Originally presented at a Conference on Labor in Nonprofit Industry and Government held at Princeton University, these studies are the first to provide an economic discussion of the public sector labor market.
Melvin Reder examines the effect of the absence of the profit motive on employment and wage determination in the public sector. Orley Ashenfelter and Ronald Ehrenberg estimate the elasticities of demand for various types of labor employed by state and local governments. Theoretical ideas about behavior in nonprofit industries are employed by Richard Freeman to study the higher education industry.
John Burton and Charles Krider try to predict the incidence of strikes in the public sector, while Donald Frey presents a model of the behavior of school boards in hiring faculty. The magnitude of the extra wage received by unionized public employees is compared by Daniel Hamermesh to that of private unionized workers in the same occupation.
Originally published in 1975.
The Princeton Legacy Library uses the latest print-on-demand technology to again make available previously out-of-print books from the distinguished backlist of Princeton University Press. These editions preserve the original texts of these important books while presenting them in durable paperback and hardcover editions. The goal of the Princeton Legacy Library is to vastly increase access to the rich scholarly heritage found in the thousands of books published by Princeton University Press since its founding in 1905.
Read an Excerpt
Labor in the Public and Nonprofit Sectors
By Daniel S. Hamermesh
PRINCETON UNIVERSITY PRESSCopyright © 1975 Princeton University Press
All rights reserved.
MELVIN W. REDER
The Theory of Employment and Wages in the Public Sector
Since no later point will be convenient, at the outset I should like to contrast the quality of the recent work in the area of this conference with the early pioneering efforts at analyzing interindustry wage differentials in the immediate post-World War Il period. In theoretical conception, econometric technique, and wealth of data analyzed, the work to which I shall refer is incomparably superior to the efforts of only a quarter century ago. However, my role is not to dwell on how far we have come, but on how far we have still to go.
I propose to organize my discussion around the following question: Why should there be a wage differential between the public and nonpublic sectors for "comparable" workers? The quick and largely correct answer is that among workers who are truly comparable there will be no such differential in the long run. The critical point, obviously, is what is meant by "comparable."
Roughly, in order to be truly comparable, workers should be alike in respect of embodied human capital, tastes, and location. Alike in respect of "embodied capital" means, in practice, alike in respect of age, years of (relevant) experience, years of schooling, and all natural endowments not fully reflected in the measures of the other characteristics. For workers to be alike in respect of tastes means that a per-hour, equalizing, pecuniary wage differential between any two jobs would be the same for any pair of workers. Location is a significant attribute of a job, and "equalizing differentials" between otherwise identical jobs may arise simply because the jobs have different locations; workers will be comparable only after such locational differentials have been taken into account.
Thus defined, hourly wage differentials between comparable workers would seem incompatible with competitive equilibrium. Yet such differentials may, and I suspect do, exist between public and other types of employment. The explanation lies in the peculiar characteristics of public employment that make jobs that are superficially similar to those in the private sector have appreciably different marginal products and supply prices.
Originally I had hoped to relate the theoretical considerations developed to observations on the differences in wage rates actually paid comparable workers in the public and private sectors. However, this intention has been frustrated. The differences in job characteristics, effort expended, and the like between superficially comparable workers in the public and private sectors are too important from the standpoint of the theory advanced to permit any useful confrontation of theory with data now available. Accordingly, I have described the data now available on public and private sector hourly earnings in the Appendix. The reader should consider it solely as background information.
This paper focuses on the public sector. The hypothesis upon which the discussion is based is that all decision makers are utility maximizers. The inferences drawn from this hypothesis are very simple and, save for one or two duly noted exceptions, follow directly from the assumption that the public-sector decision makers are successful utility maximizers operating under static conditions and with no "political rent." The assumption of static conditions means that public-sector decision makers act as though there were no difference between present and expected future values of variables subject to control. The notion of political rent is a bit more complicated and can best be discussed later.
The public-sector employer may be a federal, state, or local government or any type of special-purpose governmental unit. For convenience, let the definitive characteristic of a governmental unit be possession of some power to levy taxes on persons within its zone of authority.
The demand function of a governmental unit, for labor of any given kind, will vary with the political characteristics of the unit. Thus the demand function of the federal government will (or may) differ from that of a state government, which in turn will differ from that of a local government. Most of the extant empirical research on public employee wages refers to local governments; however, an explanation of public-private wage differentials should be as broadly applicable as possible. Our explanation of public-private wage and employment differentials proceeds under these headings: differentials arising in the production of a single product in the absence of unionism; differentials arising from differences in product-mix; and differentials attributable to unionism.
