On the Record

The Scandal of the Unemployment Insurance Payroll Tax and a Simple Remedy

Gary Burtless

Prepared remarks by Gary Burtless for “Strengthening Unemployment Insurance for the 21st Century,” a national roundtable sponsored by the National Academy of Social Insurance with the U.S. Department of Labor, the Annie E. Casey Foundation, the W.E. Upjohn Institute for Employment Research and the DirectEmployers Association

Our panel addresses some basic questions about the unemployment insurance (UI) payroll tax. Patricia Anderson has tackled experience rating in the UI tax. This is certainly a unique feature of our financing system compared with ones in other industrial countries. I think it is basically a desirable feature of our financing compared with other countries’ financing systems. It’s reasonable to charge employers – and, indirectly, their workers – for the excess costs that high layoff rates impose on other actors in the system – that is, other employers and other insured workers. Most states could certainly improve the structure of their experience rating formulas to reduce undesirable subsidies to some classes of workers and employers, but the basic idea of experience rating is a good one.

I’ll focus on another unique aspect of our UI financing system … a uniquely bad aspect … and that is the extreme regressivity of the UI payroll tax. Most of you know that employers pay a statutory UI tax for each employee on their payrolls. In principle, this tax is related to the worker’s earnings. Employers pay a tax on wages below an annual ceiling called the “taxable wage base”. The tax base is so low that it is equivalent to a poll tax on each employee rather than an earnings-related tax.

Right now the federal minimum wage is $7.25 / hour. That means a minimum-wage worker who works 2,000 hours a year earns $14,500 a year. How many states have a “taxable wage base” lower than the annual earnings of a minimum-wage worker? The answer is 32. In other words, in two-thirds of American states employers pay exactly the same UI tax for a minimum-wage worker as they do for their CEO.

In seven states, employers pay the maximum UI tax for minimum-wage workers who are on half-time schedules. That is, a person earning the minimum wage who works an average of just 20 hours a week pays the maximum UI tax. (More precisely, the worker’s employer pays the tax in the worker’s behalf. However, most labor economists, including me, believe that all (or nearly all) of the burden of the UI payroll tax falls on workers. In the absence of the UI program and the UI tax that funds it, employers would pay their workers a higher wage, and the increase in employees’ wages would be approximately the same as the current UI tax.) Interestingly, in some states if these part-time workers get laid off they would not even qualify for UI benefits, even though employers paid the maximum UI tax on their paltry wages. If a minimum-wage worker holds two part-time jobs with two different employers, each employer pays the maximum UI tax in the worker’s behalf. The minimum-wage worker who holds two part-time jobs essentially pays twice the UI taxes paid on the CEO’s wages in each of the companies he works for.

It is fair to say a tax that is identical for a minimum-wage worker and a CEO is a seriously regressive tax. A few years ago Patricia Anderson wrote a paper with Bruce Meyer in which the two authors looked into the progressivity of the UI tax. (Patricia M. Anderson and Bruce Meyer, “Unemployment Insurance Tax Burdens and Benefits: Funding Family Leave and Reforming the Payroll Tax,” National Tax Journal 59, March 2006, pp. 77-95.) In the year they looked at, 1994, workers in the bottom one-tenth of the wage distribution paid about 2.8% of their wages to the UI system in taxes. Workers in the top one-tenth of the wage distribution paid about 0.4% of their wages in UI taxes. The tax-rate ratio was therefore 7-to-1. As a percent of wages, the bottom one-tenth of wage earners paid 7 times the UI tax paid by the top one-tenth of workers.

The data used in Patricia and Bruce’s paper are 16 years old. In many states the “taxable wage base” has not increased very much in those 16 years. However, wages in the top one-tenth of the earnings distribution have increased much faster than wages in the middle and at the bottom of the earnings distribution. It is a plausible guess that the UI tax has become even more regressive since the year analyzed by Patricia Anderson and Bruce Meyer.

