Professional Documents
Culture Documents
Contract Options
for Buyers
y and
Sellers of Talent in
Professional Sports
Duane W Rockerbie
Stephen T. Easton
Palgrave Pivots in Sports Economics
Series Editors
Wladimir Andreff
Emeritus Professor
University Paris 1 Panthéon-Sorbonne
Paris, France
Andrew Zimbalist
Department of Economics
Smith College
Northampton, MA, USA
This mid-length monograph series invites contributions between 25,000–
50,000 words in length, and considers the economic analysis of
sports from all aspects, including but not limited to: the demand for
sports, broadcasting and media, sport and health, mega-events, sports
accounting, finance, betting and gambling, sponsorship, regional devel-
opment, governance, competitive balance, revenue sharing, player unions,
pricing and ticketing, regulation and anti-trust, and, globalization. Sports
Economics is a rapidly growing field and this series provides an exciting
new publication outlet enabling authors to generate reach and impact.
Contract Options
for Buyers and Sellers
of Talent
in Professional Sports
Duane W Rockerbie Stephen T. Easton
Department of Economics Department of Economics
University of Lethbridge Simon Fraser University
Lethbridge, AB, Canada Burnaby, BC, Canada
© The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer
Nature Switzerland AG 2020
This work is subject to copyright. All rights are solely and exclusively licensed by the
Publisher, whether the whole or part of the material is concerned, specifically the rights
of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on
microfilms or in any other physical way, and transmission or information storage and
retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology
now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc.
in this publication does not imply, even in the absence of a specific statement, that such
names are exempt from the relevant protective laws and regulations and therefore free for
general use.
The publisher, the authors and the editors are safe to assume that the advice and informa-
tion in this book are believed to be true and accurate at the date of publication. Neither
the publisher nor the authors or the editors give a warranty, expressed or implied, with
respect to the material contained herein or for any errors or omissions that may have been
made. The publisher remains neutral with regard to jurisdictional claims in published maps
and institutional affiliations.
This Palgrave Macmillan imprint is published by the registered company Springer Nature
Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Preface
v
vi PREFACE
vii
viii CONTENTS
5 Concluding Remarks 85
5.1 An Empirical Test 86
5.2 Extensions to the Real Options Model 89
References 91
Index 93
List of Figures
Fig. 1.1 MLB average team payroll, average team revenue and
payroll share of revenue, 1980–2018 (Source https://sites.
google.com/site/rodswebpages/codes taken on December
20, 2019. These figures are taken from Financial World
and Forbes magazine) 2
Fig. 1.2 Monopsony model of talent acquisition (Source Author’s
creation) 17
Fig. 3.1 Possible states during the two-year player contract (Source
Author’s creation) 58
Fig. 3.2 Options available to the player after one season (Source
Author’s creation) 62
Fig. 4.1 Possible states during the three-year player contract (Source
Author’s creation) 70
Fig. 4.2 Possible paths for the player after two seasons (Source
Author’s creation) 72
Fig. 4.3 Player paths that give positive put option value for the
owner (Source Author’s creation) 75
Fig. 5.1 Scatter plot of surplus to team owner versus number of
years of player contract (Source Author’s creation) 87
ix
List of Tables
xi
CHAPTER 1
1.1 Introduction
Since the beginning of the 1990 season, the date at which systematic
financial data for Major League Baseball teams became available, it has
been a boom time for the professional sport. Attendance at games has
increased from an average of 26,000 per game to over 28,000 and overall
350 0.8
300 0.7
0.6
250
0.5
$ millions
200
0.4
150
0.3
100
0.2
50 0.1
0 0
1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018
Fig. 1.1 MLB average team payroll, average team revenue and payroll share
of revenue, 1980–2018 (Source https://sites.google.com/site/rodswebpages/
codes taken on December 20, 2019. These figures are taken from Financial
World and Forbes magazine)
league attendance has grown from 54.8 million to 68.5 million in 2019.1
Between 1990 and 2016 average revenue for clubs increased more than
sixfold, and even adjusting for inflation rose 550%. At the same time
payroll for the average major league club expanded even faster. From
1990 to 2016 the average payroll paid to players’ salaries rose by 800% in
real (inflation adjusted) terms.
Major League Baseball (MLB) finance has undergone considerable
change since the 1980s, most notably the rapid increase in salaries has
coincided with the even greater increase in revenues. Figure 1.1 plots
the average team payroll from 1980 to 2018, the average team revenue
from 1990 to 2018, and the payroll share of team revenue.2 The average
annual rate of increase of average team payroll is 28.4%. The average team
7 Table Ba4280-4282 in Historical statistics of the United States, earliest times to the
present: Millennial edition, edited by Susan B. Carter, Scott Sigmund Gartner, Michael
R. Haines, Alan L. Olmstead, Richard Sutch, and Gavin Wright. New York: Cambridge
University Press, 2006. http://dx.doi.org/10.1017/ISBN-9780511132971.Ba4214-454
410.1017/ISBN-9780511132971.Ba4214-4544.
