Printer Friendly

Production costs, transaction costs, and local government contractor choice.


Declining fiscal conditions, increasing service obligations, and public clamor for more effective government have led local governments to examine alternatives to the bureaucratic supply of publicly provided services. The alternative which has received the most attention is contracting. Under this arrangement, services are financed by the local government, but production of the service is contracted to external suppliers. A growing body of work has explored the contracting decision, (1) focusing on the production choice--does a local government providing (i.e., financing) a service produce it internally or externally? The contracting decision, however, has a second part, the contractor choice--given the decision to externally produce a service, does the local government contract with other governments, with nonprofit organizations, or with for-profit firms? (2)

Contracting is expected to yield production cost savings by exploiting scale economies, overcoming input rigidities, and capitalizing on managerial and competition-induced efficiency incentives. The extent to which these production cost savings are achieved depends on the sector of the external producer. At the same time, public accountability requires contracting governments to be concerned with the proper execution of the contracting process to ensure that shirking is minimized. The transaction costs generated by this oversight also vary by the sector of the contractor. Thus the cost of service delivery depends on the sector of the external producer. Understanding the contractor choice from among the three sectors would add insight into the motivations for contracting, and ultimately enable us to more correctly predict the cost and service-quality effects of different external supply alternatives.

This paper explores the determinants of local government contractor choice. In the next section we examine the production and transaction costs associated with organizations from the public, nonprofit and for-profit sectors. We then develop a general model of the contractor choice and empirically test it using data on local government contract arrangements for three health services: hospital, drug and alcohol prevention and treatment, and mental health.


We posit that a local government in selecting the sector with which to contract seeks to maximize its utility by choosing the alternative that minimizes the costs of service delivery. There are two components of service delivery costs: production costs and transaction costs. (3)

Production Costs

Contracting is viewed as a means to overcome scale diseconomies, perverse managerial incentives, and input rigidities which generate a production cost premium for internal production. External production separates the production scale decision from jurisdictional size thus allowing scale economies to be realized. This benefit is provided by external production with any contractor of the appropriate size, and thus scale economies should not (in theory) affect the sector choice. However, the extent to which the contractor exhibits managerial incentives and input flexibility which contribute to production cost savings depends on its organizational form.

Managerial Incentives. The efficiency incentives of managers vary by the form of their organization. In public bureaus, principal-agent problems arise from the informational advantages of the agents (bureaucrats) over their principals (mayor/council) which are perceived to increase public production costs. For example, Niskanen's [1971] notion of public bureaucrats as budget maximizers is predicated on the bureau's knowledge of the actual cost of delivering services and the use of this knowledge to increase their budgets. In addition, the lack of property rights to residuals (the difference between budgets and actual production costs) reduces incentives for bureaucrats to minimize costs. In fact, there are incentives for them to create slack. This slack provides bureaucrats with either direct utility, or the resources from which to derive utility by creating incentives for subordinates to meet bureaucratic objectives.

Public organizations exposed to competitive forces, however, should respond more efficiently than those that are not. In order to bid for a contract from another government, a public organization must specify outputs and costs. Moreover, it will be more sensitive to minimizing costs since it is no longer operating as a monopolist. (4) Nevertheless, external public organizations lack many of the incentives to minimize costs that are present in their private sector counterparts.

Private sector organizations embody different managerial incentives from those that characterize public bureaus. For-profit organizations minimize the incentive problem by giving the manager a property right to a portion of the profits. The distribution of profits to managers enables owners to reward management for performing in accordance with the owner's desire to maximize profits. This provides strong incentives for production efficiency. (5)

The nonprofit organization may make profits, but as Hansmann [1980] notes, it operates under non-distribution and reasonable compensation constraints. Residuals may not be distributed directly to the board of directors or the managers, although they can be used to expand services, enhance quality, or subsidize unprofitable activities. These constraints make it difficult to devise a scheme to avert agency problems between the board and its managers. Moreover, the lack of a singular objective for nonprofit organizations analogous to profit-maximization in the for-profit sector makes it difficult to generalize about the incentives inherent in nonprofit organizations. However, in terms of the strength of managerial incentives for cost savings, the nonprofit organization would seem to fall between the for-profit and the external public organization. (6)

Input Rigidities. The ability of organizations to achieve cost efficiency depends on managerial discretion in selecting inputs and production technologies. Public organizations are subject to civil service systems and public budgeting systems which inhibit the selection of the optimal mix of inputs. Public personnel policies limit the discretion of managers in selecting the optimal combination of labor inputs for a given output. Constraints on hiring, firing, and promotion as well as compensation levels make it difficult for managers to rearrange their staffs and their skill composition to deliver services at minimum costs. In addition, public managers, faced with separate budgetary processes for capital and operating expenditures, do not always have the flexibility necessary to acquire the desired level of capital. Although the competitive pressures faced by external public organizations enhance incentives for cost minimization, they do not mitigate the rigidities within the public sector which generate suboptimal utilization of inputs.

