LIBRARY

Micluvw. Qm’re

yous

University

 

 

 

This is to certify that the
dissertation entitled

EMPIRICAL TESTS OF CONTRACT THEORY

presented by

Meng-Chi Tang

has been accepted towards fulfillment
of the requirements for the

PhD. degree in Economics

 

 

Major Professor’sTSignature

glam/1.7,. atom

Date

MSU is an Affirmative Action/Equal Opportunity Employer

 

- ---nnp.—-._.-._.-.-._.-

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 K:lProj/Acc&Pres/ClRClDaIeDue Indd

EMPIRICAL TESTS OF CONTRACT THEORY
By

Meng-Chi Tang

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILSOPHY
Economics

2009

ABSTRACT

EMPIRICAL TESTS OF CONTRACT THEORY
By
Meng-Chi Tang
Contract theory consists of two major parts: the theory of economic institutions and the
theory of incentives and information. While most of the related literature focuses on the
later part and examines monetary terms, in my dissertation I empirically test the
implications from the theory of economic institutions by using information on non-
monetary contract terms. Specifically, in the first and second chapter, I extend the current
research on the theory of economic institutions by examining the determinants of contract
length and contract structures, using the contract information from the National Football
League (NFL), the franchising industry, and the coal market. In the third chapter, I
empirically test whether contract negotiations vary as a function of an agent’s experience

in the NFL.

ACKNOWLEDGEMENTS

Five years ago I entered the States for the very first time, 8,000 miles away from
home. The loneliness hit me immediately after I woke up in the midnight, in the tiny
room of the Owen Hall. It’s the time I realized I am completely alone, and I really missed
home.

Five years later I am going home, with this dissertation and my degree. It turns
out that I was only partially right in my first night here, because I had been getting to
know people here since the second day I was here. Some fiiends have taught me things
about this profession, and some friends have shared my spare time and personal feelings.
Eventually some friends have become both, and this dissertation is dedicated to them. To
my advisor, Mike Conlin, who spent a lot of time on this dissertation and also me, I really
appreciate your guidance on my way to the academics, and I feel really lucky to meet you.
To my dissertation committee members, Thomas Jeitschko and Jay Wilson, thanks for
spending a decent time on me and teaching me how to do theoretical works. To my
fellow Chinese colleagues, Yi, Wei-Chih, and Jingjing, Professor Davidson told us once
that he learned more from his classmates than fiom his teachers, I think I have felt the
same way fiom you guys. Not to mention the fun we have during every gathering. To my
other fellow students, Wei-Siang, Simon, Nick, J aesoo, Kwang-Myoung, Nichol, Brian,
Tom, Ilya. .. thanks for every conversation we have in the class or in the office. This
lonely dissertation-writing process became not that boring because of you.

I would not be at this stage without my parents’ continuous support. There is no
word that can express my appreciation, but I just want to say I am really lucky to be your

son. The final words are saved for Pan, who has always been there for me. Thank you.

iii

TABLE OF CONTENTS

LIST OF TABLES .................................................................................... v
LIST OF FIGURES ................................................................................. vii
INTRODUCTION .................................................................................... 1
CHAPTER I

CONTRACT LENGTH: EXPECTED SURPLUS, SPECIFIC INVESTMENTS, AND
UNCERTAINTY

Introduction .................................................................................... 3
Data and Institutional Details ............................................................... 7
Hypotheses ................................................................................. 15
Empirical Strategy ......................................................................... 17
Empirical Results ........................................................................... 20
Conclusion .................................................................................. 27
References ................................................................................... 28
CHAPTER H
THE CHOICE OF CONTRACT DURATION AND PRICING PROVISION IN THE
COAL MARKETS
Introduction ................................................................................. 31
Data and Institutional Details ............................................................. 35
The Choice of Contract Duration ......................................................... 41
The Choice of Pricing Provision ......................................................... 46
The Simultaneous Choice of Contract Duration and Pricing Provision ............ 52
Conclusion .................................................................................. 55
References ................................................................................... 57
CHAPTER HI

THE EFFECT OF AN AGENT’ S EXPERIENCE ON NATIONAL FOOTBALL
LEAGUE CONTRACT STRUCTURE

Introduction ................................................................................. 59
Data and Summary Statistics ............................................................. 62
Empirical Specification and Estimation ................................................. 67
Conclusion .................................................................................. 72
References ................................................................................... 73
Appendix ............................................................................................. 74

iv

LIST OF TABLES

Table 1 Summary Statistics of the NFL Contracts during 1986-1991
and 2001-2007.. ...................................................................................................... 9

Table 2 Summary Statistics of the Franchise Contracts Information
from Bond (2007) .......................................................................... 12

Table 3 Proxies for Expected Surplus, Specific Investments, and
Uncertainty in the NFL and the Franchise Contracts .................................. 17

Table 4 Binomial Regressions: Dependent Variables
(Contract Length of the NFL Contracts) ................................................ 21

Table 5 Poisson Regressions: Dependent Variables

(Contract Length of the Franchise Contracts) .......................................... 25
Table 6 Summary Statistics: Means and Standard Deviations ................................ 36
Table 7 Summary Statistics by Coal Production Region ...................................... 39
Table 8 Contract Duration by Pricing Provision: Means and Standard Deviations. . . . . ...40

Table 9: OLS and MLE (Truncated, Poisson, and Zero-Truncated
Poisson Regressions): Dependant Variable (Contract Duration) .................... 45

Table 10 Multinomial Logit Regression: Dependent Variable (Pricing Provision). . . . . ...50

Table 11 The Choice between Pricing Provision Pairs estimated by
Multinomial Logit Model ................................................................. 74

Table 12 Seemly Unrelated Regression: Dependent Variables
(Contract Duration and Pricing Provisions) ............................................ 53

Table 13 Correlation Matrix of Residuals from the Seemly Unrelated Regression ...... 54

Table 14: Descriptive Statistics (Means and Standard Deviations) ......................... 64

Table 15 Binomial Regression and Probit Models ............................................. 68

vi

LIST OF FIGURES

Figure 1 Contract Length of the NFL Contracts ................................................ 10

Figure 2 Average Contract Length and the Proportion of Non-Division IA

(N on-DIVIA) School Graduates ......................................................... 10
Figure 3 Percentage of Building Types Allowed for Franchise Outlet ...................... 14
Figure 4 Percentage of Franchisors by Business Sector ................. I ...................... 1 4
Figure 5 Contract Duration of the Coal Contract ................................................ 38
Figure 6 Average Contract Duration by Year ................................................... 38
Figure 7 The US. Bureau of Mines Coal-producing Districts ................................ 38

Figure 8 Number of Years Agent Certified Prior to Draft for
the 2001-2007 Contracts .................................................................. 66

vii

INTRODUCTION

In this dissertation, I empirically test the implications fiom the theory of
economic institutions, using information on non-monetary contract terms. Specifically, in
the first chapter, Contract Length: Expected Surplus, Specific Investments, and
Uncertainty, I empirically test how contract length is affected by expected surplus,
specific investments, and uncertainty. Using the contract information from the National
Football League and the franchising industry, I find that contract length increases with
expected surplus and specific investments. In addition, contract length increases with
uncertainty when the seller (player) is more risk averse than the buyer (team), and
decreases with uncertainty when the seller (franchisor) is less risk averse than the buyer
(franchisee).

In the second chapter, The Choice of Contract Duration and Pricing Provision in
the Coal Markets, I look at the contract terms of the coal contracts from 1979 to 2001.
Joskow (1987) examined the importance of specific investments to the contract duration
of coal contracts. Since different price—adjust mechanisms are adopted and provide
different incentives for the contracting parties over a long-term relationship, this chapter
examines the choice of contract duration as well as pricing provision of the coal contracts.
In addition to J oskow's finding that contract duration increases with specific investments,
the empirical results show that coal contract duration increases with expected surplus and
decreases with uncertainty. Moreover, the same determinants are also found to affect the

choice probability among different pricing provisions. The simultaneous estimation

confirms the empirical results are not significantly changed by the simultaneity between
the choice of contract duration and pricing provision.

In the third chapter, The Eflect of an Agent’s Experience on National Football
League Contract Structure, we use contract information on players drafted before and
after the implementation of the NFL salary cap and finds that experienced agents are
more likely to negotiate shorter contracts and are more likely have protracted negotiations
that extend beyond the start of the NFL team’s training camp. Our results also suggest
that, prior to the NFL salary cap, experienced agents were more likely to negotiate
contracts that include an incentive clauses. While our results indicate that the contract
structure and negotiation process vary systematically with the experience level of the
player’s agent, we are unable to infer from these results whether it is in the player’s best

interest to hire an experienced agent.

