Jobs and Unemployment

Section 2. Financing: Reform financing of UI to eliminate incentives for states and employers to exclude workers and reduce benefits

Key proposals

  • Federal financing: Address chronic underfunding and incentives to reduce outlays at the expense of workers by paying for all regular and extended UI benefits with federal dollars. Absent a switch to a wholly federally financed system, reform tax rate calculations and state trust fund targets to strengthen support for workers and stabilize the system.
  • Taxable wage base: To make UI’s finances fairer and more efficient, broaden the taxable wage base for unemployment insurance to be equal to the wage base for Social Security.
  • Tax rate: In a federally financed system, reduce employer incentives to block worker claims by reforming “experience rating” to determine the employer tax rate based on changes in hours employees work, not unemployment insurance claims. In a federal–state financing system, reconsider “experience rating” and base state trust fund targets on industry-adjusted per capita targets, not “high cost multiple” systems that encourage states to slash awards to replenish trust fund accounts.
  • Recessions: In a federally financed system, acknowledge the economy-boosting benefits of unemployment insurance benefits during recessions by paying for them with general revenue. In a federal–state financing system, use automatic triggers to provide federal financial assistance to states.
  • Classification of workers: Require states to implement “ABC” tests that prevent employers from circumventing taxes by reclassifying employees as contractors, and require large businesses that use a lot of contractors to pay unemployment insurance tax on their contractors.


The problem

States and the federal government jointly finance unemployment insurance (UI) in the United States through state and federal taxes imposed on wages and paid by employers, with a portion of those taxes likely passed through to workers in the form of lower pay. During normal economic times, states pay for the benefits their workers receive, and the federal government provides grants to pay for the cost of administering the program. The federal government also finances 50% of the Extended Benefits (EB) program.1 In recent recessions, it has paid 100% of EB costs as well as the costs for additional non-EB benefit extensions.

The way we pay for unemployment insurance benefits explains many of the program’s problems today. The desire to attract businesses and jobs puts pressure on states to keep their tax rates low. To do so, they limit benefits recipiency (the share of unemployed workers receiving benefits), decrease the amount of benefits workers receive, or both. Employers also have an incentive to make it harder for workers to claim benefits—for example, by filing challenges and appeals to worker claims—because the UI taxes firms pay rise when their workers collect UI benefits, a structure known as “experience rating.”

For both employers and states, pressure to reduce the benefits flowing to workers is especially intense when benefits are needed most: during and just after major recessions. When profit margins are slim, employers may be more likely to try to save money by keeping their UI tax rate as low as possible.

Federal law requires states to save their UI funds in a trust fund account and encourages them to keep adequate funds in it. During past recessions, these accounts have run empty in a number of states. To refill their accounts and get back into balance, many states slash benefits. In the wake of the Great Recession, for example, nine states cut the number of weeks individuals could get benefits from the traditional 26 to as few as 12 (Leachman 2015).2 As of April 2021, state legislators in at least three states (Iowa, Louisiana, and Tennessee) had proposed cutting the benefit duration to 12 weeks in response to low trust fund balances following the economic crisis caused by the COVID-19 pandemic (Golshan and Delaney 2021). The consequences of inadequate program financing fall squarely on the most disadvantaged workers, disproportionately affecting communities of color, as described in Section 3, covering eligibility.

The solution

Current financing of the UI system incentivizes employers and states to obstruct the delivery of UI benefits. Redesigning the way benefits are financed could create a system that would more effectively deliver UI.


Financing UI benefits at the federal, rather than the state, level is the most logical way to ensure adequate funding for and the equitable distribution of UI benefits. If policymakers choose to leave benefit financing primarily to the states, however, significant improvements could still be made.

Federal policymakers could approximate “first-best” reforms by (a) raising the federal taxable wage base (which would require states to raise their own taxable wage bases), (b) reforming experience rating, (c) making federal financial assistance to states for funding UI benefits more predictable by tying it to automatic triggers, (d) ensuring that firms do not have incentives to outsource or misclassify workers as contractors, and (e) basing state trust fund targets on industry-adjusted per capita targets, not the current “high cost multiple” system, which encourages tax and benefit cuts.

Funding unemployment benefits with federal financing

In normal economic times, state UI benefits are the only support available to unemployed workers; during economic crises, they are often the only support preventing workers from falling into distress. But rather than ensure that these benefits are adequate, states compete in a race to the bottom, in an effort to attract firms. Employers shift production and workers to low-tax states (Guo 2020). They are especially likely to exploit differences across states in the taxation of part-time workers (U.S. GAO 1993). These moves hurt women, who are disproportionately likely to work part time, and people of color, who are disproportionately likely to work part time but want full-time work (Golden and Kim 2020).

