Permanent Job Loss

Charged benefits arise from both temporary and permanent layoffs.  Temporary layoffs are those leading only to relatively short-lived reductions in employment.  These cutbacks are reversed by new hires or recalls after fairly brief periods of time.  From the perspective of our current discussion, as introduced in chapter 1, the most important characteristic of temporary layoffs is that they of not lead to significant long-run reductions in the taxable payroll.  Permanent layoffs, by contrast, are defined as lasting curtailments in the work force and do imply a decrease in the taxable payroll.

It should perhaps be pointed out that the distinction between temporary and permanent layoffs need not necessarily correspond to the unemployment experiences of individual workers.  As an example, suppose that worker A is laid off and that worker B is hired as a replacement after a short time.  If A remains unemployed for a long period, then the layoff is regarded as temporary by the employer but as permanent by the employee.  In this study, layoffs are categorized as temporary or permanent according to the employer’s perception.  To be sure, since the rate of recall is typically very high among temporary layoffs, the latter are likely to be perceived as temporary by both the employer and by most of the affected employees.

Our subsequent analysis is facilitated by specifying a temporary layoff rate.  Which measures the proportion of the average stock of employees’ layoff in a calendar year?  Thus, if the layoff rate is 10 percent and the average level of employment is 450 (as in the preceding example), then the hypothetical employer effects 45 temporary layoffs in the calendar year.  Permanent layoffs are measured by the net reduction in the level of employment over the course of the calendar year.  In the example, permanent layoffs amount to 100 (i.e., 500-400).  Permanent layoffs are zero when the employer adds to employment over the course of the year.

Both temporary and permanent layoffs qualify for unemployment benefits.  The weekly benefit payment is determined by state law, usually at about one-half of previous earnings, subject to a maximum.  We shall use the concept of benefits per unemployment spell.  This is simply the product of the weekly benefit payment and the number of weeks that the laid-off worker receives unemployment compensation.  Thus, if the weekly benefit payment is $250 and the duration of the unemployment spell is 12 weeks, the benefit per unemployment spell is (12) ($250) =$3,000.  This would be the amount of benefits charged to the employer’s account with respect to each layoff.  In practice, the weekly benefit payment and the unemployment duration are likely to differ among employees.  Hence, the $3,000 should be considered as an average for all layoffs.

The level of charged benefits can be expressed as the following equation:

The first term in the square brackets in equation represents temporary layoffs, and the second term represents permanent layoffs.

Time Lags

With both types of experience rating, there are substantial periods of time between the year when unemployment benefits are paid and charged to an employer and the year when the tax rate changes.  Typically, the employer tax rate for 1994, for example, would have been determined between June and October of 1993, based on the layoff experience over the years 1990, 1991, 1992.  These time lags prevent the immediate response of taxes to charged benefits and may cause inequality between a firm’s charged benefits and its tax payments in any particular year Furthermore, neither method allows for interest payments or discounting.  Thus, even if the firm pays $100 in taxes for each $100 of charged benefits, the lag of the former behind the latter implies that in present value terms, tax payments fall short of charged benefits.

A comparison of a public UI system with no experience rating (i.e., a uniform tax) with one that is fully experience rated yields two key conclusions.  First, unless all employers in the system belong to the same risk class, a uniform tax will lead to a distortion of relative prices and thus the cross-subsidization of unstable by stable activities.  As a result, unstable activities are encouraged, and stable ones are discouraged.  Full experience rating avoids these problems and generates an allocation of resources similar to that of well-functioning markets.  Less-than-full experience rating leads to cross-subsidization.

Joseph Becker (1992, 1981) emphasized the distortion of relative prices and the consequent cross-subsidization in the UI system when there is less-than-full experience rating.  He referred to these problems as inappropriate “cost accounting.”  Further, his empirical investigations showed that construction firms tended to be heavily subsidized by other businesses, especially those in finance, the services, and trade.  In the long run, this situation must be expected to lead to relatively low output prices, relatively high real wages, and relative overproduction in the subsidized sectors as compared to the operation of perfect markets.  For further analysis, see Adams (1986.)