CHAPTER I
CONTRACT LENGTH: EXPECTED SURPLUS, SPECIFIC INVESTMENTS,

AND UNCERTAINTY.

INTRODUCTION

Why do some contracts have longer duration than other contracts? In a world with
unforeseeable contingencies, writing a complete contract is virtually impossible. Every
contract negotiation has to decide the duration of the relationship, considering the
tradeoffs associated with contract length such as renegotiation cost and flexibility in the
future. However, the empirical literature testing contract theory provides thin evidence on
the determinants of contract length due to the focus on incentive contracts and firm
boundaries.l Since contract length is one of the most important parameters of nearly
every contract, this paper provides evidence on how expected surplus, specific
investments and uncertainty affect the length of commercial and labor contracts.

Most of the research on contract length falls into the literature of transaction cost
theory, or incomplete contract theory.2 The central issue is how specific investments

relate to contract length. In particular, Klein, Crawford, and Alchian (1978, hereafter

 

Ii I am grateful to my advisor, Mike Conlin, for his continuous guidance and support to this project. I also
thank Thomas Jeitschko, Jay Wilson, Kent Miller, Jeff Wooldridge, and Jay Pil Choi for their suggestions,
as well as Rob Bond for his expertise on the franchising industry. All errors are mine. Financial support
from the Department of Economics at Michigan State University is greatly appreciated.
' See Chiappori and Salanié (2003) for a survey on testing contract theory, and Lafontaine and Slade (2007)
for a survey on the empirical literature of vertical integration. While the career concerns literature also
addresses the problem by relating long-term contracts with the moral hazard problem, it also falls into the
literature of incentive contract. For example, see the textbook treatment by Bolton and Dewatripont (2005,
.470)
See Guriev and Kvasov (2005) for a summary of the theoretical literature on contact length and specific
investments, and Shelanski and Klein (1995) for a survey on the empirical literature.

KCA) argued that if a long-term contract specifying the terms of fiiture transactions can
be provided ex ante, the level of opportunistic behavior can be mitigated ex post and thus
the specific investment can be made more efficient.3 The prediction has been tested in
primarily commercial contracts. For example, Joskow (1987) first tested the prediction
using the contracts between coal suppliers and electric utilities. He found that the more
important are specific investments, the longer the contract. Similar evidence is also found
in natural gas contracts (Crocker and Masten, 1988), franchise contracts (Brickley, Misra,
and Van Horn, 2006, hereafter BMV) and tenancy agreements (Bandiera, 2007), among
others.

The research relating contract length to uncertainty consists of labor and
macroeconomics literature that focuses on how longer contracts trade off the cost of
inflexibility against the cost of renegotiation. Theoretically, this relationship is
ambiguous. While Gray (1978) and Dye (1985) suggest that contract length is negatively
related to uncertainty, Harris and Hohnstrom (1987) and Danziger (1988) find that
contract length is positively related to uncertainty. The negative relationship argument is
supported by the eflicient production hypothesis, which indicates the contract wage will
be deviated away from the efficient wage when there is more uncertainty. To make the
production more efficiently, contract length should therefore decrease with uncertainty.
On the other hand, the positive relationship argument is supported by the eflicient risk-
sharing hypothesis (Danziger, 1988).4 The idea is that, when one of the contracting

parties is more risk-averse than the other, a longer contract will be more efficient in risk-

 

3 Cheung (1969) first proposed the idea that the tenancy lease duration is chosen to minimize transaction
costs, which is defined as the cost to secure and transfer the right to the income generated by private
investment, as well as the renegotiation cost.

4 Harris and Holmstrom (1987), on the other hand, emphasize the relationship between information
discovery process and contract duration.

sharing when there is more uncertainty. In addition, the empirical literature also finds
mixed evidence for the relationship between contract length and uncertainty, using
mainly the union-firm labor contracts (Wallace, 2001).

While the literature emphases the tradeoff between flexibility and renegotiation cost to
use a long-term contract, most of the empirical works focus on the measurement issue on
the future uncertainty while the renegotiation cost is given exogenously in the context of
labor-union contract.5 In this paper, I argue the amount of expected surplus from the
contracting relationship, which is defined as the willingness to pay of the buyer minus the
willingness to accept of the seller, provides an indirect measure of the renegotiation cost.
When there are more surplus to be generated from the contract, the contracting parties
will spend more resources to acquire a bigger portion of pie. The expected surplus
therefore increases negotiation cost and contract length. The relationship between
expected surplus and contract length is related to the quasi rents mentioned in KCA, but
different in the sense that expected surplus is generated even without specific investments.
Crocker and Masten (1988) empirically tested the relationship using natural gas contracts
and found contract length increases with the quasi rents at stake.

This paper empirically tests the effects of expected surplus, specific investments
and uncertainty on contract length, using contract information from the National Football
League (NFL) and the franchising industry. The benefits of using both types of contracts
are threefold. First, this allows us to examine whether the determinants of contract length
vary across labor and commercial contracts. Second, while the seller (player) is more risk
averse than the buyer (team) in the NFL, the seller (franchisor) is less risk averse than the

buyer (franchisee) in the franchising industry. Analyzing both contract environments

 

5 For example, see Murphy (1992) and Wallace (2001).

enables the comparison between two contracting environments with different relative risk
attitudes of the buyer and seller. Third, using both types of contracts enables multiple
proxies of expected surplus, uncertainty, and specific investments to be applied in the
empirical test.

The paper proceeds as follows. The next section describes the contract information and
institutional details of the NFL and the franchising industry. Section 3 proceeds to argue
why contract length is related to expected surplus, specific investments, and uncertainty.
Section 4 discusses the empirical strategy, while Section 5 presents the evidence that
contract length increases with expected surplus and specific investments. In addition,
contract length increases with uncertainty when the seller is more risk-averse than the
buyer in the NFL, and decreases as the seller is less risk-averse than the buyer in the

fi'anchising industry. Section 6 concludes.

DATA AND INSTITUTIONAL DETAILS

I use both labor contracts from the NFL and commercial contracts from the
franchising industry to test the determinants of contract length. In this section, the data

and institutional details are discussed for both industries.

National Football League Contracts

The NFL conducts their annual draft in late April. There are multiple rounds in a
draft, and in each round each team owns one pick to select a player. The order of picks is
based on the prior year's performance, with the first pick going to the worst performing
team. A team may trade its pick to another team for specific players and/or picks. After
being drafted, the draftee and his agent negotiate a contract with the team. The team has
exclusive rights on the draftee which prohibits other NFL teams from signing the player.
Almost all draftees agree to contractual terms with the NFL team holding their rights. If a
draftee is unable to reach a contractual agreement, he can either sit out a year and reenter
the subsequent year's draft or play for another professional football league (such as
Canadian Football League).

The contract information of those players drafted comes from two time periods,
1986-1991 (hereafter 86-91) and 2001-2007 (hereafter 01-07). Every contract specifies
the duration of the contract and monetary payments, which includes signing bonus and
base salary. Besides early round draft choices in the 01-07 drafts, almost all the contracts

are not guaranteed in the sense that the team has the right to cut a player and not pay his

base salary. Signing bonus, on the other hand, is guaranteed money and paid upfront.
Some contracts also include incentive clauses. After the NFL and the NFL Player's
Association (NF LPA) entered into a new collective bargaining agreement (CBA) in 1993,
the general structure of the contract has become more varied because of the introduction
of salary cap. The salary cap limits team's spending on player salaries with signing bonus
prorated evenly over the duration of the contract.

The NFLPA provided the contract data, along with players' position, college,
selection number in draft, and drafting team. The data include 1,872 contracts fi'om 86-91
and 1,782 contracts from 01-07. There were 28 teams having a single pick in each of the
12 rounds from 86-91, and 32 teams having a single pick in each of the 7 rounds from 01-
07.6 Team information is obtained from the NFL Record and Fact Book (1986-1991,
2000-2007), including team's paid attendance, stadium capacity, win-loss record, and
head coach tenure. College teams’ divisions were obtained from the 1995 Official
National Collegiate Athletic Association College Football Records Book, where the
Division IA schools are the schools with larger budgets for athletic programs than the
other schools. At last, the population in the metropolitan statistical areas (MSA) is

interpolated/extrapolated from the 1990 and 2000 US. census.