Cutting taxes requires cutting benefits, making the race to the bottom evident in every aspect of the UI system (Galle 2019). After the 2009 recession, 10 states dramatically cut the duration of their UI benefits, and others failed to adjust the maximum benefit level for inflation for more than a decade, eroding benefits (CRS 2019; Smith, Wilson, and Bivens 2014). States that have most aggressively courted business by cutting benefits the most are also those states granting UI benefits to lower shares of separated workers (Desilver 2020; U.S. DOL-ETA 2021a; U.S. Census Bureau 2019). Indeed, historical data on the recipiency rate suggest that states tightening requirements in times of economic distress have contributed to subsequent declines in recipiency rates that have never ascended back to mid-last century highs exceeding 50%, and hover around 28% today (as of 2019) (U.S. DOL-ETA 2021b). By capitulating to employer threats to move jobs out of state, local officials may see themselves as winning a competition with their neighbors. But the result is a less effective automatic stabilizer to assist displaced workers during recessions, leaving the federal government holding the bag to pay for other safety-net programs.

Policy proposal: Fund regular unemployment benefits and the full cost of extended benefits with federal financing

No solution short of federal financing will fully overcome these problems; if states can compete with one another to lower taxes, they will. If their revenue is lower, they will find new ways to reduce recipiency rates. Following the Great Recession, for example, when federal law temporarily constrained states from reducing benefit amounts, they cut the duration of benefits (CRS 2019; Smith, Wilson, and Bivens 2014). Some states also erected new administrative barriers (Wentworth 2017). For example, a new online system in Massachusetts increased one form of benefit denial by 950%, and delayed claims processing by more than six weeks on average. Florida increased one category of procedural denials by 180%, and another by 400%.

Financing at least the cost of the minimum regular UI benefits we have described with federal money—building on the existing federal financing of program administration and extended benefits—could eliminate this race to the bottom.3 Removing this competition would remove the greatest barrier to adequate benefit financing.4 A federally financed regular UI benefit could be financed in several ways, as shown below.

Increasing the taxable wage base

Currently, states and the federal government finance UI almost exclusively through taxes on employers. The total tax an employer owes is based on the product of its tax rate and the taxable wage base, the portion of each employee’s salary that is considered when calculating the tax owed. For the current federal UI tax—the Federal Unemployment Tax Act (FUTA) tax—the federal government taxes only the first $7,000 of each employee’s wages. State wage bases cannot be lower than the federal wage base; they currently range from $7,000 to about $53,000, with most of them less than $15,000 (U.S. DOL-ETA 2020c). The taxable maximums for UI wage bases are far lower than the taxable maximum wage bases for other programs. For example, the wage cap for the Social Security portion of the Federal Insurance Contributions Act (FICA) for 2021 was $142,800. The UI federal wage base has failed to keep up with inflation; it was $3,000 in 1939, or more than $55,000 in 2021 dollars, almost eight times its current value of $7,000 (Gould-Werth 2020a).

Policy proposal: Set the taxable wage base for unemployment insurance equal to the wage base for Social Security

We propose increasing the federal UI taxable wage base to a level equal to the current Social Security taxable wage base and indexing it to inflation. Raising the wage base threshold would allow for much lower rates, which would be more economically efficient, because they would reduce the incentive to attempt to avoid the tax. If UI financing were shifted to the federal level and the taxable federal wage base broadened, revenues could increase without an accompanying rate hike.

A higher wage base cap would also better align the constituencies that pay the tax and draw benefits from it. For example, a worker in Vermont who earns $10,000 a year has taxes levied on 100% of her wages and has 100% of her wages considered when setting benefits, whereas a worker who earns $50,000 has taxes levied on only 28% of her earnings, but has 92% of her wages used when determining her benefit level (U.S. DOL-ETA 2020b; Vermont DOL 2021). Such a system thus imposes a greater relative burden on the lower-wage worker.

Raising the wage cap would also reduce the incentive to prefer a small number of high-earning workers over a large number of low earners. With a low cap, an employer incurs greater tax liability if she hires two $20,000 earners than if she hires one $40,000 earner. For instance, in Vermont, where there is a 2021 cap of $14,100, an employer with a 1% rate would pay $141 in state UI tax for the $40,000 earner, but $282 for the two $20,000 earners. The system thus disincentivizes employers to employ low-wage workers.5

Under the current federal–state UI system, state wage bases cannot be lower than the federal wage base by law. Raising the federal UI cap to the Social Security cap would therefore also raise all state caps, generating all of the benefits described above.6

Changing the tax rate formulation

Average UI tax rates are low in the United States. The federal rate is 0.6%,7 and most state rates average 1%–2% (U.S. DOL-ETA 2020a). Employers pay the federal government $42 a year for each full-time, year-round employee (0.6% times the $7,000 federal wage base). In Missouri, a fairly low-tax state, the average employer would pay another 1% tax on the state wage base of $11,500, for a state-tax total of $115 per worker. Nationwide, the total average tax was about $267 per worker in 2020; the average over several recent years was a bit higher, at $350 (U.S. DOL-OUI 2021; Pavosevich 2020).

