Executive Summary

  • Updating, modernizing, and improving the Unemployment Insurance (UI) system is a necessary step for ensuring that the United States is well-positioned to protect workers and families against economic shocks – whether from pandemics, trade disputes, wars, automation, artificial intelligence, or ordinary churn in the labor market.
  • The UI system fails in three ways – it does not provide claimants with sufficient support to pay their bills while they look for a new job that matches their skills (Assistance), it does not ensure that people who lose their job can reliably access their benefits nor stop fraudulent claims (Access), and it fails to activate claimants to look for new work (Activation).
  • Recommendations for improving assistance include setting minimum weekly benefits, increasing standard replacement rates, reforming maximum weekly benefit amounts, and extending the duration of regular benefits.
  • Recommendations for improving access include the creation of a federal back-end to enhance the technical and administrative capacity of state UI agencies and the standardization of benefit calculation rules across states to simplify administration.
  • Recommendations for improving activation include replacing work-search requirements with a proactive approach emphasizing job-search planning, accountability follow-ups, and periodic counseling, and increased funding for active labor market policies.


For America’s unemployment insurance (UI) system, the early months of the Covid-19 pandemic were a combination of historic highs and lows.

On the negative side, the unemployment rate shot up to levels not seen since the Great Depression. The public health crisis gave every sign of becoming an economic catastrophe, as well. The mass hardship overwhelmed the UI system, such that many people who lost their jobs did not receive UI for months.

But there were positive developments, too. The pandemic brought out the best in our government’s ability to respond to an emergency. Within weeks of the first reports of community spread of Covid-19, Congress passed the Coronavirus Aid, Relief, and Economic Security Act, or CARES Act. That legislation addressed several big gaps in the UI system. It added $600 a week to benefits, which had previously averaged only 40 percent of lost wages. It extended UI to millions of gig workers and others who were not previously eligible for benefits. It modified work-search rules that required UI recipients to document active work search. It also allowed people to continue receiving benefits if they quit because of unsafe work conditions. Even after the initial CARES Act spending expired, many of the law’s provisions were extended by new mechanisms. Both President Donald Trump[1] and President Joe Biden[2] enacted subsequent measures to continue the added benefits, although at a lower level.

These and other pandemic relief measures saved millions of American families from extreme hardship. The economy bounced back faster than most analysts had imagined possible[3] – three times as quickly as it had recovered from the 2008 recession.[4] But not everything went smoothly. Huge programs enacted on the fly proved vulnerable to massive fraud[5]. Most importantly, now that most of the temporary pandemic measures have expired, we are left with a UI system that has been tried, tested, and found wanting.

It is time to face the fact that the UI system we have now is not ready for the next crisis. Worse than that, it cannot do its job even in normal times. The remainder of this paper will examine the shortcomings of the American unemployment system in detail and recommend numerous reforms. Making the UI system function effectively requires us to understand the current problems, and to make targeted investments in the system.

Functions and design features of the UI system

Unemployment insurance is nearly a hundred years old. It originated in the United States back in the 1930s, when as much as a quarter of the workforce was out of a job. A handful of states, starting with Wisconsin, implemented their own programs before the Social Security Act of 1935 established unemployment benefit systems across the nation.[6] This section provides an overview of the primary functions that UI serves and the key design features that determine how well it serves them.

Starting with the functions, UI is, first of all, a form of social insurance. Like disability insurance, health insurance, and old-age pensions, it protects people against what Friedrich Hayek called “common hazards of life” — risks that most people cannot adequately protect themselves against through their own means.[7]

The second function of UI is to support workers in their search for jobs that are a good match for their abilities. Work search and matching take time. Without income support, workers would be under pressure to take any available job, whether it was a suitable match or not. Although we want workers to find jobs rapidly, we also want them to find jobs that suit their skills and preferences. Better matching helps employers maintain a productive workforce and improves the labor market’s operation as a whole.

Finally, UI serves a macroeconomic function as an automatic stabilizer that helps smooth the business cycle. During recessions, when the unemployment rate rises, UI benefits support consumer demand and put the brakes on what might otherwise become a cascade of job losses. In doing so, UI not only keeps aggregate spending from falling off a cliff but also helps unemployed workers pay their rent or mortgage, so that their landlords and banks do not lose their incomes as well. During recoveries, UI benefits automatically fall off as workers find new jobs. That, in turn, reduces the risk that a healthy expansion could overshoot into inflation.

To perform these three functions effectively, a UI system must be well designed. Three key design features are the rules that govern assistance, access, and activation.

Assistance rules determine how generous UI is. One key parameter is the generosity of weekly UI benefits. This can be summarized as the replacement rate, the percentage of an unemployed worker’s previous earnings that are replaced by UI benefits. Another key parameter is the duration of benefits. As economic conditions change, additional rules may allow for automatic or discretionary changes in replacement rate and benefit duration.

Access rules determine who is eligible for benefits. As a form of social insurance, UI focuses on workers who lose their jobs for reasons beyond their control, such as cyclical downturns in demand, mergers, changes in the way the employer conducts business, or for poor performance. Depending on the state and on the specific circumstances of a termination, UI is typically not available to workers who are fired for misconduct or who quit their jobs of their own volition. This can lead to administrative difficulties, as the line between “quit” and “fired” is often unclear.[8] (Several states do recognize situations in which employees who quit are eligible, such as victims of workplace sexual harassment.[9]) New entrants to the job market, self-employed individuals, and gig workers are also typically ineligible to receive benefits. However, these are not absolutes. For example, as mentioned above, access rules for UI were considerably eased during the Covid-19 pandemic.

The third type of rules, known as activation rules, limit eligibility for UI benefits to workers who are actively looking for jobs. As such, states typically require UI beneficiaries to periodically submit documentation to demonstrate that they have been searching and applying for jobs. Furthermore, states can mandate that people accept any “suitable work,” though the state and the job seeker may differ on what jobs are appropriate. Failure to document work search or to accept an appropriate job offer can lead to loss of eligibility. In addition, states may offer voluntary or mandatory training or job counseling.

A UI system that includes a good set of assistance, access, and activation rules will be better able to serve its functions of social insurance, search and matching, and macro stabilization. However, there is no simple one-to-one relationship between functions and rules. A change in any of the three kinds of rules can affect any of the functions in ways that involve complex tradeoffs and complementarities.

For example, an increase in the replacement rate or the duration of benefits in the name of better social insurance could be seen as reducing workers’ incentives to actively search for new jobs. However, up to a point at least, more generous benefits give workers a chance to find a better job match. Similarly, a longer duration of benefits that avoided a cutoff of demand support before the recovery from a recession is complete could speed rather than slow a recovery. Also, stronger activation rules can offset any negative work incentives resulting from more generous assistance.

Even more complexities arise when we consider the relation of UI to other areas of social and economic policy. UI interacts with other social insurance programs, from food stamps to child support to housing programs to disability insurance. Sometimes the programs complement one another, but often there are perverse unintended consequences. For example, some research suggests that inadequate benefits cause workers to turn to disability insurance as a substitute and permanently drop out of the labor market.[10] Similarly, a UI system that provides inadequate automatic stabilization can complicate the conduct of both monetary and fiscal policy.

Still another source of complexity is the fact that responsibility for UI in the United States is divided between the federal and state governments. The current system suffers from a lack of uniformity in rules from state to state and imposes an excessive burden on the limited administrative capacity of state governments. We return to this problem later in the paper.

The following sections will examine the design of the American UI system in detail, keeping as many of these complexities in mind as possible. At the end of each section, we suggest potential reforms that can make the UI system as a whole function better.

Assistance: Ensuring that UI functions as a consumption-smoothing benefit

One important effect of UI is to smooth the consumption level of jobless workers while they search for new employment. As social insurance, benefits need to be high enough to keep beneficiaries out of poverty and ensure they can continue paying basic expenses. That, in turn, gives them the time look for jobs that suit their preferences and skills. At the same time, these benefits support aggregate demand for goods and services during economic downturns.