Private organizations, both for-profit and nonprofit, lack these constraints on input selection. The for-profit sector, however, with its easier access to financial capital may, as Hansmann [1980] notes, have an input control advantage over the nonprofit sector. The inability of nonprofits to sell equity shares limits their ability to finance capital acquisition. In addition, they face more difficulties in borrowing the resources necessary for capital expansion since lending institutions view nonprofit organizations as more risky than for-profit organizations. Thus the for-profit sector has an edge in terms of optimal input mix and expansion.

The magnitude of this advantage is limited to the extent that nonprofits are able to secure funding from other sources such as charitable contributions or grants. The ability of nonprofit organizations to generate resources through the donation of either time or money creates some input flexibility for the organization. In addition, foundation and government grants can provide needed resources for input acquisition by nonprofits. These sources are typically not available to for-profit organizations. (7)

In summary, the production cost savings to be expected from external production will vary by the organizational form of the contractor. Based on differences in managerial incentives and input rigidities, the for-profit firm should offer the greatest production cost savings, the nonprofit organization somewhat less, and the external public bureau the least.

Transaction Costs

Transaction costs are incurred in an effort to minimize agency problems encountered in contracting. The dilemma of public service contracting is the extent to which the principal (the contracting government) can ensure that the agent (the contractor) will behave so as to meet the principal's objectives in the presence of information asymmetries. The contracting government is not likely to have complete information on the capacity of the different bidders to perform to contract specifications, creating an adverse selection problem. To increase the likelihood of selecting the best contractor, the contracting government will incur costs to gather information. Information asymmetries also create problems at the monitoring and enforcement stage. In cases where it is technically impossible or very costly to monitor performance, there are potential moral hazard problems, i.e., the contractor may be inclined to shirk on performance.

The contracting government seeks to minimize these problems through contract design and administration, thus incurring transaction costs during contract writing and monitoring. (8) The magnitude of these costs depends on the type of service and on the contractor's organizational form.

Service factors. Contract specification requires that one can define and measure the quantity and quality of the service, and can describe the conditions under which the service will be delivered. Such tasks are facilitated when the service is characterized by relatively constant citizen preferences and relatively stable cost conditions (technology and input prices).

Once a contract is written, it must be monitored. The feasibility of measuring performance is critical. It must be technically possible to measure outputs, both quantitatively and qualitatively, and at a reasonable cost. This condition is more likely to be met when services have tangible outputs, e.g., garbage collection or road repair, as opposed to intangible outputs, e.g., mental health or child care. Although it is possible with intangible services to develop quantity measures, such as the number of individuals served, the issue of service quality is more vexing.

Organizational factors. Transaction costs also vary by organizational form. Nonprofit organizations may reduce transaction costs at the contract writing and monitoring stages. One of the alleged virtues of the nonprofit sector suggested by Salamon [1987] and Weisbrod [1977] is its responsiveness to demands for collective goods, particularly in the case of heterogeneous community preferences. If nonprofits have effectively responded to the community, then the contract writing costs for government are likely to be lower if they contract with that sector, either because the nonprofit organization's record of service provision is satisfactory or because it helps to write the contract through cooperative negotiations, as in DeHoog [1984].

Nonprofit organizations may yield even greater savings in monitoring costs. There is widespread concern that for-profit firms will cut corners on quality to increase profits. Nonprofit organizations, due to their inability to distribute residuals, are often assumed to be more trustworthy. This belief is rooted in the fact that many nonprofit organizations were created in response to a keen interest in the quantity and/or quality of a service. In addition, nonprofit organizations often receive considerable donations creating the possibility that individuals making such donations may be in a position to monitor the organization, especially when donations are in the form of volunteering. Consequently, as Krashinsky [1986] suggests, nonprofits may be preferred to for-profit firms when monitoring is difficult. In effect, choice of the nonprofit organizational form is perceived as a method of reducing transaction costs that arise from difficulties in monitoring performance.

Similar reasoning applies to external public organizations. Monitoring costs are lower in these organizations because the incentives for opportunistic behavior are significantly reduced by the pressures of accountability and the lack of property rights in the public sector.


Finally, the full array of organizational choices may not be available to the contracting government. Some services may only have suppliers in one or two sectors. For example, there are few nonprofit suppliers of public works. In addition, contracting with suppliers in a particular sector (usually the for-profit) may be prohibited by law. Thus, the contracting government may be effectively constrained in its contractor choice to particular sectors.


A contracting government may choose from among the public, nonprofit or for-profit sectors for the external production of publicly provided services. The relative advantages of the sectors with respect to both production and transaction costs are critical to the decision. Managerial incentives and input flexibility differentiate the sectors with respect to production efficiencies. Managerial incentives in the for-profit sector are most conducive to minimizing production costs, while those in the public sector are the least conducive. Similarly, for-profit managers have the most flexibility in selecting their input mix, while public bureaucrats have the least. Transaction costs are affected by the feasibility and cost of writing and monitoring contracts. These costs depend on the nature of the service and on the organizational form of the contractor. When contract writing and monitoring are difficult, public or nonprofit contractors are likely to minimize transaction costs.


