Printer Friendly

An estimate of the rent generated by a premium college football player.

I. INTRODUCTION

The National Collegiate Athletic Association (NCAA) has cartelized college athletics. It limits television appearances for college teams, it bans payments to players other than a minimum outside income they earn, and it reduces players' mobility between colleges by making them ineligible if they transfer. In the 1984 case NCAA v. The University of Oklahoma and The University of Georgia, the Supreme Court ruled that limitations on college football teams' television appearances violated antitrust legislation, effectively deregulating the output market. However, the Court upheld that many of the restrictions placed on athletes were designed to preserve the competitiveness and amateurism of college athletics, maintaining the NCAA's exclusive control as the regulator of player recruitment.(1)

The effective wage college athletes receive for their services amounts to the value of a scholarship, room and board, book allowances, and a minimal amount of outside earned income. The maximum payments are set by the NCAA and enforced by the Committee on Infractions. By limiting these payments, the NCAA reduces the competition among colleges for players. Colleges are effectively monopsony employers so players will not be paid their marginal revenue product. Therefore, colleges capture economic rents from players. Furthermore, every school has an incentive to cheat on the cartel by offering effective wages above those set by the NCAA to attract higher quality players from competing schools. This often involves breaking NCAA rules by paying athletes outside payments and gifts, or using illegal recruiting practices.

Economists agree that the abuses in college athletics are primarily a result of the monopsonistic nature of the player recruitment market, although there is no empirical literature to support the theory. Koch |1973; 1983~ first addressed the cartel behavior of the NCAA by analyzing its regulations, enforcement policies, and past reforms. Becker |1985~ provides a persuasive argument for paying athletes as a means of negating monopsonistic exploitation in the player recruitment market. Lawrence |1987~ and Fleisher, Goff, and Tollison |1992~ focus on historical and institutional behavior in the development of the cartel, as well as incentives to violate NCAA rules and capture additional rents generated by players.

This paper provides estimates of the rents generated by the most highly skilled Division I-A college football players. At a general level, it is an empirical application of the mostly theoretical rent-seeking literature. Teams able to exert more effort in locating players, evaluating their abilities, and inducing them to accept a scholarship package acquire better players and thereby produce higher levels of team skill which generate additional revenues. In this sense, teams are rent-seekers in that they expend effort in recruiting to capture the rents associated with a top recruit. Only one team acquires the player, however, so that effort expended by its competitors is a social loss.

The rent associated with a college recruit is the difference between the player's marginal revenue product and the maximum payments allowable by the NCAA. The value of an athletic scholarship typically ranges from $5,000 to a maximum of $20,000. To estimate the marginal revenue product of a premium college player, extensive revenue data were collected for a sample of Division I-A football programs. These data were supplemented with information from the National Football League (NFL) draft. Each year professional teams in the NFL search for new recruits to draft from college teams. A player who is drafted by a professional team is one of the most skilled college players. These players' marginal revenue products are estimated by regressing team revenues on the number of its players drafted, controlling for teams' market characteristics and opponent's team skill levels.

II. PLAYER RECRUITMENT

Consider a team which generates revenues from ticket sales and television broadcasts associated with maintaining any given team skill level. Revenues are a function of the home team's skill level, the quality of its opponents, and its market characteristics such as local population, the availability and quality of alternative college football entertainment, the team's past success, etc. A team is constructed with the individual players it recruits. A school must locate and carefully evaluate a player, and then attract him away from its competitors. The greater the number of premium recruits acquired, as measured along some index of individual player skill, the greater is its total team skill level.

The number of premium recruits a team acquires is a function of its market characteristics and its pool of potential recruits. First, less recruiting effort is required by teams located in markets experiencing less competition in either the number or the quality of opposing teams. Second, teams located near larger population bases have an advantage in attracting revenues and therefore are able to expend higher levels of recruiting effort. Third, a successful team gaining exposure by maintaining a top ranking during the past few seasons attracts more premium players--better teams attract better players. Fourth, teams located near rich pools of potential recruits require less effort in locating and attracting players.