1 ON THE RISE: PLAYER COMPENSATION AND MULTI-YEAR CONTRACTS 7
in 1914, and Home Run Baker, sold to the New York Yankees in 1916
for $37,500. Even a World Series appearance in 1931 could not keep the
team afloat, subsequently, Mack sold Mickey Cochrane for $100,000 to
the Detroit Tigers, Lefty Grove for $125,000 to the Boston Red Sox,
and Al Simmons, Jimmy Dykes and Mule Haas to the Chicago White
Sox for $100,000. These cash sales effectively moved the Athletics out of
contention by the 1934 season.
The reserve clause effectively eliminated the need for multi-year
contracts, even for the best players. However, there were a few excep-
tions. Data on player contracts is sparse for the early years of baseball all
the way up to the 1960s, and generally, the data that are available are
for the star players. Ty Cobb negotiated a three-year contract with the
Detroit Tigers in 1910 for an annual salary of $9000, significantly more
than the average player salary of $3000. Cobb did not report to spring
training for the 1913 season and missed a number of games while holding
out but received a one-year contract for what is thought to have been the
first five-figure salary in MLB history ($12,000). He negotiated a two-
year contract with Detroit in 1914 for annual salaries of $15,000 and
$20,000, and inked multi-year contracts thereafter until becoming the
player-manager of the Tigers in 1921. Cobb was arguably the top player in
MLB over the 1910–1920 seasons, achieving a weight-average8 batting,
on-base percentage and slugging percentage (weighted by at bats) of
0.382, 0.453, and 0.462, respectively, and was the American League (AL)
MVP in 1911.9 Cobb led the AL in batting average in nine of those
seasons and led in total hits in five. After being released by the Tigers
in 1927, Cobb signed a one-year contract for an MLB record salary of
$85,000 with the Philadelphia Athletics.10 His investments in real estate
and the stock market made Cobb a very wealthy man, far beyond his
wealth from baseball.
Yankee Stadium is still known as the “house that Ruth built,” despite
not being the original stadium in which Ruth played. That “house” was in
the Bronx, built in 1923, renovated in 1976, and closed in 2008. Ruth’s
legacy lives on through his amazing playing statistics and his financial
8 Each season’s batting average weighted by the number of at bats to form an overall
batting average for the period.
9 All playing statistics hereafter taken from http://www.baseball-reference.com.
10 To put this salary in perspective, the Dodger’s great Sandy Koufax earned the same
salary in his 1965 season, some 38 years after Cobb’s.
8 D. W. ROCKERBIE AND S. T. EASTON
rewards. Ruth signed a string of one-year contracts with the Boston Red
Sox and New York Yankees up to the 1922 season, at which point he
negotiated a three-year contract at $52,000 per year (40% of the Yankee’s
team payroll). A two-year contract followed in the 1925 season for the
same salary, followed by a three-year contract in the 1927 season for
$70,000 per season. Ruth played another four seasons for the Yankees,
each a one-year contract, until he retired. Ruth’s playing statistics over
the 1922–1930 seasons are phenomenal: a weight-average batting, on-
base percentage, and slugging percentage (weighted by at bats) of 0.351,
0.479, and 0.714, respectively, His season average of 46 home runs would
be impressive today, however Ruth played when the playing rules dictated
that hits that cleared the fences but curve outside the foul pole did
not count as home runs. He earned AL MVP honors in 1923 and the
Yankees won three World Series championships (appearing in six) over
1922–1930.
At the time of Babe Ruth, the Yankees “murderer’s row” lineup also
included Lou Gehrig. Over the 1926–1937 seasons, Gehrig averaged 37
home runs per season (weighted by each season’s at bats) and featured
a weight-average batting, on-base percentage, and slugging percentage
of 0.348, 0.457, and 0.651, respectively. He was AL MVP in 1927 and
1936 and won World Series titles in four of these seasons. Gehrig was
truly the ironman of MLB in his playing days, playing in almost every
regular season game. Despite these statistical accolades that are, in many
ways, comparable to Ruth, Gehrig was paid far less than Ruth and only
once negotiated a multi-year contract—a three-year deal in 1928 paying
him $25,000 per season. This was a significant increase in salary compared
to the $8000 he was paid in the 1927 season. Gehrig’s annual salary was
consistently in the $25,000–$30,000 range over much of his career with
the Yankees, suggesting he was severely underpaid.
Jimmie Foxx began his playing career with the Philadelphia Athletics
in 1925 and quickly established himself as one of the elite players in base-
ball. Foxx batted 0.354 in 1929, slugged 33 home runs and contributed
118 runs batted in (RBI). He was rewarded with a three-year contract
paying a total of $50,000 by Connie Mack, manager, treasurer, and part
owner of the Athletics. The team won World Series championships in
1929 and 1930 and Foxx won MVP titles in 1932, 1933, and 1938.
His career statistics compare very favorably to Cobb, Ruth, and Gehrig
in many respects, yet his annual salary averaged far less, likely due to the
financial difficulties constantly plaguing Mack and the Athletics.
