements in the usability or stability of these systems. Before investing any money in modernization, the DOL should clearly outline what metrics are important to realize for state unemployment systems. Without these metrics, money will be spent, but it will not be clear if states are actually making progress.

The nation’s unemployment systems exist to provide people with financial assistance that helps them through a period of joblessness. However, these systems, similar to other “safety net” programs like the Supplemental Nutrition Assistance Program and Medicaid, often put more emphasis on reducing fraud, waste, and abuse than expanding access to services and their timely delivery. A report prepared by California’s Employment Development Department Strike Team details how the state consistently met the DOL’s improper payment rate but never met the standard for paying applicants on time. At the height of the pandemic only three states — North Dakota, Rhode Island, and Wyoming — met the federal standard for getting benefits paid within 3 weeks to 87% of applicants.

This selective focus, whether intentional or not, has real implications for how digital services are designed and the degree to which they put the burden of proof on everyday people trying to access them. Years of focus on needless scrutiny of eligibility has left unemployment systems lacking critical identity verification capabilities that result in billions in improper payments, increasing backlogs, and delaying payments to people who actually needed and rightfully deserved assistance.

The DOL should clearly articulate the goals and measures states should work towards through the modernization of the nation’s unemployment system. It should also provide states with the necessary resources to achieve them and then hold states accountable for those outcomes.

Invest in state-level improvements tailored to federal goals

Once the Department of Labor has established that priority, it should support state digital service teams as they invest in system improvements designed to meet the goals set by DOL. These investments must be structured to improve unemployment insurance systems, not just recreate a digital version of a flawed process. That means looking at policies, procedures, and non-digital interactions all with an eye to improving the experience for unemployment insurance applicants and government staff on the way to meeting DOL priorities.

While the central priorities set by DOL should be the driving factor behind modernization plans, our experience supporting high-traffic form submission and eligibility determination services points to a couple of clear areas for improvement. States that make these investments are likely to see major improvements for user experience, processing time, and workload on office staff.

First priorities

These two investments should be the primary focus for state programs that are working with antiquated systems that are difficult for people to use and rely on outdated and inflexible technology.

Simple, modern web interfaces. Many state unemployment technology systems were first built decades ago and haven’t been systematically updated since. They tend to layer on incremental updates due to policy changes that usually add complexity, not reduce it. State program employees are required to learn complicated, non-graphical interfaces that interact with mainframe systems. The public is often interacting with older web interfaces that are not much more than thin skins over the mainframe terminal experience. These tools tend to predate modern web applications, were not built with typical conveniences such as form input validation or support for names with accent characters, and often require users to learn arcane codes and processes to accomplish simple tasks. The result is hours of lost time and workarounds that drain state resources. Something as simple as a modern web interface built with friendly user-interface components for entering and updating data records would save an enormous amount of time and effort, improve the accuracy of data, and speed up the processing time. First and foremost, states should look to provide their constituents and their internal support teams with usable interfaces that support rather than hinder the administration of these programs. Working with the Department of Veterans Affairs, for example, Ad Hoc has seen how simply enforcing validation rules on forms for claims submission resulted in fewer rejections.

Migrate away from mainframe systems. We need to have a frank discussion about mainframes and their role in government services. We ourselves have argued that it’s a mistake to blame specific technologies. Instead, we should focus on outcomes and on making better choices moving forward. Mainframes are powerful tools, marvels of engineering that can process enormous volumes of transactions in mission-critical contexts across multiple industries. COBOL can be a productive and maintainable language. But when systems are failing, we must evaluate them for what they are, not what idealized modern versions of them could be. The reality is that, when you review the landscape of state UI systems, many are held back from progress because of their legacy, primarily mainframe-based, technologies. While today’s mainframes can run Linux and developers can create applications for them in modern mainstream programming languages, many state UI technology systems were first designed and built in the 1970s and 1980s with the predominant modes of system architecture and software engineering of the day. As requirements and policies are updated, changes to these systems generally accreted over time in a piecemeal fashion. Instead of being refactored and redesigned to meet contemporary expectations and needs, these changes have led to giant, unmaintainable codebases and complex, brittle interfaces. Today, we expect digital services to be available when we want them; it’s frustrating when servers have “office hours.” And yet, it’s the batch job processing modality that these mainframe systems tend to employ that accounts for hours of downtime every day, leading to workarounds, frustration, and delays in the delivery of benefits. Speaking of frustration, UI administrators and state employees are often forced to use system interfaces to do their jobs that have not been substantially updated in decades. The goal of a modernization effort is not to decommission a mainframe, in and of itself. But it is undeniable that legacy mainframe systems are blockers to progress in UI modernization. Even Wisconsin, home of the nation’s first and oldest UI program, has issued an RFI recognizing their mainframe needs to be retired to meet the goals it has for its program. This can be accomplished, in spite of formidable challenges typically associated with mainframe and legacy modernization, with careful planning, good technology choices, and quality execution by teams working closely together to phase out the dependency on a mainframe and migrate to a modern system.