Decision Rule

The previous section conceptualizes the tradeoff between production and transaction costs in contractor choice. The external production option that minimizes production costs also maximizes transaction costs. Therefore, the contractor choice depends on how the local government weighs these two components of service delivery costs.

Production and transaction costs are fundamentally different in their effects. Production costs are visible with a direct impact on government expenditures, and thus are discussed in the context of budgetary decisions. Transaction costs, even though they have budgetary impacts, are more difficult to discern, their impacts more subtle. Therefore, in jurisdictions where fiscal concerns are important, production costs may be weighed more heavily than transaction costs. The general public should prefer the least-cost contractor for a given service output level and quality. Those who do not directly benefit from the service may prefer to minimize the direct production costs since the burden of higher costs falls on them in higher taxes or reduced levels of other services. Service constituents, however, are likely to have different preferences. They are very concerned with service quality and with the government's commitment to provision, and thus are likely to weigh transaction costs more heavily than the general public.

Thus, we posit the following decision rule. The government decision maker is assumed to maximize utility by choosing the sector that minimizes the weighted sum of production and transaction costs, where the weights represent the relative importance to the local government of minimizing the two components of service delivery cost and are functions of political and fiscal forces at work in the jurisdiction.

More precisely, let [y.sup.*.sub.i] denote the output level which local government i has decided to produce externally, and let [y.sub.ij] denote the amount of [y.sup.*] produced by sector j, where j equals 1 for bureaus of other governments, 2 for nonprofit organizations, and 3 for for-profit firms. Let PC([y.sub.ij]) and TC([y.sub.ij]) be the production costs and transaction costs associated with the production of [y.sub.ij], and let [gamma] and [delta] be the respective weights. Then the objective of local government i is to select sector j such that

min [[gamma].sub.i] [3.summation over (j=1)] PC([y.sub.ij]) + [[delta].sub.i] [3.summation over (j=1)] TC([y.sub.ij])

subject to [3.summation over (j=1)] [y.sub.ij] = [y.sup.*.sub.i].

The predicted ordering of production costs by sector for some fixed output y (with the government subscript i suppressed) is PC([y.sub.1]) > PC([y.sub.2]) > PC([y.sub.3]). The difference in production costs between the public and private sectors [PC([y.sub.1]) - PC([y.sub.2])] derives from the greater input rigidities and weaker efficiency incentives in the public sector, and the difference in production costs between for-profit and nonprofit organizations [PC([y.sub.2]) - PC([y.sub.3])] derives from greater efficiency incentives in the for-profit sector.

The predicted ordering of transaction costs by sector is TC([y.sub.3]) > TC([y.sub.2]) [congruent to] TC([y.sub.1]). The difference in transaction costs between the for-profit and nonprofit sectors [TC([y.sub.3]) - TC([y.sub.2])] derives from greater monitoring costs in the for-profit sector.

The importance of minimizing production costs, [gamma], should be greater in communities with constraints on spending. The importance of minimizing transaction costs, [delta], should be greater in communities where the service constituency is large and politically powerful.

Empirical Specification

Consider now the empirical specification of our model. The production and transaction cost characteristics of the sectors define the utility, [U.sub.ij], which each local government has for the contractors in each sector. This utility can be divided into a deterministic and random component:

[U.sub.ij] = [V.sub.ij] + [[epsilon].sub.ij].

Since the local government is assumed to select the sector that maximizes its utility (by minimizing the weighted sum of production and transaction costs), the probability that local government i chooses sector j is given by

(1) [P.sub.ij] = P([V.sub.ij] + [[epsilon].sub.ij] [greater than or equal to] [V.sub.ik] + [[epsilon].sub.ik]) [for all] j = 1,2,3, j [not equal to] k.

The deterministic portion of utility (V) is assumed to be linear in its parameters and includes both sector and jurisdictional characteristics, i.e.:

(2) [V.sub.ij] = [beta]'[S.sub.ij] = [[alpha].sub.j]'[J.sub.i]

where [S.sub.ij] are the characteristics of sector j as perceived by the decision makers in local government i, and [J.sub.i] is the vector of characteristics of the ith local jurisdiction.

If we assume the disturbances are independently and identically distributed with an extreme value distribution, then equations (1) and (2) yield the following mixed multinomial logit model: (9)

(3) [P.sub.ij] = exp([beta]'[S.sub.ij] + [[alpha].sub.j]'[J.sub.i])

/[3.summation over (k=1)] exp([beta]'[S.sub.ik] + ([[alpha].sub.k]'[J.sub.i])

Equation (3) is estimated using data on three local health services: hospital services, drug and alcohol prevention and treatment programs, and mental health programs.

