The Productivity of Schools and Other Local Public Goods Producers

I. Introduction

Vouchers, privatization, and decentralization are all being used to improve local public goods such as education and crime prevention. Nearly all studies focus on how these systems affect productivity. Interest in this area stems from the observation that budgets for local public goods are limited in developing countries and that cost increases are not complemented by improved services in industrialized countries.

It is often assumed that local public good spending is equally productive, regardless of how it is financed. Economic theory on local public goods has long concentrated on how much should each person pay for and receive. As such, a theory which balances productivity and allocative issues should be developed.

The author built a model which illustrates how imperfect information allows local public good providers to take rent. She also reveals that information is usually organized in such a way that residents and local public goods providers encounter agency problems. She arrived at a model containing the following characteristics:

• Cost conditions vary across jurisdictions but are not fully observed
• Some cost conditions are determined by the characteristics of local residents
• Moving costs are usually high enough to discourage households from moving frequently across jurisdictions
• Some local public good qualities are observable even though they are unverifiable

Social planners who wish to manage productivity would have to deal with costly information gathering. Productivity depends on the information that is collected. However, high productivity can also be attained using conventional property tax finance as long as households have enough jurisdictions to choose from. A Tiebout residential market modifies the information structure by making demand information verifiable and by abolishing the need for information about cost conditions. Local property tax finance manages productivity in a decentralized way and does not depend on the political process of determining property tax rates. Moreover, if someone were to design a process to manage productivity, local property tax finance will meet the desired conditions.

II. Productivity in the local public good literature

Tiebout’s (1956) paper described local public good providers as cost-minimizing entrepreneurs. The Tiebout process and its interaction with politics were assumed to manage the allocation issues in local public goods. However, there is no similar assumption for productivity. Also, the entrepreneurial model is quite unrealistic for local public goods like public education because quality is non-verifiable for such goods. Then again, the Tiebout process still offers useful implications for producers and consumers.

Much of the theoretical foundations of this paper are provided by the works of Laffont and Tirole (1993) on optimal incentive mechanisms under imperfect information as well as Epple and Zelenitz’s (1981) study on the crucial role of politics even in the presence of a well-functioning Tiebout process. The latter also considers incumbents to be vulnerable to rent extraction.

III. The foundations of the model: household, jurisdiction, agents and costs

A.  Households

Some households move so that their moving costs are lower relative to their choice of jurisdiction. For all others, their cost of choosing a jurisdiction is equal to their moving cost. Households are also assumed to maximize their expected utility; their utility depends on local public goods quality, services generated by property and other consumption.

B. Jurisdictions (Public School Districts)

Jurisdictions have fixed boundaries and there are different jurisdictions for each type of household. The supply of potential jurisdictions is observed to be elastic. They are classified as either commercial or agricultural.

C. Agents

There are many agents who can offer to provide local public goods to a jurisdiction. Such agents are assumed to be rent-maximizing. They can take rent through reduced effort, improved job benefits, excess salary, and gifts. Because of outside opportunities, an agent can quit if a jurisdiction offers him negative or zero rent. This rent is quantified by subtracting an agent’s effort and cost in producing the public good from the payment he receives for them.

D. Cost

 Cost is a function of local public goods quality, the agent’s effort, the type of household being served, and cost conditions specific to a jurisdiction. Ceteris paribus, per-pupil costs increase as quality increases, while it decreases as effort decreases. Cost also varies according to the type of household because it is usually more costly to provide education for households which cannot provide the complementary inputs for schooling. It is not, however, a function of jurisdiction size because it is assumed that there is no economies of scale.

IV. The information environment

Information is verifiable if it can be observed with certainty by planners or if the public can perceive it identically. For households, it is assumed that they are able to identify their own type. In addition, they can detect if households of the same type dwell in a particular jurisdiction. Social planners however, have no capacity to determine such information.

Households are capable of comparing the quality of jurisdictions because they can observe and evaluate outcomes from local public goods provision. This makes local public goods quality observable to households within jurisdictions that serve their type. Quality, however, is not verifiable because many outcomes are subjective and may be affected by household types. In contrast, budgets allotted to agents are both observable and verifiable. Households and jurisdictions also have no way to assess the quality of their efforts.

Initially, an agent is unaware of the cost conditions of a certain jurisdiction. But once he is commissioned to provide local public goods, he learns its cost conditions so he could take them into account in choosing quality.

