Southern Economic Association is collaborating with JSTOR to digitize, preserve and extend access to Southern Economic Journal.

Transcription

1 Public Sector Expansion: Stagnant Technology or Attenuated Property Rights? Author(s): W. Mark Crain and Asghar Zardkoohi Source: Southern Economic Journal, Vol. 46, No. 4 (Apr., 1980), pp Published by: Southern Economic Association Stable URL: Accessed: :27 UTC Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at info/about/policies/terms.jsp JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org. Southern Economic Association is collaborating with JSTOR to digitize, preserve and extend access to Southern Economic Journal.

2 Public Sector Expansion: Stagnant Technology or Attenuated Property Rights?* W. MARK CRAIN University of California, Los Angeles and Virginia Polytechnic Institute and State University ASGHAR ZARDKOOHI Auburn University I. Introduction There is little dispute that over the last century public sector budgets in the United States have grown, both in absolute size and in relation to GNP. There is no such general consensus over the sources underlying this growth. Recent empirical evidence does suggest that rising per unit costs in government activities account for the single most important component in expanding public sector budgets.' In this paper we examine two prominent explanations for cost differences in public versus private sector activities and offer some empirical tests between these hypothee two competing eses. The first explanation is that differential cost increases are a manifestation of the tech- nological structure of certain activities performed by the public sector. This hypothesis originates from a much-cited paper by Baumol [3]. Baumol's argument (in the original) contains two central themes that have since emerged as the basis of what can be broadly labeled the "non-progressivity" explanation for rising public sector costs. First, he suggested that economic activities can, not entirely arbitrarily, be grouped into two types: technologically progressive activities in which innovations, capital accumulation and economies of large scale all make for a cumulative rise in output per man hour and activities which, by their very nature, permit only sporadic increases in productivity. [3, ] And secondly he argued that: inherent in the technological structure of each of these activities are forces working almost unavoidably for progressive and cumulative increases in the real costs incurred in supplying them. As a consequence efforts to offset these cost increases, while they may succeed temporarily, in the long run are merely palliatives which can have no significant effect on the underlying trends. [3, 415] *We are grateful to Robert Tollison, Gordon Tullock, and an anonymous referee for helpful comments on a previous draft. Partial financial support for Crain's research time was provided by the Center for Study of Public Choice, Blacksburg, Virginia. 1. See, for example, the important studies by Bradford, Malt, and Oates [7] and Borcherding [6]. 1069

3 1070 W. Mark Crain and Asghar Zardkoohi The non-progressivity explanation for rising public sector costs (and hence budgetary expansion) thus postulates that the nature of the activities performed by the public sector determines "quite definitely" whether the productivity of labor inputs will grow slowly or rapidly.2 An alternative explanation for observed cost differences in government versus private production stresses differences in the structure of property rights associated with the respective ownership forms. The property rights explanation is similar to the non-progressivity thesis to the extent that it implies that labor productivity will be lower in public than in private enterprise. In the property rights theory, however, production inefficiencies predictably arise because of attenuated cost-reward arrangements and not because of the inherent nature of the activities performed.3 The distinction between these two competing explanations for rising per unit costs in the public sector has important and straightforward public policy relevance. The non-progressivity thesis suggests that higher per unit costs are unavoidable, re- gardless of the ownership form of enterprises performing certain activities. The property rights explanation, however, suggests that cost increases could be mitigated by alternative institutional arrangements. In Section II we discuss some institutional arrangements that are relevant to decision making in public sector productive activity. We suggest a tentative model of factor employment decisions in government operations by examining the incentives facing political-managers and the incentives of voter-taxpayers. The main inference derived from this formulation is that public sector operations will be biased toward employment of variable factors, which creates long-run production inefficiencies.4 We note that this implication is contrary to the principle conclusions reached by DeAlessi [15] and Niskanen [22], and we thus provide a brief comparison to their models in attempting to clarify the bases of discrepancy. Section III presents our empirical tests between these two explanations for the cost differences in public versus private sector activities. Specifically, we examine the pattern of input combinations employed under the alternative ownership forms in two settings; namely, for water utilities in the United States and for the Australian airline industry. Finally, Section IV offers some concluding remarks. II. The Institutional Setting of Public Sector Operations A useful conceptual analogy can be drawn between government production and natural monopolies [27]. In neither setting are there competitive suppliers during the production or 2. Beck [4] offers a quite similar argument in addressing the factors that caused the comparatively large rise in the price index of government services between 1950 and He asserts that "the productivity lag in the government sector which is more labor intensive than the private sector and which has fewer opportunities to increase productivity through technological advances" was a quite important factor in this regard [4, 16]. For a critique of Beck's analysis, see Buchanan and Tullock [8]. 3. The empirical study by Orzechowski [23] provides some counter-evidence to the Baumol-Beck arguments. Orzechowski finds that "capital per worker is highest in the federal sector and lowest in the service sector. While the service sector is labor intensive relative to all industry, the federal sector is capital intensive relative to both all-industry and the service sector. In fact, capital per employee is ten times larger in the federal sector than in the service sector" [23, 124]. We note here, however, that Orzechowski's contention that the federal government is actually highly capital intensive is somewhat misleading. The appropriate efficiency benchmark for comparing the relative extent of capital-labor intensities is private firms engaged in similar activities; that is, while the activities performed by the public sector appear to require more capital, actual capital utilization by public enterprises may still be insufficient to minimize long run per unit costs. Thus, as we argue in our subsequent empirical tests, comparisons between public versus private enterprises engaged in similar activities provide a more appropriate empirical test. 4. The implication that governments tend toward an overemployment of labor has been drawn in earlier analyses, e.g., Demsetz [17] and Tullock [28], although the underlying behavioral reasons are different from those developed in the present paper.