Current policy does not provide adequate income-smoothing

Unfortunately, the UI programs of most states fall short of adequate income-smoothing. The average UI recipient receives benefits that are less than half of their prior wages. Furthermore, in a typical year, more than 30 percent of UI recipients see their benefits expire before they secure a new job. (And even if cash from UI benefits stabilizes a family’s overall income, it does not provide them with health insurance, retirement, dental, or any other benefits that are typically employer-sponsored).

Other major countries provide bigger and longer benefits. A majority of countries in the Organization for Economic Cooperation and Development (OECD) have replacement rates over 50 percent, and rates over 60 percent are common, with benefits often lasting for more than a year.[11] That is over twice the typical duration of eligibility in the United States. Some American policymakers fear that larger benefits and longer durations will discourage work and hurt our economy. However, that concern is misguided. Keep in mind that up to a point, taking more time to search for a job is likely to lead to a better match. In the aggregate, better matching strengthens the economy by allowing workers to find jobs that maximize their productivity. That is probably one reason why countries from Scandinavia to Canada are able to provide better income support while maintaining higher employment rates than the United States.[12] We should be able to do both, too.

Some research does show that enhanced UI benefits can deter a return to work at the individual level. However, those effects are less apparent at the population level. Studies using data from the two most recent two recessions suggest that any adverse work incentives of more generous UI benefits are modest.[13] The extra benefits[14] provided to American workers during the Great Recession appear to have had only small effects, if any. More recently, the enhanced benefits awarded during the Covid-19 pandemic provided what was in essence a controlled experiment. Analyses of that experience also show small, if any, effects on employment [15] . In fact, states that chose to end the extra benefits early did not see faster employment growth than those that maintained benefits until the federal funding expired.[16]

In short, the potential negative effects of high UI benefits should not be overstated. Those effects, if any, should be balanced against the benefits of better consumption smoothing, better job matching, and stronger support of aggregate demand.

The case for higher benefit levels

Although higher benefits will be needed to ensure adequate consumption smoothing, the path to reform is made more difficult by the complexity of the decentralized American UI system.[17] Each state now has its own benefit rules, including different formulas for calculating benefits, as well as set maximum and minimum weekly amounts.[18] Adjacent states can insure dramatically different portions of prior wages.[19] This section will examine many aspects of those assistance rules. Recommended reforms include specific benchmarks for minimum and maximum benefits, and simplification and standardization across states.

One source of differences in benefits is the failure of nearly half of states to automatically adjust the maximum benefit amounts with inflation.[20] When the Social Security Act first established unemployment insurance, the typical benefit limit was set at approximately 60 percent of weekly wages.[21] As of 2021, however, benefit maximums had fallen to slightly below 40 percent of weekly earnings. This decline in benefits is partly due to gradual erosion and partly due to recent, intentional reductions. As a result, unemployment insurance in many states no longer replaces an adequate portion of middle-class incomes. Only half of states have maximums that replace at least 50 percent of $1,000 in weekly earnings. [22] It is not a coincidence that 23 of those are states that automatically adjust their benefit caps to track wage growth.

Reductions in the size and duration of benefits interfere in fundamental ways with the basic functions of UI. Consider, first, the job matching function, which is so important to overall labor productivity. When benefits decline, workers are put in increasingly vulnerable positions. As shown in Figure 1, the average American family of three spends 65 percent of its income on housing, transportation, food, healthcare, and utilities – all essentials.[23] Obviously, a 40 or 50 percent replacement rate is not enough to cover these items. Other expenditures, which include educational costs and taxes, take up an additional 17 percent. With little financial security between jobs, workers must scramble, not search, to find new employment. As a result, they end up with lower salaries and worse job matches.

The impact on job matching is confirmed by studies that compare firm-level job data in states that enact benefit reductions with those that do not. One recent analysis found that states that reduced the maximum benefit duration saw declines in starting salaries.[24] North Carolina, which cut both maximum weeks and maximum weekly benefit size, saw an especially sharp drop in the salaries of new hires. The study attributed those lower earnings partly to occupational downgrading and partly to job seekers’ weaker bargaining power, both resulting from benefit reductions.

The pressures that lead to poor job matching are not limited to the poorest workers because unemployment creates an immediate problem of “liquidity” – having readily available, or “liquid” cash.[25] Unemployed people may own substantial assets, such as cars or houses, that they cannot readily turn into the cash they need to finance consumption while unemployed. Not only is it hard to turn such assets into cash, but shelter and transportation are essential to effective work search. When benefits increase, workers can afford the time to establish better caliber job connections.[26] Workers are more likely to remain in higher-quality jobs for a longer time – which is good for both employees and employers.

The effects of inadequate UI benefits go beyond the short-term problems for work search. Data shows that the typical American adult experiences around six unemployment events by age 54.[27] Families must plan accordingly – and not only when thinking about basics like housing.

For example, when UI benefits are less generous, it is more likely that both spouses will work.[28] Although that may suit some couples well, others may have young children that they prefer to have one partner care for at home. Meager unemployment benefits make it difficult to justify doing so because the breadwinner’s joblessness could mean a catastrophic income shock. Multi-earner homes are better positioned to weather individual unemployment events: If one partner loses their job and UI replaces only half that income, they still have the other partner’s full wage, reducing the hit to household income. Increasing the generosity of UI benefits would help solve the problem by restoring parity between single- and multi-earner households. Families could then more easily sustain their family and childcare preferences in the face of periodic unemployment spells.

The case for longer benefit durations

Adequate replacement rates are not enough to ensure that UI systems are able to fulfill their functions. The duration of benefits also matters, not only if UI is to fulfill its social insurance and matching functions, but also for macroeconomic stabilization.

Although the average UI recipient requires less than the maximum number of benefit weeks[29], a third of recipients exhaust their eligibility before finding new work.[30] That percentage has risen as national investment in re-employment programs has decreased.[31] Once benefits expire, unemployed workers’ spending levels fall substantially.[32] Allowing regular benefits to last longer would provide better protection with a fairly small impact on the average duration of benefits.[33]

Everyone understands that it takes more time to find a job when the economy is in a downturn.[34] What fewer people realize is how many people are out of work for longer than the maximum duration of UI benefits even in good years. The Bureau of Labor Statistics counts people as long-term unemployed if they have been out of work, but continue to search, for 27 weeks or more. More than a million fell into that category even in years like 2019 or 2022, when overall unemployment was near historic lows.[35] For them, finding a job can be difficult no matter the state of the economy.

The longer someone is unemployed, the more challenging it is for them to find full-time work.[36] Companies discriminate against the long-term unemployed[37] and they are less likely to receive job offers.[38] Cutting off UI benefits in these situations fails to address the core problem and can make the joblessness trap even harder to escape.

Key recommendations

As the previous sections have argued, more generous assistance to the unemployed would be good for the individual worker and good for the economy. Higher and longer-lasting benefits should be viewed as complements to other policies, such as assistance with work search, that are designed to shorten the time between jobs. The long-term unemployed in particular already face an uphill climb to secure their next jobs. Making it possible for them to pay their bills while they do so will not only improve their lives, but will facilitate matching of workers’ skills with employers’ needs. What is more, enhanced UI benefits would help stabilize the economy during cyclical downturns.