We focus on health services for several reasons. First, the health care industry is one in which viable producer choices exist in all three sectors, particularly with the expansion of for-profit health care providers in recent years. Second, Ferris and Graddy [1986] found that local health services are frequently contracted, with contracts distributed across all three sectors. Finally, the delivery of health services is an important part of the service delivery obligations of local governments. Thus, these services provide an appropriate basis for an analysis of the determinants of local government contractor choice.


Data collected by the International City Management Association in 1982 provide the incidence of city contracting by sector. (10) In order to estimate the contractor choice model, the survey data were merged with information on community fiscal and demographic characteristics obtained from secondary sources. The primary source of demographic data is the 1983 County and City Data Book which only contains information on cities with a population over 25,000. Consequently, jurisdictions with populations less than 25,000 were truncated from our sample, and one is cautioned against drawing inferences about small cities from this analysis. (11)

After observations with missing data are deleted, the working sample of jurisdictions is 583. Of these cities, 107 provide hospital services, 85 percent of which contract; 179 provide drug/alcohol treatment programs, 85 percent of which contract; and 123 provide mental health services, 86 percent of which contract. (12) Observations for each health service contracted were pooled to create the data set. The pooled data set contains 309 observations. Of these, 32 percent represent contracts with other governments; 48 percent represent contracts with nonprofit organizations; and 20 percent represent contracts with for-profit firms. (13)

Measures of the determinants of contractor choice are discussed below and summarized in Table I which includes the variable descriptions, sources and descriptive statistics.
TABLE I Variable Definitions and Descriptive Statistics

 Description and
Variables Source (a) Mean (SD)

Sector-Specific Variables

Production Ordinal variable
Efficiency capturing sector
 differences in
 efficiency (higher
 values indicate more
 efficiency); value
 of 3 if for-profit,
 2 if nonprofit,
 1 if public.

Labor Costs Sector-specific
 (in $1,000) (1)
 Mean public salaries 12.12 (2.43)
 Mean private 13.50 (2.92)

Monitoring Costs Ordinal variable
 capturing service
 and sector
 differences in
 monitoring costs
 (higher values
 indicate greater
 costs); value of 2
 for intangible
 service with for-
 profit, 1 for
 intangible service
 with nonprofit or
 public, 0 for
 tangible service.

Hospital Beds Number of beds in
 hospitals providing
 service, per 1000
 population (2)
 Public 3.43 (13.37)
 Nonprofit 29.00 (39.97)
 For-profit 3.53 (11.19)

Jurisdiction-Specific Variables

Manager-Council Form Dummy variable with .660 (.474)
 value of 1 if
 government has
 form; 0 otherwise (4)

Tax Limitation Dummy variable with .557 (.498)
 value of 1 if
 property tax limit
 0 otherwise (3)

Tax Burden Local taxes per 233.72 (204.66)
 capita (1) 5.43 (2.17)

Community Wealth Median value of
 (in $10,000) (1)

Nonwhite Nonwhite proportion .159 (.144)
 of the
 population (1)

Poor Percent of 11.57 (5.35)
 population with
 income below federal
 threshold (1)

SMSA Dummy variable with .913 (.283)
 value of 1 if
 government located
 within an SMSA;
 0 otherwise (1)

* The number at the end of each variable description refers to the
data source:

(1.) U.S. Bureau of the Census [19841

(2.) American Hospital Association [1983]

(3.) Advisory Commission on Intergovernmental Relations [1984]

(4.) International City Management Association [1982b]

Production cost variables. Production cost differences across sectors are specified with two variables. First, an ordinal variable captures the expected efficiency ordering for the sectors based on their differences in managerial incentives and input flexibility. The variable assumes a value of 1 for public, 2 for nonprofit, and 3 for for-profit organizations, constant across jurisdictions. Its coefficient is expected to be positive; decision makers prefer contractors with incentives to minimize production costs.

Second, sector-specific labor costs are included to capture an important component of production costs. The mean public salaries in the jurisdiction are included as the measure of labor costs for public organizations, and the mean private service-sector salaries are used for both nonprofit and for-profit organizations. We expect a negative coefficient; cities prefer contractors with lower labor costs.

A local government's interest in production cost savings [gamma] is captured with a government form measure and with three measures of its fiscal condition. Jurisdictions with a government form or fiscal characteristics suggesting that they place a high value on production cost savings are expected to prefer private over public contractors and for-profit over nonprofit contractors.

The government form measure reflects whether the local government is administered by a professional manager. The dummy variable has a value of 1 if the city has a manager-council form of government, and 0 otherwise. Manager-council governments are assumed, as in Lineberry and Fowler [1967], to be more sensitive to reducing costs because of their professional, nonpartisan orientation.

