Property, Power and Public Choice
A. Allan Schmid. C copyright 1987, 1997
EMPIRICAL INSTITUTIONAL STUDIES
The institutional theory of this book is not intended to provide a global guide to labeling one set of rules as efficient or inefficient. The theory's primary use is in guiding empirical inquiry into the substantive consequences of alternative institutions. Comparisons are made among real alternatives and not with an abstract ideal. The purpose of the book is to provide a better basis for an empirical institutional analysis, perhaps best illustrated by a review of a sample of empirical studies. The review should demonstrate that (1) institutional analysis is a workable and useful field of inquiry and (2) the concepts of the book can be used to formulate hypotheses for testing. (3) The short summaries illustrate less well the important role of acquiring knowledge of institutional detail and history. Before we proceed to this review, the conceptual elements from the preceding chapters will be summarized into methodological steps and the meaning of institutional research clarified.
There is a parallel between production function estimation and a predictive institutional economics. In the same way that we explore the effect of alternative factor inputs on output of goods and services, we will explore the effect of alternative institutions on human behavior. The big question is, do alternative institutions make a difference?
The basic model relates situation, structure, and performance. Situation refers to the inherent sources of interdependence. Structure identifies chosen institutional alternatives in terms of varieties of property rights and their distribution. The behavior and actions of people individually and in the aggregate in firms and government agencies result in performance in terms of various intermediate products and finally the quality of human life. Policy analysis requires a total inquiry. But it is useful to distinguish between the study of institutions, which is concerned with the link between institutional alternatives and behavior, and production analysis, which studies the link between behavior with respect to resource combinations and goods and services. Such a distinction can be illustrated with an extract from an article by the author (1972, p. 894).
Consider zero population growth. It is one thing to analyze the result of ZPG on the economy in terms of income and productivity, but it is quite different to establish the connection between the alternative institutional rules that result in women having fewer babies. For example, what child-bearing behavior results from giving each woman the marketable right to bear two children? What difference does it make if the rights are initially sold to the highest bidder? ....
Or, consider research on limiting fertilizer in agriculture. We need biological data on the relationship of fertilizer runoff and aquatic life. This can be combined with information on agricultural production and demand to indicate the effect on food prices and production location as a result of different levels of fertilizer use. This is not institutional research. One institutional alternative is a legal prohibition of fertilizer use. The question is whether this in fact will obtain the given behavior (or what other behavior it also will induce). Experience with liquor and pot prohibitions indicate that this institutional form does not always produce the implied behavior.
A maximization or simulation model utilizing a production function with different constraints on fertilizer use is quite different from simulating a behavioral reaction to alternative institutions which influence the amount of fertilizer actually used. Most of the current policy models are incomplete because they begin with an assumed conduct and inquire of performance.
People seem to find it both productive and irritating to associate with other people. It seems useful then for problem-solving research to first ask what is the good or resource at issue and then categorize the inherent cause of human interdependence that the good creates (situation). What is it about different goods that influences how one person can affect another for good or otherwise? Note again that these features exist prior to the analysis and are not determined by the rights being analyzed.
The next step is to specify the alternative institutions to be chosen as well as relevant institutional variables to be held constant. Again, the institutional or property rights variables can be stated quite literally as rule X and rule Y, but it may be useful to categorize them in a more general manner in order to make comparisons among alternatives that are structurally similar even though differing in specific application. One such categorization is bargained, administrative, and status-grant alternatives were distinguished. When a certain performance is desired, different mixes of those transactions can be made.
The major institutional alternatives are often phrased in terms of government regulation versus what is euphemistically called the free market, or government provision through taxation versus market bargaining and contracts. Economists seem to gravitate toward grouping institutional alternatives in terms of market versus non-market decision systems. This probably is from habit and from argumentative strategy. Economic theory was born as people revolted against the kings, and the rules of feudalism were under pressure from a new merchant class intent on trade. Ever since, many economists have been preoccupied with rationales for the superiority of the market over government intervention (Arrow and Scitovsky 1969, p. 2). This is often an unfortunate dichotomy since there are many ways to structure the opportunity sets in a market, and the difference in performance may be as great among alternative markets as between markets and non-markets. Certainly the dichotomy between capitalism and socialism, while useful in some respects, covers more differences within each than between in many instances. In the review to follow, no one set of institutional categories will be slavishly followed. The choice is often dictated by the available studies or for future research, by the availability of actual property rights alternatives to be observed. Sometimes, type of transaction will be contrasted, sometimes other ways to categorize rights systems will be used, and sometimes the alternative will be a different interpersonal distribution of rights within a given system.
The essence of the formulation is to relate the institutional variables to performance while considering the type of human interdependence involved in the situation. The utility of the theory is to specify carefully the institutional alternatives being tested and to know when to control for other background institutional variables. In practice, it is often necessary to control for other situational economic and technological variables that might affect the performance results observed. For example, in examining the effect of alternative forms of business organization on production costs, it is necessary to separate out the effect of economies of scale from the effect of the rules on the choices by managers. There may be interaction and feedback between the two. Thus, the institutional economist often needs to combine knowledge of human behavior with knowledge of the physical production function.
The final step is the actual empirical test. Since the analyst seldom has the opportunity to set up an experiment assigning some people or states to one institution and some other people to another institution, success in this area of research depends on taking advantage of contrasts and changes as they occur. Formulation of causal tests in this context is called quasi-experimental design by statisticians. The emphasis in this review is not on statistical method, but some of the variety of designs will be noted. (The terminology used is loosely based on Campbell and Stanley [ 1963].) Since the emphasis is on model formulation and specification of the variables, no critique of methods will be made. The purpose is not to arrive at conclusions on the consequences of a given set of institutional alternatives. Thus, the statistical method and weight of the supporting evidence in all available studies are not reviewed. Still, the author cannot entirely resist some casual empiricism in some cases where the available systematic studies seem most narrowly conceived and one-sided; nor can he resist a few comments to emphasize a theoretical point discussed earlier that might illuminate policy options.
Institutional research obtains knowledge from methodologies ranging from formal quantitative hypothesis testing to what McCloskey (1985) refers to as the rhetorical methods of history, case study, analogy and introspection. The performance of alternative institutions is often sufficiently expressed in directional and qualitative terms (Simon 1978).
The cases reviewed below are loosely grouped into a number of commodity and problem areas. They are chosen to include a wide variety of situations to demonstrate the scope of possible institutional research and to test the generality of the theory.
The study of property rights in land has a long history. While landownership is no longer the major source of wealth in industrial economies, it is the basis for many current disputes over environmental policy. Land tenure remains an important issue in agricultural economies around the world (Cline 1970, Parsons, Penn and Raup 1956). Several other natural resource issues are explored below.
Use of land for cattle ranching and citrus production is incompatible with use for air-borne disposal of waste fluorides from the manufacture of super-phosphates. This interdependence situation became known in Florida in 1955.1 Apparently there was no legislation that specifically placed ownership of this dimension of land resource in the hands of farmers or phosphate manufacturers. Therefore the effective right accrued to whoever could physically appropriate the resource. Appropriation was done by the manufacturers to the detriment of the farmers.
What ultimate resource use might be expected under this set of property rules? The farmers could make market bids to the manufacturers to reduce the pollution. It seems reasonable to believe that the transaction costs of such a contract would be high relative to its value. Exclusion costs would be high, and free-rider behavior might be expected. (In the large-number case, it would be the "unwitting free riders.") It would be costly for farmers to monitor the agreed upon waste disposal level. Even if the farmers owned the resource, the transaction costs of making that ownership effective were high. The farmer would have to prove which particular plant caused the damage and prove that any observed reduction of farm production was in fact due to the pollution. Court costs would be substantial. Information costs were formidable. Hypothesis: In a situation of high exclusion and information costs, a market structure with factor ownership by industry will result in a performance of the resource being used by industry with few bids from farmers. In fact, it was observed that farmers made no bids. In turn, phosphate companies owned only 50 percent of the land affected by their operations in 1955. Court suits were filed.
In 1958, an air pollution control district was funded by government. The exact character of the rights change is not clear, but "the companies were told to buy up those lands subject to pollution damages or face the prospect of having to get their emissions down to what was termed the 'minimum technologically feasible level" (Crocker 1971, pp. 456-57). Apparently, the companies were not given the option simply to pay damages (see Chapter 6). This detail is critical to performance, as shall be noted below; it is not sufficiently descriptive just to say that ownership of the resource changed.
The companies now needed to acquire an input that they did not own. They had two major production options. They could install control equipment or purchase the land, whichever was cheaper, assuming they had a cushion of profit out of which to pay these new input costs. Apparently they chose a mixture of these inputs and increased the ownership of the land affected by their pollution from 50 to 80 percent and decreased the volume of fluoride emissions.
The relationship between exposure to fluoride and farmland productivity is a physiological one. Once people understand it, the relationship between land value and pollution should be constant over time, other things being equal such as the volume of pollution. How is land value affected by ownership? If farmers own the marketable right to be free of damage, their money income is the same whether they receive their net income in the form of current damage payments reflecting the present value of future net income or normal unpolluted farm returns. The damage payment amounts to rental of a portion of the land's productive capacity. If an equation explaining land values is constructed with a pollution variable, the regression coefficient for the amount of pollution suffered by different parcels should be statistically nonsignificant.