4 PUBLIC SECTOR EXPANSION 1071 service period, although competition or rivalry may occur among potential suppliers at the contracting or bidding stage to obtain the production rights. In other words, the absence of multiple producers does not necessarily imply the absence of a competitive process [16; 9]. This framework is useful because it emphasizes the role of competitive incentives at the "contracting" stage in determining the subsequent behavior and performance during the actual production or service period. The productive capacity (broadly defined) of "government" is owned by the citizens-although ownership shares are non-transferable-and the rights to operate the public facilities are put up to auction at periodically held elections. Under most democratic political institutions, voter-taxpayers ultimately approve or let out the production contracts to political managers, typically using simple majority rule. In an "idealized" world of democratic collective decision making, the preferences of citizens (or at least the preferences of the voting majority) would dictate the production choices in government provision. No candidate would obtain the production contract, or have his current contract renewed by doing otherwise. In such a setting, the preferences of citizen-taxpayers thus determine the details specified in the political contract [10].5 For our purposes here, then, we initially examine the cost-reward structure facing taxpayer-voters with regard to investment decisions made in publicly owned (government) enterprise. We then analyze the incentives of manager-politicians which become more relevant in determining production choices in less than idealized democratic settings, i.e., where some discretionary decision-making power is available to the individual politician-manager. In this second case we again seek to identify the relevant incentive structure facing managers in public enterprise which would affect their production and factor decisions, apart from the issue of satisfying voter preferences or strict election or reelection goals. Owner-Taxpayers and Public Investment Decisions Suppose that an individual voter-taxpayer is considering a proposal (e.g., a campaign promise) to invest resources in a government enterprise which will yield a stream of benefits over time. Assume that this project obligates the taxpayer to invest Z tax dollars in the enterprise initially, and some continual tax payment in each subsequent year to maintain the investment at its initial level, Z. What are the relevant factors that would determine whether this individual voter-taxpayer would favor (i.e., vote for) this proposal. One important factor, of course, is the comparable rate of return that the individual could obtain if this same tax obligation or investment were made in an alternative private project or enterprise. Thus, in order to gain voter approval, we posit that investments in the public sector project must be at least as attractive as its private sector alternatives.6 The property rights theory has stressed in various contexts the importance of transferability of ownership shares for enterprise efficiency [1]. For example, higher costs (if not the impossibility) of transferring ownership shares, as in the case of government enterprise, reduce owner incentives to monitor the net wealth or capitalized present value of an enterprise since no market exists whereby this capitalized value can be realized through the resale (transfer) of ownership rights in the enterprise. In this view, then, there are potential gains 5. This characteristic of an 'ideal" political democracy is discussed in Becker [5] and Schumpeter [25]. 6. We stress that alternative theories of the political process have emphasized the redistributive aspects of government activity. In our analysis we examine incentives that are unrelated to obtaining or preventing wealth transfers between individual voters or coalitions of voters. Thus, the strict assumption may be employed (although relaxing it does not alter our analysis) that all voter-taxpayers are identical.

