Yale School of Management

Program on Financial Stability

Improving our understanding and management of systemic risk.

Barriers to Wage Replacement Schemes

Access our live “financial-intervention” tracker, which is a single point-of-entry on the YPFS website to keep up with financial-crisis fighting interventions by central banks, fiscal authorities, and international organizations here.

Negotiations stalled last week on more federal relief to address the coronavirus crisis. The sticking point seems to have been unemployment insurance (UI). 

The Coronavirus Aid, Relief, and Economic Security (CARES) Act included a $600 weekly federal UI benefit on top of state benefits, called the Federal Pandemic Unemployment Compensation (FPUC). It gave two-thirds of claimants higher weekly earnings than their previous job paid. 

For these workers, Republican lawmakers say that expanded unemployment insurance (UI) offers little incentive to return to work. Republicans have therefore called for states to implement partial wage replacement rather than the Democrats’ proposal, which would simply extend the $600 supplement for the remainder of 2020. In the GOP’s wage-replacement scheme, claimants’ benefits would be equivalent to 70% of pre-pandemic wages. 

Senate Republicans included the wage-replacement scheme in the HEALS Act, a panoply of proposed bills that constitute the party’s response to the HEROES Act that the Democratic-controlled House passed in May.

Lawmakers had proposed this sort of scheme during the negotiations for the CARES Act in March. However, state unemployment administrations were not ready to implement such a program at the time due to fragmented systems, overlapping benefits, and aging information technology (IT). But some states will still struggle to implement a wage-replacement scheme by October, as the HEALS Act proposes.

Perhaps the largest problem is the federal separation of benefits administration. Unlike shared protocols such as the Uniform Commercial Code, where procedures are more or less standard across states, the United States has 52 separate UI systems -- one for each state, the District of Columbia, and Puerto Rico -- with 52 separate benefits calculations.

States also lack the earnings data that would serve as an input to such formulae. For year-round wage earners, funneling tax filings into a state’s UI system may solve the problem. But for seasonal employees or independent contractors, eligible for UI for the first time during COVID-19, determining benefit levels poses novel challenges.

UI systems nationwide have long faced disinvestment and tortuous development. Many of the computer programs used to process benefits are brittle -- some systems accept only fixed-dollar amounts, rejecting percentages -- and are written in a nearly-dead language. As a result, coders are scarce and maintenance difficult. States have reported that these technical problems alone would add up to a five-month timeframe for implementing wage-replacement formulae.

Some states began to overhaul their UI systems in response to failures during the April spike in unemployment claims and the initial federal UI supplement that the CARES Act provided. Employees in Connecticut’s Department of Labor worked 32 straight days to resolve its backlog. California’s governor appointed a “strike team” to modernize the application and distribution of UI benefits, while he prioritized communications with claimants. However, not every administration is able to marshal a timely digital response. Florida, which the New York Times reported has one of the nation’s worst UI systems, briefly reverted to paper applications.

States stand at various levels of readiness for wage-replacement schemes. Surmounting these challenges would require simultaneous (though not coordinated) action by every system. Should a third wave of federal relief mandate such a proposal, it would need to provide flexibility, additional support, or both to salvage cumbersome UI systems weighed down by millions of new and unfilled claims. However, wage-replacement features, such as those required by the HEALS Act, may end up standardizing state systems to some extent, allowing states to avoid some of the problems discussed below in future crises.

Wage replacement proposals

The Senate Republicans’ proposed HEALS Act would reduce the FPUC to $200 per week until October 5, when it would shift to funding a 70% wage-replacement scheme. The American Action Forum, a center-right Washington think tank, estimated that the average benefit under such a program would be $182 per week. The HEROES Act would simply continue the $600 FPUC through 2020.

The thinking behind the Republican proposal is straightforward: Why return to a job that pays less than UI would provide? As businesses seek to reopen, high unemployment benefits may depress demand for jobs at wages that can’t compete with claimants’ total benefits, which are made up of state and federal components. In contrast, the 70% scheme would aim to make pre-pandemic wages attractive enough that low-income workers will return to their jobs. As the pandemic continues to spread, many workers also may be reluctant to return to jobs that only exist outside their homes.

In the CARES Act, Congress expanded unemployment benefits. It extended support by 26 weeks and widened eligibility to include part-time and self-employed workers. More than 50 million people have filed unemployment claims this year. 

