Framework for Appraising Educational Reforms

1996

A Framework for Appraising Educational Reforms

Ernest R. House

Abstract. Transaction cost economics provides a framework for appraising educational reforms, i. e., whether these reforms are likely to succeed in the "real life" of schools. This framework conceives reform as a contract between reformers and stakeholders (teachers, students, parents), and these transactions are marked by bounded rationality, opportunism, and the protection of "specific assets" (e.g., accumulated knowledge and skills) on the part of participants. Not acknowledging these attributes turns reforms into unproductive planning exercises, mere promising without results, or stakeholders struggling to protect their assets, resulting in the failure of the reform. Several current educational reforms are judged on the basis of their transaction costs and consequent prospects for success.

One of the most perplexing problems in education has been the parade of school reforms that do not seem to change school practice very much. Researchers have suggested various reasons for the failure of school reforms, ranging from cyclical theories of change to maldistribution of power within schools. No doubt, these analyses have merit. In this article, however, I would like to suggest other reasons not advanced before: the failure of reform efforts to take account of the basic features of schools as organizations and institutions. In a sense, I want to conceive teachers and students as acting much like people in other organizations, an analysis based on concepts borrowed from microeconomics.

Instead of thinking of an organization, like a school, as a production function, transaction cost economics conceives organizations as structures which govern transactions among people (Williamson, 1975, 1985). Whether transactions are organized within a firm (hierarchically) or among autonomous firms (through markets) is a major consideration. Just as there are basic production costs to produce a good or service in any enterprise, there are also transaction costs, and these transactions among people and groups are highly relevant for how activities are organized. Transaction costs are critical in education because much of what schools do is determined by group and individual interactions.

Transaction cost economics can be used as a rough appraisal framework to judge the potential of educational reform initiatives. I use the term "appraisal" rather than "evaluation" advisedly. In current usage, evaluation consists of collecting data after the reform is implemented to arrive at a judgment of worth. Appraising plans for reform has the sense of estimating value and potential, as a jeweler might appraise a diamond according to its qualities. Such a framework might screen out educational reform initiatives early on that have no chance of success.

Transaction cost economics conceives people as entering voluntary contracts (explicitly or implicitly) to do something in exchange for certain benefits. In other words transactions involve agreements among people. A particular task to be accomplished can be organized in several ways, and one might ask, what are the comparative transaction costs involved? There are the "ex ante" costs of designing, negotiating, and safeguarding an agreement. And there are the "ex post" costs of maladaptation to the contract, the haggling costs of correcting defects, the set-up and running costs of governing structures, and the bonding costs of securing commitments. These costs can sometimes determine success or failure.

The focus of this branch of economics is on transactions among people at the micro level, especially the comparative costs of planning, adapting, and monitoring task completion under alternative governance structures (Williamson, 1975, 1985). If one presumes that teachers and students operate on contractual "understandings" and that reformers want to introduce new considerations or conditions into the educational work of schools, school reform is a task that can be analyzed within this framework.

Contracting itself can be conceived as an act of planning, promise, competition, or governance, depending on which cognitive competencies and self-seeking propensities are ascribed to the agents engaged in the contractual exchange. There are three key attributes in this framework, each of which I will take up in turn: bounded rationality, opportunism, and the possession of specific assets, such as specific knowledge about the situation at hand.

Transaction cost economics assumes that humans are rational, but only in a limited way. They have "bounded rationality" (Simon, 1961). They cannot know everything or be able to process information perfectly. Just because one can specify a task to be done doesn't mean that humans will be able to accomplish it. They may lack sufficient knowledge or a way of acquiring the necessary knowledge. On the other hand, people are rational enough that many of their activities are planned with forethought. Bounded rationality, limited rationality captures the notion well.

Second, humans are opportunistic: they seek their own self-interest, sometimes aggressively. "Strong" opportunism means "self-interest seeking with guile" (Williamson, 1985, p. 47) and includes blatant forms of opportunism (cheating, lying), as well as incomplete or distorted disclosure of information in attempts to mislead, obfuscate, confuse, or disguise. In other words, not all behavior is governed by standard ethical rules, and not all people are open and honest. This doesn't mean that all people behave opportunistically all the time but only that some people do so enough of the time that some defense against opportunism is necessary.

A weaker form of opportunism is simple lack of obedience. Presumed obedience is critical to "social engineering" schemes in which a central plan is expected to be implemented by obedient functionaries who identify with the overall goals, as opposed to following sub-goals or their own self-interest. Obedience of participants is a common assumption in the utopian literature. Socialist economics, for example, typically assumes both unbounded rationality and obedience. It’s presumed that people will not "take advantage" of the situation.

Third, humans often have invested specific assets of one kind or another (material assets, education, knowledge), and these vested assets can significantly affect an exchange if the investment is non-trivial. For example, financial investors must be able to seek and obtain new uses for their money when the financial arrangement they are in doesn't work. "Redeployablity" of assets is critical. Investors are reluctant to invest money in unregulated markets unless there are safeguards, regardless of potential high returns on investment. The more safeguards and liquidity (so they can redeploy their money if necessary), the more they will invest. If such safeguards are not available, either they will not invest or they will seek significantly higher returns as compensation for risk.