Within a state, different employers face different tax rates, depending on the employer’s experience with unemployment—a method of setting the tax rate known as “experience rating.” In all U.S. jurisdictions except Alaska, the rate is based on the UI benefits collected by separated workers. Every benefit recipient must be traced back to one (or more) previous employers. Employers with a higher ratio of benefits awarded to payroll pay higher rates than others up to a cap.8 In Alaska, employers instead pay a higher tax when they cut payroll, regardless of whether their employees successfully claim benefits.

The basic logic of experience rating is similar to the rationale for adjusting auto or home insurance rates upward after the insured files a claim. Layoffs impose costs on society, including the expense of providing UI benefits. Without experience rating, employers would not take account of these costs, leading to excessive job turnover (Alessie and Bloemen 2004; Karni 1999). There is strong evidence that experience rating reduces layoffs as well as seasonal employment (Albertini and Fairise 2018; Anderson and Meyer 2000; Baicker, Goldin, and Katz 1998; Ratner 2013; Woodbury 2004).

As currently designed, though, experience rating also encourages employers to prevent their employees from receiving adequate benefits. Benefit denials increased by 50%–66% in Washington state after it adopted experience rating (Anderson and Meyer 2000). Employer tactics include failing to provide information about benefits; encouraging employees not to file; challenging employee claims;9 or structuring their workforce to reduce eligibility, such as by using part-time workers or contractors, and shifting job sites to states where taxes are lower and benefits less generous (Anderson and Meyer 2000; de Raaf, Motte, and Vincent 2005; Gould-Werth 2016; Hyatt and Kralj 1995; Thomason and Pozzebon 2002; Vroman et al. 2017).

Grocery worker fired for taking health leave is initially denied benefits

Justin Grevencamp, Maine

I worked at an organic grocery-cafe in Brewer for two years. It was a beloved, locally owned small business. I developed new recipes and handled important catering orders. I took a two-week unpaid leave to deal with health issues, and when I returned I was told there was no more work for me. When I filed for UI, the store owners told the agency I had quit. At the appeal, one owner was upset that their taxes would go up because I had claimed UI. I received my benefits, but I was stunned that people I considered friends had tried to block me from receiving them.

Policy proposal: In a federally financed system, base the unemployment insurance tax rate on changes in hours employees work, not unemployment insurance claims

Because the U.S. lacks the employment protections common in most other developed economies, there is a strong case for a system that still imposes costs on employers when they terminate workers but does not encourage employers to prevent their workers from obtaining unemployment benefits. We suggest removing incentives to prevent UI claims for separated workers by using the full-time equivalent (“FTE”) or “hours-worked variation” experience rating method, similar to the method used in Alaska.10 In the FTE method, employers are rated based on how much the hours their employees worked changed, on a quarterly basis, over a three-year period. Increases in hours are not given as much weight as decreases.

Because employers are charged based on changes to payroll rather than benefit claims, these methods should make employers indifferent about whether their separated workers obtain benefits.11 Notably, until the tenure of its latest governor, Alaska’s recipiency rate was among the highest in the country (as of 2020, it had declined to around the national mean).

To calculate their tax rate in a given quarter under the FTE method, employers compute the average of the total hours worked in each of the 12 preceding quarters. They then calculate the change from each of these quarters to the next, dividing this change by the average quarterly total hours over the 12-quarter measuring period.12 For increases, the resulting product is discounted by a special weighting factor. The taxpayer’s final rating factor is the average of these 12 changes over the three-year rating period.13 Employers are then ranked and tax rates assigned as in existing experience rating systems. For more details, refer to the text box, “Calculating the tax rate under the FTE method.” Under federal financing, employers should be ranked regionally (by state or Metropolitan Statistical Area, for example), rather than nationally, to account for local variations in industry, seasonality, and other economic conditions.

Using hours worked as the base unit of measure offers a few modest advantages over Alaska’s method (based on payroll) or one based on the number of employees. Using the number of employees on the payroll cannot account for reduction in hours. It either weights full- and part-time employees equally or discounts employees who work part time, allowing employers to game the system by reducing work hours. Payroll-based systems may favor high-wage over low-wage workers, rewarding business models with a small number of highly skilled employees (Vroman et al. 2017). Such a system could result in lower levels of employment for less-educated workers.

Many states do not presently collect hours-worked data, but former U.S. Department of Labor officials have described this as only a minor obstacle (Miller and Pavosevich 2019). As explained later, expanding state collection of hours data would have other advantages as well, such as improving the data available through the Quarterly Census of Employment and Wages, and allowing for implementation of our proposal to base worker eligibility on hours, not wages.