If farmers do not own the right to be free of damage, they have a differentially productive asset depending on their exposure to pollution, just as if fertility differed. Hypothesis: The regression coefficient for the pollution variable in this case should be significant and negative. An equation can be estimated for each year before and after the change in property right occasioned by the creation of the air pollution control district.
Such a regression equation has been estimated by Thomas Crocker (1971) for citrus land. The dependent variable is land value. The equation includes a pollution variable as well as variables for tree quality, size, and land improvements. The regression coefficient for pollution is generally not significant prior to the change in property right and is significant and negative after the change, contrary to the hypothesis above. Unfortunately, with correlation as the quasi-experimental design, all other possible explanations cannot be ruled out. The nonsignificance of the pre-rule pollution coefficient might be due to lack of awareness of the causal effect between fluorides and citrus production. Damage is suffered, but no one is sure why. This awareness and the change in rules occurred together. While the potential causal effects of some variables are controlled in the regression, some remain unknown. The post-rule significant and negative coefficients need explanation. While the phosphate plants did acquire more land and did reduce emission, it is possible that some pollution remained uncompensated. Apparently the rule changed, but its administration was such that citrus growers did not have the effective right to be free of all damage. Perhaps what one observes is only a nominal rule change that should not be expected to produce a nonsignificant pollution coefficient. The size of the coefficients did decrease over time.
Crocker estimated another equation for pasture land. For the four years examined after the rule change, the coefficient for the pollution variable was significant and positive. Both the significance and the sign are contrary to the above hypotheses. This is where the form of the rule change may be critical. Somehow the ranchers were able to sell land to phosphate companies at more than agricultural rent. What right might explain this superior bargaining power? The phosphate firms, up to a point, find it cheaper to buy land than to reduce emissions. The capture of this saving is open to negotiation between phosphate firms and ranchers. If the firms had the right to use the air resource and had only to pay court-determined damages to the ranchers, the firms would capture all of the cost saving. As noted in Chapter 6, this arrangement amounts to private condemnation. But, if the firms must negotiate to secure a willing seller, they will have to share the cost-saving difference between using air as a fluoride disposal input and more expensive techniques. In fact, since the phosphate plants have immobile assets limited in location to raw material sites, the ranchers have great bargaining power. The strategic holdout has even greater power: land that was worth $150 per acre for ranching often sold for $250 to phosphate firms under the injunction rule. Alternatively, a court would base damages on the $150 value. This evidence suggests that the income distribution consequences of the right to receive court-determined damages and the right to be free of damage along with the right to be a willing seller are quite different. The seller with a choice has the right to injunctive relief as noted above in Chapter 6.
Why the different impact on citrus and ranch land values ? Why could not citrus growers also extract premium payments from the phosphate firms? Why in fact did they continue to suffer damages after the rights changed? Crocker notes that some citrus growers did receive premium payments but offers no evidence explaining the difference between the regression results of the two types of land. In another connection, he notes that in general the pasture sites were closer to the phosphate plant locations than were citrus sites. Could it be that the citrus growers had higher transaction costs in proving that they were damaged by a particular plant? Other explanations may lie in the way that air resource rights were administered. Is it possible that there is some reason for ranchers to be better organized and successful in getting their problems taken care of by government? Perhaps the air control district did not put as much pressure on plants to buy damaged citrus lands as it did for ranch land. Citrus land is considerably more expensive than ranch land, and phosphate firms may have found it cheaper to put in pollution control equipment rather than buy land. But why did some damage persist? Perhaps the citrus growers faced high transaction costs in protecting their rights.
Another type of property right is possible. In 1964, the state abandoned the above rules and instituted emission standards. It can he hypothesized that since the land-disposal input is prohibited, the premium paid for ranch land would cease. In fact, in 1965-66, the pollution regression coefficients were statistically insignificant for both citrus and ranch land. The agricultural users received less pollution but lost the right to bargain for land sale value premiums with phosphate firms. The firms supposedly incurred greater costs as they used more expensive disposal resources.
There is one other performance difference that should be noted. It is possible to design rights to receive damage payments (fees) and rights to emission standards so that the agricultural landowners' wealth positions are similar with each type of rule. But emission standards also give rights to people who use the air resource but are not landowners.2 One such group are tourists who pass through the area. If a phosphate owner reaches a market agreement with the landowner that results in use of the air for waste disposal, the interests of the tourists do not count. Those who want to pass through the area and to breathe clean air could also be given damages or injunctive relief, but the transaction costs would be high. The cost of court suit relative to individual benefit would prevent individual unilateral action.
There are many aspects of institutions relating to water use that might be subject to empirical investigation. Only one will be noted here--namely, the consequences of achieving a given level of water quality via regulatory standards versus charges or user fees. Economists are nearly united in their preference for user fees. This preference is based largely on theoretical deduction of comparative efficiency rather than any systematic empirical observation of the performance of different institutions.
Theory of the firm calls attention to the least-cost combination of resource inputs in producing a given product. Many individual products involve production of wastes. These may be disposed of by using streams or by various treatment technologies. Most research in this area presumes some political process that decides on the level of water quality and the institutional question is simply one of asking if the rules permit the given quality level to be met at the lowest cost (Freeman, Haveman, and Kneese 1973, pp. 112-15). Usually some boundary is assumed such as a river basin. This conventional formulation of the problem will be illustrated and discussed below.
Consider a basin where only two industries are located. Firm A because of its particular product and available technology produces 800 pounds of waste per day and is faced with a marginal cost of reducing waste to the stream by one pound to be 3 cents. Firm B has a marginal cost for waste reduction of 15¢ and produces 1,600 pounds of waste per day. Initially 2,400 pounds of waste are released to the stream. Suppose that government decides that the performance goal is 1,600 pounds. This can be achieved by a regulatory standard that orders each firm to reduce its waste discharge by one-third. What is the cost of achieving this standard? Firm A now reduces its waste by 267 pounds at a marginal cost of 3¢ for a total firm cost of $8. Firm B reduces its waste by 533 pounds at a marginal cost of 15¢ for a total firm cost of $79. Total cost to meet the river basin performance objective and reduce waste by 800 pounds is $87.
This cost can be compared to that of the fee approach. Suppose that after some experimentation or experience from other areas, the government finds that a charge of 3.5¢ will result in the two firms adjusting their production processes so that a total of 1600 pounds of waste is discharged. (If the government knows the marginal cost schedule of waste removal for each firm, it can compute the charge that will reduce pounds discharged to any given level; if not, it can discover the appropriate charge by trial and error.) What is the cost of achieving the same water quality with the fee approach? Firm A when faced with a fee of 3.5¢ per pound will treat all of its waste at the cheaper cost of 3¢ and no waste will be discharged. Firm A's cost is .03 x 800 pounds equals $24. Firm B will find it cheaper to pay the discharge fee than to reduce its wastes. The cost to Firm B is 0.35 x 1,600 pounds equals $56. Total basin cost of achieving the performance objective is $80, which is cheaper than the $87 cost of achieving the same performance via a regulatory rule requiring a uniform percentage reduction for all firms. This difference would not always occur but does when treatment costs facing the firms differ widely. With the regulatory standard, a high-cost-of-treatment firm is forced to use expensive treatment technology rather than the cheaper assimilative capacity of the stream. The conventional analysis stops at this point and concludes that the institution of effluent charges is more efficient than regulation (Dick 1974, pp. 67-70).
The above analysis focuses on total costs, ignores their distribution, and thus leads to a policy recommendation with differential effects on different kinds of firms. Additional perspective can be gained by looking behind the economics of resource combinations to the implied property rights and opportunity sets (Burrows 1980). A regulatory standard implies that each firm owns some right to dispose of some quantity of waste in the stream. This is also implied in the doctrine of reasonable use. As the standard is increased, the amount owned by each firm diminishes and the amount owned by third-party environmentalists increases. If the application of the standard requires each polluter to reduce by a given percentage, the relative amount owned by each firm is constant and proportional to its preregulatory use.
An effluent charge changes this relation. A charge assumes that the resource is totally owned by the government. If a charge is set to achieve a certain level of water quality, firms are thereby differentially treated. Recall that the charge means that the firms with higher treatment costs are allowed to substitute cheaper assimilative capacity of the stream for the higher-cost treatment. But where did they get the right to this stream capacity relative to the rights held under the prior rule of reasonable use? From the firms with lower cost treatment. While total cost of treatment can be less with the charges approach, Firm A's cost increases greatly (from $8 with regulation to $24 with charges). This increase represents a redistribution of income between the firms as well as between the firms and third-party environmentalists. Thus, the manner in which environmentalist rights are expanded affects the relative welfare of each industrial firm. From this analysis, one might hypothesize that firms with low treatment cost would prefer the regulatory approach and high-cost firms the effluent charges approach. This hypothesis should be subject to empirical test. (This is quite a different hypothesis as to the demand for the two types of institutions from that offered by Buchanan and Tullock 1975, pp. 139-47.)
If transactions were costless, we might expect that resource combinations would be unaffected by the choice of institution. Under regulation, if Firm A owns a certain portion of the assimilative capacity of a stream, it will sell it to Firm B whenever Firm B's offer price exceeds A's cost of treatment. Firm B will be able to substitute the cheaper resource input obtained from A rather than utilize its own more costly treatment. But, in this case, A's welfare is enhanced because it is paid for the resource instead of suffering a redistribution.