However, not all investments and assets are like financial resources, i. e., transferable in regulated markets. Education, acquired skills, and personal contacts are often asset specific in that they cannot be acquired nor transferred easily. Such investments represent "sunk costs" in that the person has already paid for them with time and effort. Such "personal knowledge" in Polanyi's (1962, p. 52-53) sense includes skills and craftsmanship so deeply embedded in personal experience that they cannot be known by others or can be inferred only with great difficulty.

Assets can be idiosyncratic:
"There exist almost unique, irreplaceable research workers, teachers, administrators: just as there exist unique, irreplaceable choice locations for plants and harbors. The problem of unique or imperfectly standardized goods...has been neglected in the textbooks." (Marschak, 1968, quoted in Williamson, 1985, p. 53)

Asset specificity typically arises over time. In some exchanges humans build up assets in conjunction with other particular persons, so that it is easier for them to continue to do business with those persons. In some contractual situations the specific identity of the contracting parties matters in that the relationship comes to be valued. As a result contractual and organizational safeguards arise to support these transactions over time in bilateral exchange relationships.

By contrast, in neoclassical market transactions, faceless buyers and sellers meet to exchange standardized goods at equilibrium prices. An open market is a contractual situation in which there is a uniform price available to all comers impersonally. But this does not characterize asset specific situations. Rather, in asset specific situations, the identity of the person with whom one is doing business matters. Sometimes the identity of the other party is the main safeguard one has.

One example of asset specific relationships is the contracting out of evaluation services by the National Science Foundation (House, Haug, Norris, 1995). Agency personnel develop close personal ties with contractors they know can do the work, and these contractors know how to get things done for the agency. The relationship develops into bilateral dependency that may not be apparent to others in the organization. The formal legal apparatus does not recognize the existence of these specific relationships, though often they are critical.

Furthermore, not just any group can enter the contractual relationship with the same ease. Both the agency and contractor build up specific assets that make it easier for them to do business with one another, assets that are not easily transferable to another partner or situation. Also, would-be rivals cannot compete at parity with the contractor on these terms. Contracting is turned into a bilateral supply arrangement over time. In such transactions personal integrity and competence come to be seen as important. One can count on certain people; it matters a great deal with whom one is doing business. These extended personal relationships may also affect the product itself.

Depending on which combination of these three attributes--bounded rationality, strong opportunism, and asset specificity--one assumes about people engaged in transactions, one can arrive at different kinds of contracting arrangements. Different combinations of these three assumed attributes are represented in the figure below. The grade of "0" is assigned if the particular attribute is not considered sufficiently. The grade of "+" is assigned if the attribute is considered sufficiently, and the grade of "?" is assigned if the evidence is conflicting about the presence or absence of the attribute because there is evidence on both sides. Grades are assigned based on an analysis of each situation.

If agents are assumed to be opportunistic and have significant asset specificity (i. e., have invested in non-trivial specific assets), but are unbounded (unlimited) in their cognitive competence, the contracting process becomes "planning," where a deal is struck at the beginning of the arrangement and adhered to. Given unbounded rationality, all parties can foresee contingent events, even those that might arise from opportunistic actions. Hence, avoidance of such events can be written into the contract by all sides. Of course, this process may assume cognitive competencies that don't exist, which is one reason why planning fails so often.

One is reminded of "strategic planning" exercises which have become popular in universities. Typically, university administrators, the board of regents, and a few faculty members go away to a retreat for several days where they endorse a plan with university goals for all to follow. Although such planning may serve a purpose for those involved, it has little salience for faculty and students, who rarely read the plan. Somehow the planning document is supposed to serve as guidance for all, but in reality it matters not (except for the symbolic value of the planning effort itself), because there is no clear conception of how the plan will be carried out. Nor does the planning consider opportunism typically. Second, if opportunism is absent, but the other two attributes are present, then contracting becomes a matter of "promising." All agents have good intentions, pure motives, and no proclivity to serve their own self-interests excessively. Hence, both sides promise to fulfill the contracted work, which is a sufficient guarantee. If there is a problem down the line, the committed parties will resolve it with good intentions. Of course, the notion that some individuals are opportunistic some of the time is a better bet on human nature. Utopian organizations in which trust and good intentions are imputed to every member can be exploited by those who don't share these qualities. Conceiving organizations as production functions in which the guiding principle is "maximize firm outcomes" does not take account of such opportunistic motivations. Why should participants maximize the goals of the organization, whether it be profits or test scores, especially if it costs them something? Why should they ignore what they think is important in order to pursue some overall goal defined by someone else? Indeed, assumptions of no opportunism mean that employees are working solely for the organization all the time, which seems unlikely. Safeguards against strong opportunism seem a prudent precaution.