Subsequent empirical work has confirmed Becker’s original research finding of substantial cross-subsidization among firms and industries.  The most recent microeconomic empirical analyses of cross-subsidization upon relative e4mployment levels was undertaken by Deere (1991), who concluded that “a 10% increase in the implicit subsidy to a layoff increases the employment share in construction by about 1.7% and decreases the employment share in services by almost 1%.”

The second basic difference between a uniform and an experience rated UI tax relates to the incentives for employers to control layoffs by changing employment practices.  Many theoretical aspects of the incentive effects of experience rating have been treated in the literature.  Two well-known articles are the ones by Feldstein (1976) and Baily (1978).  In addition, Brechling (1977) has presented a model of the incentive effects in the institutional framework of the most common U.S. method of experience rating (that is, the reserve ratio method).  This line of research has been continued by Wolcowitz (1984) and Cook (1992).

Briefly, the argument states that since experience rating implies a marginal tax cost for layoffs, a cost-minimizing employer will attempt to adjust layoff practices in response to increases in the degree of experience rating, so as to reduce the level of layoffs, benefit payments, and, hence, employer taxes.  The incentive is optimized when there is full experience rating because this would lead to the same level of layoffs that would be generated in perfectly working commodity, capital, and labor markets.  This result is especially clear in the Feldstein (1976) model.

The Case against Experience Rating

We are aware of two kinds of counterarguments to the case for experience rating.  Since these criticisms have usually been made in comments on papers or in conversation, it is difficult to credit them to particular individuals.  Hence, they are presented here without attribution.

1.  It is often argued that individual employers do not react to experience rating by altering their layoff patterns.  There are two versions of this position.  First, the shifting of the tax removes any incentive for firms to control their layoffs, and, second, even with an appropriate incentive, employers are unable to control layoffs.  Let us consider these two versions in turn.

The first line of reasoning is that since the burden of the UI tax is likely to be shifted forward via changes in output prices or backward via changes in wages, it cannot have any incentive effects; thus, experience rating cannot be effective in reducing layoffs.  This argument is based on the implicit assumption that tax shifting takes place instantaneously, automatically, and at the level of the individual employer.  Suppose, for example, that an employer is considering a change in maternity leave policy that, ceteris paribus, would reduce taxes by $X and have a marginal cost of $Y.  If the employer would invariably have to pass on the entire net marginal surplus of $(X-Y) to workers in the form of increased wages (and/or to customers in the form of reduced output prices), then the marginal return to the employer would always be zero.  Here, tax shifting has been defined in such a narrow fashion that the employer has no inducement to introduce new technology, because any net benefit in the form of higher profits is immediately passed on to others.  Thus, there cannot be a profit motive, as economists normally understand that concept.

In contrast to the narrow view of tax shifting, it seems reasonable to assume, as we have done in the illustrative two-industry model, that some industry average of the UI tax is shifted forward or backward and that the behavior of any one employer does not affect the average tax significantly.  In this case, individual employers correctly treat their output prices and wages (or compensation packages) as independent of their own attempts to reduce their tax bills; employers can reap, through higher profits, the full net benefits of any changes in layoff patterns that accrue to them through experience rating.

The second version of the argument is based on the assumption that, because employers cannot change their layoff and employment patterns, they have very limited ability to respond to the incentives of experience rating.  Hence, experience rating may be unfair for employers who operate in markets with particularly uncertain and volatile conditions.

In our view, the level of incentives of the tax system and degree to which employers can and do respond to these incentives must be determined primarily by empirical investigation, not by theoretical argument. In a very general empirical framework, it is well known that Japanese employment practices lead to substantially lower layoff rates (and unemployment levels) than are customary in the U.S., yet Japanese firms have been well able to compete successfully in international markets.  Hence, it would appear that there are some opportunities for U.S. institutional framework has shown repeatedly that employers do react to high marginal tax costs of layoffs by reducing layoffs.  The initial findings by Feldstein (1978), Brechling (1981), Clark and summers (1982), and Topel (1983) have been supplemented recently by the research of Card and Levine (1994) and Anderson and Meyer (1994).  We, therefore, reject the assertion that, in general, employers cannot control their layoff patterns under an experience rated UI tax structure.