The fiscal characteristics of the jurisdiction are captured with three variables: tax limitation, tax burden and community wealth. Legal limits on taxation are specified with a dummy variable which has a value of 1 if a jurisdiction has a property tax limit, and 0 otherwise. The existence of such limits suggests that production cost savings are important to the community. The tax burden is measured by locally raised tax revenues per capita. This variable captures the community's ability and willingness to raise revenues on its own. Its effect is difficult to predict. High per capita taxes could suggest the likelihood of constituency resistance to further government spending, making local officials aware of the need for reduced costs. Alternatively, low per capita taxes could indicate a poor tax base and thus a relative inability to raise revenues, which is also likely to increase cost-reducing pressures. To control for this latter effect, we include a separate measure of community tax base, the median value of owner-occupied housing in the jurisdiction. This measure of community wealth is a reasonable indicator of the tax base given the pivotal role, albeit diminished in recent years, of property tax in local finance. Therefore, communities with high tax burdens or low wealth are expected to value reducing production costs.

Transaction cost variables. Transaction costs are generated at the writing and the monitoring stages of the contract process. Contract writing is less costly when there is community agreement on service specifications. Homogeneous communities are likely to have stable service preferences and thus lower writing costs. We include a racial composition measure, the proportion of nonwhites in the jurisdiction, as a measure of jurisdictional homogeneity with respect to race. The quadratic form of this variable is used since both high and low proportions of nonwhites indicate homogeneity.

Monitoring costs depend on both the service characteristics and the organizational form of the contractor. Contracts for services with intangible outputs are difficult to monitor. Two of the services in our sample, mental health and drug/alcohol prevention and treatment services, have intangible outputs. In both cases, the service output is difficult to measure and assess. The outputs of hospital services are by comparison easy to define and monitor. Therefore, monitoring costs should be an important consideration in contracting for mental health and drug/alcohol services, and relatively unimportant in contracting for hospital services. For intangible services, the organizational form advantages of the nonprofit and public sectors become important. To capture the interactive influence of service and organizational form on monitoring costs, we create an ordinal variable which assumes a value of 2 if an intangible service (mental health or drug/alcohol) is contracted with a for-profit organization, a value of 1 if an intangible service is contracted with either a nonprofit or public organization, and a value of 0 for tangible services (hospital). Higher values of the variable thus represent higher monitoring costs. We expect a negative coefficient; the contracting government will prefer, when monitoring is difficult, contractors with low monitoring costs.

A jurisdiction's preferences about service quality and provision ([delta]) determine its willingness to incur transaction costs. The service constituency has the greatest interest in high quality output; thus the size of this group is a measure of these preferences. The primary constituency of local health services are the poor; thus we measure constituency strength with the percentage of the jurisdiction's population with incomes below the poverty level. Jurisdictions with large service constituencies should place a high value on reducing transaction costs, and thus should prefer public and nonprofit contractors over for-profit ones.

Availability variables. Finally, we consider the jurisdiction's choice set. Some jurisdictions do not have external suppliers in all sectors for all services. If we had precise availability measures, we could define the choice set for each jurisdiction. Unfortunately, data do not exist on all the external suppliers of these services at the local level, so we use available information to incorporate two measures of supplier availability in our model. The American Hospital Association collects data by sector on the hospitals which provide mental health, drug/alcohol, and general hospital services. Therefore, we include the number of beds in hospitals providing the service in each jurisdiction by sector. Available hospital beds do not capture all suppliers of drug/alcohol and mental health services, but the variable can be viewed as a proxy for the distribution of external suppliers by sector. There is no apparent reason to assume that nonhospital suppliers of these health services are distributed differently across sectors.

The second availability variable is a dummy variable with a value of 1 if the jurisdiction is located within an SMSA and 0 otherwise. This variable captures the option of jurisdictions located within SMSAs to cross jurisdictional boundaries to obtain external producers. This control is needed since the hospital beds measure is jurisdictionally defined.

Mixed Multinomial Logit Estimation

It is useful for an understanding of the results to separate the explanatory variables into those that are part of the S vector, which vary across sectors, and those that are part of the J vector, which vary only across jurisdictions. The S vector includes production efficiency, labor costs, monitoring costs, and hospital beds; the J vector includes government form, tax limitations, tax burden, community wealth, nonwhites, poor, and SMSA. Estimation of our mixed multinomial logit model yields a single coefficient for each sector characteristic (S vector) and a set of coefficients associated with paired alternatives for each jurisdictional characteristic (J vector). The coefficients of the S variables indicate the importance of the sector characteristics to the decision maker in making the contractor choice. The coefficients of the J variables are the usual multinomial logit parameters and give the effect of the jurisdictional characteristics on the probability of choosing one sector relative to another.


The results of the estimation of equation (3) are presented in Table II. The first column contains the coefficients associated with the sector characteristics. The coefficients presented in the second, third and fourth columns are estimates of the importance of the jurisdictional variables in the choice of a nonprofit over an external public contractor, a for-profit over an external public contractor, and a for-profit over a nonprofit contractor respectively. The results reveal the importance of production costs, transaction costs, and the available choice set in the contractor choice of local governments.