5 1072 W. Mark Crain and Asghar Zardkoohi that go uncaptured because property rights are nontransferable. However, an alternative interpretation that we propose is that this nontradeable aspect of government enterprise ownership rights can be taken into account in making investment and production decisions, which can be illustrated by returning to the choice problem of the individual voter-taxpayer described above. If the proposed investment were undertaken privately, the investor has the option of selling his share in the enterprise. He does not possess this option, however, in the case of the government enterprises. In the latter case, he is only able to consume the return on the investment, and cannot liquidate his investment should he desire to consume the initial investment. This difference in the transferability of ownership shares suggests that the return on investment in public enterprises would have to be higher than the return in analogous private enterprises in order to make the investor indifferent between the public versus private alternatives. This resulting differential between rates of return associated with private versus public investments could be substantial. For example, individuals in the United States tend to move roughly every five years.7 Movement to another political jurisdiction (e.g., across counties or states) means that the individual voter-taxpayer would lose his entire tax investment in his previous place of residence. Thus, the nontransferability of public investment as compared to private investment requires a substantially higher rate of return on the former type than on the latter type of investment in order to make the individual taxpayer indifferent between public and private projects. A second relevant aspect of nontransferability has to do with the number of years a public project renders benefits, relative to the number of years an individual (e.g., the median voter) expects to live. For example, for a given public investment, if a voter who expects to live t more years in a given jurisdiction invests Z dollars at the public sector rate of return, r, he would place a value on the project equal to PVp.b= E Zr/(l +iy, (1) j-i= where i is the market rate of interest. The present value of the public project, PVp b, in (1) implies that this voter would lose the initial amount of the investment, since he would be unable to sell the ownership right and capture the capitalized value of the investment for con- sumption (or to bequeath it to the heir of his choice). Thus, by the time this voter (of today) dies he will still have some initial investment in the public project whose benefits will accrue after his death. The extent of the perceived loss will, of course, depend upon the number of years that the median voter expects to live. The older the (median) voter, the more substantial his expected loss and thus the higher the rate of return that will be required to attract his support for a public project. By comparison, an equal private investment, if the voter expects to live t more years and invests Z dollars at the market rate of interest i, the present value of the investment will be the annual return plus the present value of the initial amount in t years, as shown in equation (2): PV~, = E Zi/(1 + iy + Z/(1 + i)'. (2) j-= In sum, an individual voter (investor) will be indifferent between the public and the private 7. See "Current Population Reports," (1972).