2.  It is sometimes argued that experience rating violates the basic principles of insurance and that it should be incomplete or even be replaced by a uniform tax.  Our discussion earlier in this section has shown that full experience rating is quite compatible with accepted insurance principles.  With experience rating, both employers and employees can be provided with some insurance.  Even if experience rating were complete, only employers would lose their coverage, while employees would remain insured.

In the absence of any type of UI program, a reasonably well-functioning labor market would generate, at least in the long run, higher average earnings in the unstable industry than in the stable one.  The absence of earnings during layoffs would be more than offset by high earnings during periods of employment: the average annual earnings in the unstable industry would have to compensate risk-averse employees for the cost and trouble to self-insure against the consequences of layoffs.  Competition in the labor market would maintain this earnings differential.

Workers in the unstable industry would have an incentive to control their unemployment duration by accepting any alternative job offers leading to an increase in earnings.  Similarly, employers in the unstable industry would have an incentive to control layoffs; if they could reduce layoffs at a sufficiently low marginal cost, they would be able to attract workers from the stable industry at lower annual earnings.  Layoff rates, unemployment duration, and earnings would be optimal, giber that risk-averse workers must self-insure against fluctuating income, in the sense that all agents optimize by equating the appropriate marginal costs and benefits.

The allocation of resources between the stable and unstable industries would also tend to be optimal in that the appropriate weight is given to instability and because workers must self-insure.  Any exogenous reduction in the degree of volatility would lead to an expansion of the unstable industry relative to the other sector.

For comparison with the cases described in the following sections, let us suppose that the smoothly distributed annual earnings in the stable industry are $25,000, while the fluctuating earnings in the unstable industry are $26,000.  The difference of $1,000 represents the cost of self-insurance.

Private Unemployment Insurance

Consider now that there is a private UI program, which can offer a constant annual income stream to workers in the unstable industry at a uniform price per employee that is less than the cost of self-insurance.  An insurance company might be able to offer such a relatively cheap contract by pooling the risk of income loss both across individual workers and over time.  In practice, the insurance company would have to charge premiums that might be substantial during periods of employment and then return most, but not all, of them as benefits during periods of unemployment.  The administrative costs, as well as normal profits for insurance provision, are covered by the difference between total premiums and total benefit payments.  Accordingly, we define two income concepts for employees in the unstable industry: gross income is the total average annual income received from the employer, while net or disposable income is the gross income plus total benefits received minus total premiums paid.  Movement by workers between the two industries equals the gross income (which is identical to net income) in the stable industry.

In comparison with the case of worker self-insurance, relative labor costs have fallen in the unstable industry, so that there will be a tendency for it to expand.  In other words, since risk pooling through insurance has removed part of the undesirable effects of instability, the activity in the unstable industry is likely to expand at the expense of the other.

The introduction of private insurance has given rise to two possible moral hazard problems.  First, since the insurance companies pay the full salary as long as individuals remain unemployed, workers have no incentive to accept any new job if they value the leisure time afforded by being unemployed.  This “free rider” situation may be avoided partially by proper policing of the insurance contract or by increasing worker participation in the insurance.  Clearly, increases in such participation may be obtained through experience rating of individual workers or through benefit payments that decline with the duration of unemployment.

The second type of moral hazard arises from the behavior of employers.  Employers have no incentive to increase the degree of employment smoothing because all employees already enjoy even incomes.  Hence, personnel would not be prepared to give up income for employment stability.  Some degree of worker participation in the insurance may, again, remove part of this moral hazard.

In terms of the previous numerical example, assume that the private insurance industry can offer insurance at a price of $500 a year.  A possible long-run equilibrium may then be established by an annual average (gross and net) income of $25,300 in the stable industry and a gross average income of $25,800 in the unstable industry, of which workers pay $500 to insurance companies and retain $25,300 as net average annual income.