Calculating the tax rate under the FTE method

Under the proposed method, the tax rate would be calculated as follows:

\begin{aligned}  &\left(\sum_{t-12}^{t-1}\left[100 \% *\left(T H_{t}-T H_{t-1}\right) / \mathrm{ATH}\right], T H_{t}-T H_{t-1}<0\right.\\  +&\left.\mathrm{W}^{*} \sum_{t-12}^{t-1}\left[100 \% *\left(T H_{t}-T H_{t-1}\right) / \mathrm{ATH}\right], T H_{t}-T H_{t-1} \geq 0\right) / 12  \end{aligned}

where TH is the total number of employee hours worked in a given quarter, ATH is the average quarterly total hours over the 12-quarter period, and W is a weighting factor that applies to quarters in which hours worked rose. Nonhourly full-time workers are counted as working 40 hours a week. The weighting factor is intended to give employers only partial credit for new hires; it should therefore be less than one. Furloughs and other temporary layoffs still incur some cost, and this cost rises the longer the layoff is. To give a larger credit when furloughs are shorter, W could be equal to (BDAF)/BD, where BD is the maximum benefit duration in days (182 for 26 weeks) and AF is the firm’s average furlough length (in days). For simplicity, AF also can be computed as an average across employers.

Policy proposal: In an experience rating system, penalize employers who consistently remain at the tax rate cap

In contemporary experience rating systems, some firms are indifferent to additional layoffs because the tax rate is capped—that is, the rate they pay for UI will not rise if additional workers are laid off. We recommend adopting the suggestion of Vroman et al. (2017) of imposing a penalty on firms that persistently remain at the cap (for three years or more, for example).

Policy proposal: In a federal–state financing system, base state trust fund targets on industry-adjusted per capita targets

In a system with significant state financial contributions, state savings targets should be reformed to be based on industry-adjusted per-capita targets, not the current “high cost multiple” system that favors cutting benefits rather than raising revenues when trust fund balances are depleted. Currently, in good times, states deposit excess UI revenues in a trust fund account, to be drawn on during recessions. Regulations encourage states to maintain an adequate balance, based on a calculation known as the “average high cost multiple” (AHCM). Because the AHCM is based on expected benefit awards, states can achieve their fiscal targets by slashing awards instead of adding revenue (ACUC 1996).14

Thus, in a system with significant state contributions, the AHCM should be replaced with a system in which state savings targets are based on industry-adjusted per capita targets. That is, state trust fund adequacy would be calculated using nationwide projected benefit costs per capita multiplied by the state population (see Galle 2019 for a complete description). This figure could then be modified to reflect each state’s mix of local industries and those industries’ historic nationwide turnover rates.

It may also be desirable to eliminate experience rating in a system that is mostly state financed. Although a portion of UI taxes are passed along to workers in the form of lower wages, in a competitive industry an employer cannot offer lower wages than its rivals, so that with experience rating higher-taxed businesses must bear the costs themselves (Gruber 1997; Anderson and Meyer 2000). This encourages business owners to seek out lower rates, contributing to the race to the bottom that federalizing the financing system would avoid.

Why not tax workers?

The suggestions in this section assume that UI benefits would continue to be financed by taxes levied on employers. The report contributors also considered the possibility of moving to an entirely worker-financed system. Taxing workers directly would make it more likely that those who pay for benefits are the ones who receive them. The Canadian model shows that such an approach is feasible (OECD 2019). An employee-side tax could also increase worker awareness of the UI program and create a sense of program ownership.

These benefits notwithstanding, we propose retaining the current employer-financing model for four reasons:

  • Coupled with the hours-worked experience rating method, employer-side taxes are a powerful tool for preventing unnecessary layoffs.
  • Experience rating prevents employers in competitive industries from reducing worker salaries to make up for the employer’s higher tax rate, ensuring that employers share in the cost of the UI system. Employers should share in UI costs because the system also benefits them by facilitating better job matching (Gould-Werth 2020b).
  • Workers who are paid the minimum wage could lose money with a switch to worker-side taxes, because employers cannot currently reduce their salaries to offset all of their accompanying UI taxes.
  • Wages might take some time to adjust. Employers are currently likely paying lower wages than they would absent a UI tax. If taxes were instead imposed on workers, it might take some time before competition drove wages up. In the meanwhile, workers would effectively be paying UI taxes twice: once in the form of lower wages and once in the form of actual taxes. Phasing in employee taxes over time might mitigate this problem.

If policymakers fail to adopt the hours-worked experience rating model, shifting to a worker-side tax could be preferable to the current model of an employer-side tax that is experience rated based on UI claims. If a switch were made to funding UI benefits and program administration through a worker-side tax, the tax should be structured progressively.