6 PUBLIC SECTOR EXPANSION 1073 investment alternatives only if the present value of his investment is the same in each case. This condition is depicted in equation (3): which can be simplified to, Zr/(l + iy = E Zi/(l + iy + Z/(1 + i)', (3) i=.ji= I r-i+ E 1/[(l + iy-' + 1]. (4) j=l Equation (4) simply restates our argument that the rate of return on a public investment, r, will have to exceed the rate of return on private investment, i, in order for the voter-taxpayer to be indifferent between an investment in the two sectors.8 And, the older the voter is or the more frequently he expects to change political jurisdictions, the greater the necessary difference between r and i to make public and private investments equally attractive. How then would this incentive to equalize rates of return across public versus private investments affect proposals offered by political-managers in the electoral rivalry? Or, stated differently, what sort of platform proposal or production contract is likely to be favored among voters? A simple numerical example is useful to illustrate this point. Suppose our representative voter-taxpayer (or perhaps "the median" voter) faces the following policy options or proposals for production in a community owned enterprise. Moreover, for the moment, let us constrain the example by assuming that voters desire an output rate of 20 units per year so, in effect, the relevant choice problem is solely concerned with which input mix will produce this output at least cost to the individual voter-taxpayer. Proposal A is to invest $200 in new capital equipment that has an expected production life of ten years, at the chosen output rate of 20 units per year. Alternatively, Proposal B is to employ the existing capital equipment, incurring a higher level of maintenance (i.e., variable or current operating) expenditure. Over a comparable ten year period, suppose this maintenance expenditure under Proposal B amounts to $300, which is more costly (less efficient) than Proposal A. That is, over the entire ten year period 200 units can be produced (consumed) at a per unit capital cost of $1 under Proposal A versus a $1.50 per unit maintenance cost under Proposal B, assuming all other costs are identical under both proposals. Proposal B, however, may not be more costly than Proposal A to our representative voter if he does not intend to be a "consumer" of this product for the entire ten year period, or if he cannot "resell" the property rights to another potential resident of the community. For example, if the voter has an expected stay of, say, five years, Proposal B is the least-cost alternative. Over a five year period 100 units of output would be consumed by the voter, yielding a per unit cost of $2 under Proposal A and (again) of $1.50 under Proposal B, with total expenses of $200 and $150 respectively. In sum, when ownership or consumption rights in the enterprise cannot be transferred, the individual voter will discount the benefits of that portion of output which accrues after his death or stay in the current political jurisdiction. This gives rise to incentives among voter-taxpayers to support production proposals that minimize "short-run" production expenses, which can be achieved by under-investing in long-run capital facilities and increas- 8. Holcombe and Zardkoohi [18] employ a similar analytical framework to explain the pattern of public and private enterprises which emerges in a democracy.

7 1074 W. Mark Crain and Asghar Zardkoohi ing maintenance expenses and variable factors. This does not minimize long-run production costs, although, we argue that the importance of the longer-run effects lose relevance to an individual taxpayer because there is no mechanism whereby such effects may be internalized into current wealth, e.g., a market for ownership shares [20]. Thus, in our initial "idealized" world of collective decision making characterized by strict voter-sovereignty, only those candidates who propose platforms for "minimizing" costs to the current constituency would be elected, which means that the "successful" production proposals would tend away from capital investment and toward variable factors and maintenance expenditures. Political Managers and Factor Choices We next examine the incentive structure facing managers in the public sector which becomes more relevant as voter influence or the desire for approval of platform diminishes. It is in this context that previous arguments seem to apply. Demsetz [17] and Stigler [26], for example, argue that political campaigns are primarily labor intensive processes and that electoral politics is essentially a continuous (as opposed to a periodic) process, which means that increasing employment in public sector operations is one means of "paying" past or present campaign workers. Relatedly, Tullock [28] and subsequently others have suggested a "selfgenerating" hypothesis that argues that expanding public sector employment is one plausible means for maintaining or increasing constituent support for public sector operations, i.e., labor can vote and capital cannot. Both arguments reinforce the implication derived above that factor mixes in public enterprise will tend toward higher usage of labor or variable factors and less capital investment, relative to counterparts in the private sector. The incentives of voters and those of political managers can be combined in the following framework. Let PL and P, be the constant objective prices of labor and capital. Let MPL and MP, be the objective marginal products of labor and capital, dependent of course on K/ L, with convex isoquants. The voter's "perceived" marginal product of capital can be represented as MP,*, equal to (MP,)?*, where / is less than unity because some of capital's marginal product accrues after the voter's expected life-span or stay in a particular political jurisdiction. The manager-politician's "perceived" MPL can be represented as MPL*, equal to (MP,? a, where a is greater than unity because of the "extra" role of employees in the electoral process. The politician will thus choose that capital-labor mix for which MP,/MP, = PL/PK. P8/a < PL/P.K (5) A Comparison to Previous Models The implication of our formulation that government enterprise will tend to be biased toward higher labor input mixes differs from the factor choices predicted in the models developed by DeAlessi [15] and Niskanen [22]. These models argue that bureaucratic production will be generally biased toward inefficiently high levels of capital investment. It is useful to note briefly the nature of this discrepancy, particularly since both stress the limited tenure or attenuated production horizon of political decision makers. First, in the DeAlessi-Niskanen models the decision maker is assumed to "maximize the