1 ON THE RISE: PLAYER COMPENSATION AND MULTI-YEAR CONTRACTS 9
Many other star players toiled in the major leagues until the 1960s,
perhaps most notably Joe DiMaggio, Ted Williams, and Willie Mays.
DiMaggio (the “Yankee Clipper”) played for the New York Yankees for
the 1936–1942 and 1946–1951 seasons, with three years of military
service during World War II. Although an excellent player throughout
his career, DiMaggio’s prewar statistics are impressive: a 0.339 batting
average (weighted by at bats), 0.403 on-base percentage, and 0.606 slug-
ging average, earning MVP titles in 1939 and 1941.11 In salary history,
DiMaggio is best known as the first six-figure player in MLB history,
earning $100,000 for the 1949 and 1950 seasons. He is also known
for holding out at the start of the 1939 season, demanding a one-year
contract at $40,000 when the Yankees offered $25,000. With nowhere
else to play, DiMaggio accepted the Yankees offer with Yankees owner,
Colonel Jacob Ruppert, quipping “I hope the young man has learned his
lesson.”12 Despite having an outstanding career and being a tough nego-
tiator, DiMaggio never negotiated a multi-year contract. Ted Williams is
generally thought to be the highest paid player in MLB during 1951 to
1960 seasons, consistently earning $100,000 or more per season. One of
the greatest hitters to have ever played, Williams played his entire career
with the Boston Red Sox and negotiated a number of short multi-year
contracts toward the end of his career. Beginning in 1951, Willie Mays
played his entire 22-year career with the New York/San Francisco Giants,
winning MVP honors in the 1954 and 1965 seasons. Mays 0.345 batting
average and 41 home runs helped power the Giants to a World Series title
in 1954, but the team achieved few other successes during his lengthy
playing career. In his best season (1954), Mays earned $12,500 and was
severely underpaid afterward with a series of one-year contracts. A three-
year contract was agreed in 1966 that paid Mays $105,000 per season.
This contract set the standard for the famous dual contract hold outs
of Sandy Koufax and Don Drysdale in 1966. Believing that they could
negotiate more effectively as a pair, Koufax and Drysdale demanded three-
year contracts totaling $500,000 each, but ultimately capitulated to the
11 In 1941 DiMaggio also hit in 56 straight games, a record that still stands.
12 D. Gaffney, “DiMaggio’s contract hold-out”. https://www.pbs.org/wgbh/americane
xperience/features/dimaggio-contract-hold-out/.
10 D. W. ROCKERBIE AND S. T. EASTON
13 The Dodgers negotiated wisely with Koufax as he retired following the 1966 season
due to arm injuries.
14 https://www.usinflationcalculator.com/inflation/consumer-price-index-and-annual-
percent-changes-from-1913-to-2008/.
1 ON THE RISE: PLAYER COMPENSATION AND MULTI-YEAR CONTRACTS 11
15 The 1970 CBA also established the 10 and 5 rule. Players with at least ten years of
MLB service and the last five years with the same club could veto any trade. This rule
was put in place in response to the Curt Flood legal challenge. Although unsuccessful,
Flood’s challenge to the reserve clause brought attention to the contract strictures.
16 https://www.latimes.com/sports/mlb/la-sp-mlb-salaries-chart-20160329-story.html.
12 D. W. ROCKERBIE AND S. T. EASTON
Bowie Kuhn vetoed the deals arguing that the cash sales violated Article
1, Section 4 of the 1921 Major League Agreement that granted the
Commissioner powers to protect the interests and morale of the game. In
response to Kuhn’s decision, Finley refused to play his three star players
but eventually yielded to a threat by the team not to play any games
until the three players were reintroduced to the lineup. Significant cash
sales have been discouraged since the Kuhn decision and none have taken
place.
17 Average annual income in the United States was about $4000 in 1960.
18 Currently the average playing life is 3.3 years in the NFL. https://careertrend.com/
how-long-is-the-average-career-of-an-nfl-player-3032896.html Accessed May 4, 2020.
14 D. W. ROCKERBIE AND S. T. EASTON
deal totaled $427,000 (with a signing bonus, a new car, and other incen-
tives), a pro football record at the time. This contract set the stage for
other star players to sign lucrative multi-year contracts in the years leading
up to the NFL–AFL merger. Heisman Trophy winner O. J. Simpson
signed a five-year deal with the Buffalo Bills for a total of $650,000
in 1969. Competition for players between the NFL and AFL drove up
salaries, particularly for rookie players who also received large signing
bonuses. After the merger in 1970, salary growth and contract lengths
were reduced until the abolishment of the reserve clause in the late 1970s.
The longest contract in NFL history is Donovan McNabb’s 12-year deal
with the Philadelphia Eagles in 2002 for a total of $115 million, and 8- to
10-year deals are not uncommon today for the league’s best star players.
However, NFL contracts are not guaranteed unless a player negotiates
guarantees in the contract.