Compared with the UI tax, both the Social Security payroll tax and the sales tax look like models of progressivity. The Social Security tax is capped when wages reach $106,800 a year. As I recall an old CBO analysis, the Social Security tax is actually a progressive tax up to family incomes of about $150,000 a year. That is, the Social Security tax represents a rising share of family income for families with incomes between $0 a year and $150,000 a year. In contrast, the UI payroll tax is lower for families in the middle income quintile than it is for families in the bottom quintile, and it is much lower for families in the top quintile than it is for families in either the bottom or the middle quintiles.

The UI tax does not look so bad if you consider it in relation to the distribution of the benefits that the tax pays for. In comparison to the income distribution of UI tax payers, the income distribution of UI recipients is pretty comparable. At least that is what Patricia and Bruce Meyer found in their analysis of 1994 data. My suspicion, however, is that the balance of tax payments and UI benefits has been getting less advantageous to people with low wages and incomes. The reason is that state UI benefit formulas more or less keep pace with the trend in wages. The average weekly UI benefit is roughly the same percentage of the average weekly wage as it was 15 years ago or 30 years ago.

The UI tax base, however, has not come remotely close to keeping up with average wages, except in a handful of states. As the taxable wage cap has fallen in relation to the average wage, a larger and larger percentage of the tax is falling on the wages of workers with the lowest wages. Some people might think this shift in the burden of paying for UI benefits is fair, but I do not.

The main factor that is generating increased regressivity in the UI tax is the very low cap on taxable wages. The low cap has two other pernicious effects:

  • First, it causes UI payroll tax revenues to climb more slowly than wages. This doesn’t make any sense when weekly UI benefits are climbing roughly in line with wage growth. In the long run, the effect of the adverse trend in tax revenues is to starve the system of revenues unless state legislatures increase the tax rate on earnings below the cap. But until they raise that tax rate, the effect of the low cap is to push state UI systems toward very low trust fund balances. This, I think, is very unhealthy for insurance protection in recessions. State legislatures facing low trust fund balances are tempted to hold down weekly benefits or tighten eligibility requirements in the middle of a recession, which is not a sensible thing to do.
  • Second, by maintaining low taxable wage caps, state UI systems tend to boost the tax rate applied to the low earnings amounts that are subject to the tax. The UI tax thus becomes the equivalent of a rising poll tax on new employment. Any job created in a year, except a job paid the very lowest wage rates, will cause employers to pay the maximum UI tax for the new job. Simple economic theory suggests the incentive created by a high tax rate imposed on a low taxable wage base is to discourage job creation while encouraging employers to add to the hours of their current employees. In states with a very low tax base, the UI tax becomes a tax penalty on job creation, at least in the short- and intermediate-runs.

In sum, the UI tax is regressive and growing more regressive each decade. The regressivity seems unfair, whatever adverse incentives are created by the tax. Second, the regressivity is caused by the very low and very slowly rising tax cap. This feature of the tax causes two other problems for the system. It slows the growth of UI tax revenue in comparison to wages given the fact that nominal wages have been rising 2½% to 4% a year for the past few decades. This in turn hurts UI trust fund balances, which reduces the insurance protection offered by UI in recessions. And finally, it pushes state legislatures to impose high taxes on a narrow tax base in order to finance benefits. This has undesirable incentive effects, because it discourages job creation in comparison with a tax that imposes lower tax rates on a larger tax base.

My policy proposal to remedy the problem is simple. The federal government should raise its own UI taxable wage base from $7,000 to one-half the wage base for Social Security, bringing it up to about $53,000 a year. It should then automatically raise the wage base every year in line with the percentage increase in the Social Security wage base. At the same time, it should require states to have a wage base that is no lower than the federal UI wage base.

It might seem astonishing from the perspective of 2010, but when UI and Social Security were created in the mid-1930s, both programs had the same taxable wage base. An overwhelming percentage of the 1930s workforce had annual wages that were below the UI and Social Security taxable wage cap. Compared with that era, my suggestion is quite modest. I am not asking that the wage bases for UI and Social Security be made the same. I only ask that the UI wage base be brought up to one-half the Social Security taxable earnings cap. This is a modest proposal, but it is a very sensible one, I think.