Government-mandated benefits, taxes, and wages.

I. Introduction

This paper contributes to an ongoing policy debate over government-mandated benefits [6; 9; 10; 13; 17; 22; 24; 27]. Current government mandates or "statuatory employee benefits," include social security, unemployment insurance, workers' compensation, maternity leave, etc. However, policy initiatives at every level of government routinely recommend new mandates. Among today's proposals, targeted employers would be required to offer new benefits from childcare to health care. Debates over proposals for new government mandates usually center on two related questions: i) do the advantages outweigh the costs?, and ii) what are the alternatives?

Assuming government intervention can be justified [17], those in favor of employer mandates contend compulsory benefits packages are less disruptive than a government tax/transfer program that offers the same benefits [22]. If employer-provided benefit mandates are fully valued by employees, wages fall to offset the cost of benefits, and there will be no efficiency cost. Alternatively, when government-provided benefits are financed with a tax, there is a deadweight loss.

A generally conceded drawback of employer mandates is that they often ignore part-time workers and the unemployed [5]. However, critics of employer mandates go further, emphasizing that higher mandated labor costs creates unemployment [11], and/or induces firms to rely more heavily on part-time workers [15; 21].

If instead of shedding labor or turning to part-time workers, targeted employers reduce wages (or wage growth) to offset the costs of new mandates,(1) then employee valuation becomes the central issue. For example, employees who prefer cash to the new benefits might keep their jobs, but their well-being declines. This paper offers a straightforward extension of the traditional utility maximization framework to help analyze employee responses, when employers offset benefit mandates with lower wages.

A careful analysis of worker and firm responses to mandates underlies any attempt to uncover advantages and costs of government-mandated benefits. As Mitchell recently concludes, an important aspect of the problem is the "need to know more about why workers differ in their demand for benefits" [17, 315]. The model provides insights into employee responses to policy proposals such as new benefit mandates (and new personal income tax rates). Recent proposals to reform health care provide an illustration?

Prominent among both federal and state health-care reform efforts are recommendations mandating employer-paid premiums for minimum "standard" health-care packages. A danger is that firms perceive such mandated increases in benefit obligations similar to a payroll tax.(3)

Given the globally-competitive nature of production, five management policy responses can be identified to preserve a firm's competitiveness: i) adopt technology that substitutes capital for labor and/or use part-time labor (to shrink the labor force affected by mandates), ii) move operations "offshore" (to lower productivity-adjusted labor costs), iii) improve the product (to justify higher prices and cover increased labor costs), iv) raise labor productivity (to offset higher labor costs),(4) or v) reduce non-mandated compensation (to offset legislated increases in benefits). The first four management responses are long-run policy options, and involve significant up-front costs. These include, respectively; capital purchases and severance pay, relocation expenses, product R&D, and training expenses. Given the deterrent of these up-front costs, and some recent empirical evidence [8; 9; 10], this study explores the impact on employees of mandated changes in benefits under the final "short-run" option.

The paper examines the prospect that targeted employers adjust wages to compensate for legislated increases in benefits. Although studies on the effect of benefit mandates on wages are fraught with data and estimation problems, a number of studies offer empirical evidence in support of this response. For example, in a careful study of the public sector, Ehrenberg and Smith [8] found a one-for-one trade-off between wages and employer-provided benefits. Moreover, noting that mandated health insurance plans are similar to existing workers' compensation programs, and that workers' compensation varies widely between states, Gruber and Krueger [10] estimate that (for five high-cost industries) 85% of workers' compensation costs were passed on to employees through lower wages. Most recently, Gruber [9] uses the Current Population Survey to study the extent to which the cost of a group-specific mandate (childbirth benefits) was shifted to the targeted group's wages. He finds "substantial shifting to wages (on the order of 100% of the cost of the mandate), with little effect on total labor input" [9, 4].

While this short-run management response, adjusting wages to compensate for increased benefits, avoids significant up-front costs to employers, it nonetheless involves sacrifice on the part of certain employees. The model suggests that while one category of employee may be indifferent (or even prefer) this adjustment, attracting and retaining another category is problematic. If employers require employees in the latter category, they have three management options: a) modify the wage adjustment, b) modify existing "voluntary" benefits, or c) offer new benefits. An exploration of these options provides a natural development of the model, and offers a starting point to help policymakers analyze ripple effects of government-mandated benefits.

The framework is developed assuming competitive labor markets where employees have alternative employment opportunities and employers pay the going market wage. Examples include the market for most professional, administrative, technical, clerical, production, and service employees. Along with a salary or "wage," business and government employers typically offer three categories of benefits: "in-kind," "lump-sum income," and "price savings."(5)

Employer-paid health insurance (or "noncontributory medical care") provides an example of an in-kind benefit. Although the cost to the employer is the same, the value of the benefit to some employees is modified if the dollar equivalent of a premium payment is disbursed directly as lump-sum income. An employer-subsidized discount on health insurance (or "contributory medical care") provides an example of price savings. An important result of this analysis is that management implications of a policy change depend on which category of benefit is impacted.

Understanding the role benefits play in compensation is critical to most companies since over the past 40 years, growth in benefits have outstripped wage increases. Employee benefit plans first grew substantially during World War II and the Korean War when a policy of wage freezes was in effect. At that time, employee benefits became an important factor in attracting and retaining employees. Since then, ". . . the rate of growth on outlays for . . . benefits has substantially outpaced that for wages and salaries . . ." [3, 2]. More recently: "nonwages grew from just under 20% of total labor cost in 1965 to over 27% in 1985," although "the growth of benefits as a share of compensation slowed during the 1970's and came to a halt in the mid-1980's" [27, 273, 291].

For purposes of this analysis it is useful to divide "benefits" into the three broad categories proposed above: (1) "in-kind," (2) "lump-sum income" payments, and (3) "price savings." Examples of the first category include "noncontributory" medical and dental care. Nonproduction bonuses and paid vacations provide examples of lump-sum income payments.(6) Finally, the benefit of price savings is enjoyed by those who, by virtue of working for an organization, have access to subsidized childcare, recreation activities, etc. Interestingly, whereas it costs employers to provide in-kind benefits and lump-sum income payments, the employer may or may not subsidize price savings enjoyed by an employee. For example, by virtue of working for a large organization, employees often receive discounts at restaurants, movie theaters, etc., at no cost to the employer.(7)

Table I. Selected Benefits and Percent Participation/Eligibility,
1989(*)

(1) In-Kind Benefits(**)
Employee Coverage                                48%

Medical Care
Family Coverage                                  31%
Employee Coverage                                34%

Dental Care
Family Coverage                                  25%
Life Insurance                                   82%
Free Parking                                     90%

(2) Lump-Sum Income Benefits(***)

Nonproduction Bonuses                            27%
Gifts                                            24%

(3) Price-Savings Benefits(**)
Employee Discounts                               54%
Employer-Subsidized Recreation                   28%
Subsidized Meals                                 23%
Employer Assistance For Child Care                5%

* Data taken from Bureau of Labor Statistics, Employee Benefits
Survey, 1990 [3].

** Percent of full-time employees participating in selected wholly
employer-financed employee benefits, medium and large firms.