II. Public-Private Wage and Employment Differentials in the Production of a Single Product
A. Utility Maximization by a Public-Sector Employer
It is tempting to regard the demand for labor of a given quality by a governmental agency as simply an application of the theory of household demand for inputs. For example, this is the spirit in which Ehrenberg (1972) proceeds. In effect, he treats a governmental decision-making unit as a household whose resource constraint reflects a process of utility maximization in which it or some higher body funding it decides the fraction of the constituency's resources to be devoted to supporting the governmental activities under discussion. The allocation of resources between the governmental unit and other uses is assumed to be "strongly separable" from the allocation of the governmental unit's resources among alternative inputs. This assumption permits focusing attention entirely on the optimal mix of publicly produced goods and services from a given governmental budget, with the resulting derived demand for a given kind of labor being determined by the maximization of some utility or objective function subject to a budget constraint with given input prices. As Ehrenberg shows, this leads to a set of interrelated derived demand functions for various kinds of labor that is formally analogous to household demand functions for goods.
This procedure precludes consideration of the political factors that bear upon government behavior. It is the premise of this paper that normally political factors are important in determining government wage-employment decisions. In some cases it may be expedient to abstract from them, but this should be done explicitly and only with specific justification. In analyzing behavior in the public sector, political considerations can never be treated as incidental.
What follows is a sketch of a model that purports to explain the politically determined behavior of a governmental decision maker. Consider first its objective or utility function that relates the utility of a government, U, to its expected number of votes, V:
(1) U = f (V, a1, ..., aq).
For simplicity, abstract from lags and consider only static relationships. (It will often be useful to measure V as a percentage of eligible voters.) In general, U'V > 0 for all V, though, as we shall see there may be discontinuities and inflection points.
The empirical characteristics of the transformation function between dollars and votes obviously is a matter of great practical interest and, in my judgment, is eminently researchable. For the present purpose I assume that the government chooses an optimal mix of disposable dollars and votes, and that its actions are selected so as to reach this optimum, given the constraint of its political resource endowment. That is, I assume the government selects a course of action such as to make its marginal rate of substitution between dollars and votes equal to the rate at which they can be exchanged for one another. It is assumed that the rate of exchange between dollars and votes is independent of wage and employment decisions. Everythingsaid of (1) refers to a political party out of office as much as to a government; however, our primary interest is in the behavior of governments, i.e., parties in office.
The parameters a1, ..., aq are "ideological parameters," which are determined exogenously. They reflect the fact that a government gets utility not only by adding to its votes but also by avoiding (engaging) in certain types of actions that it considers ideologically offensive (attractive). Normally there is some tradeoff between attracting votes and pursuing ideological objectives, but this is irrelevant to the present theme.
Depending upon the specific characteristics of the electoral system, at particular values of V, U'V may have a saltus. For example, in a two-party situation, the marginal utility of an additional vote that puts a party over 50 percent will be very much greater than that of a vote that puts it over 40 or 60 percent. Such salta sometimes cause great sacrifices of ideology to get a few extra votes.
In principle, a party will have some discretion in choosing combinations of ideological position and number of votes; it is assumed that there is some finite tradeoff between the two. In practice, this apparent discretion will be greatly limited by the exigencies of political competition, which normally compel a party to approximate vote-maximizing behavior fairly closely if it is to win elections. Parties with strong ideological commitments may survive, but their chance of winning and becoming a government is very low and therefore, for our purposes, may be ignored.
Political parties with a substantial probability of success may have some long-run options as to the combination of victory-probability and ideological position that they choose. This is analogous to the type of option that entrepreneurs may have between maximizing pecuniary income and pursuing other utility-yielding objectives in cases where they are earning rent. However, as in the case of a competitive industry, it is assumed that the range of tradeoff options compatible with continuing political contention is so narrow as to be negligible, with the result that all relevant parties are assumed to maximize expected votes.
Analogous to the situation where competitive entrepreneurs may earn large temporary quasi-rents, transitory disturbances may give a political party a comfortable vote margin for achieving victory at the next election (or put it in a position where victory is unattainable). In either of these cases an entrepreneur may temporarily deemphasize his quest for pecuniary income in favor of other sources of utility, and, analogously, political parties may relent in their efforts to attract marginal votes in favor of a more intensive pursuit of ideological purity. Although recognizing these possibilities, I abstract from them and argue as though governments normally behaved like maximizers of expected votes subject to the constraint of maintaining a certain (constant) degree of ideological purity. (The analogy to the assumption that firms behave as if they were maximizers of money profits is obvious.) The assumption of vote maximization has no greater claim to descriptive accuracy in regard to political behavior than has profit maximization as a description of ordinary business activity; indeed, it may well have less. However, it is an extension of economic theory that should at least be tried.