Policymakers should consider four major changes:

  • Set a minimum weekly benefit equal to at least 35 percent of the state median household income. That minimum should be adjusted annually for changes in incomes. For example, in Mississippi, which at $894 had the country’s lowest median weekly income from 2016 to 2020, that would translate to a minimum benefit of $313, higher than the current minimum benefit of all but one state (Washington).[39]
  • Establish an income replacement rate of 75 percent for dollars earned beyond the minimum benefit. For unemployed workers to remain financially stable between jobs, three-quarters of prior earnings that exceed 35 percent of the state median household income level should be replaced. As shown earlier (Figure 1), that level would enable the average household to cover the costs of housing, transportation, food, utilities, and healthcare, and leave a few additional dollars for expenses like education. We recommend that the benchmark be based on the income earned in either the first four or the last four of the previous five calendar quarters, whichever is greater.
  • Make the maximum weekly benefit at least equal to the state median household income. The maximum benefit level should be high enough to ensure that middle-class workers can cover their essential spending between jobs. To use the Mississippi example once again, this recommendation would raise the state’s maximum benefit level to the median household income of $894 a week. Under this formula, Mississippi workers with annual incomes up to about $95,000 would have had 75 percent or more of their previous wages replaced while searching for new employment.
  • A 39-week regular benefit duration. Recipients should be able to receive regular benefits for up to 39 weeks, a 50 percent increase over the 26-week eligibility period that most states use. This recommendation, which has been proposed for decades,[40] would bring the United States more in line with other wealthy nations. During severe recessions, benefit duration could be extended as needed for purposes of macroeconomic stabilization.

Experience rating

Increasing unemployment benefits will require increases in funding. UI is currently funded through a system of experience rating. Employers pay a tax that is a function of the UI benefits drawn by their former employees. Details differ by state. Most states use a step function where taxes increase incrementally as claims increase, and most also have minimum and maximum tax ratings.

The conceptual purpose of experience rating is to give employers an incentive to reduce the number of employees they fire and the number of unemployment claims filed by their former employees. The concept was drawn from the UI system that Wisconsin used in the 1930s when the federal system was first developed. However, experience rating has two important problems.

First, whether an employee quits, is laid off, or is fired is more of an administrative gray area than it might seem. Employers that want to terminate workers, but do not want to damage their firms’ experience rating, may try to induce that worker to quit, or challenge whether a worker is fired or laid off. Such efforts by employers to reduce their former employees’ ability to collect benefits degrade the effectiveness of the entire UI system.

Second, while experience rating incentivizes firms to limit layoffs, it also incentivizes them to limit hiring of workers they may have to lay off later. For example, they might instead prefer to handle short-term increases in work by assigning overtime to existing employees. They might also be reluctant to hire employees with marginal qualifications. That could make it more difficult for people without a college degree or with limited work experience to find jobs.

These side effects not only make life more difficult for workers, they also detract from productivity. In effect, experience rating acts as a tax on job turnover that ultimately creates a sclerotic labor market. Replacing experience rating with a flat payroll tax could significantly improve the dynamism of the economy as a whole.

Access: Reforming eligibility and reducing paperwork

The access rules that determine who is eligible for benefits are another chronic source of problems in the UI system. Two of the most serious of those problems – fraudulent claims going through and legitimate claims getting stuck – were cast into sharp relief by the temporary UI enhancements of the pandemic era.

Consider first the spike in fraud during the pandemic. Many UI claimants turned out to be unqualified. A large number of fraudulent claims involved stolen identities. Estimates by the Department of Labor’s Inspector General have suggested that nearly $46 billion was distributed fraudulently, approximately 5.2 percent of the $872 billion distributed during the pandemic.[41]

At the same time, though, many people who were qualified for benefits had difficulty actually receiving them. Payment of UI benefits fell substantially behind the rate at which claims were filed. Many of the 23 million people who were out of work in April 2020, when unemployment peaked, did not receive their benefits until months later – if they received them at all. It was a poor showing even for a system in which, under normal circumstances, only a quarter of unemployed people actually receive UI benefits.

It might seem that there are inevitable tradeoffs here. For example, more rigorous identification checks could reduce fraud, but those checks would slow down the application process and could cause some deserving applicants to miss out on benefits. In short, it might seem that reducing the number of false positives would inevitably increase the number of false negatives.

But that perception is a mirage. Ultimately, both fraud and denial of valid claims are symptoms of an ineffective unemployment insurance back-end, which pandemic legislation at most was able to patch temporarily. It should be possible to design an unemployment insurance system that can get benefits efficiently to valid applicants without vaporizing more than $40 billion in fraudulent claims.

As this section will show, making the necessary changes will require building state capacity through substantial investment in the nuts-and-bolts functioning of the UI system. At the same time, it will be necessary to reduce the complex administrative and regulatory burdens that the system now places on claimants. The prospects of doing so would be substantially improved by creating a standardized federal UI back-end for states to use at their discretion.

Processing claims and fighting fraud

The state of Florida, famously for its decisive role in presidential elections, is often a bellwether for the nation. The recent history of reforms to its unemployment insurance program is no exception. Florida’s experience with modifying its UI system in the last few decades illustrates many of the problems that have hampered other states as well.

During the Great Recession of 2008-2009, Florida’s unemployment insurance program was hampered by the lack of an online application option. In the aftermath, administrators decided to create an online portal where UI claimants could apply for benefits and submit required documentation of efforts to look for work.[42]

In May 2010, Florida’s Agency for Workforce Innovation issued a Request for Information, inviting bids to lead a UI modernization project. The project was awarded to Deloitte Consulting LLP, which proposed to do the job for $40 million. The project ran into problems immediately. Delays caused the government to issue an Intent-to-Terminate notice before a year had passed. That was later withdrawn in favor of a 10-month extension. A further extension was needed when errors were discovered as the new website went live. Eventually the cost nearly doubled to $77 million.[43]

Meanwhile the state faced difficulties in processing unemployment claims. As shown in Figure 2, before the Great Recession, the rate of approved claims to all unemployed persons was about one-third both in Florida and the United States as a whole. By 2016, the national rate had declined to about 28 percent, as states tightened standards to make up for budget shortfalls. In Florida, however, the recipiency rate fell much further, hitting a low of 10 percent.

The sharpest drop occurred shortly after Deloitte’s new system went online. Many of the delays were caused by added procedural hurdles that the new system introduced. For example, claimants were required to complete a 45-question standardized test, similar to an SAT, that covered math, reading, and research skills (this program was eventually repealed).[44]

These issues became even more pronounced during the Covid pandemic. All states faced an increase in UI claims, but Florida was less equipped to deal with the surge. National legislation expanded unemployment eligibility to gig workers, recent graduates, and other people without long work histories. In response, the national recipiency rate rose to an all-time high of 80 percent, while Florida’s lagged far behind, at 40 percent.

The collapse of Florida’s UI program led to a round of finger-pointing. Governor Ron DeSantis blamed the previous administration of Rick Scott, arguing that the UI website was “designed to fail.”[45] Scott blamed Deloitte, noting the attempts to terminate the contract. Finally, Deloitte blamed the administration of the program during the pandemic.

At some level, it does not matter who deserves the blame. Fundamentally, the system is broken. Writing more widely of America’s loss of state capacity, Brink Lindsey has pointed out[46] that expensive contracted systems of this kind have failed time and time again. Outsourcing work to contractors is supposed to bring private sector discipline to bear on complex governmental systems. In practice, though, the opposite is the case.

Part of the problem is that the private firms in question specialize not only in the contracted activities, but in winning contracts and encouraging the government to create new contracts for them to bid on. In the process, poorly paid government employees have every incentive to move to the private sector, where they can be paid more for the same work. Over time, the capacity of the state even to supervise work effectively is depleted, as no one currently on staff has the requisite skills and experience for large projects.

These issues are especially pronounced in the case of information technology projects. Budget constraints encourage states to pursue cheap solutions, which inevitably need to be patched and upgraded, often for more than it would have cost to do the job right in the first place.

A 2020 report on the UI debacle from the Florida Office of the Inspector General reached similar conclusions.[47] “Agencies should know what they want” and should “better monitor what they are getting from the vendor and build in an escape plan and financial penalties for noncompliance,” the report scolded. It further recommended that “verification processes should be independent and rigorous.” Such recommendations do not require advanced knowledge of software engineering. Rather, they are project management standards that are widely adhered to elsewhere.