*** Percent of full-time employees eligible for selected employee
benefits, medium and large firms.

Table I provides a breakdown of selected benefits into the three broad categories. The percent of full-time employees that participated in (or were eligible for) these benefits in 1989 is reported. The results are taken from a U.S. Department of Labor survey of medium and large firms [3]. In the case of In-Kind benefits, the percent of full-time employees participating in wholly-employer-financed benefit programs is reported. The percent of eligible full-time employees is reported under Lump-Sum Income payments and Price Savings.

Table II highlights in-kind medical care benefits. It reports the percent of wholly employer-financed (or "noncontributory") employee and dependent coverage by "HMO" and "OTHER" medical plans in medium and large firms in 1981 and 1989. The data reveal a decade-long decline in noncontributory medical care coverage.

Whereas noncontributory HMO coverage remained constant or increased slightly over this period, a significant decline took place in the percent of employees covered by "OTHER" wholly employer-financed medical plans. As a consequence, contributory medical plans account for an increased percentage of those offered by large and medium firms. Woodbury's [27] explanation is that employers reacted to the inflation of health insurance premiums by "shift[ing] the cost of health insurance to their employees . . . requiring workers to contribute to the monthly insurance premium, . . . and by mov[ing] away from traditional fee-for-service insurance and toward HMOs . . ." [27, 288].

In our model, employers' shares of health care premiums for optional contributory health plans, would make up the employer-provided subsidy in the category "Price-Savings Medical Care Benefits." If contributory health plans are (made) compulsory, then employers' shares of health care premiums would instead fall under the category, "In-Kind Medical Care Benefits," since in this case workers are required to accept the benefit and contribute to premium payments. The distinction relies on the compulsory nature of an employer-provided benefit. Any exploration of the impact of a particular benefit on employees depends on a proper categorization of that benefit. In the analysis that follows, it is convenient initially to focus on in-kind benefits, and subsequently to include lump-sum income payments and price savings as part of the compensation package.

Table II. Percent Employee and Dependent Coverage by Noncontributory
HMO and "OTHER" Medical Plans, 1981 and 1989(*)

                             1981              1989

In-Kind Benefit:
HMO Medical Care
Employee Coverage             35%               40%
Dependent Coverage            25%               26%

OTHER Medical Care
Employee Coverage             72%               55%
Dependent Coverage            51%               36%

* Data for 1981 taken from Employee Benefits Survey [3, 88]. Data
for 1989 taken from Employee Benefits in Medium and Large Firms
[4, 65].

Attempts to explain the post-war proliferation of wage supplements have turned to a number of factors; from unionization to the changing age composition of the labor force [26, 167]. However, the two factors that garner the most attention, and those that provide our primary focus, are employees' tastes and preferences, and income-tax avoidance. Besides workers' subjective evaluations of the utility value of specific benefits, a significant factor in employees' demands for in-kind benefits is the preferential treatment accorded many of those benefits under current federal personal income tax laws. The Tax Reform Act of 1984 also allows an employee fully to exclude "a qualified employee discount."

In contrast, a dual motivation exists for employers to supply in-kind benefits: the preferential treatment of certain benefit payments, and payroll tax avoidance. Under federal corporate income tax laws, payments by employers for insurance benefits and pension funds are deductible from the employer's gross income for tax purposes. However, and perhaps just as importantly, payroll taxes such as social security and workers' compensation are avoided when employers offer more of a worker's remuneration in-kind [19].

Since the focus of the present model is on worker responses to alternative employer-provided compensation packages, the model treats employers' wage-benefit offers as given. Moreover, it is assumed "employees are free to choose employers whose benefit mix maximizes their utility" [13, 373]. Useful variations of a different model that focuses more on employers' motivations to offer benefits can be found in Katz and Mankiw [12], Zax [29], and Scott, Berger, and Black [21].

In the present model, a representative employer adjusts wages such that, when combined with the market value of employer-provided benefits, total compensation matches that of the employee's next best alternative employment opportunity.(8) To simplify the analysis, an employee's next best alternative is assumed to be the known competitive market wage for that occupation [25].(9) Thus, the employer's decision rule in the model is straightforward. Close the pay gap between the wage and benefits package offer, and the employee's next best alternative (market) wage.(10)

The next section presents the employee's ("Kuhn-Tucker") utility maximization problem when compensation is given by the market wage. The model is modified in section III to examine employee responses to a wage and in-kind benefits package. A verbal and graphical exposition is woven together with a simple mathematical presentation of the problem. Two types of individuals (Types I and II) are distinguished as a natural outcome of the optimization. Type I's typically prefer "cash to kind," while Type II's are satisfied with the in-kind benefits provided.

Section IV presents the results of comparative statics exercises that describe the impact of changes in government-mandates, and of changes in income tax rates under new mandates. In section IV it is demonstrated that at least one of the two alternative benefits (lump-sum income payments or price savings) must accompany "in-kind" benefits to attract Type I's, or to make them indifferent between their current job and their next best alternative. Section V reveals that an inability to discriminate between Types I and II in a particular occupation, guarantees Type II's secure rents that make them better off than in their next best alternative. Section VI concludes with some policy implications of the model.

II. The Framework

Since the development of the model relies on differences in employees' tastes and preferences (as opposed to productivity differentials), individuals in a particular occupation are assumed to perform identically, each qualifying for the competitive market wage, [I.sub.o]. For purposes of illustration, after-tax disposable income available to these individuals can be spent on the consumption of only two goods, [x.sub.1] and [x.sub.2], and is given by [I.sub.o] (1 - t); where the tax rate, 0 [less than or equal to] t [less than] 1, is assumed to reflect federal, state, and local income taxes. Some [x.sub.1] is assumed to be provided by some employers as an in-kind benefit, and (in contrast with Zax [29], for example) is also available for purchase in retail markets. All other goods and services available for purchase are represented by the composite good, [x.sub.2].

It is convenient to establish a baseline derived from an initial analysis of employee choices in the absence of benefits. Once this baseline is established, the impact of benefits, and of government-mandated changes in benefits, on employers and employees, can be examined. To simplify the analysis first consider a representative employer that offers the competitive market wage, [I.sub.o], and expects employees to make independent purchase/consumption choices of [x.sub.1] and [x.sub.2]. A representative employee's response to this wage offer provides the baseline. The next step is to consider a representative employee's response to an employer-provided quantity [X.sub.1] of good 1, offered in lieu of some wages as a non-taxable in-kind benefit.

The baseline case is provided by the following example. Working for an employer that pays the competitive market wage, [I.sub.o], a representative employee's optimization involves choosing a consumption bundle ([x.sub.1], [x.sub.2]) each period that maximizes his/her (quasi-concave) utility function,

U = U ([x.sub.1], [x.sub.2]) (1)

subject to the budget constraint:

[P.sub.1][x.sub.1] + [p.sub.2][x.sub.2] [less than or equal to] [I.sub.o] (1-t), (2)

where

[x.sub.1],[x.sub.2] [greater than or equal to] 0, [P.sub.1],[P.sub.2] [greater than] 0, (3)

This is the standard utility maximization problem. The utility function, (1), translates the employee's decision to acquire [x.sub.1] and [x.sub.2] into a measure of satisfaction or "utility" derived from that particular set of choices. Choices are limited by the budget constraint, (2), such that in any given period, total spending on [x.sub.1] and [x.sub.2] cannot exceed after-tax income. Meanwhile, equation (3) simply requires that non-negative quantities of goods and services be purchased at positive prices.