A platform for UI systems

After developing modern web interfaces and moving away from mainframes, the future for UI systems should be a platform built upon modern tools that provide states the ability to adapt as new realities emerge. This new platform will be the basis for the next generation of unemployment digital services. The platform should include:

Resilient, scalable cloud infrastructure to ensure every public-facing UI website can comfortably handle the peak traffic states saw at the height of the pandemic. During the worst stretch of layoffs and closures, UI systems experienced an order of magnitude more demand than they typically handle. In one week in April 2020, that amounted to 35 claims being filed each second. 35 requests per second may be more than UI frontend systems were designed to handle, but that is a modest transaction rate for any reasonably architected cloud service.

An identity verification broker that works with federated identity providers to reduce manual verification checks. Identity verification is a commodity digital service now, used in many different industry contexts. The government needs to embrace more automated services that reduce errors and increase throughput. One caveat is that first-time job seekers, people without long employment history, and people without bank accounts or other credit histories may not have a footprint in such systems. So exceptions need to be handled in a way that does not unduly burden the applicant nor the administrator.

A consolidated design system similar to HealthCare.gov that states can elect to use and which provides a consistent look and feel. Design systems, such as the U.S. Web Design System, offer a higher level of building blocks for the design of user interfaces in the form of components of commonly-used elements. If used consistently across web and mobile digital properties, these components bake-in best practices like accessibility and responsiveness and unify discrete tools and services under a larger organization. The consistency and predictability provided by a design system gives confidence to users and helps establish trust with the digital service.

A platform of centralized services for common functions (notifications, messaging, payments) and general algorithms for UI eligibility accessible via APIs. These can act as an abstraction layer between modern digital services and legacy systems. Decoupling frontends from backends and reconnecting them via an API layer is a tried-and-true strategy for architecting digital services that are scalable, evolvable, target multiple end-user devices successfully, and extensible into other functional areas.

New tools to improve the UI experience

With a modern, stable platform upon which to build new digital services, states can begin to look at investing in new tools that will streamline existing processes for both the unemployed and state employees who administer these programs. The impact of these tools should be measured to see how well they move the needle for the state in achieving the goals set out by DOL. Some potential projects include:

An eligibility screener to triage applicants and reduce the load on the system and human eligibility reviewers. Screeners that provide an assessment — essentially an estimate — of an applicant’s eligibility based on a small amount of information provide several benefits:

  • They avoid expensive database queries or data processing, preserving system capacity.
  • They provide a higher quality user experience, in that the applicant can quickly get a sense of if they may be eligible for a benefit, and what the benefit may look like.
  • They can educate the user and prime them for the kind of info and supporting documentation that will make their claim more likely to be successfully processed, easing the burden on state administrators and speeding relief to the applicant.

A status tracker that helps people understand the status of their claim. Effective status checkers reduce uncertainty around status and can prevent duplicate applications that add to state backlogs. A tracker could provide email or SMS updates, offer a logged-in user interface with the history of a claim, and give the state more data to measure future improvements.

Document management that allows people to digitally upload documents for verification and for states to manage them centrally. Applicants should be able to upload from their mobile devices and take screenshots and photos of verifying information. Conversion of images into searchable and structured text, while still imperfect, has improved substantially in recent years thanks to commodity cloud-based models.

Additionally, states should seek to implement technology that empowers their staff by giving them more effective tools and resources and automating tasks so that they can focus on higher-value interactions in service of the public. Similarly, states should include government workers in their design processes and validate new approaches with staff and not just end users.

Build the right digital service teams at all levels

Transforming unemployment insurance systems across the country is a massive task that will need strong digital service support from the Department of Labor, at the state level, and from vendors ready to build modern government digital services.