Such issues are only compounded when fraud occurs. As Figure 3 shows, at the national level, the bulk of the fraud during the pandemic did not come from individual workers lying about off-the-books jobs while drawing benefits. Rather, it was executed by sophisticated gangs that accessed unemployment benefits with false identities. The biggest problem appears to have been individuals who applied for, and received, benefits in multiple states. Those accounted for nearly two-thirds of the fraud identified to date.[48]

The investigation has been hampered by difficulties in coordinating a response across 50 states (in addition to two territories and Washington, D.C.) that run their own UI programs. The Labor Department’s inspector-general has had to issue multiple subpoenas to states. Much of the data it has received was incomplete, needed substantial processing, or was simply unusable. In short, while fraud was committed by sophisticated actors working across multiple states, the federal response has not been nearly as coordinated.

One potential solution, not just to fraud but also to lack of capacity to keep up with normal demands, would be to federalize the unemployment system. State governments do not have the resources either to create and maintain UI systems or to adequately manage contractors. The federal government arguably has much greater capacity. In addition, the federal government would be better able to coordinate fraud prevention at the national level.

Short of full federalization, another option would be to have the federal government design and operate a customizable back-end system that states could use to run their own UI programs. Doing this would require federal reforms that harmonize how states determine UI benefits and align their business practices, including more consistent data sharing[49]. In fact, multistate UI programs already exist, such as one run by North and South Carolina.[50] Another example is the “Reemploy USA” project among Connecticut, Maine, Rhode Island, Mississippi, and Oklahoma.[51] While states will want to customize their UI programs in some respects, a good website design or software back-end looks much the same in Missouri and Massachusetts. A federal framework could allow similar customization while making it possible to process routine claims and fight fraud more effectively.

Standardizing UI replacement rates

The UI system is hamstrung by the significant variation in access rules across different states. The complexity and inconsistency of these rules make it challenging for workers to determine their benefits, let alone compare them across states. Additionally, it leads to over-complicated computer code that cannot be easily modified for emergencies.

Consider, for example, the rules that govern the replacement rate just in states that begin with “A”[52]:

  • Alabama looks at your earnings in the previous year by quarter. The state then averages the two quarters in which your wages were the highest and multiplies that by 1/26 to determine your weekly benefit.[53]
  • Alaska looks at your wages in the previous year. Your weekly benefit is set as a percentage of your annual wages, with the possible percentage ranging from 0.9 to 2.2 percent. In addition, your benefits are increased by $24 per dependent.
  • Arizona looks at your highest-quarter wage (in the four quarter “base period” before filing) and sets your weekly benefits to 1/25th of your wages in that quarter.
  • Arkansas looks at the four previous quarters, takes the average wage, and multiplies that by 1/26.

Rules in the other 46 states are similarly varied. States they might use the highest quarter’s wages for the previous year, average wages, or some combination of the two. States like Alabama and Arizona apparently disagree about how many weeks are in a quarter.

Some states also use their UI rules to pursue other social goals, however ineffectively. For example, adding $24 per dependent, as Alaska does, is a nice gesture, but the actual amount is not much for a struggling family. Furthermore, it duplicates federal and state child tax credits that are already available.[54] If the aim is to provide families with more support, it would arguably be better to do so through programs like child allowances instead of through the back door of UI.[55]

All the complexity diminishes the effectiveness of the UI system. As one glaring example, during the pandemic, complexity and coding problems made it impossible to pursue common-sense measures like raising the replacement rate to 100 percent of previous earnings. Instead, policymakers decided on the fly to top up normal weekly benefits by a flat $600 (later $300).

Another issue, less often explored, is how all the complexity affects the intelligibility of the UI system. In practice, it is often impossible for people to know what their UI benefits will be and plan accordingly. Nor can they readily compare benefit levels in other states in making location decisions or in urging their representatives to improve benefits.

Our recommendation is to wipe the slate clean. Instead of 53 states and territories using 53 different base periods, replacement rates, number of weeks in a quarter, and the manifold other small and not-so-small differences, states should have structurally similar UI programs that are easy to conceptualize, compare, and program.

The simplest way of doing this would be to have states set only one parameter: the replacement rate, defined as a percentage of average earnings in the “base period” preceding unemployment.[56] As stated in the previous section, we’d recommend a 75 percent replacement rate. But states could set their benefit level above or below that target as they see fit.

It might be objected that the differences between UI rates among states are carefully designed to optimize for each state’s unique situation. However, that does not appear to be the case. There does not appear to be any systematic relationship between economic conditions in various states and their UI rules.

For example, we can compare states that base UI benefits on “high quarter” earnings (37 states) with those that use an “average wages” approach (13 states). We might expect that the states using the highest quarter might be doing so because their state has more variable labor markets. For example, states where agricultural or tourism-based industries dominate have more seasonal variation of employment and earnings. Policymakers might plausibly take that into account when setting rules. However, a previous study[57] shows no statistically significant relationship between the variability of unemployment levels and the choice of highest-quarter or average-quarter wages as the basis for UI benefits. Instead, states appear to have made their UI calculations more complicated over time in a quasi-random walk.

It seems that when legislators decide to make benefits more or less generous, they change the baseline, add dependent benefits, multiply the whole formula by a constant, or tweak the rules in some other way to reach the desired outcome.[58] Under our proposed uniform rules, states would still have the option to make benefits more or less generous, but they would only be able to do so by directly changing one of the designated parameters.

Key recommendations

We recommend the following key changes to access rules:

  • Create a federal back-end to enhance the technical and administrative capacity of state UI agencies. Although this could be a mandatory program, it would be more realistic to offer it on an opt-in basis. It would relieve states of the massive burden of solving the “old code” problem by providing a clean-slate system that would be modern and easily updatable. It would also enormously improve the ability of states to fight fraud.
  • Standardize benefit rules across states. This reform would focus not so much on the level of benefits, which states could still determine for themselves, as on definitions of previous income and other administrative details. Standardization would best be implemented through the proposed federal back-end, but in principle it could be a stand-alone measure.

Activation: Getting people back to work

Activation is the third feature of American UI policy that needs extensive reform. Activation encompasses a broad range of specific policies, often known collectively as active labor market policies, or ALMPs.

One of the most basic activation policies is to require people to demonstrate they are actually engaged in work search. Job-search requirements may be supplemented by active advice on how to identify and apply for job openings. ALMPs also include education and training both in specific skills such as welding or programming and in more basic job skills such as showing up to work on time. Finally, ALMPs can include measures to expand job availability, for example, by ensuring that more jobs are accessible to people with disabilities. Subsidized private-sector jobs or guaranteed government jobs represent the far end of the spectrum of ALMPs.

An international comparison of activation policies shows a correlation between the generosity of assistance and the intensity of activation measures. Some countries, such as Denmark, combine generous unemployment benefits with strict activation requirements. Unemployed workers receive benefits close to what they would get on the job but must regularly meet with employment counselors regarding their work search. Other countries, including the United States, have less generous benefits and rudimentary activation policies.

Figure 4 uses data from OECD countries to illustrate the correlation between assistance and activation as of 2017.[59] Replacement rates are used as the measure of generosity of assistance. The score for activation strictness covers items such as the type of job offers that claimants need to accept, requirements for reporting on job-search efforts, obligations to participate in active labor market programs, and sanctions for failing to meet these requirements.[60] It is clear from the chart that countries with high income-replacement rates tend to have stricter strict activation requirements.

Our recommendations for reforming assistance policies included increases to both replacement rates and benefit duration. In the remainder of this section, we recommend complementary upgrades to American activation policies. Together, the two sets of reforms would move the United States up and to the right in Figure 4, toward a position more in line with other high-income democracies.

Moving to a new equilibrium that includes both more generous benefits and more robust activation requirements will require substantial investments in state capacity. Unemployment insurance should not just passively help people cope with joblessness. We will need to deploy the full toolbox of active labor market policies to maintain a dynamic economy in which entire industries may be built or disappear within a few decades.