Forming the Lagrangian function,

L = U([x.sub.1],[x.sub.2]) + [[Lambda].sub.o][[I.sub.o] (1 - t) - [P.sub.1][x.sub.1] - [P.sub.1][x.sub.1] - [P.sub.2][x.sub.2]], (4)

the first order necessary ("Kuhn-Tucker") conditions for a maximum are:

[L.sub.1] = [U.sub..1] - [[Lambda].sub.o][P.sub.1] [less than or equal to] 0; [x.sub.1] [greater than or equal to] 0 and [x.sub.1][L.sub.1] = 0 (5a)

[L.sub.2] = [U.sub.2] - [[Lambda].sub.o][P.sub.2] [less than or equal to] 0; [x.sub.2] [greater than or equal to] 0 and [x.sub.2][L.sub.2] = 0(5b)

where: [L.sub.i] = [Alpha]L/[Alpha][x.sub.i] and [U.sub.i] = [Alpha]U/[Alpha][x.sub.i]; i = 1,2, and

[Alpha]L/[Alpha][[Lambda].sub.o] = [I.sub.o] (1 - t) - [P.sub.1][x.sub.1] - [P.sub.1][x.sub.1] - [P.sub.2][x.sub.2] [greater than or equal to] 0; [[Lambda].sub.o] [greater than or equal to] 0 and [[Lambda].sub.o] ([Alpha]L/[[Alpha].sub.o] = 0.(6)

Assuming nonsatiation, and that some quantity of each good entering the utility function is consumed (i.e., [x.sub.1], [x.sub.2] [greater than] 0), equations (5a) and (5b) become equalities, and the usual utility maximization conditions hold. Specifically, employees purchase a combination of goods each period with their after-tax earnings so that their marginal rates of substitution (MRS) between [x.sub.1] and [x.sub.2] just equals the relative price of [x.sub.1] in terms of [x.sub.2]. In other words, the optimal consumption choice is one where the rate at which individuals are willing to trade off [x.sub.1] for [x.sub.2] matches the market rate of trade-off of [x.sub.1] for [x.sub.2], or from (5a) and (5b), where

([U.sub.1]/[U.sub.2]) = ([P.sub.1]/[P.sub.2]). (7)

This optimum corresponds to tangency points such as A or B in Figure 1. The two sets of indifference curves graphed in Figure 1 represent different consumption choices by two different employees in the same occupational category. In this analysis, employees in the same occupation that work for the same employer are equally productive, receive the same compensation package, and are distinguished exclusively by differences in their tastes and preferences,

III The Modified Model

Now consider an employer that offers in-kind benefits as a substitute for some cash. Economic theory has long argued that cash grants (e.g., wages) are more efficient than in-kind payments in providing a given level of utility. But Ross [20] demonstrates that, when there is an agency problem, there exist conditions under which in-kind benefits dominate cash transfers. This could occur if the head of a household, as the employer's agent, did not allocate resources in the way the employer desires. Paternalistic employer behavior offers one explanation for the provision of in-kind benefits. In this context, a government policy that legislates minimum benefit offerings (such as a "standard" medical care package) in effect mandates paternalistic behavior." Other arguments regarding government intervention and the provision of mandated benefits are discussed by Tobin [23] and Moffitt [18], for example.

Alternatively, Blackorby and Donaldson [2] argue that in-kind transfers may be explained as a means of targeting benefits towards intended recipients. Since providing in-kind benefits is more selective in its appeal to potential employees than the cash equivalent, this might influence an employer to provide specific in-kind benefits as a screening device, i.e., to select workers from an occupational labor pool with desired characteristics. For example, Mitchell [16] discusses how health insurance plans are tailored to attract and retain certain types of workers. In this context, a government policy that mandates standard minimum benefit offerings reduces the screening value of employer-provided benefits.(12)

Regardless of why the employer provides in-kind benefits, consider the consequences of providing a quantity of non-taxed in-kind benefit, [X.sub.1], of good 1. In this case the results of the optimization in section II no longer hold. The representative utility function in equation (1) must be modified to accommodate the sum of the employee's purchases, [x.sub.1], together with the employer-provided in-kind benefit, [X.sub.1]. The new utility function is given by,

U = U([x.sub.1] + [X.sub.1],[x.sub.2]). (1[prime])

An operative assumption is that "in-kind" benefits, [X.sub.1], cannot be exchanged (for cash or "kind"), or are "unconvertible" as in Tobin [23]. The case of tradeable or "convertible" in-kind benefits is discussed briefly in section V, by Tobin [23] in the context of wartime rationing, and by Moffitt [18] in the case of trafficking of food stamps.

Let [I.sub.1] [less than] [I.sub.o] represent the adjusted wage offered by the employer to accompany in-kind benefits, [X.sub.1]. Since the employer's objective is to close the pay gap between the compensation package offered and the employee's next best alternative wage ([I.sub.o]), the income level, [I.sub.1], is adjusted according to,

[I.sub.1] = [[I.sub.o] - [P.sub.1][X.sub.1]],

where [I.sub.o] represents the competitive wage, and [P.sub.1][X.sub.1] is the market value of the in-kind benefits package. Thus the wage, [I.sub.1], offered by the employer starts with the competitive market wage, [I.sub.o], and subtracts an estimate of the market value of the in-kind benefits provided.(13)

Since, according to equation (2), in any given period total spending cannot exceed after-tax income, we have

[P.sub.1][x.sub.1] + [P.sub.2][x.sub.2] [less than or equal to] [I.sub.1] (1 - t). (2[prime])

From (2[prime]) and (8), the (prospective) employee's budget constraint becomes

[P.sub.1][[x.sub.1] + [X.sub.1](1 - t)] + [P.sub.2][x.sub.2] [less than or equal to] [I.sub.o] (1 - t). (9)

The modified constraint given by (9) appears as the kinked line in Figure 1 that incorporates points [Mathematical Expression Omitted].

The quantity, [Mathematical Expression Omitted], represents the maximum quantity of good 2 that can be purchased with the cash payment [I.sub.1], given nothing is spent on good 1 (although note that at point C, while [Mathematical Expression Omitted] is the only purchase, the commodity bundle ([X.sub.l], [Mathematical Expression Omitted]) can be consumed). From (9), maximum [x.sub.2] is given by,

[Mathematical Expression Omitted]

Similarly, the quantity, [Mathematical Expression Omitted], represents the maximum an employee can consume of good 1. Together with in-kind benefit [X.sub.1], the budget given by (9) generates maximum [x.sub.1],

[Mathematical Expression Omitted]

This is greater than the previous maximum the employee could purchase (i.e., [I.sub.o] (1 - t)/[P.sub.1]) because the first [X.sub.1] units of good 1 require only [P.sub.1][X.sub.1] in pre-tax income as opposed to [P.sub.1][X.sub.1]/(1 -t). Thus, as is well known, income tax codes create a wedge between the effective prices faced by workers and the firm for the in-kind good. The impact of a change in tax policy in the model is analyzed in section IV below.