Paying for unemployment benefits during recessions

UI payouts during recessions provide large positive spillovers to the economy, including to workers and businesses that do not draw on UI (Elmendorf 2011). The cost of financing this economic stimulus should therefore be widely shared, though the method for doing so may vary depending on whether UI is wholly federally financed or is partly left to states.

Policy proposal: In a federally financed system, pay for benefits with general revenue during recessions

If funding for UI is made wholly federal, we recommend that benefits be paid out of general federal revenues during recessions, as they have been in recent episodes. As with Social Security, there are economic and political advantages to “benefit taxation,” in which taxes are explicitly linked to a particular social insurance product. In recessions, however, the UI system should be able to borrow from general Treasury funds without having to repay its borrowing via higher UI taxes later.

Policy proposal: In a federal–state financing system, use automatic triggers to provide federal financial assistance to states for unemployment insurance

Even if states are left to bear some of the costs of standard UI during normal times, the federal government should pay all of the cost during recessions. Federal dollars already pay for half of extended benefits, and in recent recessions that has been expanded to 100%, so extending federal support to base-period UI is not a dramatic shift and is economically justified. Recessions hit some regions harder than others; national financing allows states to share risk. States also face structural obstacles to countercyclical spending (Galle 2019). For example, it is politically difficult to maintain reserve funds at the state level, because today’s taxpayers may be living or doing business elsewhere when the rainy day arrives.

Federal financing for base-period UI can be expanded without major changes to the current structure. Following the Great Recession, Congress eventually granted states significant financial relief. Under current law, states accessing federal financing must risk the possibility that they will face swift repayment obligations as well as significant interest and added charges before Congress eventually provides them with relief. To eliminate this risk, we recommend that when EB is triggered (as described in Section 5, benefits duration), states become automatically eligible for low-interest or interest-free financing for base-period UI as well, repayable over a period that does not coincide with the state’s recovery from its downturn. Even greater federal support is possible; for example, severe downturns could trigger automatic loan forgiveness. That would in effect offer federal reinsurance to state trust funds, relieving some of the financial pressure on them to compete over low benefits.

Including self-employed and misclassified workers

State and federal UI taxes apply only to the wages of employees. To circumvent the tax, many employers therefore claim that their workers are contractors, not employees. Legal tests to distinguish genuine employees from contractors are usually complicated and thus costly and time-consuming to enforce. Other employers respond to UI incentives through actual (not merely nominal) changes in their workforce, shifting traditional full-time employees into more-precarious, less-rewarding contract work. The rise of large digital platform-based businesses, such as Airbnb and Uber, has drawn a growing attention to the plight of workers in these types of positions (Ravenelle 2019).

Policy proposal: Require states to implement the “ABC” test

To ensure that enterprises pay their fair share of UI taxes, rather than shift their burdens to others, federal law should make it more difficult for employers to recharacterize taxable wages as untaxed contractor payments. An important starting point is to require states to implement a version of the “ABC” test. Under this legal rule, a service provider to a business is presumed to be an employee unless the individual (a) is free from the direction and control of the business, (b) provides labor outside the usual course of the business, and (c) is customarily engaged in their own independently established business. A few states already apply versions of the ABC test, at least for select industries (Doroghazi 2019). It should be a national gold standard. Alternative legal approaches, such as the 20-factor balancing test applied by the IRS, have proven to be unpredictable, highly manipulable, and expensive and burdensome to administer.

Policy proposal: Require large 1099 payers to pay unemployment insurance tax on their contractors

As an additional disincentive to misclassification and efforts to replace the security and benefits of a steady job with the uncertainty of contractor status, we recommend taxing contractor payments at the same rate as employee hours. Imposing UI taxes on contractor payments would also bring in contributions from a self-employed population that, under our proposal in the benefits chapter, would receive greatly expanded benefits under the jobseeker’s allowance (See Section 3, on eligibility, in this report).

To reduce the administrative burden on small businesses and households, only relatively large businesses would be subject to the UI tax, and even they would be required to pay it on the wages of only some workers. Individuals who receive only very small payments from a given payer or who provide services only for the nonbusiness purposes of the payer would be exempt. The UI tax would be imposed only on payments to individuals for whom the payer is obligated to issue a Form 1099 (currently contractors paid at least $600 a year for the payer’s business purpose). For simplicity, only large firms, such as companies with more than $1 million in annual payroll or profits, might be required to pay the UI tax on contractors. If this threshold is considered too low, an additional threshold of 50 total combined employees and contractors could be applied. Both of these measures could be phased in, with the UI tax on 1099 earners starting at 10% of the full tax rate and stepping up 10% for each additional $1 million in payroll/profits to a cap of 100%. If a worker threshold is also used, the rate could step up by 10% per 25 workers, with the actual tax rate imposed being the greater of the payroll/profit- or worker-determined rate. Whichever threshold is used, related entities such as companies that share the same owners would be counted together when determining whether the firm crosses the threshold, to prevent gaming the system.