8 PUBLIC SECTOR EXPANSION 1075 total budget during his tenure as head of the bureau" [22,114]. Moreover, the bureaucrat is presumed to recognize that he will not likely be in the same position in the later years of the relevant planning period. Given this anticipated mobility, and that the appropriable gains of his budgetary expenditures are specific to his current position, the bureaucrat has incentive to select those production processes which have higher costs in the early years (which corre- spond to the bureaucrat's tenure in his current job). This formulation thus implies that capital-intensive processes, requiring large initial expenditures, will be preferred by the bureaucrat. In contrast, decision makers in our model of political enterprises are motivated by a reelection objective, which suggests that production processes will be selected which have lower costs in the early years since that is likely to enhance their support among voters. The fundamental difference between the two approaches, then, is that even though decision makers in both models perceive a limited production horizon, we assume that choices are made to extend this horizon (i.e., to increase reelection possibilities); the previous authors seem to assume that choices are made given this temporal constraint. Perhaps the degree of applicability of either approach depends on the "closeness" of a particular political decision maker to the election or reelection process. III. Implications and Empirical Evidence Our analytical model suggests that enterprises in the public sector will operate with relatively less capital and more labor than privately owned firms in performing roughly identical activities. These contrasting choices of factor usage issue from the alternative institutional structures and resulting behavioral implications associated with each respective ownership form. Again, we stress the distinction between this explanation for higher per unit costs in public sector activities versus the Baumol thesis; it is the nature of the activities performed by the public sector which dictates more labor intensive production technology. The implications of this model are readily testable, and we offer some empirical verification for two different types of activities. As we stressed earlier (see Note 3) the appropriate benchmark for examining relative factor utilization in public sector operations are private firms engaged in roughly similar activities. We first provide a comparison of the relative employments of capital and labor between publicly and privately owned water utilities in the United States. We then offer a similar type of comparison between the publicly and privately owned airlines in Australia. The Case of Water Utilities in the United States First, we offer a test of our model using data on publicly and privately owned water utilities in the United States. Stated in a null hypothesis format, the relative quantities of capital and labor used in producing given output levels should not differ significantly between public and private enterprises. The form of the model that is used to test this hypothesis is specified in equation (6). (K/L), =,o +,8Di + f2(w/r), + 33Qi + 84Di(w/r), + s5di,q + ti, (6)

9 1076 W. Mark Crain and Asghar Zardkoohi where (K/L), = the ratio of capital to labor inputs to firm i,9 Di = a dummy variable that is equal to 1 if firm i is publicly owned, and equal to 0 if firm i is privately owned, (w/r), = the ratio of the wage rate to capital costs in firm i, Qi = the output in firm i (measured in millions of gallons per year), Di(w/r)i = an interaction term that is non-zero only in publicly owned firms, DiQi = an interaction term that is non-zero only in publicly owned firms, and,i = a random disturbance term. This specification allows us to check for differences in the respective capital-labor input proportions in public versus private firms while controlling for differences in relative factor prices.'? The coefficients specified in equation (6) are estimated cross-sectionally from data on seventy-eight water utilities for The particular firms that are included in our sample and the selection methods are available on request. The estimated coefficients and their re- spective t-values are presented in equation (7). (K/L)i = Di (w/r)i Qi Di(w/r), (.74) (.58) (-.20) (1.83) (- 0.02) DQ, (7) (-1.57) N. Observed = 78 R2 = 0.10 The estimated coefficients that are of primary interest to the present study are those which indicate the relationship between output level and capital labor usage in the public and private firms. This coefficient for the private firm output variable, Qi, is significant at the.08 level. We can test the null hypothesis that no difference exists between the public versus private firms using the coefficient on the interaction variable, DiQi. Since this variable is nonzero only for the publicly owned firms, it picks up differences between the two types of ownership forms. The value of this coefficient is , which is significant at the.12 level. We can thus reject the null hypothesis at a reasonable level of statistical confidence in favor of the alternative hypothesis that the capital-labor proportions in the publicly owned firms is less responsive to output changes than for their comparable privately owned firms. The findings indicate that while the private firms tend to increase the use of capital relative to labor for output expansions, relative factor usage in public firms does not seem to exhibit a similar factor adjustment to output changes. Note that the coefficient on Qi in equation (7) is the appropriate estimate of the relationship in private firms, although for public firms it is necessary to sum the coefficients on Q, and DiQi. In order to attribute this observed difference in relative factor mixes between the two 9. Some water utilities in our sample have full- as well as part-time employees. In order to account for both categories of labor, we follow the procedure used by Nerlove [21] that the part time employees work half time. Thus, if we denote the number of part-time employees as Lp and full-time employees as LF, Li in equation (6) is defined as L, = LF + 1/2Lp. 10. For a further discussion and exposition of this type of empirical procedure see Sato [241 and Kmenta [19, ].