In the eastern United States, where riparian water law is dominant, rights to assimilative capacity (or any use rights) are not transferable. Even if they were, it would be costly to exclude waste disposal users who did not pay for the resource. It is one thing to police water quality below a firm's discharge point and determine if the reduction in quality compared with the quality above that point is within allowable limits. It is another to determine who is using the increased capacity made available when a former given user ceases and sells out. In short, there are high transaction costs for Firm B to seek out firms that might want to sell, and there would be high policing costs to keep other firms from using the resource. It can be hypothesized that regulation and charges are not likely to achieve the same resource use by industrial firms.
When the rights of third-party environmentalists are expanded, a conflict arises between two industrial firms in the sharing of the costs of redistribution. Thus, there is a conflict between achieving equal treatment between two firms and achieving lowest costs of attaining a given stream quality for both firms considered as a single firm--a conflict that would not occur if transaction costs were zero.
The above analysis is illustrated with reference to hypothetical cost data and can be extended to observation of the costs facing actual firms in a given river basin (Johnson 1967, p. 297). Such research is empirical, but whether it is empirical institutional research is another question. The research does not observe performance under two different institutions. We cannot be sure that firms will follow the orders of a regulatory body or that the quantities charged for are the quantities of waste actually discharged. While regulations and charges could be designed to obtain the same water-quality level, some interest groups may think that they could more effectively influence a regulatory body proceeding under historical regulatory procedures than a new body charged with leasing the public's water.
Arguments over the form of business organization have an ancient history. Some of
the alternatives include the degree of separation of ownership and control, stock versus mutual or cooperative firms, public versus privately owned firms, and capitalist versus worker-owned firms. Other organizational variables involve internal arrangements of individual and team work, supervisory relationships, worker participation in planning and management, partnerships (Leibowitz and Tollison 1980), and forms of productivity audit (Evan 1971). For a review of literature on how organizational structure affects the resolution of conflicts within multi-person managerial groups see Leibenstein (1979).
Manager versus Owner-Dominated Firms
Consequences of the separation of ownership and control constitute a classic issue at least since Adam Smith, who worried about the stewardship of the directors of joint stock companies (1937, book 5, chap. 1). The individual proprietor could easily see the relationship between effort and personal income. But, in the large corporation, the welfare of the manager may not be closely related to the income of the stockholder. The dispersal of ownership is hypothesized to make it difficult for stockholders to select managers who best serve stockholders. Decision and information costs work against stockholders exercising their nominal rights to determine who manages the firm. (A contrary hypothesis is offered by Armen Alchian 1969.)
It is usually assumed that the stockholder has a singleness of purpose centering on dividends and profit rates, while the nominally hired manager is served by wages and other perquisites of position including the easy life. Most studies of the consequences of the degree of owner domination focus on profit rates, testing the hypothesis that the interests of the manager are served at the expense of firm net return. While this focus seems reasonable, it should be remembered that stockholders have other interests as well. This fact was made clear in the United States during the Vietnam War, when various organizational holders of stock were forced by their members to ask if their stockholdings were in firms that made certain war materials. Fama (1980) argues that managers are disciplined by the market for their services and not directly by participating in profits. But, how do other firms get the information to know who to bid for?
Frederic Pryor (1973) defines a firm as under management control if no single individual or block controls more than 15 percent of the stock. Pryor estimates that 76 percent of major U.S. firms in 1963 were under management control. Various empirical estimates have been made on the consequences of the degree of managerial control with conflicting conclusions. David Kamerschen (1968) used multiple regression analysis to test for the relationship between managerial control and profit rates. Variables held constant included several related to market power, growth rate, and size. Managerial control had no significant effect on profit rates. Another study using a similar technique for another data set found that managerial control had only a small influence on profits (Larner 1970). Still another study using a three-way analysis of variance found that manager domination led to a profit rate 5.5 percent less than owner-dominated firms (Monsen, Chiu, and Cooley 1968). Industry type was held constant.
John Shelton (1967) compared profit rates of identical restaurants in a large chain. Profits were higher where the restaurant was operated by the franchise owner rather than by a hired manager. Casual observation suggests that outstanding cuisine is found only in small, owner-operated restaurants as contrasted to chains or large establishments with much hired help, though a consistent quality can be found in some of the chains as contrasted to the uncertainty of the local greasy spoon.
A variety of the separation of ownership and control phenomenon is contained in the mutual firm that is nominally owned by the customers. In a mutual savings and loan association, each saver may own one share with voting rights. The management is little restrained by formal voting of the nominal owners in either mutuals or stock firms with dispersed ownership. There are some differences. In the mutual, the shareholding customer does not receive a dividend but obtains benefits in the form of services and prices not available to customers of nonmutuals. Shares are not marketable so that it is not possible to capitalize future earnings (see Chapter 8). It can be hypothesized that mutual shareholders have less incentive to use their time to influence managers to initiate cost savings that would create long-term net income since these future earnings are like high exclusion cost goods and cannot be captured in present values in the market. One must stay with the firm to reap the benefits of any future income gains.
A comparison of mutual and stock savings and loan associations has been made by Alfred Nicols (1967). Performance variables included growth in assets, market share changes, expenses, gross operating income, allocations to reserves, and management nepotism. Some of these variables were tested in simple comparisons of averages between stock and mutual firms. Nicols also estimated a regression relating expenses to loan and savings activity. "When the California and Los Angeles stock data with respect to these variables are related to the regression coefficients, a mutual area with the same product mix would have had expenses of 62.6 percent .... higher than what was actually achieved by the stock associations" (p. 342). Another major finding was a slower rate of growth of the mutuals.
If mutuals do not pay a residual profits dividend to shareholders, one might expect them to pay higher interest to their saver members. On the basis of a limited two-year observation of mutuals and stock firms in the same states, this was found not to be the case. The stock firms paid a higher ratio of interest to savings capital. If we hypothesize that capital mobility limits manager discretion in stock firms, it seems logical to expect that customer mobility would limit managers in mutual firms. How do mutuals maintain their customers?3 This question raises a further question as to the relevant performance variables. Do the customers of mutual firms find some product there that they do not find in stock firms? This question remains to be explored.
Anyone with statistical training can compare the performance of alternative types of business organization. But theory is necessary in order to know what independent variables to control for. Production and marketing theory suggests that we control for product mix, firm size, and market power variables. Institutional theory suggests that we control for certain other human interdependence variables. Is the observed difference in mutual and stock firm performance due to the separation of ownership and control or to the difference in marketability of rights to profit residuals or to some other institutional factor? Is it possible that the effect observed is due to degree of ownership dispersal rather than share marketability? This question remains to be answered.
Adequate institutional theory is critical if we wish to search for changes in one property rights alternative to make it perform more like another. Is there some dominant feature of the mutual firm's structuring of opportunity sets that accounts for performance, or could a small change be made in one feature to change performance? Test of this cannot be conceived without an adequate institutional theory. Too often we are content with a plausible explanation for observed differences when the hypothesized cause itself has not been tested by separation from other potentially causal features. For example, it might be instructive to compare credit unions with mutual and stock savings and loan firms. This comparison is suggested by concepts of the role of status transactions and the sense of community. It was noted in Chapter 3 that learned behavior habits can substitute for the incentive otherwise supplied by exclusion from benefits. Credit union managers are exposed to national organizations and a history of cooperative ideology that may not be present for mutual managers. This fact might give some indication whether the observed mutual performance is inevitable or whether some factors might be changed to alter performance.
One of the problems in applying the results of empirical research is differences in scale. In experimental design, this is called a "threat to external validity." The Nicols study suggests the performance of mutual firms in an economy dominated by the stock corporation form of ownership. What would be the case if most firms were mutuals or cooperatives? Would they behave differently? Do we have any theory to guide further inquiry? One factor explored above was learned cultural habits. In a world dominated by mutual firms, these learned common expectations may be easier to create and may be sufficient to obtain cost-saving behavior. And, then again, this culture may be difficult to establish.
Since the purpose here is not to argue for any conclusions on the performance of institutional alternatives, the above is sufficient to demonstrate the type of hypothesis subject to test with a variety of quasi-experimental design methodologies. It is left to the reader to think whether all of the relevant variables have been controlled. From the list of types of human interdependencies in Part II, is there any reason to think that the ability of stockholders or customers to limit the discretion of managers might vary among different industries ? What affects information costs and thereby the mobility of capital or customers as an incentive for managers to perform in the stockholders' or shareholders' interests? When these latter interests conflict, whose interests count under alternative property rights?
Private versus Public Ownership
The theoretical hypotheses for differences in private and public ownership have already been suggested. In a sense, the public firm is a version of the separation of management and control. The essence of the conventional hypotheses is that, under public ownership, the costs and benefits of a decision are less fully borne by the decision maker than under private ownership. Or is it that the two systems may serve different owners rather than serve an undifferentiated set of owners more or less efficiently? This latter point will be pursued by raising questions about the performance variables used.
Ideal comparisons of public and private firm performance would utilize firms of the same size, producing the same product, and in the same country. Alas, natural experiments of this type do not often present themselves for analysis. Few countries have major industries that are composed of both private and public firms. (Some time series analysis might be possible in the United Kingdom, where firms are alternately nationalized and privatized with political party victories.) A comparison of socialist and capitalist countries is not a good test of alternate ownership of firms because it is also involved with a different relationship among firms as well.4 In the countries without strong central planning, public ownership is largely limited to utilities and transportation. As discussed in Chapter 4, these industries are marked by extraordinary economies of scale, and most countries prefer government regulation even if the ownership of the firms is private. Thus, the available comparisons are fewer than the amount desired to test the difference in private and public ownership since both involve direct government regulation of price and often of service. There is also the case of mixed ownership (Seidman 1975). Nominal ownership via public members of board of directors means little. However, co-determination with labor on corporate boards makes some difference (Teubner 1985).