B. Vote Maximization and Public-Sector Labor Demand
Now let us turn to the vote function:
(2) V = φ (Q(p), p, X1, X2, X3, T, C, S, w1, w2, r),
where the X are inputs, r is the cost of capital, and C are resource contributions. Let us consider the arguments of (2) seriatim; for simplicity, assume the government produces one good in quantity Q which it sells at a price, p. Assume that p and Q are selected so that their values always lie on the demand curve Q = Q(p).
It is assumed that V'p is negative. The rationale is that, ceteris paribus, a voter is more favorably disposed toward a government the lower the price at which it furnishes him a good. Hence, lowering p will tip some marginal voters toward the government. Given the assumption that the government's policy is to adjust output so as to stay on the demand curve, Q is determined once p is set. Since lowering p will increase the consumer surplus of buyers, it will increase the probability that they will vote for the government.
This line of reasoning might be thought to suggest that a rational government should set all prices at zero or even charge negative prices, but this would ignore taxes, T. The higher an individual's tax burden, the less likely he is to vote to retain a government; i.e., V'T< 0. For convenience, assume that T is a vector of tax payments for all n taxpayers, and that the tax burden of individual i, T(i), is a unique and increasing function of the general tax level, T. Let S be a vector of payments to citizens for reasons other than provision of factor services. In general, the effect of S upon V is the mirror image of that of T, hence V'S > 0.
Assume that deficits are not permitted; all government payments must be financed by sales or taxes. (This restrictive assumption prevents consideration of the political economy of inflation.) In effect, this makes government expenditure subject to a receipts constraint:
(3) E [equivalent to] pQ + T [equivalent to] X1 w1 + X2 w2 + X3 r + S,
where E is government spending.
Suppose the government goods were produced at a rate that made marginal cost equal to the price charged. To reduce its price and continue to supply the quantity demanded it would be necessary to increase outlays on factor inputs (X1w1 + X2 w2 + X3r) by more than the increment in receipts from selling the required outputs. Other things equal, this would necessitate an increase in T which would reduce V. Thus the effect of a reduction in p on V includes a favorable effect on voting through buyers of the product and an unfavorable effect through taxpayers. (Later we shall consider the impact of p on V that arises from the effect through employment and wages, but for the moment we shall assume that this is negligible.)
The impact of government decisions concerning price and output upon votes is both direct and indirect, via contributions of resources. Resource contributions, either in cash or services (which have a cash equivalent) are an argument of (2), (V'C > 0) and must satisfy:
(4) C = c1 + ... + cn.
The contribution of individual i is given by:
(4') ci = ci (Ti, Si, p),
where [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] and [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] are negative, and [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] is positive. If the government's price-output behavior injures its purchasers sufficiently, they may respond by reducing their contributions or even making them negative. The indirect effect of a price change on the government's expected vote, via contributions, may be greater than the direct effect; the relative size of direct and indirect effects depends both on the impact of the price change on contributions and on the impact of contributions on votes.
Excerpted from Labor in the Public and Nonprofit Sectors by Daniel S. Hamermesh. Copyright © 1975 Princeton University Press. Excerpted by permission of PRINCETON UNIVERSITY PRESS.
All rights reserved. No part of this excerpt may be reproduced or reprinted without permission in writing from the publisher.
Excerpts are provided by Dial-A-Book Inc. solely for the personal use of visitors to this web site.
Table of Contents
- Frontmatter, pg. i
- CONTENTS, pg. v
- CONTRIBUTORS, pg. vii
- NTRODUCTION, pg. ix
- The Theory of Employment and Wages in the Public Sector, pg. 1
- Comments, pg. 49
- The Demand for Labor in the Public Sector, pg. 55
- Comments, pg. 79
- Demand for Labor in a Nonprofit Market: University Faculty, pg. 85
- Comments, pg. 130
- The Incidence of Strikes in Public Employment, pg. 135
- Comments, pg. 178
- Wage Determination in Public Schools and the Effects of Unionization, pg. 183
- Comments, pg. 220
- The Effect of Government Ownership on Union Wages, pg. 227
- Comments, pg. 256
- COMBINED REFERENCES, pg. 264