In addition, conference revenue sharing policies allowing the home team to retain only a fraction of its revenues affect its acquisition of top recruits. Revenue sharing induces a team to recognize the effects of acquiring a top recruit on revenues generated in the opponent's market, however, it also causes a team to associate less importance to revenues generated in its own market. An increase in the fraction of home revenues retained means that the team gains, at the margin, some fraction of revenues generated in its own market and loses the same fraction of revenues generated in its opponent's market. As long as the marginal gain in its own market exceeds the marginal loss in its opponent's market revenues, then net marginal revenues increase--at the margin, the pecuniary rewards of acquiring a top player are greater for the home team than for its opponents.(2)

To estimate marginal revenue product, measures of team revenues and player performance must be available. Scully |1974~ first estimated marginal revenue product in professional baseball. He initially estimates a production function by regressing winning percentages on team inputs--batting and pitching performance--the coefficients interpreted as marginal products. He then regresses revenues on winning percentages and market characteristics. Marginal revenue products of hitters and pitchers are measured by multiplying the marginal product coefficients and the marginal revenue of wins coefficient. Atkinson, Stanley, and Tschirhart |1988~ estimated marginal revenue products in professional football following the two equation method of Scully |1974~.(3)

The primary objective of this paper is to estimate the marginal revenue product of a premium college football player. One way is to use the NFL draft data to measure college player performance and then simply regress team revenues on the number of its players drafted. This provides a measure of the marginal revenue product generated from acquiring one more of the most skilled college football players. However, the skill level of the players acquired by a college team is likely to be endogenous to its recruiting effort. Since the level of recruiting effort is a function of teams' market and recruiting characteristics, the number of players drafted from a college team would likely be correlated with the error term.

The empirical analysis proceeds as follows. To account for the endogeneity in the acquisition of recruits, I use two-stage least squares estimation. First, draft equation estimates are reported using the number of players drafted from each college team as the dependent variable. This is a measure of a premium college player's skill level, which is a function of a team's recruiting effort, market and recruiting characteristics, and conference revenue sharing policies. Second, team revenues are estimated as a function of the number of its players drafted, controlling for market characteristics and opponents' team skill levels.

III. DATA

The Higher Education Amendments Act of 1992 (Sections 103-4) requires that all Division I-A football programs report financial statements to the U.S. Secretary of Education, although the date which this takes effect has not been specified. Therefore, presently the only sources of this information are the athletic departments or institutions themselves. Accordingly, data were gathered for gate, television, and donation revenues on an individual school basis through participation in a revenue and expense survey. Private institutions are not required by the Freedom of Information laws to release this information, and due to the sensitive nature as well as the time costs involved in gathering the data, not all public institutions chose to participate. In all, revenue and expense data for the 1988-89 football season were collected for 47 of the 101 Division I-A football programs, excluding the military academies. However, only the 39 teams affiliated with a conference are included in the primary sample since independent teams negotiate revenue sharing rules on a game by game basis so that measuring a revenue retention parameter is problematic.
TABLE I

Reported Division I-A Football Revenues and Expenses for the
1988-89 Season

Means (Standard Deviations)

 Division I-A Sample Number of Respondents

Total Revenues $5,399,351 39
 (4,004,574)

Ticket Sales $3,073,510 34
 (2,516,438)

TV/Radio Revenues $543,926 32
 (487,568)

Donations $1,151,363 19
 (1,292,364)

Miscellaneous $966,111 33
 (745,963)

Expenses $2,743,215 35
 (1,223,282)


Table I reports revenue and expense summary statistics for the sample. Total revenues include receipts retained by the athletic department from ticket sales, television and radio, donations, and miscellaneous receipts from guarantees paid by other schools, bowl games, and student fee allocations. Although the total revenue figures include all of these revenue sources, interpreting each category as a proportion of total revenues may be misleading. First, programs commonly practice a two-tiered pricing policy for season-tickets where a ticket holder's donation determines the quality of the seat. These may be counted as either donations or ticket sales, depending on the athletic department. Second, several departments include donations as miscellaneous revenues rather then categorizing them separately. Third, in a few cases respondents were unable to categorize revenues and therefore only reported the total revenue figure.

The expense figures in Table I include salaries, travel, grants-in-aid, and related operating disbursements. Discretion should be used in comparing expenses across teams as well as in interpreting reported profits and deficits, since operating expenses are not uniformly defined across athletic departments. For example, several schools do not count grants-in-aid and/or salaries as operating expenses, instead these are included as general university expenses. Based on categorized data from a few schools these items may account for 30 to 40 percent of total operating expenses. Furthermore, Boreland, Goff, and Pulsinelli |1991~ report that accounting practices tend to overestimate true economic costs, primarily by figuring grants-in-aid at the inflated average, rather than true marginal cost.