Given employer-provided in-kind benefits, [X.sub.1], the new problem facing the employee is to choose a commodity bundle ([x.sub.1], [x.sub.2]) which maximizes (1[prime]), subject to (3) and (9). Forming the new Lagrangian expression,

Z = U([x.sub.1] + [X.sub.1],[x.sub.2]) + [Lambda].sub.1] [[I.sub.o](1 - t) - [P.sub.1]([x.sub.1] + [X.sub.1] (1 - t)) - [P.sub.2][x.sub.2]], (10)

the first order necessary (Kuhn-Tucker) conditions for a maximum include:

[Z.sub.1] = [U.sub.1] - [[Lambda].sub.1][P.sub.1] [less than or equal to] 0; [x.sub.1] [greater than or equal to] 0 and [x.sub.1][Z.sub.1] = 0, (11)

[Z.sub.2] = [U.sub.2] - [[Lambda].sub.1][P.sub.2] [less than or equal to] 0; [x.sub.2] [greater than or equal to] 0 and [x.sub.2][Z.sub.2] = 0, (12)

and

[Alpha]Z/[Alpha][[Lambda].sub.1] = [I.sub.o](1 - t) - [P.sub.1][[x.sub.1] + [X.sub.1] (1 - t)] - [P.sub.2][x.sub.2] [greater than or equal to] 0; (13)

where [[Lambda].sub.1] [greater than or equal to] 0 and [[Lambda].sub.1]([Alpha]Z/[Alpha][[Lambda].sub.1]) = 0.

Unlike the previous optimization, since the employer already provides some in-kind good or service ([X.sub.1]) as part of the compensation package, it can no longer be assumed that employees purchase any [x.sub.1] with their disposable income.

On the one hand, the assumption that some of each good and service is consumed implies, as before, that [x.sub.2] [greater than] 0. So equation (12) appears as an equality. Thus, at the optimum, an employee's marginal utility of income is given by,

[[Lambda].sub.1] = [U.sub.2]/[P.sub.2]. (14)

On the other hand, since the employer already provides [X.sub.1] as an "in-kind" benefit, depending on their tastes and preferences, employees may or may not choose to purchase additional quantities, [x.sub.1], with their income.(14) Thus two cases must be considered for equation (11): (i) no purchase of good 1, i.e., [x.sub.1] = 0, or (ii) purchase of good 1, i.e., [x.sub.l] [greater than] 0. The two cases reveal two different types of individuals identified here as Type I and Type II.

(i) Type I: If [x.sub.1] = 0, then from (11).

[[Lambda].sub.1] [greater than] [U.sub.1]/[P.sub.1]. (15)

Combining (14) and (15) yields the condition that (at the optimum),

[U.sub.1]/[U.sub.2] [less than] [P.sub.1]/[P.sub.2]. (16)

Employees with tastes and preferences that result in (16) will consume where the marginal rate of substitution of [x.sub.1] for [x.sub.2] is less than the relative price of [x.sub.1] in terms of [x.sub.2]. These individuals, identified as Type I's, find themselves constrained and sated by the in-kind benefits provided. In other words, Type I's subjective evaluation of [x.sub.1] is less than the value the market (and, by assumption, the employer) places on [x.sub.1], and thus Type I's tend to prefer "cash to kind."

The kink at point C is the new optimum for Type I's, and offers less utility than their next best alternative (point A). At point C, from (14) and (16), the absolute value of Type I's MRS is less than the relative price of [x.sub.1], or analogous to condition (16), ([U.sub.1]/[U.sub.2]) [less than] ([P.sub.1]/[P.sub.2]). Since Type I's are sated with the employer-provided level [X.sub.1], they use their disposable (after-tax) income to consume [x.sub.2] exclusively (i.e., [x.sub.1] = 0, and [x.sub.2] = [Mathematical Expression Omitted]). Note that employees that receive the market wage, [I.sub.o], and consume at (or in the neighborhood of) point A can now be identified as Type I's.

(ii) Type II: Alternatively, if the level of a particular in-kind benefit provided is not sufficient for the employee, then [x.sub.1] [greater than] 0, and

[[Lambda].sub.1] = [U.sub.1]/[P.sub.1]. (17)

Combining (14) and (17) yields the condition that (at the optimum),

[U.sub.1]/[U.sub.2] = [P.sub.1]/[P.sub.2]. (18)

Employees that perceive the level of "in-kind" benefits provided as insufficient, are inframarginal (i.e., consume more of the "in-kind" good or service than the level provided). Those that are inframarginal satisfy (18), and are identified here as Type II's. It is clear from Figure 1 that for Type II's, whose optimum is at (or in the neighborhood of) point E, their after-tax income, adjusted for the level ([X.sub.1]) of in-kind benefits provided, does not constrain them from their next best alternative choices (point B). In fact, since the benefits they would have purchased in any case, [X.sub.1], are now non-taxable, Type II's optimum at point E is above their next best alternative. In this case, while the usual optimization conditions apply (i.e., ([U.sub.1]/[U.sub.2]) = ([P.sub.1]/[P.sub.2])), Type II's are better off than before. Note that employees that receive the market wage, [I.sub.o], and consume at (or in the neighborhood of) point B can now be identified as Type II's.

The outcome is that Type II's will self-select to work for employers that offer the adjusted (wage plus in-kind benefit) compensation package. However, while the, package pays rents to Type II's, it implicitly discriminates against Type I's. Under this compensation package, the best Type I's can do is consume at point C, which offers less utility than their next best alternative (point A). Together, varieties of Types I and II make up the entire labor force available in a particular occupation.

IV Results

The analysis of a change in benefits policy, and in tax policy, can be performed through straightforward comparative statics exercises. The three comparative statics results reported reveal likely impacts of new benefit mandates and tax policy changes on the two categories of employees. The first result concerns government-mandated benefits, the second, personal income tax rates, and the third, a combination of the two.

(i) The first result is that the greater the in-kind benefits ([X.sub.1]) provided, the more workers reveal themselves as (worse off) Type I's, but the better off remaining Type II's (i.e., [Mathematical Expression Omitted]; [Mathematical Expression Omitted]; also [ILLUSTRATION FOR FIGURE 1 OMITTED]). As an illustration, suppose a proposal is enacted that requires large firms to provide a minimum standard medical-care benefits package for their employees. Firms that currently offer less than this package (and possibly higher wages) can be viewed as being forced to provide an increased quantity of [X.sub.1]. If firms under the mandate respond by adjusting wages to compensate for higher required benefits, the new benefits policy, while favoring Type II's, will shrink their numbers. Employees of firms currently above the mandate remain largely unaffected. The result is a growth in worse off Type I's, together with a smaller core of better off Type II's.(15)

(ii) The second result is similar, in that an increase in marginal income tax rates is likely to result in a smaller core of (relatively) better off Type II's (i.e., from (9[prime]) and (9[double prime]), [Mathematical Expression Omitted], and [Mathematical Expression Omitted]; also [ILLUSTRATION FOR FIGURE 1 OMITTED]). Thus, higher marginal tax rates actually increase rents received by Type II's when compared to the new after-tax disposable income provided by their next-best alternative market wage. Whereas the higher tax rates restrict Type II's to a lower utility level under wages [I.sub.o], the increased "benefit" of tax exempt in-kind payments under the wage-benefits package outweigh the increased "cost" of the lower [x.sub.2] purchase available with the new higher-taxed wages. For example, although the new higher income tax rates for 1993 on high income individuals hurt Type II's, they actually increase rents received by Type II's. However, whereas the "absolute" impact of higher taxes is to reduce Type I's utility, the relative impact of higher taxes on Type I's is ambiguous. Whether Type I's are worse off or better off with the wage-benefit package, hinges on whether or not the negative income effect of the higher tax rate dominates the increased tax advantage of receiving (excess) in-kind benefits.