Prior to the start of the NBA, George Mikan signed a five-year contract
with the Chicago American Gears of the National Basketball League
(NBL) for an annual salary of $12,000, a record at that time for a
professional basketball player. Mikan’s 6-foot 10-inch height made him a
dominant player in the NBL. The NBL morphed into the Basketball Asso-
ciation of America (BAA) that morphed into the NBA. The NBA played
its first season in 1949, making it the youngest of the four major profes-
sional sports leagues in North America. The 1950s witnessed unstable
franchises and low attendances for most teams. Like the NFL, the NBA
competed with the college game and was a distant second in popularity.
Bob Cousy of the Boston Celtics was one of the best players in the young
league, but only negotiated a one-year deal for $26,000 in 1959. Bill
Russell and Wilt Chamberlain each earned $100,000 for the 1966 season,
but only on one-year contracts. It was not until competition with the
upstart American Basketball Association (ABA) in 1968 that contracts
increased in value and length, despite the reserve clause. The move to the
new league started with the case of Rick Barry who signed a three-year
contract with the ABA Oakland Oaks worth $500,000 plus a 15% owner-
ship share and 5% of gate revenue. Other new players followed to the new
league for lucrative multi-year contracts, including Julius Erving (4 years,
$500,000), Artis Gilmore (10 years, $2.5 million) and David Thompson
(3 years, $1.45 million). Salaries and contract lengths continued to climb
after the NBA–ABA merger in 1979.
Kobe Bryant, Shaquille O’Neal, Tim Duncan, and Chris Webber all
share the longest contracts in NBA history at seven years. Bryant’s deal
1 ON THE RISE: PLAYER COMPENSATION AND MULTI-YEAR CONTRACTS 15
with the Los Angeles Lakers paid a total of $136.4 million commencing
in 2004, with O’Neal’s contract with the Lakers totaling $120 million in
1996. Tim Duncan signed a seven-year deal worth $122 million in 2003
with the San Antonio Spurs. The Sacramento Kings inked Chris Webber
to a deal worth $123 million in 2001. However, like the NFL, the NBA
is not friendly to longer-term contracts, perhaps due to the uncertainty
of injury and the small roster size for each team (17 players) relative to
the number of available players in college to be drafted. Many NBA star
players negotiate four- and five-year deals, but few longer than that.
The NHL (incepted in 1917) has a better-documented history than
the NFL or NBA. Again, one-year contracts were the norm in the league,
with a few exceptions for star players. Maurice Richard signed a two-
year contract with the Montreal Canadiens in 1943 worth $5000 per
season (with performance bonuses) and signed two-year deals until the
mid-1950s. The Montreal Canadiens also signed Jean Beliveau in 1953
to a three-year contract worth a total of $100,000, a large amount for its
times. Beliveau had a strong business relationship with Molson Breweries
in Montreal and could have left the Canadiens to assume an executive
position, hence his opportunity cost to play hockey was significant and
the Canadiens knew it. Bobby Hull negotiated a five-year contract with
the Chicago Black Hawks (NHL) in 1960 for $20,000 per season. The
Detroit Red Wings signed Doug Harvey to a three-year contract as a
player-coach for $27,000 per season. Bobby Orr, one of hockey’s greatest
players, signed a three-year with the Boston Bruins in 1968 for $50,000
per season. The establishment of the rival World Hockey Association
(WHA) in 1972 created a higher level of competition for star players
that increased salaries and contract lengths rapidly. The most notable of
these being the ten-year, $1 million contract signed by Guy Lafleur with
the Montreal Canadiens, the largest NHL contract up to that time, and
the ten-year $2.75 million contract signed by Bobby Hull with the WHA
Winnipeg Jets.
The longest multi-year contracts we could find in the four professional
sports leagues in North America were the 15-year NHL contracts signed
by Rick DiPietro of the New York Islanders in 2006 and Ilya Kovalchuk
with the New Jersey Devils in 2010. DiPietro was to receive $4.5 million
per season, however the Islanders bought out his contract in 2013 due
to his frequent injuries. Under the terms of his buyout, DiPietro will
receive $1.5 million per year until 2026. Kovalchuk turned down a 12-
year $101 million contract with the Atlanta Thrashers to sign with the
16 D. W. ROCKERBIE AND S. T. EASTON
Devils for a total of $100 million. The Devils initially agreed to a 17-year
contract, but the terms of the deal violated the NHL collective bargaining
agreement and was disallowed. Just three years into the deal, Kovalchuk
retired from the NHL to play hockey in his native Russia.