Individual player performance is measured using information from the NFL professional draft. A 1988 team roster might include players who will be drafted in the spring of 1989, as well as in the following 1990 and 1991 drafts. For example, a star sophomore player contributes to the 1988 team but may not enter the draft until the spring of 1991. If only draft data for 1989 were used, this player's contribution to team revenues would be ignored. The DRAFT variable is the number of all players drafted during these three years from each college team.

College teams located near rich pools of recruits are able to attract better players since less effort is required in locating recruits. Rooney |1987~ uses 1971-72, 1976-77, and 1980-81 college rosters to calculate the number of major college players produced in each state relative to the number of Division I-A teams it supports, or the percentage of players produced relative to the state's own needs. The variable POOL is Rooney's measure for the state in which the team is located.

Market characteristics are decomposed into two measures--market potential and past team success. Market potential, or ability to attract fans, is a function of population and alternative forms of entertainment within a team's market area. Market area, as distinguished from market potential, refers to a team's geographical market boundary. A team's market area is demarcated by the distance where travel to games is at a minimum. However, each team may have different market area distances from which they attract fans. Therefore, measuring teams' true market areas requires unavailable data on the distances fans travel to see games. Distances of 100, 120, and 150 miles about a team's location were used as proxy market areas to observe any sensitivity of the regression results to changing the market boundary. Furthermore, since a team's ability to attract spectators is presumably a decreasing function of distance, a distance-weighted measure of population was calculated. Specifically, the measure |m.sub.i~ was calculated,

|Mathematical Expression Omitted~

where |P.sub.j~ is population in the jth SMSA within team i's market area; |D.sub.ij~ is the distance between team i and the jth SMSA; n is the number of SMSA's in team i's market area; |Lambda~ is the distance exponent.

The measure |m.sub.i~ depends upon the value of |Lambda~ and the distance of the market area. For example, a value of zero for |Lambda~ weights each SMSA's population equally regardless of distance, while a value of 0.15 weights an SMSA 100 miles from the team's location at one-half its population--higher values attach relatively lower weights to SMSA's located within the market area. A specification search was attempted to observe the degree of sensitivity of the regression results from changing the market potential definition with respect to the |Lambda~ values and the market area distances. Market potential measures were constructed using ten values of |Lambda~ between zero and one, and using distances of 100, 120, and 150 miles about a team's location. This is discussed further in the empirical section.(4)

Alternative athletic entertainment is a function of the number and the quality of other teams within the market area. College football quality is measured using a cumulative point ranking constructed from Associated Press Top-20 weekly rankings. During any particular week a number one ranking earns a team twenty points, a number two ranking earns nineteen points, and so on. Points were calculated for a total of 137 weeks during the years 1980-89 to account for any random fluctuations. Converting the raw scores into percentiles, the highest scoring team receives an index of one and all others are indexed as a percentage of this high score. Professional teams within the market area were also given an index of one. The number of other teams in the market area, weighted by distance, was then added to the quality index yielding an alternative athletic entertainment measure. The market potential parameter, called MARKET, was then calculated by dividing |m.sub.i~ by this measure.

Past team success attracts revenues, as well as top recruits, primarily through expanded national exposure. This effect is captured by using weekly Top-20 point rankings for the 1985, 1986, and 1987 seasons. First, team success from 1985 to 1987 was presumably most influential in attracting the players on the 1988 rosters, since during this period players made their recruiting choices. Second, recent past team success also has the largest impact in attracting revenues from new season-ticket holders, television contracts, and alumni donations. The variable RANK 85-87 aggregates each team's Top-20 weekly rankings for the seasons 1985, 1986, and 1987 to conserve on degrees of freedom. Cumulative point rankings were divided by total weeks yielding an average weekly point ranking. For example, an average weekly point ranking of fifteen can be interpreted as a number five average ranking in the weekly polls for these three seasons.(5)

Conference gate revenue sharing policies fall into three categories. In the first category, a fixed fraction of revenues is shared regardless of total gate revenues. In the second, the home team guarantees the visiting team a lump-sum payment. In the third, revenues are shared equally, with a minimum guarantee and a maximum payment. Independent teams, those not affiliated with a particular conference, are excluded from the sample because of the inherent difficulties in measuring revenue retention rates. Since revenue sharing is negotiated on a game by game basis, it is impossible to calculate accurate retention rates for all games played by independents. The military academies are also excluded since revenues are generated from a variety of governmental sources.(6)

Table II compares team characteristics in the sample with those of overall Division I-A football programs. The average number of players drafted per team, 1988 Top-20 rankings, attendance, and school enrollment are slightly higher for the sample of teams as compared to all Division I-A teams. Furthermore, Table II suggests that larger schools have better programs, perhaps higher enrollments provide additional student fees as well as an attendance base necessary to support these programs. Excluding the military academies, 101 teams competed at the Division I-A level during the years 1985 to 1988, the period for which the data were collected.