(iii) The third result concerns the influence of the level of benefits on the utility impact of marginal tax rate changes. Notably, the larger employer-provided in-kind benefits, the smaller the impact of an increase in income tax rates (i.e., using the envelope theorem, from equation (10); [[Alpha].sup.2][U.sup.*]/[Alpha]t[Alpha][X.sub.1] [greater than] 0). Thus, somewhat paradoxically, government mandated increases in (non-taxed) in-kind benefits make subsequent tax-rate increases comparatively more bearable (and decreases less helpful). At first glance this result offers the prospect that government mandates reduce the apparent costs to workers of raising revenues for government programs. However, the extra revenue gained from a given tax rate increase would of course be lower, since more workers' income is now sheltered (i.e., received as mandated non-taxable in-kind benefits).

Some empirical work [13; 14; 26] suggests an alternative interpretation of the last comparative statics result. These studies observe that as marginal tax rates go up, workers appear to demand a greater share of their compensation as benefits. The model supports this conclusion since the larger employee benefits ([X.sub.1]), the smaller the negative utility impact to the employee of higher tax rates.(16) In other words, higher tax rates are relatively less harmful to employees that take more of their compensation in-kind.

Conversely, the model indicates that the lower employee benefits ([X.sub.1]), the larger the positive utility impact of lower tax rates. In other words, lower tax rates are relatively more valuable to employees that take less of their compensation in-kind. This supports recent observations [27; 28] that "declining marginal tax rates of the 1980's played an important role in the slowing growth of employee benefits." [27, 280] Thus, mandating new benefits would lower the negative utility impact on workers of higher tax rates, but would also lower the positive utility impact of lower tax rates.

Point A in Figure 1 represents Type I's baseline consumption choices ([Mathematical Expression Omitted], [Mathematical Expression Omitted]). As discussed earlier, an employer that offers (mandated) in-kind benefits [X.sub.1], and adjusts wages according to equation (9), forces Type I's to an inferior (lower utility) optimum at point C. The comparative statics results reported suggest this problem (worse off Type I's) is intensified by any policy that (further) increases in-kind benefits.

As a consequence, any attempt to attract Type I's must involve either: a) a repeal of mandates (presumably beyond an individual employer's control), b) a "cash-for-kind" exchange of (or permission to trade) "voluntary" benefits, c) a lump-sum income payment (sufficient to alter the budget constraint to the kinked line passing through [[x[double prime].sub.2], D, F]), d) price savings (sufficient to change the slope of section [C, E, [x[prime].sub.1]] of the old constraint, so that the new kink is at D), or e) some combination. Note that the proliferation of in-kind benefits offered under so-called "cafeteria plans," where employees tailor their in-kind benefits package to their needs (and are occasionally allowed cash-for-kind exchanges), can be interpreted as an attempt on the part of employers to avoid lump-sum income payment or price subsidy adjustments by driving more employees into the Type II camp.(17)

If in-kind benefits are exogenously determined (through mandates, union agreements, by Congress (for public employees), etc.), and a) not enough Type II's self-select to work for the employer (such that the employer requires some Type I's), and/or b) the employer's marginal costs of providing in-kind benefits is different than the market price of those benefits, the model offers interesting insights. This paper focuses on the former problem (not enough Type II's), and leaves the latter (the employer's marginal cost of providing benefits) for another study.(18)

If the employer requires Type I's, then the compensation package offered must provide Type I's with at least the same utility they could achieve in their next-best alternative. This utility level is given as [Mathematical Expression Omitted] in Figure 1. Moreover, assuming the employer is locked into providing the in-kind benefit level, [X.sub.1], it is clear that either a lump-sum income payment, a subsidy on the price of [x.sub.2], or some combination of the two, could be designed to allow Type I's to achieve the same level of utility.

In Figure 1, given in-kind benefits [X.sub.1], Type I's must be allowed to obtain at least [x[double prime].sub.2], in order to he indifferent between their wage-benefit package and the competitive market wage. The next section examines the case of a lump-sum income payment. Other alternatives to attract Type I's (i.e., price-savings or a combination), are then briefly discussed.

V. Supplemental Compensation Requirements

In order to examine the size of the income payment required to attract Type I's, it is useful to proceed as follows. Observe that the budget constraint passing through the points A,B in Figure 1 is tangent to point A on Type I's utility function. At point A, ([Mathematical Expression Omitted], [Mathematical Expression Omitted]) represents Type I's optimum given the market wage, [I.sub.o]. The problem is to identify the lump-sum income payment that allows a Type I to achieve the baseline level of utility, [U.sup.I], enjoyed at point A, when Type I is forced to accept part of his/her compensation as in-kind benefit [X.sub.1]. From Figure 1 it is clear that the extra (lump-sum) income must allow Type I's to consume ([x[double prime].sub.2] - [x[prime].sub.2]) additional [x.sub.2] in order for them to achieve the same level of utility, [U.sup.I].

The difference ([x[double prime].sub.2] - [x[prime].sub.2]) can be revealed using a Taylor's series approximation (see Appendix 1). The results suggest that the more averse a Type I individual is to good 1 (i.e., the less willing they are to substitute additional [x.sub.1] for [x.sub.2] in the neighborhood of the employer-provided level of in-kind benefit, [X.sub.1]), the greater the added lump-sum income payment (or cash bonus) must be to attract them. Since Type I's typically differ in their marginal rates of substitution, and employers are not generally inclined to discriminate between employees on the basis of their tastes and preferences, employers tend to offer the same lump-sum income payment to all employees in the same occupational category. Furthermore, this payment must be large enough to attract/retain the last Type I employee required, thereby providing rents to all other Type I's.

More significantly, if employers are unable (or not allowed) to discriminate between Types I and II, then Type II's benefit even more handsomely from the lump-sum income payment adjustment. For example, the Type II individual illustrated in Figure 1 would now consume at point F, receiving even higher rents than before (at point E). Thus, attempts to attract Type I's are likely to result in significant rents being paid Type II's.

As an alternative, price savings can be used to attract Type I's. Again, if discrimination (in the provision of price savings) is prohibited, in their attempt to attract Type I's, employers inadvertently guarantee rents to Type II's. However, under a price-savings adjustment, Type II's receive less rents than they would under a lump-sum income adjustment.(19) Thus Type II's are better off under the lump-sum income approach than they are under the price-savings approach, or for that matter, under some combination of the two.