 

6 There were only 31 teams in 2001 because the Houston Texans joined the NFL in 2002. The Texan's 12
contracts in 2002 are dropped in the regression analysis since there are no observations for the variables
related to the prior year.

Table 1 Summary Statistics of the NFL Contracts during 1986-1991 and 2001-2007

 

 

 

Variable 1986-1991 2001-2007

Contract length 2.74 3.81
(0.80) (1.02)

Proportion of division IA player 0.79 0.91
(0.41) (0.28)

Empty seat ratio last year 0.18 0.06
(0.14) (0.08)

Average paid attendance last year per home game ( 10,000 seats) 5.31 6.59
(0.92) (0.80)

Capacity last year (10,000 seats) 6.80 6.97
' (0.92) (0.61)

Dummy=l if stadium capacity increased more than 1,000 0_07 008
seats from the prior year (0.25) (0.27)

Dummy=l if stadium capacity decreased more than 1,000 0,02 0.04
seats from the prior year (0.15) (0.20)

Wins last yearA 7,77 7.88
(2.96) (3.05)

Head coach tenure 5.72 3.96
(6.38) (3.05)

Metropolitan population (1,000,000) 4.78 5.01
(4.94) (4.95)
Observations 1 872 1 782a

 

Note: A. Variables related to the prior year have sample size equals 1770. Standard errors in parentheses

Table 1 presents the summary statistics for each time period. As the table
indicates, the average length of the contracts in the 01-07 dataset is about one year longer
than in 86-91. Contract length ranges between one and six years in the 86-91 dataset, and
one and seven years in the 01-07 dataset.7 Figure 1 presents the distribution of contract

length for both time periods.

 

7 Section 5 of Article XVII of the 2006 CBA states: “The initial Players Contract of a Rookie, including

Figure 1 Contract Length of the NFL Contracts

5;
l

&
0|
1
i

see
4

Proportion oteontracta
8 8
l l

 

 

Contract Length

[in 36-91 I 0107;

Figure 2 Average Contract Length and the Proportion of Non-Division IA
(Non-DIVIA) School Graduates

6..
5. 73
i '5
ill ‘0
c <
o4 5

-‘1
im it
:31 2'0
8 ‘55
o 1:
o
it s
5 E
1- n.

 

Round

— Length(86-9T) LengtrilBi-or) T - - Non-DNIA(86-9t) Non-DNIA(01-07)V

 

 

any Club option, may not exceed four years in length, except that the initial Player Contract of a Rookie
drafted with a selection in the first half of the first round (e.g., the first sixteen of thirty-two selections in
the 2006 Draft), including any Club option, may not exceed six years in length, and the initial Player
Contract of a Rookie drafted with a selection in the second half of the first round, including any Club
option, may not exceed five years in length.”

In addition, Figure 2 presents the average contract length by round, as well as the
proportion of non-Division IA school draftees by round. Average contract length
decreases across rounds and the rate of the decrease is similar across time periods. The
fact that signing bonus is allowed to be prorated evenly over the duration of the contract
after imposition of the salary cap is one explanation as to why the average contract length
in every round is longer for the 01-07 data. Since teams pay more signing bonus instead
of salary to the player to save more spending space, they ask for the longer contract with
player because the signing bonus is paid upfront. 8 In addition, the number of non-
Division IA school draftees increases across rounds for both time periods, while there
were more non-Division IA school players drafted in the 86-91 drafts.

Summary statistics of the other variables are similar for both time periods, except
for the empty seat percentage. The empty seat percentage is much smaller during 01-07
because attendance has increased approximately 24 percent while there was only a three

percent increase in the average stadium capacity.

Franchise Contracts

A franchise contract specifies the agreement on the franchisee's right to use the
franchisor's trademark or the right to sell his product in a given place for a specific
duration (Lafontaine, 1992). To test the effect of expected surplus, uncertainty, and

specific investments on franchise contract duration, I obtained information from the

 

8 The tradeoff between contract length and signing bonus is discussed in Duberstein (1991 ).

11

Bond's Franchise Guide (Bond, 2007), which includes detailed survey responses from

1,005 franchisors. Table 2 presents the summary statistics.9

Table 2 Summary Statistics of the Franchise Contracts Information from Bond

 

 

 

(2007)
Variables Mean
Length of the initial contract (in year) 11.15

(6.46)
Dummy=1 if listed as Top 100 franchisors in 2006 0.10
(0.30)
Dummy=1 if allowed building type of unit are home-based or kiosk 0.21
(0.41)
Years since first franchised 19.51
( 12.58)
Required total training (in days) 22.82
(25.65)
Number of total units (in 1,000) 0-49
( 1 .98)
Dummy=1 if the franchisor is in rental, business aid, or education sectors 0.09
(0.28)
Dummy=l if headquartered in a state that restricts contract termination 0-33
(0.47)
Observations 1005A

 

Note: A. Sample size varies across variables since some franchisors did not respond to every
question in the survey. Standard deviations are in parentheses.

The mean duration of the initial contracts is slightly less than eleven years; with
51 percent having ten year durations; nineteen percent having five year durations; and 12

percent having 20 year durations. Table 2 also indicates that ten percent of the fi'anchisors

 

9 In the survey, a franchisor was asked contact information, background details, required financial and
contract terms of his franchisee, specific expansion plans in the near future, as well as the support and
training provided for his franchisee. Moreover, instead of a single value response, some franchisors
provided a range of values for some financial and contractual variables. In such case the average of the
lowest and the highest values are used as their responses. Finally, replies that are ambiguous or a non-
response are treated as missing.

12

were listed in Bond's Top 100 franchisors (Bond, 2006).10 In addition, the survey asked
each franchisor what building type was allowed for a franchise outlet. Most of the
franchisors provided several choices for prospective franchisee, with 21 percent allowing
a home-based unit or a kiosk. Figure 3 shows the proportion of franchisors who allow
each building type. Moreover, Table 2 reveals that the franchisors have average 20 years
of fi'arrchising experience and provide their franchisees an average of 23 days of on- and
off-site training.11 The franchisors have on average 492 units in a chain, where 88 percent
of the franchisors' units are franchised. Nine percent of the franchisors were in rental,
business aid, or educational sectors, and Figure 4 presents the distribution of fianchisors
by business sector. Finally, many states in the US. have statues that regulate the
franchising industry. Most of these states require good cause for termination and non-
renewal of the existing franchising relationship.12 Table 2 indicates that 33 percent of the

franchisors are headquartered in the states restricting termination of a franchise contract.

 

'0 Bond '3 Top I 00franchisors broke the franchising industry into food-services, retail, and service-based
franchises, evaluated companies on the basis of historical performance, brand identification, market
dynamics, franchisee satisfaction, the level of initial training and on-going support, litigation record, and
financial stability, etc (Bond 2006).

'1 Franchisors responded to this question using different measures, including hours, days, and weeks. All
responses are converted into days, where a training day is defined as a business day (8 hours a day, 5 days a
week). Moreover, the off-site training includes training at headquarter, training center, classroom, and
current operating unit, while the on-site training indicates training at franchisee's own site.

'2 A good cause for termination is the franchisee's failure to comply with the material terms of the franchise
contract (Blair and Lafontaine, 2005). The sixteen states which limit termination for good cause are
Arkansas, California, Connecticut, DC, Delaware, Illinois, Indiana, Iowa, Michigan, Minnesota, Nebraska,
New Jersey, Tennessee, Virginia, Washington, and Wisconsin (Klick, Kobayashi, and Ribstein, 2006,
Table 4). See also the discussion about contract duration and the termination laws in BMV, footnote 21,
and Blair and Lafontaine (2005, p. 279).

13

Proportion of Franchisors

 

Proportlon of franchisors

Figure 3 Percentage of Building Types Allowed for Franchise Outlet

 

 

Building type

Figure 4 Percentage of Franchisors by Business Sector

 

 

 

 

40
35
30 4
25
20
15 -
10 -
5
0 -

Others Food service Retail Rental, Auto products Clearing and
business aid. and services maintenance
and education

Business eector

l4

HYPOTHESES

In this section, I provide the arguments for why contract length is related to
expected surplus, specific investments, and uncertainty, using the NFL and franchising
contracts as examples.