10 PUBLIC SECTOR EXPANSION 1077 types of enterprises to the bias toward inefficiently filabor high levels of employment in public firms which is postulated by our analytical model, some additional information is required. That is, since the private water utilities in our sample are subject to rate of return regulation, it is possible that these firms are investing in inefficiently large levels of capital in order to expand the rate base, i.e., the Averch-Johnson effect [2; 11]. However, this problem can be resolved with additional information concerning relative cost differences between the public and private firms. We can report that we have estimated cost curves for both types of enter- prises in a recent paper [12] and that costs for given output levels are greater in the publicly owned firms. Hence, it is safe to surmise that even if the private firms are influenced by a "capital bias," the extent of this inefficiency is not so great as the inefficiency in the public firms deriving from an underutilization of capital, given that the latter type of firms exhibit, on net, higher costs of operation. The Case of the Australian Airline Industry Davies [13; 14] maintains that a unique feature of the Australian airline industry is that government uses the same blueprint of regulation in operating its own airline as it does in operating the private airlines. This regulatory policy is expressly designed to make the two airlines similar in most important respects. The policy has been successful. As Davies notes, the airline industry in Australia is synonymous with a monopoly of air transport divided between two massive organizations whose development is rigidly controlled by Acts of Parliamento the point where competition in the generally accepted sense of the word, is restricted to the peripheral comforts and minor variations in time-tabling [13, 154]. Along with a description of the institutional arrangement in the Australian public airline, Davies [13; 14] compares the output per unit of labor in the public airline to that of the private airline. His analysis indicates that the output per unit of labor in the public airline is far less than in the private airline. The data used by Davies, presented in Table I, can be employed further to estimate the relevant degree of overemployment in the public airline as compared to the private airline. Given the institutional arrangements described above, the two airlines must be able to employ an equal number of employees per unit of output. For example, if the private airline employs X number of laborers per passenger-mile, the public airline must be able to do the same. One problem with making the estimate, however, is that the airlines produced the two services in different ratios during (see Table I). During this period, the passengerfreight ratio was greater in the public airline than in the private airline. This implies that an estimate of the relative sizes of employment in the two airlines could create a bias against the 11. For example, the Australian government has (a) required the two airlines to acquire the same type of aircraft, and to commission the new aircraft into service at the same time; (b) set practically the same routes for the two airlines; (c) allocated equal plane capacity for both passengers and freight; (d) controlled time schedule, specific ports of call and frequencies of visits to various stopping places; (e) created equal prices for airlines' services; (f) prohibited a third airline from entering the industry; and (g) set rules to help assure that each airline possesses a comparable cost structure. Moreover, "in order to help assure that one of the airline does not cause economic harm to the other, the Commonwealth not only controls the routes, ports of call, and frequencies of stops, but also the maximum aircraft fleet capacity permitted on a given competitive route over a specified period of time. For each forthcoming six month period government officials calculate a predicted required industry capacity in units of tonmiles for all competitive trunk routes. This determination is then allocated to TAA and Ansett ANA, each receiving exactly one-half of the total allowable predicted capacity" [13, 155].