The third situation is one in which there is no asset specificity. When no specific assets are at risk, there can be pure market competition in that the parties to the contract have no continuing interest in the identity of the other party. Hence, discrete impersonal market contracting is efficacious. An impersonal market is one in which the buyer and seller do not care who is on the other side of the transaction. They care only about price and quality of the good or service. Hence, contracting is conceived as "competition," which works when asset specificity is negligible. The absence of specific vested assets applies to a wide range of human behavior so that markets do indeed work properly in many circumstances. However, when assets are non-redeployable--the owner of the assets cannot seek a different arrangement easily--then markets do not work the same way. Those with assets at risk devise safeguards of one kind or another to protect their investments.

Transaction cost reasoning assumes that each of these three contracting situations will fail often because they are based on unrealistic assumptions. Pure planning will fail because parties have limited, "bounded" rationality and cannot see how things will work out in the future. Trying to account for all contingencies in a pre-specified agreement rarely works out as planned. Simple promising breaks down because some people are opportunistic to a significant degree and will deviate from the agreement when it suits them, regardless of what was agreed to. Pure market competition fails when people have significant specific assets at risk. Those with specific (non-transferable) assets are not inclined to put those assets at risk in unprotected market situations if they can help it. Rather they develop special relationships with their contracting partners which provide safeguards. Hence, those who want to change organizations should organize transactions so as to economize on bounded rationality while guarding against excessive opportunism and recognizing specific assets where they exist. This means that different governing relationships are appropriate, depending on the situation.

Applied to Education

One can apply a similar analysis to school reform. The first thing is to conceive students, teachers, and administrators as normal, rational people who have ambitions and motivations of their own, like everyone else. They should not be idealized to have unlikely attributes (e. g., abnormal altruism, idealism, energy, obedience). Nor should they be demonized (as unintelligent, slothful, lazy, inert). Such inaccurate characterizations serve as the unspoken basis for many misguided reforms. Teachers and students are willing and able to do certain things when they are rewarded and think these things make good sense. "...investments usually are rational responses to a calculus of expected costs and benefits" (Becker, 1993, p. 17). They are likely to resist when this is not the case. Change is not cost free for them. Teachers are no more prescient than anyone else, even though more highly educated than the general population. They have formidable cognitive resources but cannot be expected to overcome very difficult situations. "Bounded rationality" captures the situation well. On the other hand, they are not miseducated or uninformed, as some critics have implied. Teachers are also like everyone else in that some are opportunistic sometimes. They are not overwhelmingly altruistic or obedient. For example, one study found that only 11 percent of Philadelphia teachers implemented all aspects of the school district curriculum policy, only 10 percent used the proper pacing schedule, and only 4 percent complied with district grading guidelines (Johnson, 1990). Such teacher behavior is typical.

Finally, and most importantly, teachers have significant specific assets which are based on their education, ability, and experience in the classroom, perhaps the characteristic most misunderstood. These skills are highly specific in that they are derived from particular events, places, and students, and often consist of tacit rather than explicit knowledge, in the Polanyi sense of knowing how to do something (like ride a bicycle) without being able to explain how to do it or teach someone else to do it. These skills cannot be explicated nor transferred easily. Most reformers greatly underestimate the craft knowledge that teachers possess. In fact, teachers themselves underestimate their own craft knowledge. It is not the kind of knowledge that can be recited on demand.

Given such specific attributes, one might expect teachers to be reluctant to risk their knowledge and skills in open competitive markets unless there are safeguards and expectations of significant gain. Why should teachers risk knowledge assets built up over many years by switching to new teaching materials or techniques of unproved quality? Their education and professional experience might be devalued.

Teachers know from experience that reformers cannot guarantee the promised outcomes of their reforms. These reforms are rarely based on careful research (not a significant source of reliable information for teachers in any case). Most reforms are the simple ideas of political and educational entrepreneurs. Almost all become fads only to disappear eventually (Elmore and McLaughlin, 1988). Why bother? And for their part, most reformers haven't the foggiest notion how their ideas will play out in schools. Distrust, and even cynicism, about new reforms are rampant among experienced teachers.

Reviewing research on the micropolitics of education, Malen says about schemes to involve teachers, By and large, teachers may be initially enthusiastic. But they get weary and wary. They get exhausted by the demands and become skeptical of the prospects for meaningful influence and suspicious of the requests for their involvement....These works suggest that teachers' professional security and integrity are at risk. Simply said, teacher are vulnerable to the criticisms of principals, peers, parents and students. They thus insulate themselves from the pressure and pain of interactions that can damage their reputations, diminish the quality of their work life and disrupt their ability to carry out their responsibilities in ways congruent with their views and values (Malen, 1994, pp. 156, 157).

The situation is different for administrators. They must attend to the public and school board. They must appear progressive (or conservative as the times demand), respond to the media, and seem to be improving the schools. That's justification for why they hold their positions. There may be small cost for them personally for adopting teaching innovations. Reform may mean extra effort but doesn't usually cost administrators asset specific resources, which for them include personal contact and knowledge of the school board. On the other hand, reforms which put their position and control at risk, such as decentralization schemes, are quite a different manner and are unlikely to be carried out as projected (Iannoccone and Lutz, 1994).