Expected Surplus. Expected surplus is defined as the willingness to pay of the
buyer minus the willingness to accept of the seller. More resources will be spent by the
contracting parties when there are more surplus to be generated from the contract.
Consequently, renegotiation cost is increased with expected surplus. Since contract length
increases with renegotiation cost (Dye, 1985), contract length is thus also positively
related to expected surplus. For example, the NFL teams often expect more surplus to be
generated from their contracts with early draftees than the later ones. It is because the
willingness to accept between the early and later draftees are not significantly different
due to their limited outside options, while the early draftees can often attract more
audience for his team. Consequently, the early draftees usually hire multiple agents to
negotiate with the draft team while the later draftees often use a friend or family member
as his representative (Conlin, Orsini, and Tang, 2009). The draft team, on the other band,
also spends more time in negotiations with the early draftees. In the franchising industry,
the willingness to pay of the buyer (potential franchisees) is higher for those good
franchisors while the willingness to accept of the seller (franchisors) is similar. Since the
good franchisors usually prepare more detailed contracts, franchisees who are interested
in joining the chains will also have to spend more time and efforts to go through the

terms and make sure of their own rights.

15

Specific Investments. One of the main focus of the transaction cost theory is a
long-term contract provides a solution to the holdup problem as mentioned in KCA. The
hypothesis is confirmed empirically using evidence mainly from the commercial contract
as previously discussed. Since the specific investments are also important to the labor
contract, such as the human asset specificity argued by Willamson (1983), contract length
is expected to increase with specific investments in the labor contract as well as in the
commercial contract.

Uncertainty. While both the efficient risk-sharing hypothesis and the efficient
production hypothesis have been proposed to explain the relationship between contract
length and uncertainty, to use both the contract information from the NFL and the
franchising industry allows us to examine both hypotheses at the same time. In the NFL
contract negotiations, teams are less risk-averse than players because teams are more
financially sufficient and diversified. Teams are thus able to provide a longer contract for
players when there is more uncertainty. Players, on the other hand, will accept the
contract in exchange of insurance.13 In the franchising industry, franchisees are more
risk-averse buyer than the seller because franchisees are often less financially sufficient
and diversified relative to franchisors. If the future development of the business is
uncertain to the franchisee, a long-term contract increases his risk to be locked-up in an
unprofitable relationship. Since the investment will be less efficient when there is more
uncertainty, while it is also costly for franchisees to breach the contract, the fianchisee

will not want to sign a longer contract when the business is more risky.

 

'3 The fact that the NFL contracts are not guaranteed does not affect the argument because of the positive
relationship between long-term contract and signing bonus (Duberstein, 1992). While the contact duration
may not provide the actual insurance for player, the more signing bonus associated with the longer contact,
which is paid upfront, suggested the compensation for player is provided when they signed a longer
contact.

16

EMPIRICAL STRATEGY

In this section, I discuss the proxies for expected surplus, specific investments and
uncertainty from the NFL and franchising datasets. Table 3 compares these proxies for

the two types of contacts.

Table 3 Proxies for Expected Surplus, Specific Investments, and Uncertainty in the
NFL and the Franchise Contracts

 

   

 

. NFL Contracts ""Frauchrs‘ i L
Expected 1. Selection number 1n the draft Whether franchisor rs lrsted as
Surplus 2. Percentage of empty seats a Top 100 franchisor

3. Team wins prior year
Specific Whether player is a quarterback 1. Whether allowable building
Investments type for outlet includes home-
based or kiosk

2. Number of days of total training

Uncertainty Whether player graduated from Whether fi'anchisor is in rental,
Division IA school business aid, and education sectors

 

National Football League Contracts

Expected Surplus. Player's selection number and team's number of empty seats
and wins last year are proxies for expected surplus. Since the team's benefit for an early
draftee is higher than a later draftee, while the player's opportunity cost does not vary by
the order of pick, the expected surplus is greater for a team and an early draftee.
Moreover, a team with more empty seats and/or fewer wins is willing to pay more for an

early draftee to make immediate contribution, compare to a team with fewer empty seats

17

and/or more wins last year. If contract length is positively related to expected surplus, an
early draftee of a team with more empty seats and/or fewer wins last year should
negotiate a longer contract.

Specific Investments. We use whether a player’s position is quarterback as a
proxy for specific investments. Compared to other positions, a quarterback requires more
team-specific training because he must learn the team's offensive system and play calling.
More team resources are thus invested in the quarterback than the other positions. Since
contract length is expected to increase with firm-specific investment, a quarterback's
contract should be longer than the contract for other positions.

Uncertainty. Whether a player attended a Division IA school is a proxy for
uncertainty, since the NFL performance of non-Division IA draftee is more uncertain to
team. Players from those small programs went through less intensive games with fewer
resources for training, compare to their bigger school counterparts.14 Because the team is
less risk averse than the player, contract duration should be greater for draftees that

attended non-Division IA school.

Franchise Contracts

Expected Surplus. Whether a franchisor is selected in the Bond's Top 100
franchisors is a proxy for expected surplus. Franchisee's benefit from joining the chains
of those Top 100 franchisors will likely be higher than other franchisors' chains. More

expected surplus is therefore generated from contracting with the Top 100 franchisors,

 

14 The same uncertainty measure has been used by Hendricks, DeBrock, and Koenker (2003). See footnote
seven in their paper for more discussion on the uncertainty proxy.

18

since the franchisor's opportunity cost is not likely to vary significantly across franchisees.
Consequently, contract duration for the Top 100 franchisors should be greater.

Specific Investments. Whether a franchisor allows a home-based unit or a kiosk
for outlet is a proxy for specific investments, since the sunk cost of a home-based unit or
a kiosk is likely less than for other building types. In addition, since much of the training
is firm-specific, the number of required training days is another proxy for specific
investments. 1 5

Uncertainty. Since a franchisee is likely more risk averse than the franchisor,
contract length is expected to decrease with uncertainty. To measure the risk incurred by
a franchisee, Lafontaine and Bhattacharyya (1995) used the yearly average proportion of
discontinued outlets between 1982 and 1986 in each sector as a measure of risk faced by
a prospective franchisee.‘6 They found the top three sectors with the highest discontinued
outlets during 1982 to 1986 were rental service, business aids, and educational products
and services sectors.l7 Accordingly, whether franchisor is in those three business sectors

is applied as a proxy for uncertainty.

 

'5 BMV used the amount of off-site training as a proxy for human capital investment. Other than the firm-
specific training, they also emphasized the amount of travel and opportunity cost the franchisee incurs.
Since the focus of this paper is investment specificity, the required total training better serves the purpose.
Similar empirical results can be obtained if the off-site training is used as the proxy for uncertainty.

'6 See Lafontaine and Bhattacharyya (1995) for other risk measures used in the literature.

'7 Lafontaine and Bhattacharyya (1995) obtained the data from Franchising in the Economy, published by
US. Department of Commerce, which is not available afier 1986.

19

EMPIRICAL RESULTS

National Football League Contracts

To test whether contract duration is affected by expected surplus, uncertainty, and
specific investments, I regress NFL contract length on player's selection number, team’s
empty seats percentage, team wins the prior year, whether the draftee attended a non-
Division IA school, as well as the team's MSA population, head coach tenure, prior year's
stadium capacity, and whether stadium capacity has increased or decreased by more than
1,000 seats. Team fixed effect, as well as player's position, draft round, and draft year
fixed effects are also included in the regression.‘8

Binomial regression is applied for the estimation, since contract duration ranges

discretely from one to seven.19 To be specific, the conditional mean to be estimated is

assumed as
E(lengthilxi)=P(Xil3)L,
where x,- denotes the set of regressors; p(x,-B) is a logistic distribution;20 and L is the

maximum contract length. Since contract length is restricted, multiplying the estimated

probability by L guarantees the fitted value will not exceed the sample range.

 

'8 The empirical results are similar if different sets of fixed effects are applied. To be specific, instead of
including all fixed effects in the regression, I have also run the regression with different combinations of
fixed effects, including 1) draft round only; 2) draft round and player's position; 3) draft round, player's
position, and draft year.

9 The other three possible choices for estimation are ordinary least square (OLS), ordered probit regression,
and Poisson regression. The problem of OLS estimation is that it may predict the fitted value to be more
than seven or less than zero. The ordered probit regression is not used since there is no data censoring
problem in the contracts used here, which is a common problem in the empirical contract duration literature
mentioned by Masten and Saussier (2000). At last, by using Poisson regression, the fitted value is assured
to be positive but larger than the upper bound of the variable. While only the results from binomial
regression are reported, they are robust to these different estimation methods.