A final observation is that, even though trading in-kind benefits is often frowned upon, if in-kind benefits ([X.sub.1]) are tradeable (unlike health insurance), and the employer requires Type I's, it is less expensive to hire Type I's by allowing them to trade benefits as they please. If Type I's receive tradeable benefits (consumption goods), they will raise their own utility by substituting some of those benefits for other goods and services ([x.sub.2]). An advantage to employers in providing tradeable benefits, and allowing employees to trade or sell those benefits, is that it reduces price subsidies and/or lump-sum income adjustments necessary to attract Type I's. Moreover, this approach also lowers rents paid Type II's. Interestingly, even though tradeable benefits work against them (since lower lump-sum income and/or price savings adjustments are required), Type II's could nevertheless be on the receiving end of some tradeable in-kind benefit(s) unloaded by Type I's. This is because Type II's generally prefer more than the in-kind benefit, [X.sub.1], provided.

Now suppose [X.sub.1] is tradeable, but that trading (or selling) in-kind benefits is prohibited. Then Type I's are placed in the awkward position of being able to increase their utility by engaging in illegal "black market" trades (sales) and, somewhat paradoxically, some of the trades are likely to be consummated with Type II's. A simple lesson is that employers (and governments) must take care not to "criminalize" workers by offering, and then prohibiting the exchange of, tradeable in-kind benefits.

VI. Concluding Remarks

This paper provides a straightforward extension of the traditional utility maximization framework to help analyze a likely employer response to government-mandated benefits. A danger is that firms perceive mandated increases in benefit obligations similar to a payroll tax. Given the globally-competitive nature of production, five possible employer responses are identified: i) reduce the labor force and/or use more part-time labor, ii) move operations offshore, iii) improve the product, iv) raise labor productivity, or finally v) reduce non-mandated labor compensation. Prior studies offer empirical evidence that the final option is a likely response to new benefit mandates. The paper investigates the impact on employees of wage adjustments that compensate for increased benefit costs. Appendix II offers some possible extensions of the model.

For purposes of illustration it can be assumed "large" employers are the target of mandates [17, 311]. Thus targeted employers are concerned that their combination wage-benefits offer is commensurate with the (next best alternative) competitive market wage offered by "small" employers. In this analysis, employees in the same occupation that work for the same employer are equally productive, receive the same compensation package (each also qualifies for the competitive market wage), and are distinguished exclusively by differences in their tastes and preferences. The sole objective of workers in a specific occupation is to maximize their utility from available compensation packages. Two general types of employees are revealed as a natural outcome of this optimization.

Whereas Type I's typically prefer "cash to kind," Type II's are satisfied with employer-provided in-kind benefits, and receive rents. If an insufficient number of Type II's self-select to work for a particular wage and benefits package, it is demonstrated that additional (lump-sum income or price savings) benefits required to attract Type I's lead to further rents being paid Type II's.

Three comparative statics results are reported. The first result is that a new benefit mandate would lead to more worse-off Type I's, together with a smaller core of better off Type II's. Thus a subset of Type II's are revealed as most likely to support such initiatives. The second result is that higher marginal tax rates increase rents received by Type II's. Thus Type II's are least likely to object to higher tax rates. Finally, and somewhat paradoxically, government-mandated increases in (non-taxed) in-kind benefits make subsequent income tax rate increases comparatively more bearable for both categories. At first glance this last result offers the prospect that government mandates reduce apparent costs to workers of raising revenues for new government programs. However, extra revenues gained from a given tax increase would be lower since more worker income is now sheltered in the new mandate.

The next step would be to attempt to identify characteristics that distinguish Type I's from Type II's. For example, heads of households with dependents likely utilize greater quantities of health services than single persons. Thus, in the context of health care, the former may be Type II's, and the latter Type I's. However, complications arise in the case of working couples since, in order to avoid double coverage, one is likely to behave as a Type I, and the other as a Type II.

If wages could freely adjust, employers would offset individual differences in expected benefit costs (higher for Type II's than Type I's) through differences in wages (lower for Type II's than Type I's). However, employers tend to offer the same wages to employees that perform equally in the same occupational category. As a consequence, in the absence of wage discrimination, employers are likely to screen workers, hiring those (Type I's) with lower benefit costs.(20) A consequence of such wage rigidity is that mandated benefit programs are likely to work against the interests of those that most desire the benefit.

If government mandates target one set of ("large") employers, and exempt another ("small" employers), Type I's will tend to move to exempted jobs (even though they are attractive to targeted firms). Type II's are more likely to seek targeted firms, who will tend to screen them. Under this scenario, targeted firms could face a problem of adverse selection. Type I's who are good risks (say in terms of health insurance) opt out of targeted firms' plans by working for exempted firms. This leaves mostly the bad risks, Type II's, at targeted firms.

Another important problem arises with a different sort of wage rigidity. In the foregoing analysis, suppose the market wage ([I.sub.o]) is also a binding minimum wage. "In this case, wages cannot fall to offset employers' costs of providing a mandated benefit" [22, 181]. So, instead of adjusting wages in response to mandated benefits, minimum wage policies tend to trigger one of the first four management responses discussed earlier. Recall that the first two responses result in unemployment. Even more worrisome is the prospect of an increase in federal minimum wages combined with new benefit mandates. This could have a devastating impact on employment opportunities among low income workers. Given that a large fraction of employees who are without health insurance are paid low wages, this problem warrants further investigation.

In an era of tight budgets, mandating employer-provided benefits is an attractive means for a government to finance its policy agenda. With the administration's recent health care proposal, there has been increased interest in mandated benefits as a tool of social policy. Although the cost of mandated benefits do not appear directly in the government's budget, the model helps uncover indirect costs such as reduced welfare for some workers, unemployment, and lost tax revenues. The problem remains to weigh the benefits of mandates against their costs, and compare the results with those obtained under alternative approaches.

Appendix I

The indifference curve corresponding to the utility level [Mathematical Expression Omitted] in Figure 1, can be written as the following function, [x.sub.2] = [Phi]([x.sub.1], [U.sup.*]), where: from the implicit function theorem, [Phi][prime] = -([U.sub.1]/[U.sub.2]) [less than] 0 is the "marginal rate of substitution" of [x.sub.1] for [x.sub.2], or the slope of the indifference curve, and [Phi][double prime] = ([d.sub.2][x.sub.1]/[(d[x.sub.1]).sup.2]) [greater than] 0 reflects the curvature, or the rate of change of the slope, of the indifference curve. Using a first order Taylor's series expansion to approximate [x[double prime].sub.2] = [Phi]([X.sub.1]) yields:

[Mathematical Expression Omitted]

where:

[Mathematical Expression Omitted].

The first term is the linear approximation of [Phi]([X.sub.1]) and corresponds to [x[prime].sub.2] in Figure 1. The error of the (linear) approximation is given by the remainder term, R, and corresponds to the distance C, D in Figure 1. It is precisely this "error" term that yields the required lump-sum income payment we seek. To recruit and/or retain Type I's, from Figure 1, the lump-sum income payment must allow Type I's to consume at least R = [[x[double prime].sub.2] - [x[prime].sub.2]] additional [x.sub.2]. Examining the remainder term carefully, it is interesting to note that [Delta]R/[Delta][Phi][double prime] [greater than] 0. The smaller the marginal rate of substitution of [x.sub.1] for [x.sub.2] (or the less negative the slope of the indifference curve becomes) at higher levels of [x.sub.1], the greater R.