Work search requirements alone are not enough

In the United States today, activation policy is strongly focused – too strongly, in our view – on work search requirements. Although that is true of all states, the precise requirements vary greatly. State work search requirements differ along several dimensions:

  • Which types of activities count as work search. Some states only count formal processes such as sending a job application to a potential employer or a job interview. Others allow a broader range of activities, including attending a job fair or creating a resume.
  • How UI recipients document their work search. Some states ask people to execute their work search through a state website. Others ask for submission of search activities to a state website. Still others require people to document work search activity at home and periodically audit their activities to ensure that they are performing as expected.[61]
  • The frequency with which work search activity has to be reported. States typically ask UI recipients to submit work search activity with the same frequency as unemployment insurance checks, typically on a weekly or biweekly basis. Others that rely on audits may check only infrequently.
  • The minimum threshold for qualified activities. Most states ask for between one and five work search activities per week. Others do not define the number of activities strictly.

Without losing sight of the state-by-state differences, we can, as a basis for discussion, use a generic model to represent the search requirements faced by a typical American UI beneficiary: A requirement to document three contacts per week with potential employers, with a low probability that the state will audit the documentation by calling the reported employers.

Such a model makes little sense in a modern job market. The requirements are, paradoxically, at the same time too loose and too strict. They are too loose because it is trivial to meet the baseline of three contacts with employers. A lazy job applicant could send three resumes to posted jobs by email in a few minutes with no intention of following up. The low likelihood of audit, driven by limited resources of state departments of labor, makes this an even lower hurdle. And even if someone from the Department of Labor does employers to follow up, employers may not be able to confirm that an application was sent in the first place. Most employers do not carefully track all applicants, especially those who have not been called for an interview or otherwise put on a list of active prospects.

At the same time, these generic requirements can be seen as too strict. Some people might actively look for work but fail to document it in a timely manner. Imagine having to fill out tax forms weekly instead of annually. Filling out the documentation itself is not a high burden in the moment, but it’s easy to forget one week, or to not keep careful records so that the forms can be filled easily. If the website for tracking job search is sufficiently unwieldly, documentation may be more of a hassle than the work search itself. A person who is applying for jobs on the internet can send hundreds of resumes or lightly-customized cover letters in a few minutes – much less burdensome than logging into a state UI site and filling in fields one-by-one.

In other cases work search strategies might not fit the three-contacts-per-week requirement. Work search requirements are not actually tracking that you are looking for work – they are tracking whether you are documenting your work search in ways that are legible to the state.[62] Many highly productive methods of looking for work will not leave a paper trail, or operate at longer cadences than the weekly requirements. For example, someone might do their job search in “batches” – sending out a large batch of resumes all in one day, and then following up over a period of several weeks. Alternatively, a person might be looking to gain a new skill before looking for work. While many states do allow training programs to count as “work search,” they can only track those that leave a paper trail. A person looking to learn a new programming language would not be able to work through online videos or side projects but would need to find a paid course that has been approved by their state. Finally, there is evidence that the most productive search strategy of all is informal networking with friends and family, even though that may not produce any documentable contacts.[63]

These tactics are the kind of things that any good career counselor would recommend, yet too much time spent on them could result in loss of benefits. Wasting jobseekers’ time just to create a paper trail could cause someone to miss a real job opportunity.

There is compelling evidence to suggest that meticulously adhering to standard work search requirements yields minimal benefits. The most persuasive evidence comes from randomized controlled trials conducted in Washington state in the mid-1980s and in Maryland in 1994. In both studies, individuals receiving unemployment insurance (UI) benefits were randomly assigned to participate in standard work search activities, have their work search requirements waived, or placed under alternative conditions (for example, mandatory career counseling, or more stringent work search requirements). For the sake of simplicity, we will focus solely on the disparities between “no work search requirements” and “standard work search requirements” during the first year of each study.

In the Washington experiment, the impact of work search requirements was negative, but small. Within the first year of receiving UI benefits, 86.7 percent of individuals without work search requirements found employment, compared to 88.3 percent of those with standard work search requirements. Additionally, the former group earned an average of $11,384 per year, while the latter earned $11,064 per year (approximately $31,180 and $30,304 in 2023 dollars, respectively). None of these differences proved statistically significant, indicating that these slight variances are not distinguishable from random noise.

In the Maryland experiment, waiving work search requirements yielded slight but positive effects. Within the subsequent year, 80.8 percent of individuals exempt from work search mandates secured employment, compared to 80.0 percent of those required to document their job search activities. Moreover, the former group earned an average of $8,754 ($17,246 in 2023 dollars), while the control group earned $8,407 ($16,563 in 2023 dollars) – a statistically significant difference.

One experiment demonstrated that work search requirements somewhat impeded the speed of reemployment, while the other revealed that work search requirements slightly accelerated it. However, in both cases, the effects observed were relatively minor. Although some differences were statistically significant, such as the increased earnings for the group without work search requirements in Maryland, they were not necessarily economically meaningful.

Even the hypothetical gains from enforcing strict work requirements come at a considerable cost. States must establish an infrastructure for job seekers to report their work search activities, and individuals seeking employment must invest time in entering their information into the website. Department of Labor staff must also dedicate their time to maintaining the website and auditing the data, instead of focusing on facilitating job placement. These efforts represent significant investments for minimal returns.

A better activation system would retain the benefits of work search requirements while eliminating the tedious paperwork. The next section explains what such a system might look like.

A new approach to activation

Instead of a demoralizing focus on creating a paper trail, a new approach to activation should emphasize opportunities to help unemployed workers build capacities and keep themselves motivated. Requirements should be flexible, allowing job seekers to choose the activities that best match their needs and abilities.

Above all, such a system should be proactive. A properly forward-looking activation system would include a work search plan and regular accountability follow-ups. It would also need adequate funding. Here is how it could work.

The work search plan. As part of registering for unemployment insurance, people would be asked to create a “work search plan.” The plan would begin with general, open-ended questions, such as, “What type of career or industry are you interested in?” It would also include time-staged prompts, such as, “In June, you will have been unemployed for three months. How many hours a week will you plan to spend looking for work? What days will you focus on looking for a job?”

These prompts have several goals. First, they encourage job seekers to create a realistic plan. Some people who have lost their jobs feel adrift and do not know where to start. Others may have overly optimistic beliefs about how fast they will find a job.[64] Simple planning prompts can be remarkably effective at getting people to think through the barriers to finding work.[65] Finally, the personal, self-directed goals in the plan set the stage for assessment of progress.

Regular accountability follow-ups. Proper accountability requires regular communication between job seekers and the state Labor Department to assess progress in looking for a job and maintain motivation. Ideally, that would take the form of direct conversations with employment counselors at a job center. However, that sort of personal touch requires resources.

In practice, it would probably be necessary to rely heavily on automated systems and to require the unemployed workers themselves to take the initiative.

The least complex system would be a simple automatic message system. Text messages would remind job seekers of their previous commitments. They would ask them whether they had met those commitments or whether they want to revise them going forward.

Alternatively, a system could be set up to connect job seekers with one another. They could be assigned to “accountability groups” where they report on their general work search progress and on key milestones, such as a first callback, first interview, and so on. Part of the goal would be to reframe the work search as one in which people try to improve their skills, not just report back on failures.

Ultimately, the goal would be to invoke job seekers’ own desire to find a job, as well as their desire to meet the goals they set for themselves. While connecting with coaches or other job seekers can strengthen those goals, the operating principle should be to refine and augment what already exists in the individual.

Although in some ways this would be similar to the current work requirement framework of weekly or biweekly reporting of search activity, it would differ in crucial ways. Most importantly, we do not recommend making these reporting activities mandatory for receipt of UI benefits. Benefits should be available to everyone and not conditional on paperwork. Instead, we advise that reporting be linked to a small supplementary payment, say $20, added to the basic UI benefits.

Mandatory periodic counseling. That said, activation requirements should become more rigorous over time. A failure to find a job after several months is an indication that the job seeker needs more assistance. Similarly, although UI benefits should not be contingent on a weekly paper trail, over a longer time, it is not unreasonable to ask recipients to demonstrate effort.