20

Table 4 Binomial Regressions: Dependent Variables (Contract Length of the NFL

 

 

 

 

Contracts)
, ~ .- i ' 1986—1991 , 2001:2007
Independent . ' -. ,. - ' . q .
variables (1) (2) (3.). (4) 7 (5) .’ a (6), . l_' ,
Selection number -0.004*** -0.003** -0.005*** -0.002* -0.001 —0.002**
in the drafi (0.001) (0.002) (0.002) (0.001) (0.002) (0.001)
Empty seats 0.112 0.380” 0.108 -0.215 0.346 -0.196
percentage in home (0.132) (0.166) (0.132) (0.185) (0.286) (0.183)
games last year
Empty seats -0.002*** -0.004***
percentage in home (0.001) (0.002)
game last year"
selection number
in the draft
Number of team 0.000 0.001 -0.008 -0.007* -0.008** -0.029**"‘
wins last year (0.005) (0.005) (0.007) (0.004) (0.004) (0.008)
Number of team 0055* 0157*"
wins last year prior (0.029) (0.044)
to drafi“ selection
number in the draft
divided by 1,000
Dummy=1 if -0.054“'** -0.055**"‘ -0.055"‘** -0.027 -0.026 -0.026
graduated from (0.019) (0.019) (0.019) (0.031) (0.031) (0.031)
Division IA school
Dummy=1 if -0.031 -0.030 -0.032 0.097“ 0.095” 0.084“
player is (0.049) (0.048) (0.048) (0.047) (0.046) (0.046)
QuarterbackA
Metropolitan -0.l61*** -0.159""""l -0. 160*" 0.028 0.028 0.027
population (0.054) (0.055) (0.054) (0.033) (0.033) (0.033)
(1,000,000 person)
Number of head 0.013*** 0013*" 0013*" 0.007 0.007 0.006
coach tenure at (0.003) (0.003) (0.003) (0.004) (0.004) (0.004)
drafi year
Stadium capacity -0.071 -0.054 -0.069 0.016 0.020 0.016
of team last year (0.044) (0.046) (0.044) (0.037) (0.049) (0.036)
(10,000 seats)
Stadium capacity -0.097 -0.062
of team last year (0.090) (0.235)
(10,000 seats)*
selection number
in the draft divided
by 1,000

 

 

2° The marginal effects are similar if the probability function is assumed to be Gaussian.

21

 

 

Table 4 (cont’d)
Dummy=1 if -0.040 —0.042 -0.040 -0.099*** -0.103*** -O.102***
stadium capacity (0.049) (0.048) (0.049) (0.032) (0.032) (0.032)
increased more
than 1,000 seats
from the prior year
Dummy=1 if 0105* -0.109* -0.106* -0.051 -0.050 -0.050
stadium capacity (0.058) (0.057) (0.058) (0.039) (0.039) (0.039)
reduced more than
1,000 seats from
last year
Indicator
variablesB YES YES YES YES YES YES
Observations 1872 1872 1872 1770 1770 1 7 70
R-squaredc 0.65 0.66 0.66 0.68 0.68 0.68
Ljog-likelihood -2255.13 -2254.14 -2254.83 -2292.87 -2292.00 -2291.23

 

 

Note: A. The coefficients were estimated relatively to the position of defensive linebacker.
B. The indicator variables include the draft round, player’s position, draft year, and draft team fixed
effects.
C. R-squared is the squared correlation coefficient between the observed and the fitted values of the
contract length variable.
The coefficient estimates are not marginal effects. Robust standard errors presented in parentheses. *
significant at 10%; ** significant at 5%; *** significant at 1%.

Columns (1) and (4) of Table 4 present the coefficient estimates from this baseline
regression.21 The coefficient estimates support the conjecture that, when the buyer is less
risk averse than the seller, contract length increases with expected surplus, uncertainty,
and specific investments. First, the selection number of player has a negative and
statistically significant coefficient for both time periods. If a player was drafted 20 picks
later within a round, the duration of his contract will be reduced by about 0.1 years for
both time periods. Notice the coefficient estimates were obtained conditional on the
round fixed effects. The round dummy coefficients suggests that, compared to players

drafted in other rounds, the first round draftees obtained an average of 0.57 year longer

 

2‘ The coefficient estimates reported in Table 4 are not marginal effects. To calculate the marginal effect of
a continuous variable from the binomial regression, the partial derivative of the probability function with
respect to the regressor is evaluated at the sample mean of all regressors. For a dummy variable, the
marginal effect is evaluated as the difference between the estimated probability when the dummy equals
one and when it equals zero.

22

contract during 86—91, and 2.17 years during 01-07. The result indicates contract length
increases with expected surplus. In addition, the coefficient on empty seats percentage
and number of team wins in prior season vary in Sign with the only statistically
significant coefficient being the negative coefficient associated with wins for the 01-07
drafts. 22 Second, players from small schools obtained on average 0.08 year longer
contracts than their big school counterparts during 86-91, and 0.04 year longer contracts
during 01-07. The coefficient estimates indicate that, when the buyer is less risk averse
than the seller, contract length increases with the uncertainty. The result is not
statistically significant in the 01-07 dataset because only nine percent of draftees
graduated from small schools during 01-07, while 21 percent of draftees were from small
schools during 86-91. Third, comparing to other positions, a quarterback had the second
shortest contract in the 86-91 dataset, but the longest one in the 01-07 dataset. Since
players rarely changed teams unless via a trade prior to the 1993 CBA, the holdup issue is
more of a concern in the later years.23

Regressions (2) and (5) of Table 4 include an interaction of empty seats and
selection number. An interaction of the stadium capacity last year and selection number is
also included as a covariate. As discussed in section 4, the effect of prior year wins and
empty seats on contract length may differ for early and late draftees becauSe early

draftees are more likely to make an immediate contribution to team performance. The

empirical evidence supports this conjecture. All else equal, the empty seats percentage in

 

22 All else equal, one percentage increase of empty seats in the prior year is related to a 0.17 year longer
contract in the 86-91 dataset, but a 0.38 year shorter contract in the 01-07 dataset. On the other hand, one
more winning game in the prior year relates to 0.002 year longer contract in the 86-91 dataset, but 0.012
year shorter contract in the 01-07 dataset.

3 For example, the baseline regression indicates a quarterback has an average of 0. l 3 year longer contract
than other positions during 01-07. By using the same specification but excluding the position fixed effect,
the regression shows a quarterback has on average 0.05 year shorter contract than other positions during
86-91, but 0.07 year longer contract during 01-07.

23

prior season is positively related to contract length, but the effect decreases with the
selection number.24 Similarly, regressions (3) and (6) of Table 4 include the interaction
term between the number of wins last year and selection number. Compared to other
teams, a team with fewer wins last year signed longer contracts with early draftees but
shorter contracts with later draftees.25 These results provide further supports that contract

length increases with expected surplus.

Franchise Contracts

To test whether franchise contract length is affected by expected surplus, specific
investments, and uncertainty, I regress contract length on whether the franchisor was
listed in Bond's Top 100 fianchisors, whether the franchisor allows a home-based unit or
a kiosk as an outlet, the number of days of total training, and whether the fi'anchisor is in
the retail, business aid, or education industry. The other covariates are the number of
units in a chain, the number of years since the franchisor first franchised, and whether
franchisor is headquartered in a state that restricts termination of a franchise contract.26

Since lengths are often much longer for franchise compared to NFL contracts,
Poisson regression is an appropriate estimation method.27 In particular, the conditional

mean is

 

2“ For example, longer contracts were signed by the teams with more empty seats last year in the 86-91
dataset, but the same teams signed shorter contracts with the draftees after the 190th pick.

25 For example, shorter contracts were signed by the teams with more wins last year in the 01-07 dataset,
but the same teams signed longer contracts with the draftees after the 185th pick.

26 BMV applied the number of total units and the number of years since first franchised as the proxies for
learning. They argued franchisor with more experience and units in a chain is able to learn more about the
optimal contract terms. The contract length will be longer since the franchisor does not need the flexibility
to adjust the contract terms.

27 Similar results can be found by using OLS or binomial regressions.

24

E(lengthilxi)=exp(xifl),

where x,- denotes the set of regressors, and length,- given x,- has a Poisson distribution.

Table 5 Poisson Regressions: Dependent Variables (Contract Length of the

Franchise Contracts)

 

 

 

Independent variables (1) (2)
Dummy=1 if listed as Top 100 franchisors in 2006 0133*" 0179*"
(0.061) (0.053)
Dummy=1 if allowable building type of unit are home-based or kiosk -0149":- 41113"
(0.049) (0.057)
Number of days of required total training 0002*" 0.002"
(0.001) (0.001)
Dummy=l if the franchisor is in rental, business aid, or education sectors 0227*" .0,420***
(0.060) (0.082)
Number of total units divided by 1,000 0,007 0,014“
(0.007) (0.007)
Number of years since first fianchisedl 0004*" 0,002
(0.001) (0.001)
Dummy=l if located in the state requires good cause for termination 0.089" 0100*"
(0.040) (0.037)
Business sector fixed effect NO YES
Observations 848 848
R-squaredA .09 .25
Log-likelihood -2815.21 -2620.45

 

Note: A. R-squared is the squared correlation coefficient between the observed and the fitted values of the

contract length variable.