Appendix II

The model could be extended in several interesting directions. For example, suppose: (1) government-mandated benefits consist of a single in-kind benefit and the representative employee happens to hold another in-kind benefit in his choice set (i.e., not exactly the same as the one being offered), or (2) government-mandated benefits consist of multiple new in-kind benefits and the employee does not have in-kind benefits at all.

In the first case, the firm now has the opportunity to adjust to the new mandate by reducing wages, reducing the previously provided (different) in-kind benefit, or both. The impact on the model would be to add another in-kind benefit in the utility function and modify the budget constraint according to the firm's likely response to the mandate.

In general, the model can be modified to accommodate either (1) or (2) as follows. Assume there are N goods and services in the individual's utility function and that a subset 0 [less than or equal to] n [less than] N are already provided as in-kind benefits, and/or are offered as multiple new in-kind benefits. Then, a representative individual's problem can be written as:

[Mathematical Expression Omitted]

subject to:

[summation of] [P.sub.i] [[x.sub.i] + [X.sub.i](1 - t)] where i = 1 to n + [summation of] [P.sub.j][x.sub.j] where j = n + 1 to N [less than or equal to] [I.sub.o](1 - t); [for all]x, P [greater than or equal to] 0.

If an employer (mandate) provides n goods and services ([X.sub.i]) as part of a compensation package, then it is likely some individuals will choose [x.sub.i] = 0 for some of the i = 1,..., n goods. In fact, two cases must be considered for each of the n goods; either (a) [x.sub.i] = 0, or (b) [x.sub.i] [greater than] 0. From the Kuhn-Tucker conditions, at the optimum, for any two goods i = 1,..., n and j = n + 1,..., N, case (a) yields [U.sub.i]/[U.sub.j] [less than] [P.sub.i]/[P.sub.j], while case (b) yields [U.sub.i]/[U.sub.j] = [P.sub.i]/[P.sub.j].

The typology of individuals is now more complex in that it depends on the particular good(s) that is (are) provided in-kind, the tax status of each benefit, the individual's marginal rates of substitution for any two goods, and the particular manner in which the firm adjusts to mandates. Future extensions of the model along these lines could offer additional insight into this important area of research.

My thanks to Richard Ault, Don Bellante, Robert Ekelund, David Henderson, Guillermo Owen, and an anonymous referee for helpful suggestions. I am also grateful to the Economics Board at the University of California at Santa Cruz for offering facilities and valuable faculty contacts during my sabbatical, and to DRMI for financial support. The usual disclaimers apply.

1. Recent evidence that firms reduce wages to compensate for the costs of providing fringe benefit mandates is provided by Gruber and Krueger [10]. Their research suggests that increased costs of workers compensation mandates was largely shifted to wages with little effect on employment. In the context of health care mandates, the chief economist of the Employment Policies Institute observes that "workers who receive health insurance through their place of employment pay for that coverage, either directly through deductions from their paychecks or indirectly through lower wages" ("The Price of a Health Care Mandate," Wall Street Journal, 8/20/93 p. A12).

2. The administration's proposal was similar in many respects to that first suggested by Davis [6]. An important component of Davis's plan is an employer mandate under which: "Coverage of all full-time workers and their dependents under employer health plans would be required by law, and minimum benefit standards would be prescribed" [6, 350]. Moreover, the proposal would have employers "pay a specific percentage of premium costs for the mandated plan (say, 80 percent) . . . [although] higher premium shares could, of course, be permitted" [6, 350]. Interestingly, the premium share suggested is the same one that appeared in the President's proposal.

3. In fact, the Congressional Budget Office agreed that under the administration's proposal; "the health-care 'premiums' that employers would pay for their staff are, for budgetary purposes, a tax" [The Economist, 2/12/94, p. 25]. Although not all employers would pay more under the plan, the mandatory nature of the premiums makes them similar to a payroll tax.

4. Ideally, labor productivity would increase as workers are better cared for under mandated medical care plans, thereby offsetting higher labor costs to employers.

5. These definitions of benefits roughly correspond to those used by Leibowitz [13] for example, who distinguishes between "nontaxable substitutes for private consumption" (In-kind benefits), sick leave and paid vacations, etc. (Lump-sum income), and "discounted taxable substitutes for private consumption" (Price-savings). Productivity-related awards or bonuses are not considered explicitly in this analysis, since the model focuses on the utility value of different forms of labor compensation and not on the incentive value of different forms of labor compensation. The model studies workers in a specific occupational category with comparable "experience" or labor productivity.

6. While the category of benefits called "lump-sum income" can also be used to capture the discounted present value of "deferred compensation," such as retirement or pension plans, a dynamic extension of the static model developed here would extend the analysis, allowing an interesting differentiation of individuals according to their relative time rates of preference. The specific role of retirement pay in compensation has recently been investigated by Asch and Warner [1] in the context of a large hierarchical organization.

7. Thus the term "price savings" (which can result from third degree price discrimination) is more accurate than "price subsidies" (which presupposes employer provided discounts).

8. As Mitchell observes: "mandated-benefits proposals . . . often exempt some portion of the labor market from coverage . . . [and] for this reason small businesses are frequently allowed to avoid participating or . . . are permitted to be less comprehensive . . ." [17, 311]. Thus, targeted "large" employers are concerned that their combination wage-benefits offer, is commensurate with the (next best alternative) wage-benefits employees can earn from "small" employers.

9. It is also assumed the characteristics of different jobs within an occupation are sufficiently similar (or that perceived costs and benefits of different jobs offset one another) such that compensating wage differentials are not required.

10. As labor markets become more flexible and competitive with cross-border labor flows (under NAFTA or the EEC for example), the model could be used to anticipate impacts of exclusive new government mandated benefits in one country, when the market wage is available in one or more other countries. Alternatively, the model could be applied within the U.S. to an exclusive state-mandated benefit, when the market wage is available in other states, or as may be the case with health care, to mandates that target "large" firms and largely exempt (or subsidize) "small" firms.

11. Of course in the case of strong negative externalities that arise from the underutilization of certain available health care procedures, such as immunizations against communicable diseases, "paternalistic behavior" might increase overall social welfare.

12. The value of benefit offerings as a screening device is evidenced in a recent survey by the Kessler Exchange in California. As recently reported in the Wall Street Journal, 60% of the approximately 500 small-company owners questioned said they offer health insurance to attract good workers. In fact many small-business owners expressed concerns "they will lose a competitive edge in wooing skilled workers if mandates force everyone else to offer health care too" ["Mandated Health Coverage Finds Surprising Opponents," Wall Street Journal, 5/17/93, p. B2].

13. In her analysis of the trade-off between wages and fringe benefits, Leibowitz [13] found that full-time workers are able to choose between lower-paying jobs ([I.sub.1]) with benefits, and higher-paying jobs ([I.sub.o]) without.