Australia provides an opportunity to contrast the performance of a privately and publicly owned interstate airline. The government regulates prices and routes and allocates needed capacity equally between them. They have nearly equal equipment and wage rates. Per employee performance measures were tons of freight and mail carried, the number of paying passengers carried, and the revenue earned. A simple comparison by David Davies (1971) of the two companies over a number of years shows that the private company was more efficient with respect to these measures. A question might be raised as to the comparability of outputs, such as timeliness of service or labor relations.
The United States provides another example with its regulated privately and publicly (municipal) owned electric utilities. A commonly used experimental design is to formulate a regression including private or municipal ownership as a dummy variable and including variables that production theory would suggest might affect costs and outputs such as size of firm. The results of various studies have been summarized by Louis De Alessi as follows:
Municipal firms, relative to privately-owned regulated firms, in
general will: charge lower prices; have greater capacity; spend more on plant construction; have higher operating costs; engage in less wealth maximizing price discrimination, including fewer peak-related tariffs; relate price discrimination less closely to the demand and supply conditions applicable to each group of users; favor business relative to residential users; offer a smaller variety of output; change prices less frequently and in response to larger changes in economic determinants; adopt cost-reducing innovations less readily; maintain managers in office longer; exhibit greater variation in rates of return. (1974, p. 36)
The theory behind each of these results will not be reviewed here again except to note the theory related to rate structure. As noted in Chapter 7, different groups of customers have different demand elasticities, and separation of these groups is relatively cheap with respect to electric service. Business customers can be charged a different rate than residential when resale is relatively difficult. Massachusetts voters in 1976 turned down a referendum to prohibit industrial and residential rate differentiation. Price differentiation increases total revenue for any given level of output. As noted above, there is empirical evidence to support the belief that, if public managers have less incentive to maximize profits, they will practice less price differentiation than private managers. (This hypothesis is complicated by regulation of total returns. Regulatory bodies have traditionally given more attention to controlling the overall rate of return than to allocation of charges among different users. This focus is now changing.) Is this difference due to a simple lack of incentive to maximize returns, or is it in response to a different set of "owners" whose concept of fairness does not permit all the discrimination that is possible? The answer to this would require more elaborate tests.
The results summarized above were obtained to a large extent from regressions containing a dummy variable for private or public ownership, which is a very gross distinction that hides many variations in rules within each. In the earlier discussion of different kinds of transactions, the point was made that the market is not a single alternative. Many alternative rules and opportunity sets are possible within the market-bargained transaction. The same is true for administrative transactions. Much work remains to be done in specifying the institutional variable. Some progress has been made in this regard in a study of private and public electric utilities by Marc Roberts (1975). He focused on environmental practices as a performance variable. He found no consistent differences in these practices. Some private firms had clean stacks and protected the environment while some public firms did not and vice versa. His sample was comprised of three public and three private firms. Roberts suggests that performance is not determined by any simple private-public distinction but is a function of a complicated set of differences in history, fuel prices, external pressures, the strength of internal control mechanisms, and the balance of group perspectives within each organization (pp. 425-26). Alternative property rights apply not only to different groups of the public but also to different groups within the firm. These rights are not necessarily heavily circumscribed by public or private ownership. The evidence is mixed. Roberts finds slack (ability to serve manager's personal preferences) regardless of private or public ownership while Nicols in the savings and loan firms does not.
Roberts does not regard the firm as a single profit-maximizing or utility-maximizing central computer. The following gives a flavor of the type of hypothesis he suggests: "In electric utilities, power plant operators value plant reliability. Hence, the forced outage rate of a system's facilities will be inversely related to the seniority and influence of the company's chief operating official and to the percentage of the organization's technical personnel who were operators at the time the plant was designed" (p. 426). This type of institutional data on administrative and status rights cannot easily be obtained for a large sample of firms. Roberts demonstrates the utility of the case study approach where a large number of subtle institutional variables can be obtained, but only for a few firms. A similar approach has been taken by David Granick (1974) in a study of labor productivity as related to managerial performance in France, England, and the United States.
In the United States, fire protection is provided by municipal departments. One exception to this in Scottsdale, Arizona, presents an opportunity to test the performance of public and private firms. The city collects taxes and contracts with a private supplier for the service. Private contracting between individual homeowners and firefighting companies was practiced in colonial America but has now disappeared. An analysis of this transformation from individual contract to municipally owned departments serving all might be instructive particularly of the role of transaction costs. The primary impact of fire is damage to the individual property owner. If you do not buy the service, it is relatively cheap to exclude you from the service. However, if your neighbor does not buy protection, you may also be exposed to damage if the fire spreads. This could be accounted for by rules of liability, but most countries choose to have a tax-supported service equally available to all.
A comparison of the privately owned firefighting company of Scottsdale and municipally owned departments has been made by Roger Ahlbrandt (1973). He hypothesizes that the private company will provide lower-cost service. Production theory suggests that cost of service is related to size and wage rates. It might be possible to find several cities with public departments that are of the same size and wage level as Scottsdale, which would allow one to test the effect of ownership. A more sophisticated methodology was used by Ahlbrandt. He estimated a regression equation with data from 44 municipal departments with per capita costs as the dependent variable. Independent variables included population, area, assessed value, housing condition, wage index, equipment, and the fire insurance rating index. (One additional variable was a dummy for paid, volunteer, or mixed departments. This might be interpreted as an institutional variable affecting some aspects of performance, but it is used by Ahlbrandt as a variable affecting factor costs.) The insurance rating variable gives statistical control to output quality, since cost might be expected to vary with quality. This estimated cost function allows Ahlbrandt to plug in the appropriate variables for Scottsdale and derive what the cost would be if it were under municipal ownership. This cost can then be compared to the actual cost of the privately owned company. Such a comparison shows a 47 percent cost saving to the advantage of the privately owned company. In effect this figure represents a comparison of the cost for the average public firm with one private firm. It would be useful to note the variability in costs of public firms of the same size, and so on. Perhaps the skills of the managers are not held constant or the institutional alternatives affecting incentives are not fully specified by the public-private dichotomy.
The private company has made two key choices not typical of public departments. The private firm has made better use of new technologies developed by public research (U.S. Navy) and has built a new design of cost-saving trucks. It also has an arrangement with the city to utilize other city employees with additional pay to supplement the full-time firefighting employees. Firefighting is a service with peak loads, and employees are not utilized continuously. This cost saving, however, has not entirely escaped the notice of other city governments, but it has been resisted by firemen's unions. Failure to utilize police and other city employees as supplemental firefighters is a property right enhancing the employment of full-time firefighters.
There is some trade-off between cost saving and fire damage risk. City employees who are called to aid full-time firefighters may create some time delay in service. Some citizens may prefer the trade-off available in Scottsdale, rather than the one chosen by the typical city council and fire chief. One can imagine conflicting preferences in the same community. In any case, the researcher must exercise care in making cost comparisons. Difference in institutions sometimes means difference in product. One quality of the good preempts another.
It was hypothesized that private firms could teach disadvantaged children better than the usual public schoolteachers if payment was tied to learning gains (Gramlich and Koshel 1975). The performance variable was reading and math achievement. The experiment involved six different companies at 18 sites. The experimental design utilized matched groups of low performing students. The companies used a variety of teaching methods. The results were equivocal. In areas where it is not clear what the best production methods are, an institutional evaluation of final performance is ambiguous. Do the results prove that private performance contracting does not work or that the educational production function is unknown regardless of institutional incentives? Most of the cases examined here involve formal legalistic rights, but the concepts can be applied to informal rights as well.
Human Blood Supply
One of the most precious goods in any society is human blood for transfusions. It can be transferred from producer to user under a variety of types of transactions. Various countries have different systems for provision of blood. Data on these systems have been assembled by Richard Titmuss (1971). No systematic experimental design is possible, but some evidence nevertheless emerges from international comparisons over time. Blood has some special characteristics as a product. It can be a carrier of a killing disease called serum hepatitis. However, the quality of blood is difficult to determine. Information costs are thus high. The most reliable determination of quality is the truthful reporting of medical history by the person producing the blood. Thus, the motivation of the individual from whose vein the blood is drawn in a critical factor in the health of the person who needs a transfusion. Japan represents countries who rely heavily on the market transaction to produce blood. Most of the blood is bought and sold. It is hypothesized that, in market transactions, there is a minimum of feeling of community between the parties. The producer is motivated by the possibility of payment and not the needs of the buyer. Thus, it is possible that people who need money will sell blood and not reveal information about themselves that could determine its quality. The data reveal that Japan has one of the highest rates of post-transfusion hepatitis in the world.
An alternative relationship is that of the status-grant transaction, which Titmuss calls the gift relationship. Here the producer is not motivated by the exchange of goods but by an identification with the recipient. Where the welfare of the recipient is uppermost, there is no motivation to withhold information. In Great Britain, blood is given by voluntary donors. The hepatitis rate is very low.
The holder of status (who needs blood) has a property right in a status-grant society just as the owner of income and factors has a right in a bargained economy. It is no less a right for being a matter of custom and cultural practice. The empirical data suggest that a person who needs blood in a society utilizing status-grant transactions has a more valuable right than a person who owns income to buy blood in a bargained exchange economy.