1. According to the U.S. Department of Labor, “Extended Benefits are available to workers who have exhausted regular unemployment insurance benefits during periods of high unemployment. The basic Extended Benefits program provides up to 13 additional weeks of benefits when a State is experiencing high unemployment. Some States have also enacted a voluntary program to pay up to seven additional weeks (20 weeks maximum) of Extended Benefits during periods of extremely high unemployment. Extended Benefits may start after an individual exhausts other unemployment insurance benefits (not including Disaster Unemployment Assistance or Trade Readjustment Allowances). Not everyone who qualified for regular benefits qualifies for Extended Benefits…. The weekly benefit amount of Extended Benefits is the same as the individual received for regular unemployment compensation. The total amount of Extended Benefits that an individual could receive may be fewer than 13 weeks (or fewer than 20 weeks).” U.S. DOL-ETA 2021c.

2. The states were Arkansas, Florida, Georgia, Illinois, Kansas, Michigan, Missouri, North Carolina, and South Carolina.

3. We do not mean to rule out the possibility that states would voluntarily impose their own taxes to pay for benefits more expansive than those we propose. Added benefits might also be fully or partially federally funded.

4. If the taxable wage base were raised, workers in high-wage states would cost their employers more in UI taxes than workers in low-wage states. But equal tax rates would ensure that this difference was driven by labor costs, not taxes. The marginal dollar of salary would cost the same no matter what state it was paid in.

5. The other side of this argument is that raising the cap could discourage pay raises and potentially full-time work. In many jurisdictions, the cap is now so low that it is effectively a fixed tax per employee, leaving employers indifferent to wage levels. (For a summary of the evidence that rated taxes affect only the “extensive,” or hiring, margin, see Guo and Johnston 2020.) With a higher cap, pay raises potentially lead to a UI tax increase. Depending on other UI rules, this principle can also translate into a preference for part-time employees: Once it is no longer advantageous to hire one $40,000 salary worker instead of two $20,000 salary workers, an employer might split a single position into two, perhaps to take advantage of other provisions that turn on full-time versus part-time status (Guo and Johnston 2020).

6. If states continue to pay UI benefits through state trust funds, expanding the wage base could help strengthen state finances (ACUC 1996; Vroman et al. 2017). Although states could cut rates in response to an expansion of their wage base, analysts predict that many states would still bring in additional revenue on net (CBO 2012).

7. The full federal rate is 6%, but it is reduced to 0.6% in states that are in compliance with federal law. The Department of Labor has never found a state to be so out of compliance, so the 6% rate has never been triggered.

8. Most states still use an antiquated variant (the “reserve ratio” method), in which long-standing employers can become insensitive at the margin to further changes in employment. Under this system, high-turnover firms are strongly incentivized to outsource work to contractors (Pavosevich 2020).

9. It does not appear that there are any strong reasons to encourage employers to challenge employee benefits. McLeod and Malcolmson (1989) argue that if employers are not rated based on their employees’ benefits, they will not be incentivized to share with government officials information that could indicate a fraudulent claim. We could not identify any evidence that this is the case. Moreover, fraud reduction is small compared with the direct impacts of experience rating. For example, recipiency rates vary dramatically between high- and low-recipiency states, with workers in high-recipiency states receiving benefits more than twice as often as workers in low- recipiency states (Desilver 2020). It is unlikely that differences in fraud detection explain a meaningful portion of these differences.

10. Our proposal builds on work by Vroman et al. (2017) and Miller and Pavosevich (2019). Miller and Pavosevich (2019) propose alternative systems modeled on Alaska’s payroll decline method. Vroman et al. (2017) conclude that none of the alternatives they modeled was clearly superior to the best options now in practice, but they did not consider the refinements Miller and Pavosevich propose, such as the use of the “average quarterly hours” measure we describe below.

11. Municipal and nonprofit employers often are not formally experience rated, but instead may opt to “self-insure,” which means that they pay into the system only when their employees claim benefits. Our proposal would therefore leave these employers with strong incentives to prefer lower benefits and to contest claims. It is likely all employers should be taxed by the same rules.

12. Although the use of average quarterly total hours—as Miller and Pavosevich (2019) propose—adds some complexity, it solves a key problem Vroman et al. (2017) identify. Consider a 10-worker firm that lays off all 10 workers in Q1 and rehires them in Q3. There is a 100% decline from Q1 to Q2 and an infinite percentage increase between Q2 and Q3. If the firm lays off and rehires only nine workers, there is a 90% reduction between Q1 and Q2 and a 900% increase between Q2 and Q3. Both these scenarios would produce highly inaccurate results for the rehiring employers. Using average total hours smooths these changes, so that in the second scenario, for instance, there is about a 130% reduction (9/[(10+1+10)]/3)) in the number of employees followed by a 130% increase. Applying the discount factor, W, to the increase would result in a net reduction in hours, reflecting the fact that the firm made some use of the UI system.