In this chapter, we sketch the background to the central analysis presented in the remainder of the book.  First, we discuss briefly the major principles of insurance and the manner in which experience ration operates.  Second, systems of unemployment insurance (UI) without and with experience ration are outlined and analyzed.  Third, the two chief experience-rating methods currently used in the U.S. are described.  We then briefly review some of the relevant recent theoretical literature on experience rating.

Some Basic Principles of Insurance

Insurance is a contract in which an individual pays premiums to an insurer, who promises to compensate the insurant for a loss caused by some unpredictable circumstance covered by the agreement.  Thus, insurance is a hedge against the costs of uncertain events.  Risk aversion on the part of the insurant and the ability of the insurer to pool risks make insurance possible, likely, and generally efficient from an economic perspective.

Insurance may, however, entail some inefficiencies.  Most notably, the so-called moral hazard problem arises when the insurant has some control over the insured event.  For example, an insured business owner facing bankruptcy might hire an arsonist; if the owner could easily arrange for the burning of the business, insurers would be unwilling to provide insurance.  Thus, for private insurance markets to exist, the insurer must be able to control moral hazard at a low cost.

In well- functioning insurance markets, the price of insurance, namely, the premium, is set at a level such that the present value of the stream of the insurant’s premiums is just equal to the expected present value of the individual’s insurance claims plus the cost of administering the contract, all measured over an appropriately long period of time.  Insurants with similar risks of loss are grouped together into a risk class and are charged the same insurance premium.  Thus, within a risk class, insurance claims and premiums tend to offset one another, both over time and across different insurants.  In other words, risks are pooled both over time and among different members of the same risk class.  As a consequence, over an appropriate span of time, the members of the risk class finance their own claims, producing neither a surplus nor a deficit.  In any particular year, any member of the risk class may, of course, have a deficit or a surplus with the pool.

It follows that an insurant’s weekly, monthly, or annual insurance premiums will differ according to the risk class to which that individual is assigned.  For example, teenagers pay higher automobile insurance premiums than do adults, house insurance is more expensive for dwellings located near frequent mud slides or in a hurricane area, life insurance is more expensive for people with severe medical conditions,  and so on.

In some types of insurance, in which the insured event occurs repeatedly, the insurant may b reassigned periodically to a different risk class, depending on the person’s prior claims experience.  This procedure is caked experience rating, and it is particularly helpful when the factors, such as age, sex, and marital status, are not very accurate predictors of the insurant’s risk class.  Thus, if an insurant’s claims were very low over a period of years, the individual may be assigned to a lower risk class with a decreased premium.  Perhaps the best known example of experience rating occurs in automobile insurance in the form of “good driver discounts.” Experience rating may also be affected through a special levy, such as deductibles or copayments, when an insurance claim is paid.

Two related aspects of experience rating are of special interest.  First, experience rating implies incentives for the insurant to reduce claims by either controlling the uncertain events or by not submitting claims.  Experience rating in automobile insurance may induce insurants to drive more carefully and/or to pay for minor accidental damage out of their own pockets, so as not to lose the “no claims bonus.” In this way, experience rating tends to reduce the various forms of moral hazard in insurance.

Second, experience rating implies that there is participation by the insurant in the insurance against the occurrence of the uncertain event.  This is the coinsurance aspect of experience rating.  The extent of this participation depends upon the degree of experience rating.  As the degree of experience rating increases, the insurant’s premium (or copayment) corresponds more and more closely to his/her recent claims experience, so that the insurant pays more of the bill out-of-pocket.

Suppose, for instance, that the insurant has a claim for $5,000 and that consequently this individual’s premium either shows no change or rises immediately by $500 or $5,000.  If the premium does not increase, there is no experience rating.  If it rises by $5,000, experience rating is extreme in that the insurant suffers the entire loss.  Since the total insurance premium includes a charge for administrative expenses, the insurant will reject contracts with such severe experience rating conditions: they impose costs (the $5,000 plus administrative fees) yet offer no insurance protection.

Experience rating appears to be particularly effective in discouraging relatively small and frequent claims against which insurants are quite willing to self-insure.  Further, since these types of claims tend to have especially high administrative costs, their elimination is likely to reduce average insurance premiums.