The United States is in an intermediate position with a mixed system. In 1965-67, about 30 percent of the blood supply was purchased while 10 percent was from voluntary donors with the balance some mix of payment and other inducements. Its hepatitis rate is also intermediate. (In passing it can be noted that one-half of the blood supply in the Soviet Union is purchased.) Studies of hospitals that switched from commercial to voluntary resources show dramatic drops in hepatitis rates (Sapolsky and Finkelstein 1977). But it should not be concluded that all volunteer blood is good and all paid blood is bad. There are other factors that interact with this type of transaction dichotomy to affect performance.
Again, questions can be raised about specification of the institutional variable. There are alternatives within the market system. A system of labeling might be instituted. Disease rates of different hospitals might be published so consumers can put pressure on hospitals to be more careful in choosing their source of supply. There are also alternatives within the status-grant system such as group size and type of sanction applied.
The Titmuss data do suggest a relationship between the type of transaction and blood quality. But it is another matter to predict how quantity supplied would be affected if a country now using a mixed system permitted only grants. This requires more than casual observation of psychological variables.
It would appear that the historical experience of a people may be important in shaping behavior within a grant system. If a grant system were relied upon entirely in the United States, is the sense of community strong enough for supplies to be adequate, as it is in Great Britain? Has the United States had the same experiences as Great Britain such as the common purpose created in the context of wartime survival under attack? The utilization of psychological variables at the level of national character is difficult, but smaller-scale experiments might be possible. For example, it might be instructive to determine if communities with volunteer fire departments produce a character of people with a greater sense of community; or the direction of causation may run the other way: If people have a sense of community, volunteer fire departments are easier to organize. Or, perhaps, the two feed back and reinforce each other. These are very difficult to untangle in a formal experiment, and judgments will have to be made from historical and case analysis.
Presence versus Absence of Direct Government Regulation
What the government can regulate is a question that must be asked even when a group does not like the results of market transactions. Just because the group has a reason to regulate does not mean that direct regulation can be made effective or that the group will want all of the associated consequences.
Inflation is a common feature of market economies and has characteristics of a social trap. One approach to controlling inflation is the rules of competition and monetary and fiscal policy. Another approach is price control such as that imposed in the United States in August, 1971.
The first phase of price control lasted 90 days and was an absolute freeze (while the administrative agency was being established). Phase II lasted for 14 months, during which selected price increases were allowed by the administrative agency. The testing of its effect might appear to be a simple matter of checking the time series record and comparing prices before and after the controls. But, in order to determine the effect of the controls, we must establish what would have happened in their absence. However, there are no data to be observed where the same conditions occurred in the economy with and without price control. The only alternative is to project and estimate what prices would have been without price control. This has been done in a study by Edgar Feige and Douglas Pearce (1973), who constructed a simple time series model that uses past rates of price change to predict future prices. More sophisticated econometric models were used by others to provide statistical control for a numher of other economic variables, but the general experimental design is the same. Feige and Pearce found that the actual observed price changes were less than the estimated price changes during Phase I for both the consumer and wholesale price indexes. In Phase II, the consumer index seemed unaffected by price controls and the wholesale index seems to have increased faster than was estimated. The analysts suggest that imposition of controls may actually have caused some otherwise forbearing sellers to raise prices in reaction to the controls.
When we leave aside the questions of experimental design, what can be concluded from tests where the institutional variable is a dichotomy between two very broad alternatives--presence or absence of controls in this case? Does the institutional variable need further specification? Previous discussion has suggested that public versus private ownership is too gross a distinction to capture all of the relevant property right variables that structure the incentives and opportunity sets of the actors. Before we can draw conclusions on the effect of price controls, we would need to test for the effects of alternative systems of incentives for the government administrators. How is their performance monitored, and how are costs and benefits assessed to the bureaucrats? Perhaps it would be useful to compare the administrative structure and procedural rules used in the wartime price control agency with that of the 1971 agency. There can be as much institutional variation within the broad category of price control as between price-control and so-called free markets.
To what extent must institutional models contain behavioral variables for those being regulated? To what extent are history and the current environment relevant in influencing how people will react to a change in the rules such as imposition of price controls? A possible hypothesis has already been suggested that the controls may have caused some firms to raise prices which would otherwise have not. The same kind of problem arises in the administration of traditional monetary controls by the Federal Reserve Board. The board does not influence the rate of investment and borrowing by direct controls. In theory, it can affect these rates by influencing interest rates. But, in periods of inflationary psychology, a rising interest rate may be interpreted as a signal that rates will increase even more and that one should borrow now, rather than being interpreted as a signal to postpone borrowing. Apparently neither economic models nor government policy has successfully incorporated these contextual psychological variables.
Consider the performance variables chosen for test. Feige and Pearce chose to restrict themselves to examination of the general price level. But inflation of the general price level is only a mild irritant if everyone's income moves together (Bazelon 1959, p. 107). Inflation is probably the most significant form of income redistribution at work in market economies. It may overshadow most overt changes in property ownership or redistribution via the tax system. Thus, an important performance variable is relative price change. Who are the groups whose real incomes are gaining at the expense of other groups? In this regard, the rules of access to the price controllers become critical.
Administrative transactions constitute one of the most direct ways that rights shape the content of opportunity sets. Prohibition of a certain action with appropriate fines as sanctions might appear to have an assured effect on performance. Yet the prohibition of liquor (1920-33) did not eliminate drinking, and contemporary fines for speeding and drunk driving have not eliminated these practices. One response is to increase fines and surveillance.
For example, in 1955 after an increase in highway accidents and deaths, Connecticut implemented a crackdown on speeders. The number of driver's license suspensions increased sharply. Analysis of accident rates using an extended time series design with neighboring states as "control" groups suggests that the greater degree of enforcement did reduce accidents (Ross, Campbell, and Glass 1970). In a sense, the increased fine (and probability thereof) can be conceptualized as a typical problem of estimating the price elasticity of demand for a product (fast driving). But the institutional question includes much more such as the level of administration. The law may be passed by the state, the state police may make arrests, but the cases are heard by a local elected judge, who may be responsive to local demands that differ from those expressed at the state level (Broder 1981). Thus, boundary issues affect how a nominal right is actually applied. The results cannot be extended to other states unless the institutional variable is carefully specified.
Another institutional specification problem arises from exclusion cost (quite literally, policing cost, in this case). Just because a person does not have the right to drive fast does not mean that exclusion from that activity is cheap. The fine level is one thing, but police resources are another. In the case of drunk driving prohibition, a complicating feature is the cost of obtaining information that proves a person was driving under the influence of alcohol if the driving was not actually witnessed by police. There are physical tests of blood and breath that can provide such evidence, but civil rights have often prevented their use (cf. contemporary drug testing). The effect of allowing breath tests for determining driving impairment potentially lowers information costs to law administrators. The impact on traffic fatalities was analyzed when these rights changed in Great Britain in 1967 (Campbell and Ross 1968).
To summarize, institutional theory suggests that there is more to regulatory performance than the nominal prohibition and fine level. Interdependence created by exclusion and information costs, for example, is controlled by other types of rights.
Competition and Market Structure
The emphasis of the book has been to supplement the institutional variables suggested by the model of pure competition. However, there is no denying the importance of the degree of competition as a variable controlling the distribution of power. This has been an area of much empirical work and the subject of an entire subdiscipline in economics --industrial organization. Since this subject has been so thoroughly analyzed elsewhere, it is not necessary to repeat any case studies here. For some examples, see Bain (1966); Philips (1971); Scherer (1970); Weiss (1971); Mueller (1983).
PUBLIC SERVICES: POLICE
The point has been made that the simple dichotomy of privately and publicly owned firms ignores many possible alternative rules within each type. In this section, variation of property rights within the public sector will be examined.
There has been a strong reform tradition in political science favoring the political consolidation of various units of local government in the United States. This tradition borrows the economist's concept of economies of scale and argues that the larger political jurisdiction will be more efficient (see Chapter 4). It also is related to the notion of enlarging the size of the firm to internalize externalities. These issues have been combined here under the concept of boundary issues as discussed in Chapter 8. What is the substantive result for different groups of people when they live in small political units as opposed to larger, consolidated units? One dimension of this has been tested with respect to police services by Eleanor Ostrom and associates (1973). Also see McDowell (1975) and Toma (1979). The performance variable chosen is citizen satisfaction with the service received and opinion on the desirability of political consolidation. Data were obtained by personal interviews. A comparison was made of a number of matched communities--one a small independent suburb and another a community within a large central city. The two areas were matched on the basis of the usual socioeconomic characteristics such as income, education, and racial mix. It was assumed that the two contiguous areas generally had the same set of preferences. The test was conducted in a number of cities, including Indianapolis, Indiana, and Grand Rapids, Michigan. In general, the studies show that the people in the small, politically independent suburbs with their own police force were more satisfied with police performance than were the people in the neighboring area who received the services of the central city police. The people in the independent suburbs generally opposed consolidation.
The source of the difference in satisfaction appears to be the kind of technology and service provided. The independent, small police departments tend to provide a highly visible service with many patrols, while the central city police are organized into specialized bureaus with less visibility. Are the differences there because different group interests prefer one type of service over another, or is the city service the only type possible, given a large department? The connection of the test above does not answer these questions. However, casual observation suggests that some large cities do put a great deal of discretion in management at the local precinct level, and there seems no technical production reason why a large department cannot have many patrols if that is how they wish to use their resources. An alternate hypothesis is that the preferences of largely middle class fringe areas of central cities are not met because the resources are used to meet the preferences of other groups. When crime rates are highest in the core area, an urban police department may choose to concentrate its forces there at the expense of its fringe areas.