The coefficient estimates are not marginal effects. Robust standard errors presented in parentheses.

* significant at 10%; ** significant at 5%; *** significant at 1%.

Table 5 presents the coefficient estimates from this specification. These estimates
are similar whether or not business sector fixed effects are included. The empirical results
support the hypothesis that, when the buyer is more risk averse than the seller, contract

length increases with greater expected surplus, less uncertainty, and more specific

25

investments. First, all else equal, franchisors listed in the Bond's Top 100 franchisors
provided about 18 percent longer contract than other franchisors, and this difference is
statistically significant. The results also indicate that franchisors offered longer contract
when there was greater expected surplus. Second, the franchisors in rental, business aid,
and education industry provided 42 percent shorter contract than other franchisors. This
statistically significant difference suggests that, when the buyer is more risk averse than
the seller, contract length decreases with the uncertainty. Finally, if a franchisor allows a
home-based unit or a kiosk as a franchise outlet, the franchisor on average provides a 12
percent shorter contract. If a franchisor reduces the required total training by a day,
contract length on average shorten by 0.2 percent. These estimates are statistically
significant and provide the evidence that contract length increases with specific

investments, which is consistent with the literature.

26

CONCLUSION

The existing research testing contract theory provides limited evidence on the
determinants of contract length. The effects of uncertainty and specific investments on
contract length have also been tested separately on labor and commercial contracts,
respectively. This is the first paper to address the relationship between expected surplus
and contract length, and tests the determinants of contract length for both commercial and
labor contracts using NFL and franchising data. The empirical evidence indicates that
contract length increases with expected surplus and specific investments. In addition,
contract length increases with uncertainty when the seller (player) is more risk averse
than the buyer (team), and decreases with uncertainty when the seller (franchisor) is less
risk averse than the buyer (franchisee). Further research is required to see if these

relationships hold for other types of commercial and labor contracts.

27

REFERENCES

Bandiera, Oriana. "Contract Duration and Investment Incentives: Evidence from Land
Tenancy Agreements," Journal of the European Economic Association, 2007,
5(5), pp. 953-86.

Blair, Rogert D. and Lafontaine, Francine. The Economics of Franchising, 2005,
Cambridge University Press.

Bolton, Patrick and Dewatripont, Mathias. Contract Theory, 2005, The MIT Press.

Bond, Robert E., et al. Bond '3 Franchise Guide, 2007, Source Book Publications.

--------- . Bond's Top 100 Franchises, 2006, Source Book Publications.

Brickley, James A.; Misra Sanjog; and Van Horn, R. Lawrence. "Contract Duration:
Evidence from Franchising," The Journal of Law and Economics, 2006, 49(1), pp.
173-96.

Cheung, Steven N.S. "Transaction Costs, Risk Aversion, and the Choice of Contractual
Arrangements," The Journal of Law and Economics, 1969, 12(1), pp. 23-42.

Chiappori, Pierre-Andre and Salanié, Bernard. "Testing Contract Theory: 3 Survey of
Some Recent Work," CESifo Working Paper Series No. 738, 2002, SSRN:
http://ssrn.com/abstract=3 l 8780

Collective Bargaining Agreement (CBA): between The NFL Management Council and
The NFL Players Association, March 8, 2006.
http://wwwnflplayers.com/user/template.aspx?fmid=l 8 l &lmid=622&pid=0&typ
e=l

Conlin, Mike; Orsini, Joe and Tang, Meng-Chi. "The Effect of an Agent's Experience on
National Football League Contract Structure," Michigan State University
Working paper.

28

Crocker, Keith J. and Masten, Scott E. "Mitigating Contractual Hazards: Unilateral
Options and Contract Length," RAND Journal of Economics, 1988, 19(3), pp.
327-43.

Danziger, Leif. "Real Shocks, Efficient Risk Sharing, and the Duration of Labor
Contracts," Quarterly Journal of Economics, 1988, 103(2), pp. 435-40.

Duberstein, M. J ., “On the Sidelines: An Annual Economic Analysis of the National
Football League Prepared for Members of the NFL Players Association,”
National Football League Association, 1992.

Dye, Ronald. "Optimal Length of Labor Contracts," International Economic Review,
1985, 26(1), pp. 251-70.

Gray, J o A. "On Indexation and Contract Length," Journal of Political Economy, 1978,
86(1), pp. 1-18.

Guriev, Sergei and Kvasov, Dmitriy. "Contracting on Time," American Economic
Review, 2005, 95(5), pp. 1369-85.

Harris, Milton and Holmstrom, Bengt. "On the Duration of Agreements," International
Economic Review, 1987, 28(2), pp. 389-406.

Hendricks, Wallace; DeBrock, Lawrence and Koenker, Roger. "Uncertainty, Hiring, and
Subsequent Performance: The NFL Draft," Journal of Labor Economics, 2003,
21(4), pp. 857-86.

J oskow, Paul L. "Contract Duration and Relationship-Specific Investments: Empirical
Evidence from Coal Markets," American Economic Review, 1987, 77(1), pp. 168-
85.

Klein, Benjamin; Crawford, Robert G. and Alchian, Armen A. "Vertical Integration,
Appropriable Rents, and the Competitive Contracting Process," The Journal of
Law and Economics, 1978, 21(2), pp. 297-326.

Klick, Jonathan; Kobayashi, Bruce H. and Ribstein, Larry E. "The Effect of Contract

29

Regulation: The Case of Franchising," George Mason Law & Economics
Research Paper No. 07-03, SSRN: http://Ssrn.com/abstract=951464.

Lafontaine, Francine. "Agency Theory and Franchising: Some Empirical Results,"
RAND Journal of Economics, 1992, 23(2), pp. 263-83.

--------- . and Bhattacharyya, Sugato. "The Role of Risk in Franchising," Journal of
Corporate Finance, 1995, 2, 39-74.

--------- . and Slade, Margaret. "Vertical Integration and Firm Boundaries: The Evidence,"
Journal of Economic Literature, 2007, 45(3), pp. 629-85.

Masten, Scott E. and Saussier, Stephane. "Econometrics of Contracts: An Assessment of
Developments in the Empirical Literature on Contracting," Revue d'Economie
Industrielle, 2000, 92, pp. 215-36.

Murphy, Keven J. "Detemimants of Contract Duration in Collective Bargaining
Agreements," Industrial and Labor Relations Review, 1992, 45(2), pp. 352-65.

Vézquez, Luis. "Determinants of Contract Length in Franchise Contracts," Economics
Letters, 2007, 97, pp. 145-50.

Wallace, Frederick H. "The Effects of Shock Size and Type on Labor-Contract
Duration," Journal of Labor Economics, 2001, 19(3), pp. 658-81.

Williamson, Oliver E. "Credible Commitments: Using Hostages to Support Exchange,"
American Economic Review, 1983, 73, pp. 519-40.

30

CHAPTER II
THE CHOICE OF CONTRACT DURATION AND PRICING PROVISION IN

THE COAL MARKETS‘

INTRODUCTION

In his seminal paper, Joskow (1987) provided the empirical evidence supporting
the importance of specific investments to contract duration, using 277 contracts fi'om the
coal markets. His paper confirms the hypothesis proposed by Klein, Crawford, and
Alchian (1978) and Williamson (1983) that long-term contracts provide the incentive for
more efficient relationship-specific investment by reducing the possibility of ex post
opportunism behavior or the holdup problem. However, J oskow (1987) did not consider
the simultaneity between the choices of contract duration and pricing provision. Different
pricing provisions are used in the coal market (Joskow, 1985, 1988), and different price-
adjustrnent mechanisms reduce the costs of signing a long-term contract while preserving
the benefits of such contract. To be specific, a long-term contract saves renegotiation
costs and induces more efficient specific investments, but also reduces the flexibility of
the contracting parties when facing fiiture contingencies. It is therefore important to apply
appropriate pricing provision to maximize the expected revenue generated from the
contract over a long-term relationship, subject to the flexibility constraint. For example, if
a fixed price is negotiated for trading a certain good over multiple years, it will not be

easy for the contracting parties to accommodate the future contingencies. However, if a

 

. I thank Mike Conlin, Thomas Jeitschko, Jay Wilson, and Kent Miller for their comments and suggestions,
as well as George Warholic for his expertise in the coal industry.