Data appeared to confirm the Republicans’ argument that the CARES Act subsidy, combined with existing state unemployment benefits, exceeded many workers’ earnings. In a paper published in May, researchers at the University of Chicago estimated that “68% of unemployed workers eligible for UI will receive benefits which exceed lost earnings,” with a fifth of those eligible receiving more than twice their lost earnings. The authors hypothesized that the $600 supplement could promote compliance with stay-at-home orders, but would dampen policies that sought to reignite hiring.

However, more recent studies have added to our understanding of the FPUC’s effects. For example, the May paper only compares wages to earnings, where relatively granular data are available. The National Employment Law Project analyzed data on UI recipients to estimate that claimants’ pre-pandemic compensation was, on average, 28 percent higher than the Chicago researchers had assumed because they excluded non-wage benefits (page 4). If so, then the effects of the CARES Act subsidy on rehiring may not be as important. Early scholarship on hiring and work scheduling fits this conclusion.

By a range of measures—hours worked, applications per job vacancy, employment probabilities—researchers have shown that the employment effects of the FPUC was likely not significant. This result seems to hold both in the aggregate as well as at the individual level. The first and third papers linked leveraged the Current Population Survey, which tracks tens of thousands of households over a four-month period. In this case, following individual households allowed researchers to distinguish between new hires and rehires; they concluded that higher rates of wage replacement, on average, did not affect decisions to return to work. Rather, they suggested that decreased demand for labor is the limiting factor.

Challenges to implementing a wage-replacement scheme

Even before the COVID-19 crisis, UI benefits faced several challenges. First, the federal structure of UI systems means that each state has its own, idiosyncratic program. Second, the existence of overlapping state and federal systems creates diverse categories of workers, each with distinct benefits programs, funded by different sources. Third, the systems that process unemployment claims are aging and frequently underfunded.

Problems with federalism

Each state administers UI benefits through its own program. This leads to considerable variation among states in the length of UI benefits, the minimum and maximum benefit, the method of calculation, and the data sources. Some states use weekly wages, while some use quarterly wages or annual wages.

The length of the basis period for calculating the highest quarter varies from state to state. Most states count the first four of the last five completed calendar quarters for calculating weekly benefits, but some do not. In 40 states, more recent earnings can be counted towards the weekly benefit calculation, in some circumstances. Forty states do not pay a benefit in the first week of unemployment, while 10 states do. Below is a table of selected states to illustrate the range of formulae.

Table 1: Sample range of formulae for UI benefit calculation

Method of calculation

Min benefit

Max benefit

Length of benefits, weeks

% of claims filled by June 15

How long does it take to decide on claim? (% decided on in <21 days)

Arizona

1/25 of highest quarterly wage (HQW)

$187

$240

8-26 

46.85%

10.8%

Delaware

1/46 of total wages in highest two quarters

$20

$400

24-26 

6.93%

32.6%

Nebraska

1/2 average weekly wage

$70

$440

10-26

51%

53.1%

New York

1/25 to 1/26 of HQW

$104

$504

26

64.93%

26.6%

Virginia

1/50 of total wages in highest two quarters

$60

$378

12-26

130.82%2

8.6%

Sources: U.S. Department of Labor, Significant Provisions of State Unemployment Insurance Laws; the Century Foundation, Unemployment Insurance Data Dashboard

Problems with overlapping systems

The existence of overlapping federal and state UI programs pose unique difficulties to updating the systems and implementing a wage-replacement scheme. Combining pre-existing UI benefits systems and new pandemic-related programs, there are four different programs. The last two programs on the following list were created by the CARES Act:

  1. Standard state systems 
  2. Extended state benefits 
  3. Pandemic Unemployment Assistance (PUA)
  4. Pandemic Emergency Unemployment Compensation (PEUC)

Standard state systems pay benefits for at least eight and not more than 26 weeks. These benefits are funded in full by states. When a state’s unemployment rate is higher than a certain level, unemployment benefits can be extended up to 13 additional weeks. The federal government pays 50% of extended benefits. Upon the expiration of extended benefits (EB), workers are eligible for Pandemic Emergency Unemployment Compensation (PEUC), which is funded in full by the federal government but still administered by states. 