What is empirically observed may have little to do with size itself but with the fact that larger size usually means a more heterogeneous set of preferences. The drawing of a boundary affects whether a given group is in the minority or majority. The middle-class fringe areas of a central city may simply be outvoted by the residents and business interests of the core area. This speculation might be tested by comparing the police service preferences of each of these groups. The property rights issue may turn on who is responsible for the costs incurred in the high crime areas. Should suburban tax dollars be used only for suburban service? The independent suburbs have been able to acquire this right subject only to some redistributive taxation and attempts to encourage regional planning. (See Hanf and Wandesforde-Smith 1972 and Schmid and Faas 1980.)
PUBLIC CHOICE RULES
There are rules for making rules which affect the outcome of public choice. What determines which group gets rights favorable to itself? One arena where this question can be asked is with reference to tariffs. A theory which pays no attention to situation might predict that there would be few tariffs since the losses to consumers are larger than producer gains because the consumer pays the tariff and the increase in product price due to smaller supply while producers receive only the benefit of higher prices. But, in fact, producer pressure is more prevalent and successful than consumer interests.
Jonathan Pincus (1977) suggests that there is an asymmetry of response to tariff possibilities. There are many consumers relative to producers, and the gain to any consumer from opposing a tariff is small relative to the cost. As shown in Chapter 3, this would make no difference if exclusion costs were low and the small but numerous bids of consumers could be aggregated. So Pincus tests the proposition that the degree of exclusion and associated transaction costs affect what products have the most tariff protection. Exclusion cost is less for smaller groups of producers of intermediate goods than for larger groups of producers of final goods. This is supported by a regression model explaining the relative tariff rates passed in the U.S. in 1824.
Producers of different products have different costs of communication, discovery of group interdependence and effectiveness of peer pressure related to the total number of proprietors and their dispersion. Pincus hypothesized that duties would vary positively with industrial concentration measured by market share and inversely with the geographic dispersion of sales by counties having the same kind of products, and this was confirmed by the evidence.
However, other measures of group size produce conflicting evidence. While industrial concentration was positively correlated with tariff rates, the number of establishments was also positively correlated, contrary to hypothesis. The Pincus model is not a test of the impact of alternative institutions combined with a given situation, but rather the effect of situational variation with a particular and largely unspecified institution. The intrinsic effect of high exclusion and transaction costs can be modified by structures. Casual observation suggests many examples where concentrated interests are more powerful than dispersed but individually small interests. Yet, the political rules place some limits on this. Further confirmation of the rules-situation interaction awaits the opportunity to compare tariffs for different products with different costs of exclusion in different countries with different public choice rules.
Environmental legislation is an area where dispersed interests in high exclusion cost goods have won over concentrated interests. Kalt and Zupan (1984) studied the Congressional vote on surface mining reclamation and inferred a substantial degree of institutional slack allowing political representatives to escape the concentrated mining interests and vote for what they conceived as a broad public interest.
Government settles conflicts among members of the public, and the rules of public choice influence whose tastes count. Just as factor ownership determines whose preference counts for incompatible use goods, the rules of public choice determines whose preference counts in making the rules of factor ownership. One rule making arena is that of the courts. Among the institutional rules affecting court decisions are those related to judge selection and boundaries. Smaller jurisdictions are more likely to have people of homogenous tastes than larger jurisdictions. A group which is in the majority in a small jurisdiction may be an outvoted minority in a large jursidiction. That is to say that a group's effective "ownership" and control depends on its boundaries.
There has been a movement in the United States to consolidate and enlarge local government units to achieve economies of scale (Warren 1966). The pricing issues created by economies of scale will not be explored here but rather the issues of changing product quality. A study of local courts in Georgia by Broder and Schmid (1983) recorded the satisfaction of residents with court decisions with respect to drunk driving and other traffic violations. Prior to consolidation, the largest city in the county had its own court and the Spearman rank correlation between fine levels for different violations given by the court and preferred by the city inhabitants was higher when they had their own court than after consolidation. The satisfaction of county citizens outside of the city increased. In fact, a regression discontinuity model showed that the fines given by the court for different violations were significantly different before and after consolidation. A change in the "ownership" of the court produced a different performance. The "efficiency" of achieving economies of scale is meaningless without the institutional context which determines the product quality which all must consume. The issue is always one of efficient for whom.
Civil Rights and Ballot Design
Some civil rights are a part of the rules for making rules. These rights determine who counts and participates in public choice of the rights that in turn determine who participates in resource-use decisions. The paradigm can be used to ask how a given right controls conflict growing out of a particular variety of interdependence. There are some striking parallels between the categories of human interdependence observed in participation in government and market decisions.
Chapter 6 illustrated the potential role of contractual costs in determining outcomes of market bargaining. The consumer's effective bid is reduced by the amount of the transaction costs, and this reduction often works to the advantage of other parties. Other areas that have been empirically studied are the determinants of campaign expenditures, political participation, advertising, and vote influence. (For a bibliography see Mueller 1976.) There are also costs associated with the exercise of voting rights. It takes time and energy to get to the voting place as well as to understand the available choices and their probable consequences (the same is true for consumer information in the market). Some very small differences in procedural detail can make a big difference in performance.
One factor affecting transaction cost is the design of the ballot. The party column ballot, often referred to as the Indiana ballot, lists each office and the given party's nominees therefore in a single column. A straight-ticket alternative is offered whereby a single mark gives the individual's vote to all of one party's nominees. This alternative saves time in the voting booth as well as time in acquiring candidate information. One can learn something about the general performance of the party and not bother to get information on specific candidates.
An alternative property right and opportunity set is contained in the so-called Massachusetts office-bloc ballot. The names of the candidates of all parties are listed after each office to be filled. There is no provision for voting a straight ticket with only one mark or lever. An opportunity to test the consequences of these alternatives was provided by Ohio in 1949 when the Indiana-type ballot was replaced with the Massachusetts-type ballot. The results have been analyzed by Donald Zauderer (1972). An extended time series quasi-experimental design was used. Observation of performance was made over the period 1934 to 1964 with the change in institution occurring in 1949. The dependent variable chosen was what political scientists call the percentage of vote roll-off. This is the difference in the number of votes received by the office that received the greatest number of votes and the office receiving the least, expressed in percentage terms. There is differential interest and information available on different offices. The office of U.S. senator may receive many more votes from a given area than does a local official's, say city mayor or county sheriff. The percentage difference is the roll-off. If everyone's subjective evaluation of transaction costs and net benefits were the same, then a change in their absolute magnitude might be of little interest. However, if some groups fail to participate in the voting for some offices, the candidate is chosen by fewer people of a different group. The roll-off can be large enough so that the number of voters voting for some offices (A) but not for others (B) may be large enough to have determined the winners of those B offices.
Zauderer found that generally the ballot design change, which increased transactions costs, resulted in a substantial increase in vote roll-off. When some wards and precincts were compared with differences in education and race, it was found that the increase in roll-off was progressively higher as one moved from upper to lower educational groupings, and within educational groupings the roll-off was higher in black districts than white. This implied that the Democratic party, which has proportionately more lower educated voters, was placed at a competitive disadvantage by the change in voter rules.
Apparently it is not enough to be in favor of democracy; where transaction costs are differentially perceived, the particular rules for an election seem to make a difference. Just as there are many variations of rules within what most would define as a market system, there are many alternative rules within a democratic government system. The differences within may be as great as between systems.
In the earlier discussion of goods with high exclusion costs, it was noted that government taxes and spending can solve the free-rider problem at the expense of creating unwilling riders who have to pay for something they do not want. Voting rules are important property rights affecting real wealth distribution. This fact is especially clear where one group can vote a tax that may be disproportionately paid by another. Several empirical studies have noted a divergence in voting patterns on school tax referendums between property owners and renters who were otherwise similar (Peterson 1975; Wilson and Banfield 1965; Sproule-Jones 1974).
American cities provide a wide variation in governance institutions which might affect whose preferences count. One cluster of representative structures might be termed Jacksonian democracy including an elected mayor, election of council members by wards in partisan elections, and long ballots. Another cluster is the Wilsonian structure including an appointed professional city manager, non-partisan council elections at large, and short ballots (Maser 1985, Miller 1985).
It can be hypothesized that the above structures would affect demand articulation by those with higher transaction costs such as the poor and uneducated similar to that already noted in the case of ballot design. The voter registration and turnout rate is lower for poor blacks which means that they are sometimes poorly represented even in cities where they have an absolute majority. With at-large elections, Blacks can be outvoted by whites who turnout at a higher rate. With elections by wards, exclusively black wards can be expected to elect a black representative even with low turnout. Similarly, non-partisan elections mean that voters must obtain information on individuals rather than just party ideology. This is hypothesized to favor highly educated elites.
Empirical research supports these hypotheses. Cities with Wilsonian structures hire fewer black employees than Jacksonian cities (Stein 1984). Cities with at large elections elect fewer black city-council members with respect to their numbers than cities with wards (Engstrom and McDonald 1981). In general, cities with Wilsonian structures have lower correlations between socio-economic variables and policies than Jacksonian cities and give more power to elites.