What we need is a way to push people to search actively for new jobs while minimizing the administrative burden. One way to do that is to move from the current paradigm of frequent, superficial work search requirements to infrequent, intensive counseling sessions.

The model we have in mind is something like the current Reemployment Services and Eligibility Assessment (RESEA) program the U.S. Department of Labor offers. RESEA is a meeting between job seekers and job counselors, where the counselor reviews work search activities, provides custom labor market information, and introduces other programs.

Until recently, RESEA was only provided to specific applicants who have been identified as statistically likely to exhaust their benefits. As the algorithms used to make these determinations are woefully out of date, this requirement was relaxed in 2018. Today, roughly 20 percent of UI recipients go through RESEA at some point.

Randomized trials of RESEA have demonstrated its effectiveness. One study found that within a year, employment rates for a treatment group of job seekers assigned to RESEA were 5.3 percentage points higher.[66] The average time of unemployment was reduced from 18.7 weeks to 16.6 weeks. In fact, the impact of receiving RESEA may be bigger. Even though failure to attend an assigned RESEA session can result in a loss of benefits, only 55 percent of those assigned actually attend (a big reason is simply that the notification is confusing or may not be read).[67] Because the randomized trial only looked at the impact of being assigned to receive RESEA, the effect on those who actually participated could be almost twice as large.

The need for adequate funding. Scaling up RESEA will require additional funding for reemployment programs. The United States spends far less on active labor market programs than other OECD nations, and, as Figure 4 shows, funding has decreased over time.[68]

Yet, even as funding has fallen, the number of programs has increased rapidly. Today there are at least 43 different federal employment and training programs.[69]  It would be one thing if these programs were highly specialized to provide training or support for specific occupations. In practice, though, they are largely duplicative except for the way in which people gain access to them. Consolidating these programs would make it much easier to fill the funding gap.

Key recommendations

Abolishing work search requirements to replace them with planning prompts and infrequent meetings may seem radical. However, the reforms we suggest are not dissimilar to the policies implemented now in some states and in other countries.

New Hampshire, the “live free or die” state, simply asks people to look for work the way a “reasonable person” would, without mandating a specific quota. The “honor-system” group in the Washington state experiment discussed above found jobs almost as fast as those with rigid documentation requirements, and that was without the planning and follow-up mechanisms we recommend. Finally, all states waived work search requirements during the Covid-19 pandemic, which was followed by one of the most rapid employment recoveries the United States has ever seen.

Specifically, we recommend the following key reforms:

  • Replace work-search requirements with a proactive approach emphasizing job-search planning, accountability follow-ups, and periodic counseling. Some aspects of RESEA policy could serve as a model.
  • Increase funding of active labor market policies. Consolidation of duplicative programs would make it easier to achieve adequate funding levels.

For all practical purposes, the United States, at present, does not have active labor market policies like those of our wealthy, democratic peers. What we have might better be described as a “passive labor market policy.” In the old Soviet economy, there was a standing joke: “We pretend to work, they, pretend to pay us.” Without extensive reforms, we could describe American activation policy with a twist on that: “We pretend to look for work, they pretend to help us.”


This paper has focused on three functions of unemployment insurance: social insurance, search and matching, and automatic stabilization. Each of these functions reinforces the others. The social insurance function provides the income smoothing needed for unemployed workers to meet their basic needs. With temporary support for meeting those needs, workers have time to search for a job that best matches their skills, and as a result, employers get a more productive workforce. A productive workforce means a more stable and prosperous economy, while adequate income smoothing ensures faster recoveries when downturns do occur.

But in order to serve these functions, the unemployment insurance system must be well designed. Sadly, the American UI system falls short in three key design areas: assistance (adequate income smoothing), access (coverage for all who need it), and activation (motivating and supporting work search).

To fix this broken system, we have recommended eight key reforms:

  1. A minimum weekly benefit of at least 35 percent of the state median household income.
  2. An income replacement rate of 75 percent for dollars earned beyond the 35 percent minimum.
  3. A maximum weekly benefit equal to the state median household income.
  4. A 39-week regular benefit duration.
  5. Creation of a federal back-end to enhance the technical and administrative capacity of state UI agencies.
  6. Standardization of benefit calculation rules across states.
  7. Replacing work-search requirements with a proactive approach emphasizing job-search planning, accountability follow-ups, and periodic counseling.
  8. Increased funding of active labor market policies.

The reforms outlined in this paper should be able to achieve the goals we have set. They should provide families with a financial cushion when a breadwinner loses their job. They should ensure that people have the time and financial capability to look for work that suits them. Finally, they should help to stabilize the economy by automatically supporting consumer demand during recessions and phasing out the support when recovery makes it no longer needed.

While any one of these reforms would have value as a standalone measure, they are stronger together than individually. The political prospects for comprehensive reform are bolstered by the fact that its effects would be welcomed across the ideological spectrum. Conservatives will welcome better support for a productive labor force and a stable economy – not to mention the safeguards against fraud and abuse. Progressives will find favor with stronger social insurance and improved opportunities for the unemployed to find steady, high-paying jobs. Enhanced benefits, stronger protections against fraud, and stronger activation policies all fit together to promote faster growth and fairer growth.

The experience of the Covid-19 pandemic should serve as a warning sign. The pandemic almost immediately made it clear that American UI policy was not operating as intended. Heroic measures by Congress and the executive branch staved off disaster. But the pandemic measures, which have been compared to changing the engine on an airliner while it is in mid-flight, were far from perfect. No one wants to go into another crisis with the same antediluvian computer systems and the same loopholes in hastily drafted legislation that created a fraudster’s holiday.

America needs and deserves an economy that grows rapidly, with benefits that are widely shared. As we look to the future, there will inevitably be new labor market shocks in store – other pandemics, trade disputes, wars, automation, artificial intelligence, or something completely unexpected. We need a UI system that works, and that is embraced across the entire political spectrum, to prepare for those shocks.

[1] Zoya Teirstein, “Trump Wants to Use FEMA to Pay for Unemployment. It’s Not as Crazy as It Sounds,” Grist, August 15, 2020, https://grist.org/politics/trump-wants-to-use-fema-to-pay-for-unemployment-its-not-as-crazy-as-it-sounds/.

[2] “American Rescue Plan,” The White House, Accessed March 2023, https://www.whitehouse.gov/american-rescue-plan/.

[3] The Congressional Budget Office’s July 2020 report “An Update to the Economic Outlook: 2020 to 2030” predicted that Gross Domestic Product would be $20.8 trillion in Q1 2021. The actual number was $22.3 trillion, a $1.5 trillion difference. “An Update to the Economic Outlook: 2020 to 2030,” Congressional Budget Office, accessed March 2023, https://www.cbo.gov/system/files/2020-07/51135-2020-07-economicprojections.xlsx; “Gross Domestic Product [GDP],” Federal Reserve Bank of St. Louis, accessed March 2023. https://fred.stlouisfed.org/series/GDP.

[4] Ben Winck and Andy Kiersz, “The Economy’s Recovery From COVID Is 3 Times Faster Than the Rebound From the Great Recession,” Business Insider, August 6, 2022, https://www.businessinsider.in/policy/economy/news/the-economys-recovery-from-covid-is-3-times-faster-than-the-rebound-from-the-great-recession/articleshow/85104076.cms.

[5] Chokshi, Niraj. “Pandemic Unemployment Fraud Estimate Rises to $45.6 Billion.” The New York Times, September 22, 2022, https://www.nytimes.com/2022/09/22/business/unemployment-insurance-fraud-pandemic.html.

[6] Katz, Lawrence F., Claudia Goldin, and Katherine Baicker. “A Distinctive System: Origins and Impact of U.S. Unemployment Compensation.” In The Defining Moment: The Great Depression and the American Economy in the 20th Century, edited by Michael Bordo, Claudia Goldin, and Eugene White, 227-264. National Bureau of Economic Research and University of Chicago Press, 1998; Social Security Administration. Office of Research, Evaluation and Statistics. “Social Security Programs in the United States.” SSA Publication No. 13-11758. July 1997.