IV. EMPIRICAL RESULTS

Two-stage least squares estimation is used to account for the endogeneity in the acquisition of recruits. First, the draft equation estimates are reported from regressing the number of players drafted from each college team on the conference revenue retention parameter, as well as the team's and its opponents' market and recruiting pool characteristics. Second, using two-stage least squares team revenues are estimated as a function of the number of its players drafted, controlling for opponents' team skill levels, market characteristics, and conference revenue retention policies.

Two-stage least squares is an appropriate estimation method assuming linear relationships in the first-stage and second-stage regressions. However, the qualitative TABULAR DATA OMITTED nature of the future NFL draftees variable may suggest a nonlinearity in the first-stage draft equation which would affect the second-stage estimates.(7) This issue is addressed by testing the normality of the DRAFT variable--if the number of future draftees is normally distributed, then the first-stage regression satisfies a classical linear regression model. The DRAFT variable is distributed with a mean of 8.03 and a standard deviation of 5.24, ranging from a minimum value of zero to a maximum of eighteen, reflecting a normal distribution. (DRAFT is censored at zero for only one team.) The Bowman and Shenton |1975~ test statistic for normality, ||Chi~.sup.2~(2)=0.199, supports the hypothesis that the DRAFT variable does not significantly depart from a normal distribution. Therefore, linear estimation in the first-stage is appropriate so that two-stage least squares is a valid estimation method.(8)

The draft equation estimates are reported in Table III. The variable MARKET is the market potential parameter measured in units of 100,000; OPPONENT-MARKET is the average market potential of scheduled opponents measured in units of 100,000; RETAIN is the revenue retention parameter, the fraction of revenues retained by the home team; POOL is Rooney's percentage of college players produced in the state relative to the number of teams it supports, and OPPONENT-POOL is this measure averaged for all opponents; RANK 85-87 is the team's average weekly Top-20 point ranking during TABULAR DATA OMITTED the 1985, 1986, and 1987 seasons, and OPPONENT-RANK 85-87 is the average weekly Top-20 point rankings of opponents.

The coefficients for MARKET, OPPONENT-MARKET, and RETAIN in specification (i) suggest that teams located in better markets and in conferences practicing less revenue sharing, other factors constant, acquire more premium players. A 100,000 increase in a team's own market potential is worth approximately one more future NFL draftee. The same change in opponents' market potential results in 1.531 NFL caliber players. The RETAIN variable may be picking up conference effects, and is likely endogenous if conference revenue sharing policies are determined by member team characteristics. Revenue sharing policies may attempt to distribute team skill levels to maximize conference revenues, or alternatively, to lessen any intermarket differences and more equally distribute team skill.

Including RANK 85-87 as a past success measure in attracting players substantially reduces the size and significance levels of the MARKET and RETAIN coefficients. The RANK 85-87 coefficient suggests that past Top-20 rankings as a measure of national exposure play an important role in attracting top recruits. A jump of about five points in the weekly Top-20 rankings attracts approximately one more future NFL draftee. The coefficients associated with POOL and OPPONENT-POOL are both negative and insignificant. Perhaps a reason for this is the ability of many teams to recruit nationally, so that regional considerations are not as important. Excluding the pool variables in specification (iii) does not notably change the results.

A specification search was attempted to examine the sensitivity of the results to changing the definition of the market potential parameter. Overall, increasing |Lambda~ from zero to 0.70 in specification (i), for example, reduces the RETAIN coefficient by around 1.00, the MARKET and OPPONENT-MARKET coefficients increase by about 50 percent and their standard errors by approximately 70 percent. The degree of sensitivity was similar in specification (ii) except that the OPPONENT-MARKET and OPPONENT-RANK 85-87 coefficients nearly double. The reported results use a |Lambda~ value of 0.15 and a distance of 100 miles. These values satisfy Amemiya's |1980~ prediction criterion for choice among regression models. Furthermore, the results did not prove to be sensitive to either changing the market area distance or excluding professional teams from the market potential parameter.