Mixed Multinomial Logit Parameter Estimates

Variables Nonprofit/
Production Efficiency 1.86 ***
Labor Cost -.075 *
Monitoring Costs -2.45 ***
Hospital Beds .0023
Manager-Council Form .852 **
Tax Limitation -1.08 ***
Tax Burden .00099
Community Wealth -.084
Nonwhite -.356
Nonwhite-Squared -.048
Poor .052
SMSA -1.51 **
n 309

[X.sup.2] 134.52 ***

Variables For-profit/ For-profit/
 Public Nonprofit
Production Efficiency

Labor Cost

Monitoring Costs

Hospital Beds

Manager-Council Form 1.41 *** .554
 (.453) -0.431
Tax Limitation -.812 * .266
 (.406) (.360)
Tax Burden .00044 -.00055
 (.0012) (.0011)
Community Wealth -.110 -.025
 (.123) (.093)
Nonwhite -1.26 -.904
 (4.80) (4.45)
Nonwhite-Squared -.0099 .038
 (.086) (.080)
Poor -.023 -.075
 (.062) (.049)
SMSA -2.16 *** -.649
 (.728) (.564)


Note: Standard errors are in parentheses.
Significance levels in two-tailed tests are indicated by
asterisks as follows: *** denotes .005 level; ** denotes .01 level;
* denotes .05 level.

The two sector-specific production cost variables have statistically significant coefficients with the expected sign. As predicted, sector differences in production efficiency affect the contractor choice. Local government decision makers prefer organizational forms that are conducive to minimizing production costs. This observed preference suggests that government decision makers perceive the production cost ordering that we hypothesized. (14) Labor cost differences also affect the contractor choice, with decision makers preferring contractors with lower labor costs. Both results are consistent with our theoretical model.

Two of the variables which measure the importance of production cost savings to the jurisdiction also have statistically significant coefficients. As predicted, local governments with a manager-council form of government prefer private sector contractors, both for-profit and nonprofit, over public contractors. This result is consistent with the argument advanced by the local government "reform" movement of the 1920s, that professional administration of cities would yield more efficient service delivery.

The effect of a tax limitation, however, is not as predicted. The results indicate that jurisdictions with legal limits on taxing prefer other governments as contractors over the nonprofit and for-profit sectors. A preference for private contractors was hypothesized. This result may reflect the existence, in fiscally conservative areas such as California, of contract cities--cities created to provide services, but which rely exclusively on their counties for service production.

Transaction costs also affect the contractor choice. When contracting intangible services, cities prefer nonprofit or public organizations over for-profit ones. This result supports our contention that monitoring costs are important to the contractor choice. (15)

The results also indicate the importance of producer availability in the contractor choice. Local governments located within SMSAs prefer to contract with other governments over both for-profit and nonprofit organizations. The unimportance of available hospital beds (the sector-specific availability measure) is disappointing, nevertheless the importance of the jurisdictional variable is suggestive of the role of producer availability in this decision. Cities in SMSAs have other governments, both other cities and counties, from which to choose, and they prefer these contractors over private ones.

In summary, the empirical analysis of the sector variables indicates that both production and transaction costs affect the contractor choice of cities that choose to externally produce health services. All three measures of these costs have statistically significant coefficients. The results for the jurisdictional variables, although somewhat weaker, also lend support to our theory, indicating that the value jurisdictions place on production cost savings as well as supplier availability affect the contractor choice.

The distinctions local governments make between public and private contractors and between for-profit and nonprofit ones can also be addressed. The significance of the two sector-specific production cost variables suggests that local governments perceive a cost difference between the public and private sectors. The jurisdictional variables indicate that cities with a manager-council form of government prefer private-sector contractors, while those with tax limitations prefer public-sector contractors.

The significance of the monitoring cost and production efficiency measures indicates that local governments differentiate the two private sectors. Both the expected advantage of the for-profit sector in production costs and of the nonprofit sector in transaction costs are valued by local government decision makers. The jurisdictional variables, however, do not discern the choice between for-profit and nonprofit contractors, suggesting either that the characteristics which define the relative importance of production and transaction costs to the local government do not generate a preference between the private sectors, or that we have not adequately captured community differences in service output preferences. Better community-level measures of transaction-cost preferences are needed to yield a definitive interpretation.


This initial effort at understanding the contractor choice by incorporating the organizational economics literature with the public service delivery literature has been fruitful. The recognition that service delivery costs entail both production and transaction costs is essential to understanding the balancing act that local governments face between cost savings and accountability, and explains the importance of organizational form in the local government's contractor choice.