[7] “Nor is there any reason why the state should not assist the individuals in providing for those common hazards of life against which, because of their uncertainty, few individuals can make adequate provision.”  F.A. Hayek, The Road to Serfdom (Chicago: University of Chicago Press, 1944).

[8] Matthew Darling, “Rethinking Unemployment Benefits Under Vaccine Mandates,” Milken Review, February 10, 2022, https://www.milkenreview.org/articles/rethinking-unemployment-benefits-under-vaccine-mandates.

[9] Judy Conti, Kristi Del Gallo, and Lisalyn Jacob, “Unemployment Insurance & COVID-19: What Domestic/Sexual Violence Attorneys & Advocates Should Know,” webinar from the Center for Survivor Agency & Justice, June 17, 2020, https://csaj.org/wp-content/uploads/2021/10/CSAJ_Unemployment-Insurance_HANDOUT.pdf.

[10] Stephan Lindner, “How Do Unemployment Insurance Benefits Affect the Decision to Apply for Social Security Disability Insurance?” Journal of Human Resources 51, no. 1 (2016): 62-94.

[11] “OECD Employment Outlook 2018,” Organisation for Economic Co-operation and Development (, OECD Publishing, accessed March 2023, https://doi.org/10.1787/empl_outlook-2018-en.

[12] OECD. 2023. “Employment Rate (Indicator).” doi: 10.1787/1de68a9b-en. Accessed April 19, 2023.

[13] Arindrajit Dube, “A Plan to Reform the Unemployment Insurance System in the United States,” Hamilton Project Policy Proposal, no. 3 (2021).

[14] Christopher Boone, Arindrajit Dube, Lucas Goodman, and Ethan Kaplan, “Unemployment Insurance Generosity and Aggregate Employment,” American Economic Journal: Economic Policy 13, no. 2 (2021): 58-99.

[15] Peter Ganong, Fiona Greig, Max Liebeskind, Pascal Noel, Daniel Sullivan, and Joseph Vavra, “Spending and Job Search Impacts of Expanded Unemployment Benefits: Evidence from Administrative Micro Data,” University of Chicago, Becker Friedman Institute for Economics Working Paper (2021-19); Fiona Greig, Daniel Sullivan, Peter Ganong, Pascal Noel, and Joseph Vavra, “When Unemployment Insurance Benefits Are Rolled Back: Impacts on Job Finding and the Recipients of the Pandemic Unemployment Assistance Program,” SSRN 3896667 (2021).; Arindrajit Dube, “Aggregate Employment Effects of Unemployment Benefits During Deep Downturns: Evidence from the Expiration of the Federal Pandemic Unemployment Compensation,” National Bureau of Economic Research, working paper no. w28470 (2021).

[16]Peter Ganong, Pascal Noel, and Joe Vavra, “Aggregate Short-Term Employment Effects of Terminations of Pandemic Unemployment Programs,” University of Chicago (2021),  https://voices.uchicago.edu/ganong/files/2022/09/pandemic_ui_terminations.pdf.

[17] Will Raderman, “Include Unemployment Insurance Reform in the Labor Policy Arsenal,” Niskanen Center, August 22, 2022, https://www.niskanencenter.org/include-unemployment-insurance-reform-in-the-labor-policy-arsenal/.

[18] “Significant Provisions of State Unemployment Insurance Laws Effective July 2022,” U.S. Department of Labor Employment and Training Administration, Office of Unemployment Insurance, accessed March 2023, https://oui.doleta.gov/unemploy/content/sigpros/2020-2029/July2022.pdf.

[19]Jin Zhang and Matt Darling, “A More Legible UI Policy.” Niskanen Center, February 7, 2023. https://www.niskanencenter.org/a-more-legible-ui-policy/.

[20] “Unemployment Insurance Data Dashboard: What Do Unemployment Benefits Look Like in Each State?” The Century Foundation, August 18, 2022, https://tcf.org/content/data/unemployment-insurance-data-dashboard/.

[21] Daniel N. Price, “Unemployment Insurance, Then and Now, 1935-85,” Social Security Bulletin 48 (1985): 22.

[22] “Significant Provisions of State Unemployment Insurance Laws Effective July 2022.” U.S. Department of Labor Employment and Training Administration, Office of Unemployment Insurance, accessed March 2, 2023. https://oui.doleta.gov/unemploy/content/sigpros/2020-2029/July2022.pdf.

[23] “Table 1600. Number of Earners in Consumer Unit: Annual Expenditure Means, Shares, Standard Errors, and Coefficients of Variation, Consumer Expenditure Survey, 2019,” U.S. Bureau of Labor Statistics, accessed March 3, 2023. https://www.bls.gov/cex/tables/calendar-year/mean-item-share-average-standard-error/cu-earners-2019.pdf.

[24] Gordan Dahl and Matthew M. Knepper, “Unemployment Insurance, Starting Salaries, and Jobs,” National Bureau of Economic Research, working paper no. w30152 (2022).

[25] Raj Chetty, “Moral hazard versus liquidity and optimal unemployment insurance,” Journal of Political Economy 116, no. 2 (2008): 173-234.

[26] Ammar Farooq, Adriana D. Kugler, and Umberto Muratori, “Do Unemployment Insurance Benefits Improve Match and Employer Quality? Evidence from Recent US Recessions,” National Bureau of Economic Research, Working Paper no. 27574.

[27] “National Longitudinal Surveys NLS FAQs: Frequently Asked Questions,” U.S. Bureau of Labor Statistics, Accessed January, 2022, https://www.bls.gov/nls/questions-and-answers.htm#anch42.

[28] Julie Berry Cullen and Jonathan Gruber, “Does unemployment insurance crowd out spousal labor supply?” Journal of Labor Economics 18, no. 3 (2000): 546-572. .

[29] “Monthly Program and Financial Data,” U.S. Department of Labor Employment & Training Administration, accessed January 2022, https://oui.doleta.gov/unemploy/claimssum.asp.

[30] “ET Financial Data Handbook 394,” U.S. Department of Labor Employment & Training Administration, accessed January 2022, https://oui.doleta.gov/unemploy/hb394.asp.

[31] Samuel Hammond and Brink Lindsey, “Faster Growth, Fairer Growth,” Niskanen Center(2022), https://www.niskanencenter.org/faster_fairer/agenda.html.

[32] Peter Ganong and Pascal Noel, “Consumer spending during unemployment: Positive and normative implications,” American Economic Review 109, no. 7 (2019): 2383-2424.

[33] Dube, “A Plan to Reform the Unemployment Insurance System in the United States.”

[34] Pascal Michaillat, “Do matching frictions explain unemployment? Not in bad times,” American Economic Review 102, no. 4 (2012): 1721-50.

[35] “Duration of Unemployment,” U.S. Bureau of Labor Statistics, accessed January 2022. https://www.bls.gov/charts/employment-situation/duration-of-unemployment.htm.

[36] John Robertson and Ellyn Terry, “Looking Beyond the Job-Finding Rate: The Difficulty of Finding Full-Time Work,” Federal Reserve Bank of Atlanta, Policy Hub: Macroblog, June 9, 2014, https://www.atlantafed.org/blogs/macroblog/2014/06/09/looking-beyond-the-job-finding-rate-the-difficulty-of-finding-full-time-work.aspx.; Kory Kroft, Fabian Lange, and Matthew J. Notowidigdo, “Duration dependence and labor market conditions: Evidence from a field experiment,” The Quarterly Journal of Economics 128, no. 3 (2013): 1123-1167.

[37]O’Brien, Matthew. “The Terrifying Reality of Long-Term Unemployment.” The Atlantic, April 13, 2013. https://www.theatlantic.com/business/archive/2013/04/the-terrifying-reality-of-long-term-unemployment/274957/.