Marginal Revenue Product Estimates

The marginal revenue product of a premium college football player is measured by regressing 1988 team revenues on the number of its players drafted using two-stage least squares to account for the endogeneity of the draft measures, holding constant a team's market potential and opponents' team skill levels. Cumulative point rankings from the Associated Press Top-20 rankings control for the total team skill levels of scheduled opponents. The variable OPPONENT-RANK 88 is the average weekly point ranking of opponents during the 1988 football season. Tables IVa and IVb report the OLS and two-stage least squares marginal revenue product estimates.

A team's market potential and opponents' Top-20 point rankings are insignificant factors in attracting revenues. RANK 85-87, controlling for past success as a means of attracting revenues through new season-ticket holders, alumni donations, and perhaps television contracts, proves to be significant when added to the estimation.

This paper is primarily concerned with the two-stage least squares estimates of the DRAFT coefficient.(9) Table IVb suggests that highly skilled college football players generate revenues well in excess of their effective wage set by the NCAA. Recruiting an additional player with NFL capabilities is worth $646,150 in annual revenues for his college team. Including RANK 85-87 into the estimation reduces this marginal revenue product estimate to $586,790 and including RETAIN reduces it to $538,760. Over a four-year college career a premium player could therefore generate over two million dollars in revenues for his college team.(10) In any case, the value of an athletic scholarship package is limited to $20,000 annually, suggesting that substantial economic rents are transferred from the premium college players to other agents.(11) Moreover, even if the total operating TABULAR DATA OMITTED TABULAR DATA OMITTED expenses reported in Table I were attributed to the costs of recruiting and supporting players, the average cost for the allowed ninety-five scholarship players would be less than $30,000 annually per player.

The estimated rent of a premium player is defined as the difference between the revenue generated by the player and his effective compensation (i.e., value of a scholarship and costs of supporting a player.) However, this may overstate a player's true rent since it does not include future lifetime income received as a professional player. For example, if teams and players could bargain over compensation, a player may be willing to forego some current income to play for a team providing the best training and exposure, and therefore, the highest lifetime income. Alternatively, inferred estimates of the rents generated by premium college football players may be biased downward for two reasons presented by Boreland, Goff, and Pulsinelli |1991~. First, the true marginal costs of scholarships are 10 percent to 50 percent of the reported scholarship values. Second, revenues tend to be under-reported or allocated to non-athletic accounts. For example, some or all of the concession and parking monies are often not allocated to athletics.(12)

The empirical results quantify the argument made by economists concerning the incentive for college football teams to capture rents by offering higher effective wages to attract better players. This often involves breaking NCAA regulations by offering outside payments and gifts to players, as well as committing other recruiting and academic abuses. Penalties for violating NCAA rules typically include a reduction in the number of allowable scholarships and sanctions on television and post-season bowl game appearances. The marginal revenue product estimates reported in Table IVb combined with the Table I summary statistics allow for informal comparisons of sanctions imposed on NCAA violators. For example, the NCAA allows Division I-A football teams twenty-five new scholarships per year and a total of ninety-five. Assuming the best players receive scholarships, on average over 8 percent of these will be drafted. If a one-year 10 percent reduction in scholarships leads to a decrease of one NFL caliber player then the cost imposed upon violators may be up to $600,000 in lost annual revenues, or perhaps over two million dollars for a four-year career. (Although a decrease in allowed scholarships does not imply that less are allocated to top recruits, likely players of lesser quality will not be offered scholarships.) Alternatively, a one-year sanction on television appearances costs a team on average $543,925 in lost revenues according to Table I. In any case, a more rigorous analysis must include other factors such as revenue sharing which provides a form of insurance against sanctions, as well as the effects of sanctions on future team performance.