13. For small employers, the quarterly components could be weighted by the quarter’s total hours, a step that would tend to reduce the volatility of the measure (Vroman et al. 2017).

14. For example, suppose a state is projected to average $1 billion in UI spending during high-cost periods but it has only $750 million in its trust fund. A target AHCM of 90% would require the state to hold at least $900 million in its trust fund. If the state cuts benefits, its projected spending might fall to $800 million, allowing the target trust fund balance to be only $720 million. The state has cut its way to its target.


Advisory Council on Unemployment Compensation (ACUC). 1996. Collected Findings and Recommendations: 1994–1996. Washington, D.C.: Advisory Council on Unemployment Compensation.

Albertini, Julien, and Xavier Fairise. 2018. “Layoffs, Recalls and Experience Rating.” TEPP Working Paper 2018-10.

Alessie, Rob, and Hans Bloemen. 2004. “Premium Differentiation in the Unemployment Insurance System and the Demand for Labor.” Journal of Population Economics 17, no. 4 (December): 729–765.

Anderson, Patricia M., and Bruce D. Meyer. 2000. “The Effects of Unemployment Insurance Payroll Taxes on Wages, Employment, Claims and Denials.” Journal of Public Economics 78, nos. 1–2 (October): 81–106 .

Baicker, Katherine, Claudia Goldin, and Lawrence Katz. 1998. “A Distinctive System: Origins and Impact of U.S. Unemployment Compensation.” In The Defining Moment: The Great Depression and the American Economy in the Twentieth Century, edited by Michael D. Bordo, Claudia Goldin, and Eugene N. University of Chicago Press.

Congressional Budget Office (CBO). 2012. Unemployment Insurance in the Wake of the Recent Recession. Publication no. 4525, November 2012.

Congressional Research Service (CRS). 2019. Unemployment Insurance: Consequences of Changes in State Unemployment Compensation Laws. R41859, updated October 23, 2019.

Elmendorf, Douglas W. 2011. “Policies for Increasing Economic Growth in 2012 and 2013.” Testimony of Douglas W. Elmendorf, Director, Congressional Budget Office, before the Committee on the Budget, United States Senate, November 15, 2011.

de Raaf, Shawn, Anne Motte, and Carole Vincent. 2005. “A Literature Review of Experience-Rating Employment Insurance in Canada.” SDRC Working Paper Series No. 05–03, May 2005.

Desilver, Drew. 2020. “Not All Unemployed People Get Unemployment Benefits; In Some States, Very Few Do.” FactTank Blog (Pew Research Center), April 24, 2020.

Doroghazi, Lauren. 2019. Beyond California AB 5 States Address Independent Contractors. MultiState, November 2019.

Galle, Brian. 2019. “How to Save Unemployment Insurance.” Arizona State Law Journal 50: 1009–1064.

Golden, Lonnie, and Jaeseung Kim. 2020. Underemployment Just Isn’t Working for U.S. Part-Time Workers. CLASP, May 2020.

Golshan, Tara, and Arthur Delaney. 2021. “The Looming Showdown Over Unemployment Benefits.” HuffPost, April 12, 2021.

Gould-Werth, Alix. 2016. “Workplace Experiences and Unemployment Insurance Claims: How Personal Relationships and the Structure of Work Shape Access to Public Benefits.” Social Service Review 90, no. 3 (June): 305–352.

Gould-Werth, Alix. 2020a. “Fool Me Once: Investing in Unemployment Insurance Systems to Avoid the Mistakes of the Great Recession During COVID-19.” Washington Center for Equitable Growth, April 30, 2020.

Gould-Werth, Alix. 2020b. The Long-Run Implications of Extending Unemployment Benefits in the United States for Workers, Firms, and the Economy. Washington Center for Equitable Growth, December 2020.

Gruber, Jonathan. 1997. “The Incidence of Payroll Taxation: Evidence from Chile.” Journal of Labor Economics 15, no. 53 (July): S72–S101.

Guo, Audrey, 2020. The Effects of Unemployment Insurance Taxation on Multi-Establishment Firms. MPRA Paper 97919, University Library of Munich, Germany.

Guo, Audrey, and Andrew C. Johnston. 2020. “The Finance of Unemployment Compensation and Its Consequence.” IZA Discussion Papers 13330, Institute of Labor Economics (IZA). June 2020.