31

price-adjust mechanism is included in the contract, such as designing the price as a
fimction of current consumer price index, the contract price can therefore automatically
reflect the current market situation. In this paper, I examine the choice of contract
duration and pricing provision simultaneously, using more than 5,000 observations from
coal markets.

Most of the related literature focus on the choice of pricing provision, while
relatively fewer researches look at the choice of contract duration and the joint decision
of pricing provision and contract duration. 28 The choice of contract type has been
examined in multiple industries, including timber-harvesting (Leffler and Rucker, 1991),
natural gas (Crocker and Masten, 1991), jet engine procurement (Crocker and Reynolds,
1993), and tenancy (Bandiera, 2007).29 The main question is how to design a long-term
contract that incorporates the flexibility for future adjustment and reduces the possibility
of potential opportunism, and retains the benefits of maintaining a long-term
relationship.30 While the empirical results are industry-dependent, the available evidence
in general confirms the importance of uncertainty and specific investments to the design
of pricing provision of the long-term contracts.

Contract duration has been examined separately on commercial contracts and
labor-union contracts, focusing on the effect fiom specific investments and uncertainty,
respectively. The relationship between contract duration and specific investments has
always been the focus of the transaction cost theory, and the idea has been tested using

the commercial contracts, such as coal contracts (J oskow, 1987) and natural gas contracts

 

28 Crocker (1996) reviewed some related works on the choice of contract duration and pricing provision by
public utilities.

9 For more detailed review, see Masten and Saussier (2000) and Masten (2000).
3° Cheung (1969) argued that the choice of contractual arrangement is made so as to maximize the gain
from risk dispersion subject to the constraint of transaction costs.

32

(Crocker and Masten, 1988). The consensus from the existing evidence is that contract
duration increases with specific investments. The labor-union contract, on the other hand,
has been used mainly by the macroeconomics literature to test how the future uncertainty
affects the negotiated contract duration (Wallace, 2001). While the empirical evidence on
the effect is mixed, Tang (2009) argued the relationship between contract duration and
uncertainty depends on the relative risk attitude between the buyer and seller. Using
contracts from the franchising industry and the National Football League, he also found
the expected surplus generated from the contracting relationship is the other factor that
positively relates to the contract duration.

While Masten and Saussier (2000) note that a few papers address the simultaneity
concern between contract duration and pricing provision design, Bandiera (2007) is a
notable exception. In addition to the typical focus of the literature on the tenancy
contractual choice, Bandiera (2007) estimates the determinants of contract duration
jointly with contract type. She found both the contract duration and contract type are
affected by the wealth of the tenant and the type of the crops, which indicates the
importance of specific investments and flexibility to the tenancy contract design. On the
other hand, instead of jointly estimating the duration and pricing provision, Crocker and
Masten (1991) also incorporates the effect fiom contract duration when looking at the
pricing provision choice of the natural gas contract. They addressed the simultaneity
issue using an instrumental variable regression and found strong correlation between
contract duration, uncertainty and pricing provision. Nevertheless, they found no

significant effects from specific investments to the contract design.

33

This paper examines the choice of coal contract duration and pricing provision in
the coal markets, and proceeds as follows: the next section describes the data. Section 3
discusses and empirically tests the determinants of contract duration. The results indicate
coal contract duration increases with expected surplus and specific investments, and
decreases with uncertainty. Section 4 argues the different costs and benefits behind
different pricing provisions over a long-term relationship, and empirically tests the
determinants of the pricing provision choice. I find the determinants that affect contract
duration also affect the choice of pricing provision. Section 5 estimates the choices of
contract duration and pricing provision simultaneously. The empirical results with the
simultaneity consideration are not significantly different from the estimates obtained in

the previous sections. Section 6 concludes.

34

DATA AND INSTITUTIONAL DETAILS

In this paper, I focus on the contract arrangement of coal delivery between electric
utilities and coal suppliers.31 I use the data from the Coal Transportation Rate Database
(CTRD).32 The CTRD contains coal supply contract data between each utility and coal
supplier, includes contract terms, plant, route, and transportation-related information.33
The primary source for the CTRD is from Federal Energy Regulatory Commission's
Form 580, which requires responses fiom all jurisdictional utilities that either operate or
have ownership interest in at least one steam-electric generating station with a capacity of
50 megawatts or greater}4 The CTRD contains the information of 2,980 contracts for 141
utilities and 384 plants thru 1979-2001, provides an origin-to-destination record for every
route over which that plant's coal flows. Because one coal supply contract can serve
several plants and contain different routes for the same coal delivery with different
contract terms, I use each coal flow as the unit of observation rather than the contract per
se. Twenty three percent of the observations were signed before 1979, indicates some

contracts were executed before the sampling period. In addition, a coal flow can be

observed repeatedly across the sample years but only the first record is used for

 

3 ' For detailed overview of the coal industry, see Joskow (1985).

32 The data can be obtained from http://www.eia.doe.gov/cneaf/coal/ctrdb/database.html. For detailed
description, see Warholic (2004).

33 An observation in the CT RD identifies the parties involved in the contracting relationship (the utility
company, the utility plant, the coal supplier, the mines that provided coal), the agreed duration of the
relationship, the upper and lower limits of the agreed quantity to be shipped, the agreed coal quality
specification (the upper and lower limits of Sulphur, Ash, and Moisture), the coal rate information, and the
transportation information (the geographic information of the delivery, the number of routes used, the
modes of transportation used for each route, and the rate for each transportation mode). Each observation
also indicates whether the plant is a mine-mouth plant, and whether the utility company owns the
transportation mode that is being used for delivery.

3" Jurisdictional utility is defined as the facilities involved in the transmission of electric energy in interstate
commerce and the sale of electric power at wholesale in interstate commerce. Also see
http://www.ferc.gov/docs-filing/hard-filing/form—580/overview.asp for more information about Form 580.

35

analysis.35 I also exclude 1,669 observations with incomplete information on contract

length and 61 observations with incomplete information on the coal production region or

the coal is obtained outside the North America.

. Table 6 Summary Statistics: Means and Standard Deviations

 

Variable

 

 

Contract Duration (years) 8.493
(9.647)
Mine-Mouth Plant (=1 if the plant is a mine-mouth plant, 0 otherwise) 0.006
(0.078)
Total Number of Contracts Signed between the Same Plant and Supplier 3.027
(2.921)
Minimum Annual Contract Quantity (trillion Btu’s) 15.171
(19.960)
Percentage of Sulfur of the Coal Shipped 1.333
(0.943)
Percentage of Ash of the Coal Shipped 9.452
(3.1 10)
Percentage of Moisture of the Coal Shipped 9.925
(8.3 14)
Total Distance of A Route (1,000 miles) 0.439
(0.428)
Total Number of Observations 5,835

 

 

Table 6 presents the summary statistics. There are 5,835 observations from 2,649

contracts with 116 utilities, 344 plants, and 730 coal suppliers. During the sample period,

about 65 percent and 73 percent of the observations are the first contract observed

between the same utility and supplier and the same plant and supplier, respectively. The

 

35 A coal flow in the sample has been observed on average six times during the sample period. 13,071

observations are dropped.

36

average contract duration is 8.5 years. About half of the observations have a length equal
or less than five years, while five percent of the observations have contract length less
than a year. The later is treated as spot purchase and recorded with zero duration. Figure
5 shows the distribution of the contract duration. In addition, Figure 6 indicates there is a
decreasing trend of contract length across sample years. 36 In addition, 0.6% of the
observed coal flows are directed to mine-mouth plants, which are the plants that located
next to specific mines in anticipation of taking all or most of their requirements from that
mine.37 A plant on average signed three contracts with the same supplier during the
sample period in the sequential years. The minimum annual contract quantity is on
average 15.17 trillion Btu's. Table 6 also presents the characteristics of the coal shipped. I
use the shipped coal quality rather than the agreed coal quality because more than half of
the observations do not have the latter information. At last, the average distance of a
route for coal delivery is 439 miles. Up to four transportation facilities used for delivery
are recorded in the CTRD, and more than 68% of the coal deliveries used rail.38 Sixty-
three percent of the coal deliveries used only one transportation mode. About 10% of the

utility companies own the transportation facility for coal delivery.