14. This provides an important contrast with most other models, such as Zax [29] for example (whose model is based on Katz and Mankiw [12]), that assume "that individuals receiving nonwage compensation [cannot] purchase additional benefit goods in retail markets, . . ." [29, 172]

15. Note that this result is mitigated to the extent "managed competition" succeeds in reducing health care costs (i.e., lowering [P.sub.1]). However, the result would be reinforced if, instead, managed competition increases costs (i.e., raises

16. Scott, Berger and Black [21] also observe that higher income workers (that face higher marginal tax rates) tend to prefer a compensation package more heavily weighted toward fringe benefits. They study a different form of labor market segmentation than that described in this paper. They show that a "nondiscriminatory" provision in federal tax law to ensure all (high and low-income) workers (in any occupation) with the same experience level at a given firm receive the same fringe benefits, leads to labor market segmentation. The result is that "workers tend to sort themselves across the economy with high-income workers dominating some sectors and low-income workers dominating others" [21, 216]. In contrast, the model in this paper focuses on workers in a particular occupation with the same experience level and in the same general tax bracket. Sorting in this model results exclusively from differences in tastes and preferences given different employer offerings. For example suppose lawyers have a choice to work at large law firms that offer a wage and in-kind benefits package, or at small law firms that only offer the "market wage." Labor market segmentation would result in Type I's tending to sort themselves into small firms, and Type II's into large firms.

17. However, this begs the important question of the employer's costs of providing an increased menu of choices. To circumvent this transactions cost, experiments with so-called "flexible benefits plans" allow employees to make contributions deducted before taxes to dedicated accounts, thus reducing the employee's taxable income. [3, 160].

18. The latter problem is also interesting since an important assumption in the mandated health-care debate is that "managed competition" will (eventually) reduce the cost to employers of health-care premiums.

19. This comes from the fact that the price subsidy must be such as to shift the budget constraint from the kinked constraint which goes through points [[x[prime].sub.2], C, [x[prime].sub.1]] in Figure 1, to a line from the point [x[prime].sub.1] to D and from D to [x[double prime].sub.2]. Since the lump-sum income approach results in the budget constraint passing through [[x[double prime].sub.2], D, F] and the portion of that constraint which passes through D, F is everywhere above the portion of the price-savings budget constraint which passes through [x[prime].sub.1], D, Type II's enjoy higher utility under a lump-sum income adjustment than they do under a price-savings adjustment.

20. Of course this result would be offset to the extent Type II's have desired (productivity-related) characteristics that distinguish them from Type I's. Then their attraction to certain benefits might signal employers to hire them.

References

1. Asch, Beth and John Warner. "Compensation and Personnel Management in Hierarchical Organizations," Paper presented at Western Economic Association meetings, July 12, 1992.

2. Blackorby, Charles and David Donaldson, "Cash versus Kind, Self-selection, and Efficient Transfers." American Economic Review, September 1988.

3. Bureau of Labor Statistics. Employee Benefits Survey. U.S. Department of Labor, Bulletin 2362, June 1990.

4. -----. Employee Benefits in Medium and Large Firms, 1989. U.S. Department of Labor, Bulletin 2363, June 1990.

5. Chollet, Deborah, "A Profile of the Nonelderly Population Without Health Insurance," in Government Mandating of Employee Benefits, edited by D. Salisbury. Washington: Employee Benefit Research Institute, 1987.

6. Davis, Karen, "National Health Insurance: A Proposal." American Economic Review, May 1989.

7. Deaton, Angus and John Muellbauer. Economics and Consumer Behavior. New York: Cambridge University Press, New York, 1980.

8. Ehrenberg, Ronald and Robert Smith. "Who Pays for Pensions in the State and Local Sector: Workers or Employers?" IRRA 32nd Annual Proceedings, IRRA, 1979.

9. Gruber, Jonathan. "The Efficiency of a Group-Specific Mandated Benefit: Evidence from Health Insurance Benefits for Maternity." National Bureau of Economic Research, Working Paper #4157, 1992.

10. ----- and Alan Krueger. "The Incidence of Mandated Employer-Provided Insurance: Lessons from Workers' Compensation Insurance," in Tax Policy and the Economy, edited by D. Bradford. Cambridge, Mass.: MIT Press, 1991.

11. Hamermesh, Daniel. "The Demand for Workers and Hours and the Effects of Job Security Policies: Theories and Evidence," in Employment, Unemployment and Labor Utilization, edited by R. Hart. Boston: Unwin Hyman, 1988.

12. Katz, Avery and Gregory Mankiw, "How Should Fringe Benefits Be Taxed?" National Tax Journal, March 1985.

13. Leibowitz, Arleen. "Fringe Benefits in Employee Compensation," in The Measurement of Labor Costs, edited by J. Triplett. Chicago: University of Chicago Press and NBER, 1983.

14. Long, Jim and Frank Scott, "The Income Tax and Nonwage Compensation." Review of Economics and Statistics, May 1982.

15. Mangam, Garth, Donald Mayall and Kristin Nelson, "The Temporary Help Industry: A Response to the Dual Internal Labor Market." Industrial and Labor Relations Review, July 1985.

16. Mitchell, Olivia, "Fringe Benefits and Labor Mobility." Journal of Human Resources, Spring 1982.

17. -----. "The Effects of Mandating Benefits Packages," in Research in Labor Economics, edited by L. Bassi and D. Crawford. Greenwich, Conn.: JAI Press, 1990.

18. Moffitt, Robert, "Estimating the Value of an In-kind Transfer: The Case of Food Stamps." Econometrica, March 1989.

19. Mumy, Gene, "The Role of Taxes and Social Security in Determining the Structure of Wages and Pensions." Journal of Political Economy, June 1985.

20. Ross, Thomas, "On the Relative Efficiency of Cash Transfers and Subsidies." Economic Inquiry, July 1991.

21. Scott, Frank, Mark Berger, and Dan Black, "Effects of the Tax Treatment of Fringe Benefits on Labor Market Segmentation." Industrial and Labor Relations Review, January 1989.

22. Summers, Lawrence, "Some Simple Economics of Mandated Benefits." American Economic Review, May 1989.

23. Tobin, James, "A Survey of the Theory of Rationing." Econometrica, October 1952.

24. Uwe, Reinhardt. "Should All Employers Be Required by Law to Provide Basic Health Insurance Coverage for Their Employees and Dependents?," in Government Mandating of Employee Benefits. Washington: Employee Benefit Research Institute, 1987.

25. Wessels, Walter, "The Effect of Minimum Wages in the Presence of Fringe Benefits: An Expanded Model." Economic Inquiry, April 1980.

26. Woodbury, Stephen, "Substitution Between Wage and Nonwage Benefits." American Economic Review, March 1983.

27. -----. "Economic Issues in Employee Benefits," in Research in Labor Economics, edited by L. Bassi and D. Crawford. Greenwich, Conn.: JAI Press, 1990.

28. ----- and Daniel Hamermesh, "Taxes, Fringe Benefits and Faculty." Review of Economics and Statistics, May 1992.

29. Zax, Jeffrey, "Fringe Benefits, Income Tax Exemptions, and Implicit Subsidies." Journal of Public Economics, November 1988.