Pandemic Unemployment Assistance (PUA) adds an additional $600 per week for workers who do not typically receive state UI benefits. These include sole proprietors, gig workers, part-time employees, and those furloughed as a direct result of COVID-19. It also pays a minimum weekly benefit equal to 50% of each state’s average weekly benefit, although states are mandated to phase in fuller payments

While state systems vary considerably, all those who qualify for pre-pandemic state-provided UI benefits submit wage records. So, in principle, states could calculate a 70% wage replacement rate for each worker under existing UI programs. 

But the CARES Act does not require self-employed workers to supply wage records to state agencies to receive the minimum benefit. Therefore, a wage-replacement scheme would, at present, be impossible to implement for this category of worker.

In the week ending August 8, 831,000 people filed initial claims for state-provided UI, while 489,000 people filed initial claims for PUA. 

Technical problems

Technological disinvestment and the state of information technology procurement have afflicted UI benefits programs for decades. Between 2001 and 2020, nationwide funding for UI systems declined in real terms from $3.27 billion to $2.14 billion (p. 4). And much of the remaining funds flow to failed projects: high-profile failures anchor estimates that, worldwide, only 13% of government IT projects larger than $6 million succeeded. Some states, such as Michigan, have successfully fixed broken systems, but the appropriations process that rules UI funding make iterative development—the core of modern software development—difficult to execute.

A recent report from the Century Foundation revealed that only 60% of unemployment claims filed between March 1 and June 15 were paid by the end of June. The unprecedented recent strains have laid bare fractures that have grown for years. Between 2007 and 2017, the false denial rate has more than doubled from 8% of unemployment claims to 17%. Michigan’s system offers an egregious example: after the system’s deployment in 2013, “The MiDAS system flagged more than 40,000 workers for fraud, and it was 93 percent inaccurate.”

Yet, Michigan’s commitment to its system catapulted it to among the fastest states for processing unemployment claims during COVID-19. Governor Gretchen Whitmer’s executive orders early in March are part of the reason for this improvement. Following the spate of false denials that marred the system’s deployment, lawsuits and legislation mandated improvements to the state’s system. But continuous improvement is the exception, not the rule, in government-procured IT. Government contract administration discourages the testing and iterative development standard in the private sector. These incentives result in big-ticket procurements that fail when scaled up to millions of users.

The appropriations cycle has also exposed IT procurement to procyclical funding, which poses problems for UI systems that face, by definition, countercyclical demand. Michigan suffered, along with other states, fiscal austerity following the Global Financial Crisis and Great Recession. In that case, the revelation of massive errors in MiDAS spurred pressure to reform. States with quieter failures still languish at the bottom of the claims-processing totem pole. Florida, which neglected important reforms to its $77 million system, briefly reverted to issuing paper applications, and, as of June 30, had paid only 52% of claims filed since March.

Only 16 states have fully modernized their UI systems, according to the National Employment Law Project. Aging systems often rely on COBOL, a 70-year-old programming language used to access mainframe computers. Only 37 schools in the world offer a mainframe computing course. In April, New Jersey’s governor appealed to retirees fluent in COBOL’s syntax to help fix its system.

But, even if states could rewrite their code to define benefits in terms of prior earnings, the PUA has expanded eligibility to self-employed workers and other workers who don’t report their earnings. Depending on the level of integration between UI administrations and revenue services, the ease of importing these data will vary. But the case of Virginia shows how technological problems can compound those of federalism: because many Virginia citizens worked in DC or Maryland, the state has handled many claims originating outside its borders; to use the earnings data of these citizens, it would have to coordinate with the DC’s and Maryland’s UI systems.

Responses to proposal

Taking together the existence of 52 unique state-level unemployment programs and two major federal programs, a 70% wage replacement program is likely to take weeks or months to implement. A memo from the National Association of State Workforce Agencies (NASWA) estimates that changing FPUC to be tied to individuals wage records would take from “four to twelve weeks or more” to implement. 

The debate over the next fiscal spending bill continues, and there is some indication that the final bill won’t include the Republican wage-replacement proposal. The President recently signed an executive order extending a $300 per week federal benefit through December 2020, and demanding that states cover an additional $100. 

Conclusions 

A number of logistical and administrative problems have plagued unemployment benefits programs for decades. State UI systems collectively faced some 33 million claims by the end of May. This strain exposed deep flaws in states’ technical capacities and reporting. The federal programs tailored to the pandemic-caused recession have imposed new standards on old, fragmented systems, further complicating the prospects for a speedy implementation of wage replacement.