The wage rate per unit of talent is Z i and fixed costs are FCi . Maxi-
mizing profit with respect to stock of team talent gives the first-order
condition
∂ Ri ∂wi d Zi
− ti + Z i (ti ) = 0 (1.2)
∂wi ∂ti dti
1 ON THE RISE: PLAYER COMPENSATION AND MULTI-YEAR CONTRACTS 17
The first term in (1.2) is the MRP of talent and the bracketed term is
the MRC that includes the wage rate per unit of talent plus the increase
in the wage rate paid to all units of talent when one more unit of talent
is acquired (the total being the MRC). Figure 1.2 demonstrates that a
profit-maximizing owner will page a wage rate per unit of talent Z i when
acquiring a team stock of talent Ti , much less than the MRP, hence players
are “exploited.” It is important to note that it is a mistake to conclude
that each player will be paid his MRP in the absence of monopsony
power. Each player contributes a different MRP of each unit of talent,
but the wage rate per unit of talent is determined by the last unit of
talent acquired. The player’s salary is then the profit-maximizing value of
Z i multiplied by the amount of talent units ti j that player j contributes to
the team stock of talent. More talented players will earn higher salaries,
but in equilibrium, the MRP is the same for all units of talent and thus
players. If all players earned their MRPs, there would be no surplus left
for the owner, however in the case of the monopsonist, the owner earns
an additional area given by (M R Pi − Z i )Ti .
Assuming specific functional forms and estimating slope coefficients for
the independent variables allows for the computation of the MRP for each
player on the roster. The first step is to estimate the team revenue func-
tion. This is straightforward given the availability of data. Unfortunately,
18 D. W. ROCKERBIE AND S. T. EASTON
comprehensive revenue data was difficult to obtain 50 years ago, and the
Forbes magazine estimates were not produced at that time. Scully (1974)
computed gate revenue by multiplying an average ticket price by atten-
dance for each team, then added the share of national broadcast revenue.
This introduces two problems. First, the average ticket price should be a
weighted average of ticket prices, the weights being the share of seats in
each seating section. It is not clear if the original Scully (1974) study used
weighted average ticket prices or just some sort of other average. Second,
Scully admits that his revenue data were not post-revenue sharing. Up to
the late 1990s, visiting clubs received a fixed share of gate revenue in the
American League and a fixed price per ticket sold in the National League.
Team revenues that are available to the owner to pay players will be over-
estimated for teams above the average team revenue and underestimated
for teams that fall below. This is true whether the revenue sharing formula
is the current pooled system, or the old gate-sharing system (Rockerbie
2009a), however in the latter system, the ranking of each team in the
league revenue hierarchy can change using team revenue after sharing.
Scully specified a revenue function and a winning percentage function
that are each linear in their components. This could be an issue if one is
concerned with concavity of the revenue function (diminishing marginal
revenue), however if the winning percentage function is nonlinear and
concave, a downward MRP schedule is still assured. The A-vector in the
Scully method is composed of the same independent variables that we
employ in Chapter 2 and summarized in an appendix.19 Scully found that
a one percentage point increase in the team winning percentage was esti-
mated to increase team revenue by $10,330 and the regression model fit
reasonably well with an adjusted R2 of 0.75.
The winning percentage function is also linear in its components with
the same independent variables that we employ in Chapter 2 and summa-
rized in an appendix. The fit was quite good with an adjusted R2 of 0.88
and slope coefficients for the team slugging average and team strikeout to
walk ratio of 0.92 and 0.90, respectively. The linearity of the revenue and
winning percentage functions result in a perfectly elastic MRP schedule, a
problem that has been corrected in later studies described in Chapter 2.20
above their estimated MRP. This does not fit with the standard monop-
sony model. One explanation is that some star players bring more to a
team than just their on-field performance. Acquiring a superstar player
could result in higher local media revenue, apparel revenue, promotion
revenue, or even result in a team developing a world brand. In other
sports, there are players who bring a sort of cachet to an organiza-
tion to which fans respond. Christiano Ronaldo and Neymar’s move to
Juventus and Paris-St. Germain in European football came with record
transfer fees (112 million euros and 222 million euros) that give an indi-
cation of their worth to their clubs beyond their performance on the
field. David Beckham’s move from Real Madrid to LA Galaxy brought
tremendous exposure to a previously mediocre Major League Soccer in
North America. The top European clubs can earn large bonus revenues
from competing deep into the annual Champion’s League competition.
Shirt sales often garner enough revenue to recover a significant portion
of salaries for these superstars.
Today there are few examples in MLB of these players whose popu-
larity extends well beyond their performance on the field. Babe Ruth,
Joe DiMaggio, and Ted Williams are players from the distant past now.
The skew in the salary–MRP relationship is more systematic in MLB,
suggesting that there is an important feature of bargaining that did not
exist in the reserve clause period. One factor at the aggregate level is
that the talent market cannot be characterized as a monopsony for free-
agent players. Competitive bidding between clubs drives a player’s salary
much closer to their expected MRP and could drive it above that if
owners are willing to bid away some or all of their expected surplus
(Rockerbie 2009b). This sort of “winner’s curse” argument could hold
for the very best free agents, but not likely for the large number of free
agents observed to be “overpaid” in Chapter 2.
References
Haupert, M. (2012). MLB’s annual salary leaders since 1874. Society for Amer-
ican Baseball Research (SABR) Newsletter. https://sabr.org/research/mlbs-
annual-salary-leaders-1874-2012.
Kahn, L. (2000). The sports business as a labor market laboratory. Journal of
Economic Perspectives, 14(3), 75–94.