Similar reasoning applies to students and parents. For example, students have less investment in the educational system, at least in the early years. So they are more receptive to new ideas--up to a point. However, when the rules are changed significantly, they can become extremely contentious. For instance, contemporary student riots in France have been caused by university students protesting democratization of the higher education system. Such reforms threaten the specific assets they have accrued within the system. Ordinarily, it is the socialist unions of students and teachers who lead the resistance to these democratizing reforms.

Parents also have investments in moving their children along in the educational system to achieve social mobility. Since WW II social mobility has been achieved primarily by people entering the professions, for which higher education is the gatekeeper. For example, a reform like doing away with grades (the giving of marks) in the elementary schools seems a simple reform that most would applaud. Students should be overjoyed to be relieved of grading pressures, and teachers would not have to deal with an odious task. However, grades are an important control mechanism for some teachers. Furthermore, most middle-class parents are adamant about maintaining grades. It provides them with a way of knowing whether their children are on the right track and gives their children a differential advantage in the competition for entry to higher education. Eliminating grades usually results in a strong backlash.

Although it sometimes appears that the schools don't change because of a paucity of ideas and energy, that they are inert, their lack of innovativeness is more often caused by vested interests pulling in different directions. The schools are the way they are partly for historical reasons and partly because they are the result of constant pressures from different directions. The groups doing the tugging--teachers, students, parents, and taxpayers--share attributes of bounded rationality, opportunism, and asset specificity. Policies to change the schools must take these things into consideration at risk of failure.

Goals, Standards, and Decentralization

Consider how some educational reforms might stack up against criteria from the transaction cost framework. Most reforms ignore critical attributes. The chart below indicates whether the three factors are considered sufficiently in various reform proposals. Again, the grade of "0" is assigned if the attribute is not considered sufficiently. The grade of "+" is assigned if the attribute is considered sufficiently by the reformers. And the grade of "? " is assigned if it is not clear whether the attribute is considered sufficiently, either because there is conflicting evidence of its presence or absence or because the reform in question has more than one version. One form may consider the attribute while the other doesn't.

For example, national standards reforms do not generally consider the question of opportunism sufficiently because teachers may teach the tests or ignore the standards, and most reforms offer no protection against this. Usually, national standard reforms do recognize that teachers have assets and expect teachers to use their knowledge to attain the standards. On the other hand, some standards reforms recognize bounded rationality and provide help through specification of standards or special training; other schemes don't. Hence, national standards are given a "?" on the bounded rationality because of different versions, a "0" on opportunism and a "+" on recognition of asset specificity. Grades are assigned based on judging the reforms as they have been explicated by leading proponents. No doubt, some proponents might argue for different assignments.

National Goals. One prominent plan for educational reform is America 2000, begun in the Bush administration and continued by the Clinton administration as Goals 2000 (U S Department of Education, 1991). This plan combines the establishment of national goals with national standards, albeit voluntary ones. However, there are other goal programs without such standards, so that one can advocate goals without explicit standards. The basic idea of national goals is that schools are unfocused. There are too many goals, too many interest groups, and too many purposes for the school to serve. The result is lack of direction.

Hence, if one can focus on fewer goals, the productivity of schools will improve significantly. The critical question is, why would teachers and others pursue the national goals rather than state, local, or personal goals? Announcing a set of goals does nothing to establish those goals in schools and classrooms, unless one assumes that teachers are especially altruistic or obedient. So motivation is lacking, if one assumes teachers are like everyone else. Would professors pursue national goals for universities? Physicians for medicine? Businessmen for business? It seems unlikely. In fact, Goodlad (1984) found that teachers and parents think schools should be comprehensive, which means caring for student well-being across the board, a view which would work against fewer goals.

Furthermore, if fewer goals were specified, would teachers be able to achieve them? Drug-free schools? First in the world in math and science education by the year 2000? Every child coming to school ready to learn? The achievement of these goals demands resources beyond the reach of teachers and administrators. They wouldn't know how to accomplish these goals, even if they wanted to. National goal specification does acknowledge that teachers have asset specific abilities and skills. New goals do not require teachers to abandon their own skills and abilities. But the plan is deficient in not recognizing opportunism and bounded rationality.

National Standards. Another reform is when goals and standards are combined into content standards schemes. The national standards might be subject matter standards developed by professional organizations, such as the National Council of Teachers of Mathematics. The presumption is that teachers don't have the proper focus, nor the correct content. Content standards specify in detail what students should be expected to know. And teachers may facilitate this knowledge in their students however they deem fit.

This approach assumes that the motivation for teachers adopting such standards will be provided by publicity given to test scores based on the standards and to the prestige accorded material developed by professional organizations. Such an approach overestimates the degree to which teachers will adopt standards and miscalculates how teachers will react if their students test scores are made public (Cohen, 1995). The history of such attempts is that teachers teach the test items under conditions of strong accountability and manipulate or distort scores (Glass and Ellwein, 1986). The standards approach probably also misjudges teachers' ability to achieve the standards. There is some attention to teacher motivation, but it is insufficient. And there is attention to providing what to teach (through the standards) and how to teach, but it is probably insufficient as well, as it now stands in most schemes. More is needed than specifying standards and testing for them.