The above material is included to suggest the versatility of the paradigm and in no way exhausts the field. Access to government includes much more than voting rights. Some groups are more effective because they work through the courts or the bureaucracy (Viteritti 1973). One of the uses of the paradigm is to suggest alternatives to a group that finds its interests blocked in a particular avenue of power.
Some scholars reserve the term "public choice theory" to refer to alternative forms of government and such issued as voting, constitutional, social contract, decentralizations, and bureaucracies (Tullock 1979 and Mueller 1976). The term is used here to refer to the impact of rules for making rules as well as the resulting rules directing private transactions.
All of the various institutional alternatives noted above with respect to natural resources, business organization, public services, and so on combine to have an impact on general patterns of economic development. In this section, several studies of developing countries will be examined along with a U.S.-Japanese comparison.
There is a great deal of discussion on the consequences of a highly centralized, planned, and directed economy versus what is misleadingly called the "free market" economy (it will be made clear below why this is a misleading term). The histories of Kenya and the Gold Coast (now Ghana) during the period between 1900 and World War II provide a natural experiment to test the consequences of alternative institutions. Both were British colonies, but Kenya was marked by much direct intervention of the government in the economy that was absent in the Gold Coast. The following comparison has been made by Robert Seidman (1973).
Kenya was actively colonized by white settlers. The best land was kept for the whites while the Africans were forced to live on reserves. On the other hand, in the Gold Coast, whites were forbidden to own land. Factor ownership was thus restricted in both cases but to the advantage of different groups. Institutions with respect to labor, however, were vastly different. It was in the interest of the white settlers to obtain African labor at low wages. Short of slavery, it is difficult to attract workers from their subsistence agriculture without higher than subsistence wages. Instead, the Kenyan colonial government instituted a poll or head tax. The only way for most to raise this cash was to work for the whites at whatever they wished to pay. The wage bargain was nominally voluntary, but in effect the white employers had greater bargaining power as the Africans had few alternatives. The rights of labor were even more restricted by passage of master-servant ordinances that prohibited an employee from leaving a job without the employer's permission and enforced certain work standards by criminal sanctions.
In the Gold Coast, there was no direct state intervention compelling the Africans to work for whites. There was already an indigenous export-oriented agriculture on small farms, primarily growing cocoa. Whites did not try to obtain ownership of the land and make the Africans work for them. The Africans were free to sell their cocoa in the market. However, bargaining power was very unequal. The crop was financed, purchased, stored, and shipped by a few European firms. Their fewness led to frequent price-fixing agreements, which went unopposed by the non-interventionist government. The government, of course, protected the buyers' property in docks, storehouses, and so on.
From the discussion in Chapter 8, we might hypothesize that large groups of producers will face high transaction costs in organizing any group activity. These differences in costs between the few buyers and the many sellers can create unequal bargaining power. The producer owns his farm and is free, but the freedom is limited to selling at the buyer's price or not selling at all. Producers did organize withholding strikes in 1930 and 1937 during periods of dramatically low prices but were not able to sustain an effective counter to the buyers' market power. This is consistent with experience in agriculture the world over. Farmers have not been able to organize for collective bargaining without the aid of government in reducing transaction costs and the free-rider problem.
The two countries differed greatly in other aspects. The government of Kenya intervened in the economy with transportation subsidies, credit programs, research, price support, and marketing boards. The Gold Coast had little of these. For example, there was no government credit program, and producers were continually in debt to the cocoa buyers. The fact that the government did nothing to clarify confused and conflicting land titles that made the use of land for bank credit difficult may also have contributed to the cocoa-buyers' power, but this is not explored by Seidman. The power of a particular group is a function of a large variety of rights that individually may appear of little consequence.
What is the result? The performance variable chosen was average per capita income, which Seidman concludes was not substantially different in the two countries during this period. The profits of agricultural production went to Europeans and not native Africans (p. 569). Seidman's methodology is a simple cross-sectional comparison over a period of time. No statistical tests are possible. The performance variable may not be available for exactly the same years or in the same terms, and thus care is necessary in interpretation. The institutional variable is often multidimensional. One cannot be sure just which of the several property rights' differences may have caused the results, or even if one rule is working at cross purposes with another. Also just because we have a theory as to why a certain performance should follow from a given property right does not mean that when we observe the performance the explanation is proved. Performance may have been due to an uncontrolled-for simultaneous event. This is the reality of much institutional research. It is not possible randomly to assign different countries to different institutions. The great bulk of evidence of institutional performance will be this type of historical case study.
The results of such studies as that of Seidman do indicate that there is more than one way to achieve a given performance. The Seidman study suggests the need to specify the institutional variable as carefully as possible. Those who hope to achieve a changed performance via changed factor ownership (land reform) must be aware of the total institutional framework, which can offset the effect of change in factor ownership. The market is not a single institutional alternative. There are alternatives within it that can make it perform in a variety of ways including that identical to a given set of rules of an administered governmental system (and, probably, vice versa). These market alternatives are chosen by government; government is never neutral. The performance of the Gold Coast economy no less than the Kenyan economy was the result of government policy. The method can vary, but government choice is inevitable. The absence of government in doing nothing to help producers overcome their transaction costs or to prevent buyers from engaging in price fixing is no more or less coercive or interventionist than a head tax or subsidy. Both shape the real opportunity sets of the parties and thus the performance. Where interests conflict, there is no such thing as a neutral noninterventionist government.
The adoption of new agricultural technologies is the source of great increase in output. The ability of different groups of farmers to utilize these technologies has a great deal to do with relative income distribution. How do alternative property rights and their distribution affect technological adoption? An opportunity to test the performance of alternative institutions is provided by a comparison of Pakistan and Bangladesh (formerly East and West Pakistan).
The characteristics of the technology under consideration should be noted first to determine what sorts of human interdependence its use involves. Tube wells for irrigation can be installed by an individual farmer, in contrast to a large dam and ditch distribution system. Still, the tube has economies of scale (see Chapter 4). Asian farms are very small, and it is not practical for each farmer to have a well. There are several institutional alternatives to accommodate the character of the technology. One is a market in water where a large farmer would put in a well and sell water to smaller neighbors. Another is for small farmers to organize a cooperative and in effect become joint owners of the well.
Carl Gotsch (1972) has explored the institutional conditions under which either of the above alternatives is possible. He contrasts the institutions of Pakistan and Bangladesh. This quasi-experiment allows for control of relevant engineering and economic factors. Irrigable areas, costs, and crop response are similar in the two areas, but institutions are quite different. Pakistan begins with a more unequal size distribution of landownership than Bangladesh. There are many more farms that are of sufficient size to install their own wells. Since there are no organizational problems for large farmers, it might be expected that the rate of technological adoption would be faster in Pakistan. It was. If one were interested only in the number of acres irrigated, a small number of large farmers would be preferred.
But what of the smaller farmers? Does being small in a society of unequal distribution have different results than being small when others are also small? In Pakistan, 70 percent of the tube wells were installed by farmers who had more than 25 acres, while the great majority of Pakistani farmers have less than 13 acres and only installed 4 percent of the wells. A market in water did not develop. Gotsch suggests several reasons. In addition to the economies of scale, there are geographic factors that create a monopoly. A small farmer is not likely to be able to buy water from a number of potential suppliers. Without something like public utility regulation, the small farmer is faced with a monopoly price. The owner of the closest well also may not be in a position to offer credit. This might be solved by other market changes or by a government credit program, but it was not.
The role of malevolence cannot be ruled out. There is deep enmity between large and small farmers. The large farmer may not sell to a small farmer if it will make the small farmer better off, even though it also makes the large farmer better off.
If the market does not develop, what about the cooperative alternative. Gotsch offers several hypotheses as to why cooperatives did not materialize in Pakistan, but did in Bangladesh. Rural politics in Pakistan is marked by extreme factionalism. Gotsch observes that individuals in the village do not form alliances even though they have a common goal. Contending political groups are dominated by the large landowners (patrons), and they recruit followers (clients) depending on their alternatives (Coyer 1973). Those who are tenants or need the goodwill of a particular moneylender have little choice. Others are recruited by any means available, including force. Gotsch concluded that class alignment is virtually impossible.
In Bangladesh, factionalism and dependence are less severe. There are fewer tenants, and people are more equal. It seems doubtful from the earlier discussion of the problems of organizing large groups that these qualities would be sufficient for success. Government assistance in organizing the cooperative can be hypothesized as a key factor in overcoming transaction costs.
The ultimate performance variable was the distribution of increased farm income. This follows directly from the adoption of the new technology. Only the large farmers in Pakistan adopted the tube wells while many small farmers participated in Bangladesh. Possible cumulative effects should also be noted. New income from use of tube wells by small farmers may break their dependence on large farmer-moneylenders. Also, the experience with the production cooperative may also be utilized to organize for political action. Property rights operate at various interacting levels. One set of institutions (for example, tenancy and distribution in landownership) can affect the development of market and cooperative institutions, which in turn may affect the ability of different groups to obtain political power to change the rules of the game. Political and economic power are interdependent.
While this book has emphasized the impact of property rights on performance, it is not argued that property rights are always the decisive factor. Power is not the only variable. Differences in psychological variables may account for more difference in performance in this case example than differences in property rights. Of course, the two can be related and feed back upon each other.