Rockerbie, D. (2009a). Free agent auctions and revenue sharing: A simple
exposition. Journal of Sport Management, 23(1), 87–98.
1 ON THE RISE: PLAYER COMPENSATION AND MULTI-YEAR CONTRACTS 21
2.1 Introduction
One of the important issues in the sports economics literature is whether
players are paid according to what economic theory predicts: profit-
maximizing firms pay a salary equal to a player’s marginal revenue product
(MRP). Scully (1974) tackled the problem decades ago during the last
remnants of the reserve clause era in MLB. His method and results are
reviewed in the last chapter, and his paper sparked a burgeoning literature
1 These revenue data are not without their limitations. See Zimbalist (2010) for an
in-depth analysis of salary and revenue data. Forbes revenue estimates are net of revenue
sharing of local revenues and net of debt payments on stadiums paid by the teams. See
Ouzanian and Badenhausen (2020) for a brief description of the methodology of revenue
calculations and team valuations.
2 The Designated Hitter (DH) position for American League teams was not used since
most teams did not employ a full-time DH, rather they split time among several players
at different fielding positions with the DH position.
26 D. W. ROCKERBIE AND S. T. EASTON
2000 8 3 6 3 5 9 34
2001 7 5 4 5 4 13 38
2002 6 5 5 4 5 21 46
2003 7 7 6 5 3 14 42
2004 12 9 9 8 7 32 77
2005 11 7 6 5 9 19 57
2006 8 9 7 5 6 16 51
2007 9 9 14 10 3 25 70
2008 7 3 4 5 3 14 36
2009 7 8 5 4 7 17 48
2010 12 11 8 7 7 19 64
2011 15 8 4 6 6 19 58
2012 9 9 9 3 7 20 57
Total 118 93 87 70 72 238 678
that higher-priced free agents, defined as those paid above the league
average salary for all players, tend to be overpaid (or their expected MRP
overestimated) while other free agents are underpaid or paid appropri-
ately. Berri et al. (2015) found that NBA players were overpaid over
the 2001–2011 sample period, resulting in negative surpluses for team
owners. They suggest that the MRP rule works to some extent and that
players earn part of the owner’s surplus that arises from fixed revenues
through a bargaining process. Krautmann and Solow (2018) found that
most MLB free agents were overpaid using a different methodology to
estimate player MRPs. These results contrast sharply with the belief that
the player’s market is a monopsony (Humphreys and Pyun 2017; Leeds
and Leeds 2017). Krautmann et al. (2009) found that players with low
bargaining power tend to be underpaid in the National Football League,
the National Basketball Association, and MLB. Our empirical results in
this paper substantiate this claim and provide a number of further insights,
the most important being that free agents signing multi-year contracts
are overpaid relative to those signing one-year contracts. In Chapter 4 we
provide a theoretical structure to help explain this apparent anomaly.
2 THE PUZZLE OF OVERPAID AND UNDERPAID PLAYERS 27
R = X α (W P)β (2.1)
where
3 Some of these are the Blue Ribbon Report and follow-up report commissioned by
MLB in 1999 and 2000. Deadspin leaked financial statements for several MLB teams for
specific seasons but not for long enough periods of time to be used in our study. See
https://sites.google.com/site/rodswebpages/codes for all of the Deadspin releases.
4 New or extensively renovated stadiums since 2001–2012 included Cincinnati, Miami,
Milwaukee, New York (NL), New York (AL), Philadelphia, Pittsburgh, San Diego, St.
Louis, Kansas City, and Minnesota. New stadiums opened before the start of the 2000
season were not included, even though the 5-year period extended into the sample period.
28 D. W. ROCKERBIE AND S. T. EASTON
2005), and three dummy variables (TV1, TV2, and TV3) that take on
the value one in each season during or after a new national television
contract (2001, 2007, and 2012, respectively). Finally, team revenue, per
capita income, and the average ticket price were deflated by an individual
state consumer price index (2000 = 100). A list of sources for all variables
is contained in Table 2.2.
The revenue function in (2.1) is a departure from typical linear revenue
functions utilized in the literature, however we see it as an improve-
ment. Revenue functions that are linear in the winning percentage assume
strong separability and a completely elastic marginal revenue schedule.
The more general revenue function in (2.1) allows for a downward
sloping marginal revenue schedule that shifts with changes in the inde-
pendent variables. Since market conditions can differ between cities and
can change over time, it is reasonable to assume that the MRP schedule
is not constant.
The function relating team winning percentage to player performance
is given by the logistical function:
where
These stadiums included Detroit, Houston, San Francisco, and Seattle. It was thought best
not to include these as the variable would not capture what their revenues were prior to
the new stadium opening.