Decentralization. In analyzing the movement to push decisions to the school or classroom level, Ferris (1992) noted, "A key premise of the decentralization movement in education policy is that those at the school site have information necessary for the decisions to enhance educational performance, information that is not available at the district level or above" (Ferris, 1992, p. 333). The comparative advantage of decentralization is based on the idea that those closest to students are in the best position to judge their needs. However, just because the requisite information is not available at the district or higher levels doesn't mean it is available at the school or classroom level. It is one thing to be freed from bureaucratic strictures. It is quite another to know what to do with this freedom.

Authentic decentralization takes full advantage of the asset specificity of the teachers skills and knowledge. On the other hand, one must be concerned about whether teachers and administrators know how to improve things and also about opportunism, whether they will do the appropriate things rather than pursue subgoals. Ferris (1992) has applied "principal-agent" theory to decentralization efforts. Principal-agent theory treats decentralization as a contract between the granting authority, the "principal" (school district or state) and the "agent" (school or teacher). The principal must worry about whether the agent is performing properly. The cost of enforcing such a contract depends on the measurability of the agent's behavior (e. g., effort) and performance (outcome). The more uncertain the relationship between behavior and outcome, the more the principal must invest in gathering information. The more expensive the cost of collecting information, the more the contract is likely to be based on behavior rather than outcome. One problem with measuring performance is the lack of adequate performance measures. Standardized tests are the usual way of measuring school outcomes but there are long-standing problems with tests as accurate measures of outcomes (Porter, 1990). There have been efforts to develop "alternative assessment" measures, but good alternative tests are not available currently. Furthermore, few teachers would be willing to have their performance judged on the basis of student test scores. In fact, a recurrent problem is that teachers under pressure teach students the test items (opportunism).

Markets and Incentives

Two ideas given great currency in the media are markets and incentives applied to schools, including school vouchers, schools of choice, merit pay, performance incentives, "contracting out" services, and business-operated schools. Market Reforms. The most influential work to propose that public schools abandon control through elected school boards and resort to market mechanisms in which parents have freedom to send their children to whatever schools they want is Chubb and Moe (1991). They contend that parental choice would free schools from bureaucratic control and that student achievement would improve because schools would be more autonomous and able to increase achievement. The basic impediment is democratic control. Even though we know what makes for a good school (from the "effective schools" literature), we are not able to incorporate such features.

School goals are countless in number, lack coherence, and are watered down to please numerous constituencies. Democratic control results in bureaucratic mandates and edicts being imposed from above, leaving little room for teachers and administrators to do what they know best. The imposition of higher order values weakens school effectiveness because it takes decision making away from workers at site level. Education is a process that requires that critical decisions be made on the spot, in the school and classroom.

In Chubb and Moe's analysis, schools cannot be controlled effectively from above because measuring output properly is so problematic as to be impossible. "The only way to measure performance adequately is to rely on the discretion of those who work in the school--which is precisely what the unions and democratic authorities are strongly inclined not to do, and what the whole system is built to prevent" (p. 197). The technology of schooling is local ("bottom heavy") in that people in the school must decide what to do if effective education is to occur.

By contrast, in a market system of school choice, school personnel would be able to exercise their knowledge more effectively, and schools would have to produce what parents want. Those who operate the schools would have to please the "clients" (parents and students) because parents could switch their children to other schools. Schools that did not perform satisfactorily would go out of business from loss of clients. Bureaucratic strictures would lessen because the overgrown apparatus of administrative control would not be needed.

However, Chubb and Moe have not gone far enough in their analysis. Schools are indeed impeded by the bureaucracy and having to serve many constituencies. But does bureaucracy stem from democratic control? Business corporations also have developed large bureaucratic organizations, and they are not democratically controlled. In fact, schools have become bureaucratic partly from emulating corporations (Callahan, 1962). Responding to constituencies does increase the administrative bureaucracy by adding services to deal with these constituencies. However, bureaucratic escalations have occurred in both private and public institutions. Bureaucratic structures are partly a response to the nature of the transactions they undertake.

Chubb and Moe are correct that school technology is "bottom heavy" in that school personnel must make critical decisions that cannot be dictated effectively from above. Furthermore, the nature of the educational process is so complex as to defy quality measurement that can provide a simple reading. They assume that teachers and administrators, or at least researchers, know what to do if only they were not restrained by bureaucratic impediments. This is not the case. Decentralized schools operate much like other schools.

Then there is the more critical problem of the asset specificity of teachers and students. Putting teachers in an unrestricted open market would threaten their knowledge and skills. According to transaction cost economics, such a situation is not stable. Either workers will demand high wages to compensate for their exposure or they will demand safeguards to protect their assets. Another difficulty is the asset specificity of students. Market schemes presume that if parents and students don't like the school, they can leave, as customers might cease shopping at a particular store. However, such an impersonal market prevails only where there is no asset specificity. Students actually develop a heavy investment in their school over a period of time. They invest in particular friendships, participate in particular social structures, and learn from a particular curriculum. Their investment in a particular school is high, as those who have tried to get their children to move or attend another school know.