IV. CONCLUSION

Organizations such as the NCAA, the Knight Foundation, and the United States Congress have recently proposed a number of additional regulations and restrictions aimed at maintaining the competitiveness and amateurism of college athletics. Some examples include reductions in the number of permissible grants-in-aid for all Division I sports, limiting the hours per week spent on athletic competition and practice time, and raising standards for admissions and academic progress for all NCAA institutions. Federal legislation has been passed requiring all Division I-A public institutions to disclose the graduation rates and SAT scores of all athletes competing in revenue producing sports, as well as to report operating revenues and expenses (Student Right-to-Know and Campus Security Act of 1990; Higher Education and Amendments Act of 1992). Economic theory predicts that incentives exist to violate NCAA rules as long as there are rents to be captured by offering athletes effective wages above the NCAA limits. This paper shows that these rents may be substantial for premium college football players.(13) Accordingly, policies aimed at reducing NCAA violations and other abuses must lessen these incentives and/or raise the costs of cheating by sufficiently monitoring teams and penalizing offenders. There are reasons to be believe that present reform policies may fail on both grounds.

First, the NCAA Committee on Infractions is responsible for enforcing all rules and regulations. Actual investigations of any of the over 800 institutions monitored are conducted by the Compliance and Enforcement Department, consisting of fifteen full-time NCAA investigators and an additional forty privately contracted part-timers. Given the limited staffing of the Compliance and Enforcement Department, the NCAA alone cannot effectively monitor all schools.(14) Second, by concentrating on additional restrictions, recent reforms do not mitigate the incentives arising from the monopsonistic nature of the college player recruitment market. If the probability of detection is expected to be low, then more stringent regulations mean that the gains to violating the rules are greater. As a result of the incentive structures, ineffective monitoring devices, and sizeable rents to be captured, recent reform packages will likely fail to discourage recruiting and academic abuses.

APPENDIX
TABLE V

Simple Correlation Coefficients

 OPPONENT-
 DRAFT MARKET POOL POOL RANK 85-87

MARKET 0.31

POOL 0.10 0.54

OPPONENT
POOL 0.01 0.11 0.23

OPPONENT-
RANK 85-87 0.35 -0.12 0.07 -0.28 0.33

RANK 85-87 0.57 0.25 -0.04 -0.11 1.00
TABLE VI

Yearly Top-20 Ranking Correlation Coefficients

 RANK RANK RANK RANK RANK RANK RANK
 1987 1986 1985 1984 1983 1982 1981

RANK 1987
RANK 1986 .689
RANK 1985 .555 .662
RANK 1984 .644 .624 .650
RANK 1983 .478 .620 .595 .614
RANK 1982 .531 .490 .230 .535 .563
RANK 1981 .485 .595 .362 .490 .684 .706
RANK 1980 .642 .595 .507 .651 .706 .637 .666


1. Fleisher, Goff, Shughart, and Tollison |1990~.

2. See Brown |1992~ for a treatment of revenue sharing incentives in college football. There it is shown empirically that equilibrium team skill levels are positively related to the fraction of home revenues teams are allowed to retain.

3. Other estimates of marginal revenue products in professional sports follow similar methods. See Rottenberg |1956~, Medoff |1976~, Cairns, Jennett, and Sloane |1986~, and Kahn |1991~ for a survey of discrimination in professional sports.

4. See Bradford and Kent |1986~ for a comprehensive presentation of spacial interaction models.

5. See Table VI in the appendix for Top-20 point ranking correlations over time.

6. A detailed discussion of measuring revenue retention rates is available in Brown |1992~.

7. I thank an anonymous referee for this comment.

8. See Greene |1990~ and Maddala |1992~ for discussions of nonnormality tests. For a nonlinear first-stage regression the fitted values would be used as instruments in the second-stage. Since the number of future draftees are ordered, but are not categorical, a Poisson regression model would be the most appropriate procedure in the first-stage.

9. In consideration of the case of nonlinear first-stage results, a Poisson draft equation and the instrumental variables marginal revenue estimates were generated. The DRAFT coefficient was consistently positive and significant with marginal revenue product estimates between $722,070 and $482,640.

10. The significance level of the DRAFT coefficient did not change for different values of |Lambda~. However, the coefficient decreases, for example, in specification (i) from 850,082 for |Lambda~=0 to 604,541 for |Lambda~=0.70. The reported marginal revenue product results use a |Lambda~=0.35 which satisfies the Amemiya |1980~ criterion.

11. Agents receiving the rent transfers may include coaches and administrators in the form of higher salaries, the university general fund, or non-revenue producing athletic programs, among others.

12. Boreland, Goff, and Pulsinelli |1991~ also report that the largest missing revenue is usually money that athletic foundations pay universities for scholarships and is not reported as revenues since it goes into the general fund. This may exceed $3 million at large programs.