2 THE PUZZLE OF OVERPAID AND UNDERPAID PLAYERS 29
The independent variables OPS, K/BB, and ERA are measures of hitting
and pitching quality that have been justified by Scully (1974) and others
elsewhere. The independent variables CONT and OUT were also used by
Scully (1974) and others as proxies for team morale or perhaps manage-
rial ability. They represent a shift in the team production function based
on factors other than player performance. We have retained their use to
allow for comparison of our results to previous work. An earlier version
of the model also included fielding percentage (outs per fielding attempt)
30 D. W. ROCKERBIE AND S. T. EASTON
θ W P(1 − W P)). Teams with a 0.500 winning percentage will show the
largest marginal effect, holding player performance and team morale
constant. We chose to use the fitted value for the current season’s winning
percentage for each team as the baseline winning percentage. This places
each team on a logistic function at different positions with different
marginal effects. This could not be done with a linear probability model
and we think it is an improvement.
Taking the natural logarithm of (2.1), the calculation for MRP is
straightforward once (2.1) and (2.2) have been estimated. For a specific
hitter denoted as player i playing on team j, MRP is given by (where AB
denotes at bats):
M R Pi = (∂ Ri /∂ W Pi )(∂ W Pi /∂ O P Si )d O P Si j (2.3)
1 α β
dR = dX + dW P
R X WP
The vector X is not a function of WP by assumption so dX = 0. Multi-
plying by R and dividing by dOPS gives the first required derivative in
(2.3), d Od RP S = R WβP ddOWPPS . Totally differentiating (2.2) and focusing only
on the OPS term gives
1−WP 1
dW P = θ d O P S
WP (1 − W P)2
1
dW P = θ d O P S
W P(1 − W P)
dW P
= W P(1 − W P)θ
dOPS
Inserting the necessary derivatives into (2.3) gives
ABi
M R Pi j = β θ R j 1 − W P j O P Si j (2.4)
AB j
6 OPS is on-base percentage plus slugging percentage. The former is based on total plate
appearances, while the latter is based on total at bats. Total plate appearances includes
walks, while total at bats does not. We chose to use total at bats for the player contribution
to the team OPS, although we admit the choice is arbitrary.
7 Since the team OPS is calculated by T eam O P S = N (AB /T eam AB) × O P S
i=1 i i
we use the right-hand side value for each player as the player’s contribution. The Scully
32 D. W. ROCKERBIE AND S. T. EASTON
method uses (ABi /T eam AB) × T eam O P S which does not utilize any information about
the individual player and is not reflective of how the team OPS is calculated. The original
Scully (1974) paper used team batting average, not team OPS, but the team calculation
is the same.
2 THE PUZZLE OF OVERPAID AND UNDERPAID PLAYERS 33
left in place on the reasoning that each owner can more or less predict
team revenues after revenue sharing fairly accurately and base their talent
acquisition and payroll decisions on available revenues after sharing. Team
fixed effects were included to account for variations in team revenue
not explained by the independent variables in (2.1). Team fixed effects
were included, but year fixed effects were not.8 A weighted least squares
method was used to account for heteroskedasticity across teams. Although
Scully (1974) included a dummy variable to distinguish between National
League and American League teams, we did not. The National and Amer-
ican League used the same revenue sharing formula over our sample, so
we see little reason to include the dummy variable in (2.1). After testing
for and rejecting nonlinearity, we report the linear regression results (t-
ratios appear in parentheses) in Table 2.3. To test for robustness of the
team revenue model, results are also reported for the 2000–2008 and
2004–2012 subsamples.
The team fixed effects coefficient estimates are omitted from Table 2.3
for the sake of brevity. The high significance of the F-statistic and the very
acceptable adjusted R2 suggests that the log-linear model is appropriate
for explaining variations in team revenue. All the regression coefficients
are statistically significant at 95% confidence with the exception of the
state unemployment rate. The unemployment rate is a proxy for economic
conditions and the results suggest that MLB team revenues are not sensi-
tive (a-cyclical) to the business cycle. The team winning percentage is
the key variable in the marginal revenue calculation and its elasticity esti-
mate of 0.523 suggests that a ten percent increase in winning percentage
(say from 0.5 to 0.55) raises team revenue by about 5.23% indicating
that revenue is quite inelastic with respect to winning percentage. Given
that approximately 90% of teams in any given season fell within winning
percentages of 0.4 and 0.6, large revenue swings due to team performance
are not a frequent occurrence in MLB.
A new stadium has no significant effect on team revenue, but has a
significantly negative effect on revenue in the 2004–2012 seasons. This
result is counter to the short-run effects on attendance found by Coffin
(1996), Clapp and Hakes (2005), and Coates and Humphreys (2005).
8 The inclusion of year fixed effects created vectors of zeroes and ones whose combi-
nation was perfectly correlated with the combination of vectors from the NEWSTA, TV1,
TV2, and TV3 independent variables. Rather than omit one of these variables, it was
thought best to omit the year fixed effects.