When parents first consider which school their children should attend, the situation may resemble an open market. Once the investments begin, however, the relationship quickly turns into bilateral dependency. Friendships develop, teachers come to know the students, including their personalities, learning styles, and parents. Withdrawing from school because of poor performance of the school is not a viable option for most, even if poor performance is ascertainable. Rather in such asset specific relationships, the parties to the contract will try to find special safeguards to protect their investments. The mechanism of withdrawal and competition will not work the way stipulated because the situation is not a neo-classical market--which is not to say that choice and competition may not work in other ways. As it stands, Chubb and Moe's market reform proposal seems deficient on recognizing asset specificity. People with specific assets will not risk those assets in a competitive, impersonal market.

Performance Incentives. Hanushek (1994) focuses more on the classroom than on governance.
"If a single, glaring lesson is to be learned from past attempts at school reform, it is that the ability to improve academic performance using standard, uniformly applied policy is limited....The most appropriate, indeed, the only place to begin promoting diversity is at the basic unit of the school: the individual teacher in an individual classroom." (Hanushek, 1994, p. 85)

In his view, it is impossible to define "best practice" among teachers because there is none. Teachers succeed and fail in many different ways. The reason decentralized management schemes have failed to improve education is that they lack clear incentives for teachers to improve. Improvement depends on individual teachers being rewarded for good performance and punished for bad. This requires performance evaluation systems which are not simplistic.

Hanushek proposes three strategies: increasing the efficiency of resource use; using performance incentives; and learning from experience. Evaluation is critical to performance incentives. If one cannot accurately assess the performance of teachers, then how can one reward them for achievements? He insists on using student achievement measures for this purpose. He recognizes that student achievement is derived from many sources and that a performance assessment mechanism must assign responsibility to teachers in proportion to what they have accomplished.

Unfortunately, Hanushek seriously underestimates the difficulty of measuring individual teacher performance by standardized student tests. There are no standardized tests available for this purpose and no way of parsing out achievement attributable to individual teachers (Meyer, 1994). Such a technology does not exist. It would be possible to evaluate teacher performance using non-standard methods not based on student test scores, but Hanushek rejects non-standard methods, even though universities evaluate faculty performance in this fashion.

In summary, standardized means of performance assessment are lacking. In other words, there is bounded rationality not only on the part of teachers, which Hanushek recognizes, but also on the part of the measurement and research communities, which he does not. These groups cannot supply the evaluation and assessment techniques needed, except perhaps at great cost. On the other hand, Hanushek's scheme recognizes the asset specificity of teachers.

This information cost problem is pervasive, exemplified by Stiglitz's (1994) analysis of the failure of market socialism. For public programs there are multiple objectives, measuring outputs is problematic, and measuring the contributions of individuals and suborganizations to the overall productivity of an organization is all but impossible, in his judgment. Even measuring inputs is difficult. Such information is not available for anything like reasonable costs. In fact, the centralized Soviet economy eventually collapsed under such an information burden (combined with distorted incentives). The neo-classical market paradigm overestimates the availability of information and underestimates the costs of acquiring it. It assumes that variables of interest are perfectly observable and that the employer can costlessly ascertain levels of effort and make payment contingent. In fact, production is more a process of negotiation than of "price-taking." It is more feasible to compare one situation to another in gross terms, rather than try to manage precise cost production comparisons. Government planners do not possess such information, and the capacity of any central body to gather, process, and disseminate such information is severely limited. These limitations apply to indicator systems as well.

Asset Policies

Teachers learn to teach through direct experience and participation. From direct experience teachers draw cause-and-effect inferences as to what works and doesn't work for themselves. This knowledge is personal and particular to the actual situation, and much of it is tacit: The teacher knows how to do things he or she cannot explain. One way to proceed is to increase the specific skill assets of teachers in some way.

Self-Assessment . Teacher evaluation problems include the fact that evaluation systems are developed for reasons of accountability external to teachers and not used to improve practice. Second, teaching is so complex that a multitude of relevant factors must be evaluated, making results difficult to interpret (Good and Mulryan, 1990). And, of course, it is difficult to find consistent factors that work in most situations.

Given the autonomy of teachers and strong contextuality of teaching, an evaluation system should be context specific, ongoing and formative, flexible and evolutionary, and personally and institutionally relevant. It might improve practice by bringing tacit knowledge to consciousness, remove isolation and enhance communication, promote a professional culture, and enable educators to learn from practice (Holly, 1989). These criteria take into account the asset specificity of the teacher's investment by building on the teacher's knowledge and skills.

One way to offer teachers the opportunity to improve the quality of their work is through self-assessment (Barber, 1990). This process requires that teachers monitor selected aspects of their own instruction. A number of data collection techniques can be used as the basis for self-assessment, including audio or video tapes of classroom instruction or student conferencing, questionnaires or interviews administered to students, interviews administered by one teacher to another, and teacher logs and journals (Lapan and Hays, 1992).