Both production and transaction costs were found to affect contractor choice. The importance of the production-cost variables reinforces the notion that contracting is a method for reducing the costs of producing publicly provided services. However, the importance of the transaction-cost variable suggests that contracting cities are concerned with the costs of monitoring contracts and may prefer public or nonprofit contractors, particularly for hard-to-monitor services.

The implications for the current contracting debate are twofold. First, the cost advantage claimed for for-profit contractors may be overstated. Contracts with other governments or nonprofits may yield lower service delivery costs by minimizing transaction costs. Second, since limited availability of contractors in certain sectors may adversely affect service delivery costs, governments should consider contractor availability in their production choice. Governments interested in external production may want to go further and encourage the development of suppliers in particular sectors. (16)


Advisory Commission on Intergovernmental Relations. Significant Features of Fiscal Federalism, 1982-83. Washington, 1984.

American Hospital Association. American Hospital Association Guide to the Health Care Field. Chicago, 1983.

Bays, Carson W. "Cost Comparisons of For-profit and Nonprofit Hospitals." Social Science and Medicine 13C, 1979, 219-25.

Becker, Edmund R. and Frank A. Sloan. "Hospital Ownership and Performance." Economic Inquiry, January 1985, 21-36.

Borcherding, Thomas E. "Some Revisionist Thoughts on the Theory of Public Bureaucracy." European Journal of Political Economy 4, 1988, 47-64.

Borcherding, Thomas E., Werner W. Pommerehne, and Friedrich Schneider. "Comparing the Efficiency of Private and Public Production: The Evidence from Five Countries." Journal of Economics, Supplement II, 1982, 127-56.

DeHoog, Ruth Hoogland. Contracting Out for Human Services: Economic, Political and Organization Perspectives. Albany: State University of New York Press, 1984.

Ferris, James. "The Decision to Contract Out: An Empirical Analysis." Urban Affairs Quarterly, December 1986, 289-311.

Ferris, James and Elizabeth Graddy. "Contracting Out: For What? With Whom?" Public Administration Review, July/August 1986, 332-44.

--. "Production Choices for Local Government Services." Journal of Urban Affairs 10(3), 1988, 273-89.

Hansmann, Henry B. "The Role of Nonprofit Enterprise." Yale Law Journal, April 1980, 835-901.

--. "The Effect of Tax Exemption and Other Factors on the Market Share of Nonprofit Versus For-profit Firms." National Tax Journal, March 1987, 71-82.

International City Management Association. Alternative Approaches for Delivering Public Services. Washington, October 1982a.

--. Profile of Alternative Service Delivery Approaches-1982. Washington, 1982b.

Krashinsky, Michael. "Transaction Costs and a Theory of the Nonprofit Organization," in The Economics of Nonprofit Institutions, edited by Susan Rose-Ackerman. New York: Oxford University Press, 1986, 114-32.

Lineberry, Robert and Edward Fowler. "Reformism and Public Polices in American Cities." American Political Science Review, September 1967, 701-16.

McFadden, Daniel. "Conditional Logit Analysis of Qualitative Choice Behavior," in Frontiers in Econometrics, edited by Paul Zarembka. New York: Academic Press, 1974, 105-42.

Mehay, Stephen L. and Rodolfo A. Gonzalez. "Economic Incentives under Contract Supply of Local Government Services." Public Choice 46(1), 1985, 79-86.

Niskanen, William A., Jr. Bureaucracy and Representative Government. Chicago: Aldine-Atherton, 1971.

Robinson, James C. and Harold S. Luft. "The Impact of Hospital Market Structure on Patient Volume, Average Length of Stay, and the Cost of Care." Journal of Health Economics, December 1985, 333-56.

Salamon, Lester. "Partners in Public Service: The Scope and Theory of Government Nonprofit Relations," in The Nonprofit Sector: A Research Handbook, edited by Walter W. Powell. New Haven: Yale University Press, 1987, 99-117.

Schulz, Rockwell I., James R. Greenley, and Robert W. Peterson. "Differences in the Direct Costs of Public and Private Acute Inpatient Psychiatric Services." Inquiry, Winter 1984, 380-93.

Sonenblum, Sidney, John J. Kirlin, and John C. Ries. How Cities Provide Services: An Evaluation of Alternative Delivery Structures. Cambridge: Ballinger, 1977.

Stein, Robert. Urban Alternatives: Public and Private Markets in the Provision of Local Services. Pittsburgh: University of Pittsburgh Press, 1990.

U.S. Bureau of the Census. County and City Data Book, 1983. Washington: Department of Commerce, 1984.

Watt, J. Michael, Steveo C. Renn, James S. Hahn, Robert A. Derzon, and Carl J. Schramm. "The Effects of Ownership and Multihospital System Membership on Hospital Functional Strategies and Economic Performance," in For-Profit Enterprise in Health Care, edited by Bradford H. Gray. Washington: National Academy Press, 1986, 260-84.

Weisbrod, Burton A. The Voluntary Non-Profit Sector. Lexington, MA: D.C. Health, 1977.