Japan and the United States
Most of the studies noted in this chapter try to isolate the effect of specific institutional alternatives on a particular performance measure. They suggest that what we mean by economic development is controlled by a large number of rights as related to a large number of sources of interdependence. Since one might expect that there are many complements and substitutes among these institutions, the explanation of different rates of growth (however measured) among countries is a complex matter. Succinct specification of institutional structures is difficult and attempts to explain growth by reference to such aggregates as capitalist and socialist or planned or unplanned are difficult if not misleading. And, the interaction of situation, structure and performance often does not lend itself to neat econometric tests (or summary in a few pages).
There is a growing literature trying to identify the institutional and other sources of Japanese success and lessons for the U.S. (Vogel 1981, Johnson 1982, Thurow 1985, Solo 1984). It is not possible here to summarize or evaluate these studies. But, some of the debate on industrial policy can be better understood in SSP terms.
One of the things that is different in Japan compared to the U.S. is institutions which can make a long-run commitment to develop and utilize a new technology or product. Japan's Ministry of International Trade and Industry is designed to commit capital to a new area and provide supporting infrastructure. This is addressed to the transaction costs stemming from cost discontinuities and investment coordination. the Japanese have had success in reducing the uncertainty of suppliers who can achieve economies of scale only by being assured of a market for their output which in turn depends on achieving low prices. The Japanese firm will build large plants before the market develops thus pre-empting competitors. American firms on the other hand have short planning horizons partly because of the credit system and personal incentive systems internal to the firm.
Interdependence stemming from interdependent utilities and its effect on labor productivity is addressed by different rights structures in the two countries (Cole 1979, Aoki 1984, Ouchi 1981, p. 25). The difference in incomes of workers and managers is less in Japan. Job security is higher in Japan so that workers can sacrifice current income to help the firm compete and be sure they will be employed if the firm succeeds. The effect of these and other institutional differences are controversial and depend on qualitative and directional evidence, but it does remind us that there is nothing automatic about institutional change. The property rights structure makes a difference and there is no assurance that benign neglect of institutional design will maintain a country's economic position. The best argument that something better is possible in the fact that something better exists (even if its instrumentalities are uncertain).
This review indicates that empirical institutional analysis is both a workable and fertile field of study. Researchers need not restrict themselves to the traditional fields of research in production economics, monetary policy, or international trade in fear that institutional studies can only be polemical treatises of philosophy, history, and social criticism.
The review further indicates that the categories of human interdependence developed in Part II do provide a starting place in formulating useful hypotheses for relating institutional alternatives to performance. There is more to interdependence than is controlled by factor ownership and competition (or the right to vote). The key utility of this theory is in better specification of institutional variables. Research must move beyond the simple dichotomies of public versus private ownership, market versus nonmarket systems, capitalism versus socialism, or democracy versus oligarchy. Further theoretical development is needed in formulating models that can control for the relevant background institutional variables other than the main instrumental variables being tested. Such further work is especially critical where random assignment of the main institutional alternatives is not possible. With random assignment, other possible causal factors are controlled for (even when unknown) in the experimental design. But in quasi-experimental designs, the competing causal factors must be specifically accounted for. The researcher cannot know what to try to control for without adequate theory.
Adequate theory is not the only barrier to useful empirical research. One problem is the lack of institutional variation. In many countries, there is only one kind of institution or right applied to a particular commodity or governmental function. There are simply no institutional alternatives to observe. This explains in part why axiomatic and deductive reasoning is so popular. It is easier to logically prove that something is possible, impossible, or Pareto-better than it is to find two different institutions and accompanying performance to contrast.
Little reference has been made here to empirical research which does not identify instrumental variables which could be manipulated to change performance. For example, there is a large literature which has as its purpose to prove that both private and public decision-makers are self seeking constrained rationalists and that this is consistent with observed outcomes. Knowledge of modes of information processing and decision making are certainly necessary in hypothesizing how legal change in opportunity sets will be utilized. But, one wonders if the repetition of this work has another purpose. The Pigouvian literature of market failure seemed to provide a legitimacy for an enlarged government (more accurately a shift in whom government serves). If it can be demonstrated that politicians and bureaucrats are self seeking, rather than serving the public interest (whatever that is), this provides a government failure antidote to market failure. But, if one wants to escape the whole notion of presumptious failures and stick to substantive prediction of performance, this empirical typing seems less useful even if one were convinced that self-seeking was meaningful without asking how people define themselves.
There is another point about instrumental variables. Explanation has several implications. One can find an association between two variables, but if the explanatory variable is not instrumental and can't be changed, it may be intellectually satisfying but of little use to institutional design.
Morris Fiorina (1979, p. 49) critiques a study purporting to explain U.S. voter turnout over time by reference to information cost. First, Fiorina suggests that the dependent performance variable should be eligible electorate voting instead of percent voting among those of voting age. The authors utilized the proportion of households with radios and televisions as a proxy for information cost. They concluded that the availability of free information about candidates was positively correlated with voter turnout. Firoina suggests that many other variables would show the same correlation. If his preferred dependent variable of eligible electorate voting were used, "This series shows a general decline since 1868 and will correlate negatively with any time series which captures technological advancement (radios, television, rifles, airplanes, even deodorant), the rational theory of voting behavior to the contrary not withstanding." For all of our econometric sophistication, model specification and interpretation remain troublesome.
Let us raise our sights and not let our explanation stop short. What good is it to know that radios are correlated with voter turnout? Perhaps government could manipulate the number of radios to achieve given performance, but this is not too useful. However, if researchers go on to ask how institutions and rights instrumentally interact with a given information cost (or differential information costs of different groups) we may be of more practical use.
In that regard, Robert Rabin (1979, p. 991) says that "Documenting the proposition that law has an impact, without linking the findings to behavorial patterns and institutional change is a dead end." He illustrates with respect to the institution of a liability rule for charitable hospitals that were formerly exempt. A finding that such hospitals then increase room prices to cover liability insurance is not very surprising. More interesting would be such questions as, did the legal change cause changed safety practices in the hospital, cause cutbacks in service to different kinds of patients, or change the level of claims consciousness by patients. Hopefully, better theory will let us raise our sights.
The available empirical studies seem to suggest that prediction of performance can be made with knowledge of how rights affect opportunity sets in the context of a particular variety of interdependence combined with informal estimates of psychological variables. In other cases, the institutional analyst will require more formal estimates from behavioral scientists.
The cases examined primarily involve formal, legalistic institutions, but some cases suggest that the paradigm can apply to rights that are informal and embedded in the culture. In practice, one of the tough empirical problems is distinguishing the prevailing rule from the nominal.
The performance variables used in many of the studies seem quite narrow. They focus on some traditional items of interest to economists such as profit rates, growth in income, and measures of certain costs. But many people who are not satisfied with the performance of current institutions have additional measures in mind. Institutional researchers have a major interest in development of the field of social indicators.
The time scope of the studies noted here is short. Observation is usually limited to a few years of comparison. But, for some purposes, we want to know the performance of alternative institutions over many years of changing conditions. The test of many institutions is their ability to adapt over time rather than to achieve a given performance at any particular moment. Such adaptation is particularly relevant if some people are to avoid what they would consider a social trap. The point is made by S. V. Ciriacy-Wantrup (1967), who says that, in institutional research, "Emphasis is on determining conditions for economic growth rather than on locating peaks, on avoiding dead-end streets rather than on computing the shortest distance, and on adaptability rather than optimum adjustment." For example, he suggests that we should look at 80 years of experience in California's irrigation districts and compare that state's performance with other states and nations that have different rules. This presents formidable methodological problems, however. In this length of time, many peripheral institutional and other changes occur, and it is difficult to control for them. Nevertheless, this is the reality of what we have to work with. There is an historical literature relating institutions to performance that is too vast to review here. (See, for example, Hurst 1956 and 1960 .) Despite the statistical control problems, we shall have to learn what we can from it because it is all the evidence that is available in many instances; "the only alternative to seeking insights out of experience .... is to rely upon the absolutes of ideology, dogma, or revelation" (Parsons 1974, p. 737).
Where institutional variation exists (or has existed), there is an opportunity to learn from it and to derive some predictions of how the duplication of the institution might perform in another time, state, or perhaps country. Theory helps to suggest what situational variables must be similar for the same results to be expected. In many cases of interest, however, policy makers ask the analyst to suggest the impact of a new institution (right) that has never been tried in a particular community or problem area. At such times, theory is especially necessary to utilize experience gained in observing the application of a given type of right to a given species of human interdependence to predict how a similar type of right would perform in the context of a similar sort of interdependence, even though the particular commodities differ. Thus, for example, experience with one high exclusion-cost good is part of the ability to predict how a given right might affect performance with another high exclusion-cost good. The only real test, of course, is to try it; all any scientific theory can do is to allow the process of experimentation to proceed less blindly.
1. The following factual account is from Crocker (1971). The interpretation here departs somewhat from Crocker's.
2. Crocker (1971, p. 461) ignored this fact and made the erroneous assertion that "the subsequent imposition of emission standards was at least redundant and probably uneconomic."
3. Stock savings and loan firms are prohibited in some states. For example, in 1973 Michigan had no stock companies and had only recently passed a law allowing them. But this example is irrelevant to the above comparison, which is only made for each state where both forms exist.
4. There is a large literature comparing the performance of socialist and capitalist countries. This is well summarized in Pryor (1973, pp. 322-35). His general conclusion is that there is about as much variation in performance within each system as between. Also see Wiles (1977).
If you have any questions or comments, please email firstname.lastname@example.org
Back to Main Page