The U.S. unemployment insurance (UI) system is unique in the world in that it is financed by an experience-rated payroll tax.  This means that individual firms pay higher or lower UI taxes depending on whether they cause more or less unemployment__and unemployment benefit payments.  Experience rating internalizes the costs of unemployment to individual firms, thereby inducing them to stabilize employment patterns.  Hence, experience rating may lead to a reduction in layoffs and in unemployment.

Since the inception of the federal-state UI system in the mid-1930s, the experience-rating provisions have been vigorously debated among politicians, trade unionists, business people, and academic economists.  The last comprehensive treatment of all aspects of experience rating was undertaken by Joseph M. Becker, whose seminal book on this topic was published in 1972.  His views were restated, with minor modifications, in his 1981 monograph.  Becker’s work was followed by the contributions of Feldstein (1976), Brechling (1977), and Baily (1978).  The early literature was surveyed by Hamermesh (1977, 1978), Gustman (1982), and Topel and Welch (1980).

The Need for Research

In the research underlying this book.We havereexam-ined  the role of experience rating in the UI system.  Such reconsideration is needed because of recent changes in labor market conditions, including the growth in permanent employment reductions due to major downsizing by some large employers, increased numbers of plant closures, and bankruptcies.  Large structural shifts in economic activity and employment took place in the 1980s, and they are likely to continue.  Thus, the question arises as to whether the present UI system and, in particular, its experience-rating provisions are capable of coping adequately with the consequences of the substantial reallocation of labor.  Our research is aimed at providing some answers to this important economic and political question.

Current systems of experience rating seem to be well designed to allocate the costs of unemployment caused by temporary, easily predictable, and recurrent layoffs.  Such layoffs do not lead to significant permanent employment changes.  There are valid arguments for allocating the costs of temporary layoffs to the firms that caused them.  Further, in most states, it seems that the existing experience-rating systems can be modified to allocate these costs to the appropriate firms.

Most theoretical and empirical investigations of the UI system have been based on models of temporary layoffs.  For instance, the influential article by Feldstein (1976) and later contributions by Wolcowitz (1984) and Cook (1992), all employ an approach, similar to that of early implicit contract theories, in which employees have a lasting attachment to a particular firm, but are laid off periodically and later recalled in a fairly predictable manner.  In these models, UI benefits are a means of raising workers’ incomes during periods of temporary unemployment, and, so the argument goes, these benefits ought to be regarded as part of the firm’s labor costs.  If, by contrast, UI benefits were financed by a general payroll (or other) tax not based on an experience rating, there might be more layoffs, and high-layoff firms would receive a permanent subsidy from low-layoff firms.  As a result, high layoff activities would be expanded.  Experience rating clearly leads to increased efficiency and social well-being, at least from a long-run perspective.

In much of the UI research, experience rating is modeled in a fairly abstract manner.  For example, following Feldstein’s original contribution, many researchers have described experience rating simply by the ratio of total benefits charged to the tax payments of employers.  The reserve ratio method of experience rating, which is the most commonly used approach, has been modeled by Brechling (1997) and, more recently, by Wolcowitz (1984) and Cook (1992).  Cook has extended the work to the benefit ratio method, the other important technique of experience rating.  Both approaches imply that experience rating is imperfect in the sense that (1) there are substantial lags between the payments of benefits to workers and the corresponding receipt of UI taxes, and (2) there are maximum and minimum tax rates that curtail or even suspend the relationship between benefits and taxes.  Although these features of the UI tax systems have been modeled in an insightful manner by both Wolcowitz and Cook, neither examines the effects of permanent employment reductions.

Suppose now that a substantial proportion of total layoffs is permanent, necessitated by some structural development, such as changes in tasted, new technologies, or competition from imports.  Some plants may have to close completely, some may go into bankruptcy, and others may experience substantial downsizing.  In any case, employment in the industry must contract substantially.  From a social point of view, who should bear the unemployment costs of these layoffs?  Is experience rating a desirable property of the UI system?  So present experience-rating methods allocate these costs appropriately?  These types of questions have not been considered in the previous literature on experience rating.  Consequently, we have addressed these issues in our research for this book.