34 D. W. ROCKERBIE AND S. T. EASTON
However, Rockerbie and Easton (2019) find the same result on team
performance using a much longer sample period and a model that incor-
porates the availability of talent. The lack of statistical significance is likely
due to the inclusion of the real average ticket price in the team revenue
function. The previously cited papers did not include the real average
ticket price. A regression of the real ticket price on the opening of a
new stadium suggests that the average real ticket price increased by an
average of 20.8% (p-value = 0.001) in the first season. If ticket demand
is inelastic, the revenue increases noted by the previously cited papers will
show up as an increase in the average ticket price. Removing the real
average ticket price from the regression of (2.1) reveals a coefficient for
NEWSTA of 0.088 with a t-statistic of 4.15, suggesting that real team
revenue increased by 9.2% following the construction of a new stadium.
We chose to include the real average ticket price since it provides a useful
elasticity estimate.
2 THE PUZZLE OF OVERPAID AND UNDERPAID PLAYERS 35
Real per capita income has the largest positive effect on revenue
with an estimated elasticity 0.873, confirming that baseball’s composite
commodity is a normal good. Ticket price has a positive effect on revenue
that is inelastic (0.457), confirming that marginal revenue is less than the
ticket price. Each new national television contract increased team revenue
independently of the other variables, with percentage increases of 0.158,
0.339, and 0.439, respectively.9
The coefficients for the 2004–2012 subsample are all smaller than
those for the 2000–2008 subsample, although the statistical significance
is virtually unchanged. This can be explained by the declining annual real
revenue growth that occurred over the 2000–2012 seasons. Revenues
grew by an annual average of 7.3% in 2000–2008 but fell to 5.6% in
2004–2012. With the weighted average ticket price (P ) held constant,
revenue growth could have slowed from a decrease in attendance growth
and/or decreases in the growth of other revenue sources (concessions and
other local revenues).
The winning percentage function (2.2) was also estimated using fixed
effects and weighted least squares to account for heteroskedasticity. The
model was estimated separately for the National and American Leagues
owing to the use of the designated hitter in the American League. Team
hitting statistics are somewhat higher and pitching statistics are somewhat
lower for the American League relative to the National League where
pitchers are required to hit. The results for each league are given in
Table 2.4. Again, the fixed effects estimates are omitted for the sake of
brevity.
The degree of fit was quite high for both leagues and the F-statistic
for each regression suggests that the model specification is appropriate.10
All of the coefficients carried the expected signs and all of the coefficients
were statistically significant with the exception of the strikeout–walk ratio
for the National League. The marginal effects can be found dividing each
coefficient by 4 since the linear marginal effect is given by βi W P(1 − W P)
and W P = 0.5 over the entire sample. Since we focus on the perfor-
mance of hitters, the OPS variable is of key interest. A 100 point increase
in team OPS is predicted to increase the team winning percentage by
9 The effect on revenue expressed as a percentage change when using a dummy variable
is eb − 1 where b is the slope coefficient.
10 This measures the degree of fit for the nonlinear dependent variable, but not for the
predicted winning percentages themselves.
Another random document with
no related content on Scribd:
hauskoja muistoja noilta vuosilta, jolloin opetin sinua, ja olet sanonut
minulle useammasti kuin kerran, että elämäsi iloisin päivä tulee
olemaan se, jolloin Heven maakunnan jokainen asukas osaa lukea
ja kirjoittaa».
Andráksen ote heltisi äkkiä, hän laski vanhan ystävänsä käden irti
ja hänen katseensa synkistyi huomattavasti.
Mutta sen sijaan olikin Ilonka, joka nähtävästi oli vielä heikko ja
sairas kaikesta kokemastaan, kieltäytynytkin kertomasta sanaakaan
lyhytaikaisesta olostaan miehensä kodissa. Hän oli ottanut vaitiollen
vastaan vanhempiensa suutelot, vaitiollen oli hän istuutunut entiselle
paikalleen pöytään ja tyynesti ruvennut jälleen asumaan entisessä
huoneessaan yön poissaolon jälkeen ja jatkamaan tuskin
keskeytyneitä hommiaan.
Kreivillä oli nykyään aina runsaasti rahaa, ja tämän vuoden sato oli
ollut niin runsas, että se oli täydellisesti korvannut viimevuotisen
tulvan aiheuttamat vahingot.
Keményn András näytti todellakin ymmärtävän, miten maatilaa oli
hoidettava. Ja sehän olikin luonnollista, sillä kaikki talonpojat
ymmärsivät aina, miten vehnää ja turnipseja oli kasvatettava. András
hoiti kaiken eikä kreivin tarvinnut muuta kuin iloita saavutuksista ja
vuosituloistaan, joita hän ei enää jaksanut kokonaan tuhlatakaan.
»Odotan».
»Ketä sitten»?
»Erästä liiketuttavaani».
»Isä, etkö halua kertoa minulle, mikä tuo 'kiitollisuuden taakka' on,
jonka minä ja me muut kaikki olemme velkaa miehelleni»?
»Ilonka, olet kuunnellut»!
»Tiedät aivan hyvin, isä, ettei hän kerro minulle mitään. On aivan
hyödytöntä taistella sitä vastaan, rakkaani, sillä en aio poistua tästä
huoneesta, ennenkuin saan tietää, mitä tahdon».