Self-assessment is best implemented as a collaborative effort where teachers learn to collect data about their teaching and follow a protocol for discussing representations of their instruction. Teachers are encouraged to discuss their own teaching actions and reflect on how they might improve. As they examine their own data, they are guided by questions from other teachers. For example, a teacher might teach a lesson and record it on video tape. A small group of four or five colleagues would analyze the tape using criteria provided in advance. This process of visiting and revisiting the data allows the teacher to reflect on differing interpretations. The teacher could move again to the planning and acting phases.

Judged by the criteria of transaction cost economics, self-assessment does fairly well. It takes account of asset specificity by investing in the teacher's abilities. On the other hand, it does not assume teachers already know the best way to do things, but rather provides a process for learning, thus addressing bounded rationality. One weak point is the reward structure. What would impel teachers to engage in such activities? Although there are intrinsic rewards of working with colleagues once the process starts, why begin? So control of opportunism is lacking. Furthermore, teacher norms of privacy often prevent group work from occuring. Also, teachers exposing their classroom practice to others might be seen by some as infringing on their specific assets by making those assets organizational property. Central Park East . Another strategy is to change the school itself to maximize specific knowledge. Central Park East is perhaps the best known school in the country. Although the school exists in a largely minority community, the school's graduates have a high college attendance rate. Its success has been widely acclaimed, and it needs no endorsements from this appraisal scheme. However, how does the appraisal scheme itself measure against CPE? Using CPE as an anchor test, can this appraisal scheme explain CPE's success with the transaction cost concepts?

In founder Meier's (1995) judgment, the key to success is that schools be small, with elementary schools having 300 students and secondary schools about 400. Small size is critical because faculty, parents, and students must find enough time for discussion and argument in order to reach consensus as to what the school should do, and these discussions must be face-to-face. The agreement reached provides a vision for the school and one voluntarily entered. Through on-going discussions, persuasion takes place which results in changes in practice. Furthermore, teachers work together collectively and collaboratively in a small group. Such collaboration is essential for a strong school ethos.

The faculty is held accountable collectively to produce the overall school effect. This means they must have access to each other's work, and the teacher work group must be small enough to allow this to happen, for teachers to visit each other's classes and engage in peer critique. Above all, teachers must get to know the students and their work, even the way individual students think. Students must get to know each other and the teachers. With such a small school, accountability can be a matter of access, not of monitoring. There is no need for cumbersome measurement systems to inform parents since they can come see, as can administrators. Finally, small schools immerse students in an intense school culture that adults have a role in shaping deliberately, rather than abandoning students to an autonomous peer culture shaped by the mass media.

To accomplish all this, the school must have autonomy, controlling budget, staffing, scheduling, curriculum, and assessment. Also, according to Meier, creating this in New York City would have been impossible without school choice. Choice allowed opportunity for change. Mandated change results in unwilling, unready parents and professionals, with high likelihood that the school will return to the status quo.

The Central Park reform attends to critical features of the school as an organization. By being small enough that parents and teachers can talk through problems in face-to-face conversation, it allows everyone to build specific assets, which are heavily vested in particular and personal knowledge. This collective personal knowledge is the accumulated wealth of the school. Small size facilitates agreement and criticism among the faculty, a way to develop their knowledge further--the economies of small scale. Hence, the reform takes account of bounded rationality, opportunism, and asset specificity in a total package. The high transaction costs required are managed by the small size.

No doubt, Debbie Meier and the CPE developers could care less about transaction cost economics and had nothing like this in mind as they developed the school. However, this appraisal scheme can explain CPE's success using concepts intrinsic to the scheme, not a conclusive demonstration of the scheme's utility perhaps, but it would be telling if the scheme could not account for CPE.

Summary

The conceptual scheme presented here for appraising school reforms is no doubt incomplete and insufficient, but I believe that accommodating bounded rationality, controlling opportunism, and recognizing the value of specific assets are essential if reforms have any chance of success. Most popular reforms do not meet even these limited criteria, though admittedly I have constructed only brief sketches of reforms for purposes of illustration, perhaps even parodies. No doubt, actual reforms are more complex. It is also true that teachers and schools differ considerably among themselves, and I have ignored this complication. Nonetheless, politicians in particular have a propensity for advocating extremely simple reforms which have no chance of success, even by these limited criteria. Educators need a way of critiquing such plans other than saying, "It won't work."

The appraisal framework does not preclude the possibility of combining complementary reforms, according to the criteria of bounded rationality, opportunism, and specific assets, or adding other elements that might address such deficiencies. For example, one might blend the advantages of self-assessment, strong in bounded rationality and asset specificity, with performance incentives, strong on opportunism and motivation. However, one can see problems that might result. If one evaluated performance on the basis of data generated by self-assessment, that might corrupt the information. There may be unproductive interactions among approaches, as well as productive ones.

In my judgment, the best reforms would focus on greatly reducing the administrative hierarchy, transforming the purpose and structure of the central staff to a small strategic staff, reducing the size of schools to 300 or 400 students, making schools relatively autonomous and unregulated, providing opportunities for schools of choice, encouraging the internal flexibility of a Central Park East, and maintaining good staff development, such as self-assessment. Such reforms recognize the grass-roots technology that constitutes personalized education, fix decisions at the school and classroom levels, and enhance the specific knowledge and skills on which good teaching depends. However, they do not begin to exhaust the possibilities consistent with transaction cost criteria.