Hyatt, Douglas E., and Boris Kralj. 1995. “The Impact of Workers’ Compensation Experience Rating on Employer Appeals Activity.” Industrial Relations 34, no 1 (January): 95–106.

Karni, Edi. 1999. “Optimal Unemployment Insurance: A Survey.” Southern Economic Journal 66, no. 2 (October): 442–465.

Leachman, Michael. 2015. “States Cutting Jobless Benefits Hadn’t Adequately Prepared for Recession.” Off the Charts Blog (Center on Budget and Policy Priorities), May 28, 2015.

McLeod, W. Bentley, and James M. Malcolmson. 1989. “Implicit Contracts, Incentive Compatability, and Involuntary Unemployment,” Econometrica 57: 447–480.

Miller, Michael, and Robert Pavosevich. 2019. “Alternative Methods of Experience Rating Unemployment Insurance Taxes.” Public Budgeting & Finance 39, no. 4 (Winter): 28–47.

Organisation for Economic Co-Operation and Development (OECD). 2019. The OECD Tax-Benefit Model for Canada: Description of Policy Rules for 2019.

Pavosevich, Robert. 2020. “The Cost of Layoffs in Unemployment Insurance Taxes.” BLS Monthly Labor Review, April 2020.

Ratner, David D. 2013. “Unemployment Insurance Experience Rating and Labor Market Dynamics.” Board of Governors of the Federal Reserve System Finance and Economics Discussion Series No. 2013-86, December 2013.

Ravenelle, Alexandrea J. 2019. Hustle and Gig: Struggling and Surviving in the Sharing Economy. Oakland: University of California Press.

Smith, Joshua, Valerie Wilson, and Josh Bivens. 2014. State Cuts to Jobless Benefits Did Not Help Workers or Taxpayers. Economic Policy Institute Briefing Paper #380, July 2014.

Thomason, Terry, and Silvanna Pozzebon. 2002. “Determinants of Firm Workplace Health and Safety and Claims Management Practices.” Industrial and Labor Relations Review 55, no. 2 (January): 286–307.

U.S. Census Bureau. 2019. “Table DP05. 2019 American Community Survey 1-Year Estimates Selected Population Profiles.” Accessed June 2021.

U.S. Department of Labor Employment and Training Administration (U.S. DOL-ETA). 2020a. “Estimated Employer Contribution Rates Calendar Year 2020” (data table). September 2020.

U.S. Department of Labor Employment and Training Administration (U.S. DOL-ETA). 2020b. “Monetary Entitlement” in Comparison of State Unemployment Laws 2020.

U.S. Department of Labor Employment and Training Administration (U.S. DOL-ETA). 2020c. “Significant Provisions of State Unemployment Insurance Laws,” January 2020.

U.S. Department of Labor Employment and Training Administration (U.S. DOL-ETA). 2021a. UI Replacement Rates Report (online database). Accessed June 8, 2021.

U.S. Department of Labor Employment and Training Administration (U.S. DOL-ETA). 2021b. Unemployment Insurance Chartbook (online database), “Category 12 Regular Program Insured Unemployment as a Percent of Total Unemployment.” Accessed June 2, 2021.

U.S. Department of Labor Employment and Training Administration (U.S. DOL-ETA). 2021c. “Unemployment Insurance Extended Benefits” (web page), accessed May 14, 2021.

U.S. Department of Labor Office of Unemployment Insurance Division of Fiscal and Actuarial Services (U.S. DOL-OUI). 2021. State Unemployment Insurance Tax Measures Report 2020. April 2021.

U.S. General Accounting Office (U.S. GAO). 1993. GAO/HRD-93-107, Unemployment Insurance: Program’s Ability to Meet Objectives Jeopardized. Report to the Chairman, Senate Committee on Finance, U.S. Senate, September 1993.

Vermont Department of Labor (Vermont DOL). 2021. Quarterly Reporting & Taxable Wage Information (online database). Accessed May 21, 2021.

Vroman, Wayne, Elaine Maag, Christopher O’Leary, and Stephen Woodbury. 2017. A Comparative Analysis of Unemployment Insurance Financing Methods. Washington, D.C. Department of Labor, Chief Evaluation Office, prepared by the Urban Institute, December 2017.

Wentworth, George. 2017. Closing Doors on the Unemployed: Why Most Jobless Workers Are Not Receiving Unemployment Insurance and What States Can Do About It. National Employment Law Project, December 2017.

Woodbury, Stephen A. 2004. Layoffs and Experience Rating of the Unemployment Insurance Payroll Tax: Panel Data Analysis of Employers in Three States. October 2004.


See related work on Unemployment insurance | Unemployment | Black Americans | Latinx Americans | Stimulus/stabilization policy | Women | Nonstandard work arrangements | Gig economy | Contingent workforce | Coronavirus