13. As pointed out by Brian Goff, the rents may be dissipated to some degree on expenditures for training facilities, equipment, coaches salaries, as well as illegal recruiting practices.

14. Lawrence |1987~ details NCAA monitoring procedures and ineffectiveness, and also Koch |1983~. Fleisher, Goff, Shughart, and Tollison |1990~ show that inferences and investigations on illegal practices are made from observing team performance. Greater variability in performance leads to suspicion, increasing a team's chances of being investigated.

REFERENCES

Amemiya, T. "Selection of Regressors," International Economic Review, 21, 1980, 331-54.

Atkinson, Scott A., Linda R. Stanley, and John Tschirhart. "Revenue Sharing as an Incentive in an Agency Problem: an Example from the National Football League." RAND Journal of Economics, 19(1), 1988, 27-43.

Becker, Gary S. "College Athletes Should Get Paid What They Are Worth." Business Week, September 1985, 18.

Boreland, Melvin V., Brian L. Goff, and Robert W. Pulsinelli. "College Athletics: Financial Burden or Boon?", Working paper, Western Kentucky University, 1991.

Bowman, K. O., and L. R. Shenton. "Omnibus Test Contours for Departures from Normality Based on b1 and b2." Biometrika, 62(2), 1975, 243-50.

Bradford, M. G., and W. A. Kent. Human Geography: Theories and their Applications. Oxford: Oxford University Press, 1986.

Brown, Robert W. "Revenue Sharing and Incentives in College Football." Working paper, University of North Texas, 1992.

Cairns, J., N. Jennett and P. J. Sloane. "The Economics of Professional Team Sports: A Survey of Theory and Evidence." Journal of Economic Studies, 13(1), 1986, 3-80.

Fleisher, Arthur A., Brian L. Goff, William F. Shughart II. and Robert D. Tollison. 1990. "Crime or Punishment? Enforcement of the NCAA Football Cartel," in Sportemetrics, edited by Brian L. Goff and Robert D. Tollison. College Station, Texas: Texas A&M University Press, 1990, 153-71.

Fleisher, Arthur A., Brian L. Goff, and Robert D. Tollison. The National Collegiate Athletic Association: A Study in Cartel Behavior. Chicago, Illinois: The University of Chicago Press, 1992.

Greene, William H. Econometric Analysis. New York, N.Y.: Macmillan Publishing Company, 1990.

Kahn, Lawrence M. "Discrimination in Professional Sports: A Survey of the Literature." Industrial and Labor Relations Review, 44(3), 1991, 395-418.

Koch, J. V. "A Troubled Cartel: The NCAA." Law and Contemporary Problems, 38(Winter/Spring), 1973, 39-69.

-----. "Intercollegiate Athletics: An Economic Explanation." Social Science Quarterly, 64(2), 1983, 360-74.

Lawrence, Paul R. Unsportsmanlike Conduct. New York, N.Y.: Praeger Publishers, 1987.

Maddala, G. S. Introduction to Econometrics. New York, N.Y.: Macmillan Publishing Company, 1992.

Medoff, M. H. "Racial Discrimination in Professional Baseball', Atlantic Economic Journal, 3(April), 1976, 37-44.

Rooney, John F. Jr. The Recruiting Game. Lincoln, Nebraska: University of Nebraska Press, 1987.

Rottenberg, S. "The Baseball Player's Labor Market." Journal of Political Economy, 64(3), 1956, 242-58.

Scully, Gerald W. "Pay and Performance in Major League Baseball." American Economic Review, 64(6), 1974, 915-30.

ROBERT W. BROWN

Assistant Professor, University of North Texas. The author gratefully acknowledges the universities and colleges which released the financial data necessary for this study. Valuable comments on earlier drafts were received from Jon Sonstelie, Stephen Bronars, Steven Cobb, Brian Goff, William Shughart II, Stephen Trejo, Eleanor Brown, and anonymous referees. Research assistance was provided by Munawar Piracha.
COPYRIGHT 1993 Western Economic Association International
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1993 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Brown, Robert W.
Publication:Economic Inquiry
Date:Oct 1, 1993
Words:5640
Previous Article:Does monitoring increase work effort? The rivalry with trust and loyalty.
Next Article:The dark side of the force: Western Economic Association International 1993 Presidential Address.
Topics:

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