If one were to develop a more complete appraisal scheme, one would have to take account of other important factors as well. It has been suggested that research on educational reform occurs from three basic perspectives--the technological, political, and cultural (House, 1981). In a sense, transaction cost economics is an up-dating of the technological perspective. Political factors, such as conflicting interests, and cultural factors, such as the organizational culture, would expand the comprehensiveness of the appraisal scheme, but also make it more complex.

The three basic perspectives rest on concepts drawn from economics, political science, and anthropology. Since disciplinary knowledge is abstracted from the institutions under study, such knowledge can never fully explain institutional reality, any more than physics can map every physical event. Combining all three perspectives would provide a more complete picture, but a picture still necessarily incomplete (House and McQuillan, in press). The most conceptual schemes can provide is a rough map of reality, hopefully one that identifies critical features. They cannot substitute for reality, through which one must still negotiate. It is one thing to analyze educational reform and quite another to do it.

References

Barber, L. W. (1990). Self-assessment. In J. Millman and L. Darling-Hammond (Eds.). The new handbook of teacher evaluation, Newbury Park CA: Sage, 216-228.

Becker, G. S. (1993). Human capital. 3rd ed. Chicago: University of Chicago Press.

Callahan, R. E. (1962). Education and the cult of efficiency. Chicago: University of Chicago Press.

Chubb, J. E. and Moe, T. M. (1990). Politics, markets, and America's schools. Washington, D. C.: Brookings Institution.

Cohen, D. (1995). What standards for national standards? Phi Delta Kappan. 76, 10, 751-757.

Elmore, R.F. and McLaughlin, M. W. (1988). Steady work: policy, practice and the reform of American education. Santa Monica, CA: Rand Corporation.

Ferris, J. M. (1992). School-based decision making: A principal-agent perspective. Educational Evaluation and Policy Analysis, Winter, 14, 4, 333-346.

Glass, G. V and Ellwein, M. C. (1986). Reform by raising test standards. Evaluation Comment. Los Angeles: Center for the Study of Evaluation.

Good, T. L. and Mulryan, C. (1990). Teacher ratings: A call for teacher control and self-evaluation. In J. Millman and L. Darling-Hammond (Eds.). The new handbook of teacher evaluation, Newbury Park CA: Sage,191-215

Goodlad, J. L. (1984). A place called school. New York: McGraw-Hill.

Hanushek, E. A. (1994). Making schools work: Improving performance and controlling costs. Washington, D. C.: Brookings Institution.

Holly, M. L. (1989). Perspective on teacher appraisal and teacher development. In Simons, H. and Elliott, J. (Eds.). (1989). Rethinking appraisal and assessment. Open University Press, 100-118.

House, E. R. (1981). Three Perspectives on Innovation: Technological, Political and Cultural. In Rolf Lehming and Michael Kane (Eds.) Improving Schools: Using What We Know. Beverly Hills: SAGE Publications, pp. 17-41.

House, E. R., Haug, C. and Norris, N. (1995). Evaluation policies and issues in the National Science Foundation. School of Education, University of Colorado. 146 pages.

House, E. R. and McQuillan, P. (In press). Perspectives on Innovation. In A. Lieberman, M. Fullan, and A. Hargreaves, and D. Hopkins (Eds.). International Handbook of Educational Change. Dordecht, The Netherlands: Kluwer. Roots of Change, Vol. 1.

Iannaccone, L. and Lutz, F. W. (1994). The crucible of democracy: the local arena. Journal of Education Policy, 9, 5 & 6: 39-52.

Johnson, S. M. (1990). Teachers at work. New York: Basic Books.

Lapan, S. D. and Hays, P. A. (1992). Developing special programs for minority gifted youth: Northern Arizona University's teacher training program. Potential, 18 (7), November-December, 1-3.

Malen, B. (1994). The micropolitics of education: Mapping the multiple dimensions of power relations in school politics. Journal of Education Policy , 9, 5&6, pp. 147-167.

Meier, D. (1995). The power of their ideas. Boston: Beacon Press.

Meyer, R. H. (1993). Can schools be held accountable for good performance? In E. P. Hoffman (Ed.). Essays on the economics of education. Kalamazoo, MI: W. E. Upjohn Institute for Employment Research, 75-109.

Polanyi, M. (1962). Personal knowledge. New York: Harper Row.

Porter, A. C. (1990). Assessing national goals: Some measurement dilemmas. The assessment of national educational goals. Proceedings of the 1990 Educational Testing Service Invitational Conference. Princeton, NJ: Educational Testing Service.

Simon, H. (1961). Administrative behavior. New York: Macmillan.

Stiglitz, J. E. (1994). Whither socialism? Cambridge, MA: MIT Press. U. S. Department of Education. (1991). America 2000: An education strategy. Washington, D. C. 62 pages.

Williamson, O. E. (1975). Markets and hierarchies. New York: Free Press.

Williamson, O. E. (1985). The economic institutions of capitalism: Firms, markets, and relational contracting. The New York: Free Press.