11 1078 W. Mark Crain and Asghar Zardkoohi Table I. Statistics on Airline Operations Freight and Passengers Year Mail (tons) (000's) Employees The Public Airline: TAA The Private Airline: Ansett Transport Industries, Air Group Source: Davies [13]. Original Source: Stanley Brogden, Australia's Two-Airline Policy 214 (1968); Trans Australian Airlines Ann. Rep. ( ); Ansett Transport Industries Ltd. Ann. Rep. ( ). productive efficiency of one or the other airline; that is, as long as one output is relatively more labor intensive than the other, the difference between the output ratios across the two airlines naturally would bring about a difference between the number of laborers necessary in the production processes. The equality between the output ratios is strictly required in order to prevent the bias arising from differences that could exist between the input intensity of the two outputs. One way to account for the labor intensity of each service is to use information on the relative marginal products of labor across the airlines. Unfortunately, this information is not available. An alternative way to estimate the overemployment in the public airline is to make an adjustment in the figures listed in Table I, in a manner that would reflect an equality between the output ratios in the two airlines. Moreover, in order to prevent any measurement bias against the public airline, adjustments will be made in a manner that would indicate more output for the public airline than the airline actually produced.

12 PUBLIC SECTOR EXPANSION 1079 Table II. Statistics on Airline Operations: Adjusted Values Freight and Mail (Tons) Passengers Year Adjusted (000's) Employees The Public Airline: TAA The Private Airline: Ansett Transport Industries, Air Group Source: See Table I. Table II lists the adjusted figures. The method used to make the adjustment is first to determine the output ratio in the private airline. Then, the absolute amount of the service in the public airline is adjusted upward such that the ratio of the amount of services produced in the public airline is equal to that in the private airline. For example, during the period the ratio of the amount of freight per passenger in the private airline is ( = 42401/ ). The same ratio for the public airline is 0.02 ( = 17696/871000). In order to make the ratios equal, the amount of freight in the public airline (i.e., 17696) is adjusted upward such that the new output ratio in the public airline is also Similar calculations are performed for the remaining periods as listed in Table II. The information in Table II can be used to estimate roughly the extent of overemployment in the public airline. We first determine the quantity of labor per unit of one of the services in the private airline. Then, using this figure, we determine the maximum number of laborers in the public airline required to make this airline as efficient as its private

13 1080 W. Mark Crain and Asghar Zardkoohi Table III. The Actual and the Required Maximum Manpower in the Australian Public Airline Actual "Maximum Excessive Excessive Year Employment Requirement" Employment Employment (%) Source: See Table II. counterpart.'2 Finally, we make a comparison between the number of laborers employed in the public airline and the maximum number of laborers necessary to make the public and private airlines equally efficient. The absolute and the percentage differences between these two sets of figures are listed in Table III for the period The figures listed in Table III indicate that the public airline employed more laborers than it needed to make the airline as efficient as the private airline. In particular, the public airline employed from 7 to 32 percent more than the "required maximum" during the period.'3 Summary The results presented in this section support quite well the implication that for similar activities enterprises in the public sector employ less capital and more labor per unit of output than privately owned firms. Our findings thus suggest that non-progressivity or productivity lags in some public sector activities may well be attributable to the underlying institutional structure of property rights rather than the nature of the activities, per se, as the Baumol thesis predicts [3]. IV. Concluding Remarks The conceptual model developed in this paper stresses the effects of institutional arrangements in government enterprise on the incentives and behavior of voter-taxpayers and man- 12. In other words, we determine the number of employees per unit of, say, freight in the private airline for (i.e., 3965/42401). This ratio multiplied by the amount of freight in the public airline should give us the number of employees required in the public airline (3965/42401) (33033) = In other words, the result, 3089, shows the maximum number of employees for which must have been hired in the public airline in order to make the airline as efficient as the private airline. 13. The degrees of overemployment in the public airline which are listed in Table III are probably too low because the actual figures for the production of the public airline are adjusted upward in order to account for the difference between the output ratios (see Table III).