Williamson, Oliver E. The Economic Institutions of Capitalism: Firms, Markets, Relational Contracting. New York: The Free Press, 1985.

Wilson, George W. and Joseph M. Jadlow. "Competition, Profit Incentives, and Technical Efficiency in the Provision of Nuclear Medicine Services." Bell Journal of Economics, Autumn 1982, 472-82.

(1.) For example, see Sonenblum, Kirlin and Ries [1977], Ferris [1986], Ferris and Graddy [1988], and Stein [1990].

(2.) The for-profit and nonprofit organizations considered here are privately controlled. Publicly controlled corporations or nonprofit organizations represent very different supply arrangements that typically involve off-budget activities and rely to some extent on private financing.

(3.) The literature on cost comparisons across sectors has tended to focus on production costs, with the strong presumption that private production is less costly than public production. This presumption is buttressed by a number of service-specific cost comparisons across sectors surveyed by Borcherding, Pommerehne and Schneider [1982]. These studies, for the most part, fail to compare costs in actual contracting environments and do not assess the transaction costs involved in external production. Borcherding [1988] suggests that incorporating transaction costs may reduce the cost savings often attributed to contracting and may even outweigh the production cost savings.

(4.) Mehay and Gonzalez [1985] found that public organizations that supply services to other governments have lower costs than those that do not.

(5.) Such a solution is typically not available in the public bureau for two reasons: (1) profits are not easily measured due to the absence of explicit prices and outputs; and (2) designing civil service systems to reward individuals based on performance is difficult. These two factors are reinforcing; it is difficult to devise a performance-based salary scheme when outputs and costs are difficult to measure.

(6.) The empirical literature is inconclusive. For example, the relationship between organizational form and cost has been exhaustively studied for hospitals, with mixed results. Bays [1979], Wilson and Jadlow [1982], and Robinson and Luft [1985] found that for-profit hospitals had lower costs than nonprofits, and public hospitals had higher costs than nonprofits. In contrast, Becker and Sloan [1985] and Watt et al. [1986] found no relationship between ownership and cost. For other health services, there is far less information. In a rare study of organizational form and cost in mental health services, Schulz, Greenley and Peterson [1984] found that direct costs were significantly lower in private community general mental health units compared to county-operated mental health units.

(7.) Differential tax treatment also has consequences for production costs. For-profit firms are generally subject to taxes, while public and private nonprofit organizations (to the extent the activities are related to the organization's primary mission) are tax exempt. This gives nonprofit contractors an advantage over for-profit ones, as evidenced by the positive and significant relationship between the nonprofit market share and preferential tax treatment found by Hansmann [1987].

(8.) Transaction costs also include the reduced competition that results over time from asset specificity. See Williamson [1985] for a comprehensive exposition of transaction cost economics.

(9.) See McFadden [1974] for the derivation.

(10.) Between March and June 1982, the chief administrative officers of 3130 cities were surveyed about the services they provide and how they are delivered to citizens. Forty-nine percent of the cities responded, yielding information on the service delivery arrangements for approximately sixty public services. For additional details on the survey, see International City Management Association [1982a].

(11.) Cities with populations below 25,000 have a slightly higher rate of contracting for local health services than those with populations above 25,000.

(12.) Contracting may involve no public inputs (exclusive external production) or may include public employees used in conjunction with contracts (mixed production).

(13.) The dependent variable is constructed as the most private form of contractor. For example, if both nonprofit and for-profit organizations receive contracts for a service, the contractor is recorded as for-profit. This specification is necessary because the estimation technique does not allow the selection of multiple alternatives by a given decision maker.

(14.) Our results do not address whether this perceived ordering is in fact correct. Private health providers are widely perceived as being more efficient than public providers, and for-profit as more efficient than nonprofit, despite mixed empirical evidence.

(15.) The results are robust with respect to a specification of monitoring costs that treats drug/alcohol services as having tangible outputs. The magnitude of the coefficient is reduced as is the overall fit of the model. If separate mental health and drug/alcohol monitoring variables are specified, their coefficients are very similar in magnitude, and the model fit is comparable to the reported specification.

(16.) This is consistent with Salamon's [1987] view of government-nonprofit relations at the federal level and with DeHoog's [1984] at the local level.


* Professor and Associate Professor, University of Southern California. We would like to thank Thomas Borcherding, H. E. Frech III, Steven Rathgeb Smith, Richard Steinberg, Fred Thompson, and an anonymous referee for helpful comments on earlier versions of this paper.
COPYRIGHT 1991 Western Economic Association International
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1991 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Ferris, James M.; Graddy, Elizabeth
Publication:Economic Inquiry
Date:Jul 1, 1991
Previous Article:Efficiency wages and equilibrium wages.
Next Article:Customer discrimination and affirmative action.

Terms of use | Copyright © 2016 Farlex, Inc. | Feedback | For webmasters