Centralized rules for financing usually start from cost or outcomes. Cost-based rules are poor at managing productivity because they do not give agents the incentive to minimize cost for a given quality. Outcome-based rules, on the other hand, are enacted occasionally and are usually short-lived. For schools, outcome-based rules encourage agents to focus on outcomes that are tested like student scores on state-wide tests; while other outcomes are neglected. These rules can be improved by investing in more information on objective outcomes, households, and the relationship between them.

V. Local property tax finance

Conventional local property tax finance give incentives to produce services of high quality. For non-moving households, property tax rate is chosen by means of a local political process. The outcomes of such conventional political processes are determined by the preference of incumbents. They choose the appropriate tax rate if they thought they were in a jurisdiction with the worst cost conditions. Such choices must always be above zero-rent rates of agents to prevent agents from quitting from producing the needed services.

Each agent receives his revenue from local property tax. Each set their marginal cost of producing higher quality equal to the marginal revenue they earn by producing at such value. They also set the reduction in costs from higher effort equal to the reduction in money utility from higher effort.

Agents who have better cost conditions and those who offer high quality often receive positive rent. This is limited, however, because households keep some of the gains associated with unique cost conditions. In additon, there are certain budget-quality combinations where no agents will produce the required services because of insufficient compensation.

The housing market gives incentives for agents with good cost conditions to produce higher quality. These agent rewards are not necessarily drawn from tax rate changes. Most of the time, they are pulled from incumbents so that they do not have to rely on political processes to reflect their desires. Although incumbents want a complicated tax schedule to realize capital gains from the appreciation of their properties, such tax rates might cause agents to quit. For administrative simplicity, they stick to a particular optimum tax rate.

VI. What a social planner can achieve with a generous amount of information

Outcomes achieved by local property tax finance are similar to those that are produced when a social planner has perfect information on quality, household’s type, outcome, cost, and functional forms of rent and utility functions.

Social planners have no capacity to prescribe the effort, quality, and costs that agents should employ. As such, they instead offer budget-quality, rent-minimizing packages, constrained by the fact that agents will not compromise with zero-rents and that households can move anytime they want. An incentive compatible option will be produced if it gives an agent enough rent for offering high quality.

Social planners find the incentive-compatible option that maximizes rent by determining how much an agent’s rent must increase as his cost conditions fall. By doing so, an agent will have an incentive to report his true cost conditions and make the effort appropriate for those cost conditions. The social planner then minimizes rent relative to the incentive compatibility constraint and the agents’ participation constraint.

VII. A comparison of the local property tax solution and the social planner’s solution

Productivity differences between the local property tax solution and the social planner’s solution are due to the assumption that the social planer knows the form of agents’ rent functions. When the budget-quality curve is below the rent-minimizing incentive-compatible curve, the financial incentives for improving quality are not enough to cover the costs. Otherwise, households will only be limiting, but not minimizing rent.

If we assume that households had enough information about agents’ rent function, they will have a complex tax schedule. If not, they will settle with a flat tax rate. Also, when social planners are not aware of the said information, they will not be able to find the rent-minimizing incentive-compatible curve. They will need to settle with the local property tax solution – a solution which may not eliminate rent but can reduce it significantly.

Outcomes under centralized finance are comparable with those under local property tax finance in a residential market with many jurisdictions. It can produce results that are similar to the product of social planners who have all the necessary information. It is able to manage productivity efficiently because 1) it makes information verifiable by providing a measure of quality for cost – property prices; and 2) it partials out other necessary details.

Local property tax-based finance is automatic and relatively cheaper to use. It removes the disadvantages of moving costs and enables households to use private information to identify households of their own type. This allows household type to be cancelled out of the productivity problem. In choosing and assessing schools, parents only need to make simple comparisons using the people they know. They do not need to quantify their observations or develop a complex analysis with it. This cannot be the case under centralized finance.

In contrast, centralized finance does not automatically employ information from individual households’ actions. Required information is often unverifiable, costly, and difficult to use. It also allows producers to extract rents by minimizing unverifiable quality or by exaggerating costs.

Source:

Caroline M. Hoxby, “The Productivity of Schools and Other Local Public Goods Producers,” Journal of Public Economics, Vol.  74 (1999) pp. 1-30.