14 PUBLIC SECTOR EXPANSION 1081 agers. We have identified incentives on the part of these two groups which influence the choice of productive inputs in publicly owned firms. Moreover, the bases of distinction between our approach and the previous models developed by DeAlessi and Niskanen have also been explored. The effect of limitations on the production horizon of political decision makers and the inability to transfer ownership shares in our model is to bias downward longterm capital investments in favor of a higher employment of variable factors, e.g., labor. The empirical results for water utilities in the United States provide supporting evidence for the implication that the usage of capital and labor factors differs between public and private firms. The findings for this industry support the underlying proposition that publicly owned enterprises employ relatively more labor and less capital per unit of output than their privately owned counterparts. Our examination of the publicly and privately owned airlines in Australia indicated a similar type of behavior regarding input choices. The attenuation of property rights in public enterprise operations, thus appears to be a source of higher per unit costs in some public sector activities and hence expanding public sector budgets. References 1. Alchian, Armen A., "Some Economics of Property Rights." II Politico, December 1965, Averch, H., and L. Johnson, "Behavior of the Firm Under Regulatory Constraint." The American Economic Review, December 1962, Baumol, W. J., "Macroeconomics of Unbalanced Growth: The Anatomy of Urban Crisis." American Economic Review, June 1967, Beck, Morris, "The Expanding Public Sector: Some Contrary Evidence." National Tax Journal, March 1976, Becker, Gary, "Competition and Democracy." Journal of Law and Economics, October 1958, Borcherding, Thomas E. "One Hundred Years of Public Spending," in Budgets and Bureaucrats: A Public Choice Approach to Government Growth, edited by T. E. Borcherding. Durham, N.C.: Duke University Press, 1977, pp Bradford, D. F., R. A. Malt, and W. E. Oates, "The Rising Cost of Local Public Services: Some Evidence and Reflections," National Tax Journal, June 1969, Buchanan, J. M. and Gordon Tullock, "The Expanding Public Sector: Wagner Squared." Public Choice, Fall 1977, Crain, W. Mark and Robert B. Ekelund, "Chadwick and Demsetz on Competition and Regulation." Journal of Law and Economics, April 1976, , "Deficits and Democracy." Southern Economic Journal, April 1978, and Robert Tollison. "Managerial Incentives in Private and Regulated Firms." Mimeograph, Virginia Polytechnic Institute and State University, and Asghar Zardkoohi, "A Test of the Property Rights Theory of the Firm." Journal of Law and Economics, October 1978, Davies, David G., "The Efficiency of Public Versus Private Firms, The Case of Australia's Two Airlines." Journal of Law and Economics, April 1971, , "Property Rights and Economic Efficiency: The Australian Airlines Revisited." Journal of Law and Economics, April 1977, DeAlessi, Louis, "Implications of Property Rights for Government Investment Choices." American Economic Review, March 1969, Demsetz, Harold, "Why Regulate Utilities." Journal of Law and Economics, April 1968, "The Growth of Bureaucracy." unpublished paper presented at Public Choice Society Meetings, New Orleans, March Holcombe, Randall G. and Asghar Zardkoohi. "Public Investment in a Democracy, A Theory of Public Enterprise." Mimeograph, Auburn University, Kmenta, Jan. Elements of Econometrics. New York: Mcmillan, Manne, Henry G., "Mergers and the Market for Corporate Control." Journal of Political Economy, April 1965,

15 1082 W. Mark Crain and Asghar Zardkoohi 21. Nerlove, Marc. "Returns to Scale in Electricity Supply," in Measurement in Economics, edited by C. Christ. Stanford: Stanford University Press, 1963, pp Niskanen, W. A. Bureaucracy and Representative Government. Chicago: Aldine, Orzechowski, W. P., "Labor Intensity, Productivity, and Growth of the Federal Sector," Public Choice, Fall 1974, Sato, Ryuzo, "Homothetic and Non-Homothetic CES Production Functions," American Economic Review, September 1977, Schumpeter, Joseph, Capitalism, Socialism, and Democracy, 3rd ed. New York: Harper Colophon Books, Stigler, George, "The Theory of Economic Regulation." Bell Journal of Economics, Spring 1971, Tullock, Gordon, "Entry Barriers in Politics." American Economic Review, May 1965, Politics of Bureaucracy. Washington, D.C.: Public Affairs Press, 1965.