Proposed bills to improve unemployment insurance and modernize state and local technology systems provide funding and access to federal resources like USDS, 18F, and GSA’s Technology Transformation Service to support improving local technology. Having technologists in government at the state level, in addition to the federal level, is an important ingredient to the success of improving unemployment insurance and systems.

Looking back at successful models of digital services teams in government, we’d propose that the Department of Labor and current digital service teams consider the following approach to establish a network of digital service teams that work together to further the priorities and outcomes set by the Department of Labor.

The Department of Labor UI Digital Service Team

At the federal level, the DOL should create a cross-functional team staffed with technology, policy, and regulatory experts to set the priorities for unemployment insurance systems and direct federal investment into components that can best achieve DOL’s desired outcomes. The complexity of the policy and the technology behind these systems means it’s critical that this team have a broad, cross-functional approach to their work.

Guiding standards and outcomes: As the nature of people’s jobs, lives, and locations change, it’s important for our systems to work together and adapt to change. DOL should define standards and outcomes for state systems that guide implementation and purchasing decisions, and enable an interconnected system of data that each state can contribute to and use. With product management, research, engineering, and user experience experts at the table with policy leadership, DOL can design priorities that are technically feasible and ready to translate into product North Stars at the state level.

Reusable components: Federally developed services don’t currently have the support and investment needed to compete with and support implementation across each of the 53 jurisdictions for unemployment. Instead of developing products that states purchase, DOL should develop reusable, shared components that vendor teams can implement in support of the states. For example: instead of developing a document management system that states would purchase/use from the federal government, the DOL digital service team could develop and make available a reference code base and outcomes that states could use to guide and jumpstart their efforts.

A centralized unemployment experience: Our economy has changed dramatically over the past decade with the rise of remote work and the gig economy. The current model for unemployment insurance is based on the premise that people work near where they live. This is no longer the case for a significant portion of the workforce. With these changes come new realities for people that are left unaddressed even with modernized state unemployment systems. The DOL digital services team should consider the implications of the changes in our economy along with emerging policy from the administration and how that might be realized in a centralized, federal unemployment service.

Teams at the state level

The bulk of modernizing unemployment will happen at a state level. Beyond priorities and components, the Department of Labor can support state governments as they build and expand their digital services teams with access to federal acquisition vehicles, procurement professionals with experience buying IT services, and implementation consultation with technologists.

These procurement and technology experts could embed at the state level in a model similar to Presidential Innovation Fellows. These fellows, embedded at the agency and office level, use their expertise to help agencies address challenges and build capacity in areas they haven’t developed yet. This same model could guide the procurement and implementation of systems and services towards a successful modernization of unemployment. It could also provide a beneficial feedback loop from folks in the states back to DOL to refine and update priorities as conditions change.

Regional sharing and support communities

The Department of Labor should establish or strengthen regional sharing and support communities for those working to modernize unemployment insurance. Today, groups like Code for America create an informal opportunity to share ideas and solutions across similar challenges. Creating specific opportunities for state teams to come together with DOL will help build capacity across the community, share feedback and findings, create solutions to shared challenges, and pool procurement and technology resources.

Access to digital service vendors

Even with strong product priorities and a central DOL digital services team, both the DOL and state agencies will need support from digital service vendors with native skills in human-centered design, DevOps, product management, and agile software development. If states continue to hire the traditional vendors who built the unemployment insurance systems that failed during the pandemic, they’re risking a similar outcome.

With procurement support from DOL, 18F, and USDS, state agencies should design contracts that put the government in day-to-day control of the product and use vendor evaluation tools that require teams to demonstrate real capabilities. The DOL should also help state agencies access federal contract vehicles to increase the pool of vendors to give states a better opportunity to hire competent vendors with consumer-industry skills. If state systems make it possible, states should allow for work to be completed off-site to allow a more competitive pool of vendors to bid on the work.

Ad Hoc has helped agencies like the Department of Veterans Affairs and the Centers for Medicare & Medicaid Services transform how they collect information and provide benefits and services to the public. We have experience working within existing systems, diving deep into the user needs of particular populations, and building robust, flexible digital services ready to meet any kind of demand. Our teams are ready to help states seize this opportunity to modernize their unemployment insurance services and better serve their constituents.

Thank you to Larry Bafundo, Ben Kutil, and Andre Francisco for their contributions to this post.