[38] Michael R. Dalton and Jeffrey A. Groen, “How Do Jobseekers Search for Jobs? New Data on Applications, Interviews, and Job Offers,” U.S. Bureau of Labor Statistics, Employment & Unemployment Beyond the Numbers 9, no. 14 (November 2020), https://www.bls.gov/opub/btn/volume-9/how-do-jobseekers-search-for-jobs-new-data-on-applications-interviews-and-job-offers.htm; “Worker Displacement: 2019-2021,”U.S. Bureau of Labor Statistics, Economic News Release, August 26, 2022, https://www.bls.gov/news.release/disp.nr0.htm.

[39] “Median Household Income (in 2021 dollars), 2017-2021,” U.S. Census Bureau, QuickFacts, accessed January 2022, https://www.census.gov/quickfacts/geo/chart/US/INC110220.

[40] Saul J. Blaustein, “Job and Income Security for Unemployed Workers: Some New Directions,” W.E. Upjohn Institute for Employment Research, January 1, 1981.  

[41] Larry D. Turner, “Pandemic Response and Accountability: Reducing Fraud and Expanding Access to COVID-19 Relief through Effective Oversight,” testimony before the U.S. Senate Committee on Homeland Security and Governmental Affairs, March 17, 2022, https://www.oig.dol.gov/public/testimony/20220317.pdf.

[42] Eric Levitz, “Florida GOP Realizes Deliberately Impoverishing the Unemployed Has Downsides,” Intelligencer, April 3, 2020, https://nymag.com/intelligencer/2020/04/florida-unemployment-benefits-desantis-trump-rick-scott.html.

[43] Office of the Chief Inspector General, “Review of the Department of Economic Opportunity Florida Connect System,” Executive Office of the Governor, March 4, 2021, https://www.floridajobs.org/docs/default-source/reemployment-assistance-modernization/chief-oig-review-of-connect.pdf?sfvrsn=d1b253b0_2.

[44] Kenneth Quinnel, “Florida Leads Race to the Bottom on Unemployment Compensation,” AFL-CIO, October 22, 2012, https://aflcio.org/2012/10/22/florida-leads-race-bottom-unemployment-compensation.

[45] Matthew Chapman, “Ron DeSantis admits GOP put up “pointless roadblocks” so fewer people would sign up for unemployment,” Salon, August 5, 2020, https://www.salon.com/2020/08/05/ron-desantis-admits-gop-put-up-pointless-roadblocks-so-fewer-people-would-sign-up-for-unemployment_partner/.  

[46] Brink Lindsey, “State capacity: What it is, how we lost it, and how to get it back,” Niskanen Center, November, 2021, https://www.niskanencenter.org/wp-content/uploads/2021/11/brinkpaper.pdf .  

[47] Office of the Chief Inspector General, “Review of the Department of Economic Opportunity Florida Connect System.”

[48] Turner, Larry D. “Testimony before the U.S. Senate Committee on Homeland Security and Governmental Affairs. Hearing Title: ‘Pandemic Response and Accountability: Reducing Fraud and Expanding Access to COVID-19 Relief through Effective Oversight.’ Testimony of Larry D. Turner, Inspector General, Office of Inspector General, U.S. Department of Labor. Number 19-22-003-03-315. March 17, 2022.”

[49] Matthew Darling and Will Raderman “Improving the ‘Protecting Taxpayers and Victims of Unemployment Fraud Act’”, Niskanen Center. May 1st, 2023. https://www.niskanencenter.org/improving-the-protecting-taxpayers-and-victims-of-unemployment-fraud-act/

[50] “Southeast Consortium Unemployment Benefits Integration,” Intergovernmental Software Collaborative, accessed January 2022, https://softwarecollaborative.org/cooperatives/southeast-consortium-unemployment-benefits-integration.html.

[51] “ReEmployUSA.” Intergovernmental Software Collaborative, accessed January 2022. https://softwarecollaborative.org/cooperatives/reemployusa.html.

[52] “Significant provisions of state unemployment insurance laws effective July 2022,” U.S. Department of Labor Employment and Training Administration, Office of Unemployment Insurance, accessed March 2023, https://oui.doleta.gov/unemploy/content/sigpros/2020-2029/July2022.pdf

[53] 1/26 is a recurring number in UI benefits. With roughly 13 weeks per quarter, a 1/26th replacement rate sets weekly benefits at approximately half your wages).

[54] Joshua McCabe and Robert Orr, “Strengthening Family Tax Benefits in Massachusetts: An Analysis of State CTC Reform Options,” Niskanen Center, April 14, 2022, https://www.niskanencenter.org/strengthening-family-tax-benefits-in-massachusetts/.

[55] Samuel Hammond and Robert Orr, “The Conservative Case for a Child Allowance,” Niskanen Center, February 4, 2021, https://www.niskanencenter.org/report-the-conservative-case-for-a-child-allowance.

[56] The base period is generally the “first four of the last five completed calendar quarters preceding the filing of the claim”. Some states will use alternative base periods (the last four completed quarters) or extended base periods that go back further. This, too, should be standardized. Office of Unemployment Insurance. Comparison of State Unemployment Insurance Laws, 2022 ed. U.S. Department of Labor, Employment and Training Administration, OUI/DL, 2022. https://oui.doleta.gov/unemploy/statelaws.asp.

[57] Zhang, Jin and Matt Darling. “A More Legible UI Policy.” Niskanen Center, February 7, 2023. https://www.niskanencenter.org/a-more-legible-ui-policy/.

[58] Louisiana, for example, has done this twice – its UI formula is to set benefits at 1/25th of average wages, then multiply that number by 1.05, then multiply that by 1.15. This is equivalent to setting a replacement rate of 4.83 percent of average wages.

[59] “Strictness of Activation Requirements,” Organisation for Economic Co-operation and Development (OECD), accessed January 2022, https://stats.oecd.org/Index.aspx?DataSetCode=SBE.

[60]“How Demanding Are Activation Requirements for Jobseekers?” Organisation for Economic Co-operation and Development (OECD), accessed January 2022. https://www.oecd.org/social/strictness-benefit-eligibility.htm.

[61] This is also a driver of UI “improper payments” which are often improperly categorized with fraud. As UI recipients are expected to maintain a regular job search log, and many do not, or are unable to show documentation months later, their UI payments would get categorized as an improper payment.

[62] James C. Scott, Seeing Like a State: How Certain Schemes to Improve the Human Condition Have Failed (New Haven, CT: Yale University Press, 1998).

[63] Karthik Rajkumar, Guillaume Saint-Jacque, Iavor Bojinov, Erik Brynjolfsson, and Sinan Aral, “A causal test of the strength of weak ties,” Science 377, no. 6612 (September 2022): 1304-1310.

[64] Johannes Spinnewijn, “Unemployed but optimistic: Optimal insurance design with biased beliefs,” Journal of the European Economic Association 13, no. 1 (April 2014): 130-167.

[65] Martin Abel, Rulof Burger, Eliana Carranza, and Patriizio Piraino, “Bridging the intention-behavior gap? The effect of plan-making prompts on job search and employment,” American Economic Journal: Applied Economics 11, no. 2 (April 2019): 284-301.

[66] Day Manoli, Marios Michaelides, and Ankur Patel, Long-term effects of job-search assistance: Experimental evidence using administrative tax data, National Bureau of Economic Research,no. w24422.

[67] Matthew Darling et. al, “Using behavioral insights to improve take-up of a reemployment program: Trial design and findings,” (Mathematica Policy Research, 2017), https://research.upjohn.org/cgi/viewcontent.cgi?article=1079&context=externalpapers.

[68] Sam Hammond and Brink Lindsey, “Faster Growth, Fairer Growth.” https://www.niskanencenter.org/faster_fairer/agenda.html

[69] United States Government Accountability Office. Employment and Training Programs: Department of Labor Should Assess Efforts to Coordinate Services Across Programs, GAO-19-200, report to the Permanent Subcommittee on Investigations, Committee on Homeland Security and Governmental Affairs (U.S. Senate, March 2019),  https://www.gao.gov/assets/gao-19-200.pdf.