 

3" As Warholic (2004) pointed out, coal customers preferred shorter terms of contracts in recent years
because of the expectation of declining coal prices.

37 Joskow (1987, pp. 170) provided the definition of mine-mouth plant. The CT RD identifies 36 coal flows
to mine-mouth plants.

33The other choices are barge, collier, truck, conveyor, and pipeline.

37

Figure 5 Contract Duration of the Coal Contract

 

 

Figure 6 Average Contract Duration by Year

3...

a8“ -

C

3 o. o

E .

o_

o w

U

E .'. . s

o o " q.

:8" . ..-....
0..
‘00,.

0
ca m

 

 

l l l l l l l I l
1920 1930 1940 1950 1960 1970 1980 1990 2000
The Year Contract Signed

Figure 7 The U.S. Bureau of Mines Coal-producing Districts

 

38

The characteristics of coal production and transportation availability vary across
different regions. Using the U.S. Bureau of Mines (BOM) coal-producing districts
definition shown in Figure 7, production occurs in three regions: East, Midwest, and
West.39 In general, coal quality is more uniformly distributed in the East than the West.
The East also has more abundant transportation alternatives for delivery than the West. In
both cases, the Midwest is somewhere in between. 40 Table 7 shows the summary
statistics by region. Twenty percent of the coal supply in the data is from the West, which
has the longest average contract length and distance per route. Most of the mine-mouth

plants in the sample are located in the West.

 

Table 7 Summary Statistics by Coal Production Region

 

 

Region
West Midwest East
Average Contract Duration (years) 10.938 9.186 7.573
(13.054) (9.297) (8.213)
Total Number of Mine-Mouth Plants 18 8 10
Average Distance of A Route (1,000 miles) 0.934 0.245 0.327
@490) (0.362) (0.280)
Total Number of Observations 1,195 834 3,806

 

 

Note: Standard deviations are in parenthesis.

Various pricing provisions of contract are used in the coal industry. 41 The
simplest versions are fixed price contract with the price fixed over the life of the contract,

and market price contract with the delivery price fluctuates with spot market. The other

 

39 Figure 3 is obtained from http://www.eia.doe.gov/cneaf/coal/page/districts/coaldistricts.htrnl. I use the
definition by Joskow (1987, pp. 184). The Energy Information Administration names the regions as
Appalachia, Interior, and Western. Sixty-one percent of total observations are dropped due to lack of this
information or the source mines are not included in those regions.

4° For more details, see Joskow (1985, 1987).

4' Federal Energy Regulatory Commission provides the definition of the coal contract types in the Form
580, which is obtained from http://www.ferc.gov/docs-filing/hard-filing/form-S80/overview.asp. See also
Joskow (1985, 1988) for more details on the types of coal contracts.

39

contract types, which incorporate some price adjustment mechanisms, are cost plus
contract, base price plus escalation contract, and evergreen contract. By using a cost-plus
contract, the buyer agrees to pay all the seller's costs. Some cost-plus contracts provide
for payment of both a management fee and a profit, which can be fixed or tied to various
productivity and cost reduction incentives. In addition, a base price plus escalation
contract uses different components of the price escalate as a fiinction of changing
economic conditions or indices.42 At last, an evergreen contract allows the price to be
renegotiated at predetermined intervals, usually one year. It may also contain provisions
for price adjustments between renegotiations. Table 8 presents the distribution of the
contract types, along with their average durations. Most of the contracts in the CTRD are
designed as the base price plus escalation contract. Among the contract types, fixed-price
contracts have the shortest average duration while cost-plus contracts have the longest
one.

Table 8 Contract Duration by Pricing Provision: Means and Standard Deviations

 

m
. -‘\ ‘1'

 

 

   

 

ovmmn ' ; e '
Base Price Plus Escalation
Fixed-Price 20. 14
Evergreen 6.05
Cost-Plus 2.25 (16.71)
. 9.1 1
Market-Price 1.1 1 (949)
Unknown or Others 10 16 8'59
' (1.10)
Total Number of Observations 5,835

 

 

 

 

"2 Joskow (1985, pp. 69) gave specific examples on how base price plus escalation contract ties to the
economic indices.

40

THE CHOICE OF CONTRACT DURATION

To build a new generating plant using coal, there are various decisions to make
(Joskow, 1985). First, the utility has to choose the location of the plant. It can build a
plant that closes to the service territory, or build a mine-mouth plant. The utility also has
to choose the type of coal the plant will utilize, and design the generating unit's boiler
accordingly. Moreover, the utility has to consider how the coal will be transported from
the mine to the plant. This section empirically tests the determinants of contract duration

under these considerations.
Empirical Strategy

Joskow (1985, 1987) suggested that asset specificity is an important factor
affecting the contract structure of the coal contracts. In his seminal paper, J oskow(1987)
examined three types of relationship-specific investments proposed by Williamson
(1983), including site specificity, dedicated asset specificity, and physical asset
specificity. He found that contract duration of the coal contracts are longer when the
relationship-specific investments are more important. Using contracts from the
franchising industry and the National Football League, Tang (2009) also argued that
expected surplus and uncertainty are the other important determinants of contract
duration besides specific investments. His empirical evidence suggested that contract
length increases with expected surplus and specific investments. Depending on the

relative risk attitude of the buyer and seller, contract length increases with uncertainty

41

when the buyer is less risk averse than the seller, and decreases with the buyer is more
risk averse than the seller.43

To apply the above ideas on the coal contracts, different proxies are applied for
expected surplus, uncertainty, and specific investments in the empirical analysis. First, I
use the mine-mouth plant indicator variable as a proxy for expected surplus. Since a
mine-mouth plant is designed in anticipation of taking coals fiom the mine that's located
next to the plant, the willingness to pay of a mine-mouth plant to acquire coal from its
supplier will be greater than to transport coal fi'om the other plants. The same variable is
also used by Joskow (1987) as a measure of site specificity. Similarly, the total distance
over a route is also used as a proxy for expected surplus, because a plant is willing to pay
more for the supply from a closer mine than the other mines. Second, I use the
accumulated number of contracts signed between the same plant and supplier across the
sample period as a proxy of uncertainty, because it indicates the familiarity between the
contracting parties.44 At last, I use the coal supply region as a proxy for specific
investments because the coal quality is more heterogeneous in the West than in the
Midwest and the East. Coal suppliers are therefore more likely to exploit the lock-in
effect associated with boilers designed to burn coal with specific characteristics in the
West than in the Midwest and the East (Joskow 1985, 1987). Joskow (1987) used the
same proxy to measure the physical asset specificity. The contract quantity is applied as
the other proxy for specific investments because the larger the coal supply is contracted,

the more difficult for the buyer to replace supplies when the seller breaches the contract

 

“'3 For more comprehensive review of the contract duration literature, see Guriev and Kvasov (2005) and
Tang (2009).
‘4 Empirical results are similar if the interactions between utility and supplier are used instead.

42

in the future. Joskow (1987) used the same proxy as a measure for dedicated asset

specificity.

Empirical Results

To test whether the coal contract duration is affected by expected surplus,
uncertainty, and specific investments, I regress contract length on whether the plant is a
mine-mouth plant, the total distance of a route, the accumulated number of contracts
signed between the same plant and supplier across the sample period, the annual contract
quantity and the square of the annual contract quantity, whether the coal production
region is in the Midwest or the West, and whether the observation is among the observed
initial contracts signed between the same plant and supplier. The quality measures of coal
and the year fixed effects are also controlled.

Four methods are applied for estimation. First, I use the ordinary least square
(OLS) estimation as J oskow (1987). However, as J oskow noticed, the contracts that were
not in force during or after 1979 are not observed. Specifically, a contract i signed in year

T is observed only if
DurationiTZ(l979-T),

which results in a sample selection problem that may bias the estimation. Therefore, I
also apply the truncation regression as used in Joskow (1987), which accounts for the
truncated nature of the sample and estimates accordingly.45 In addition, since contract

length is a count variable with a lower bound of zero, while the upper limit of contract

 

'5 For the estimation purpose, the lower bound is set as zero. It thus excludes 264 spot purchases in the
sample. Estimates obtained using different lower bounds are similar.

43

length in the sample is not restrictive, poisson regression is another appropriate choice
that provides positive fitted value with non-constant marginal effect estimates. The zero-
truncated poisson regression is also applied to account for the sample selection problem.
Table 9 presents the empirical estimates. The signs of the coefficient estimates are
comparable with Joskow (1987) and consistent between different estimation methods.
The statistically significant results indicate the coal contract length increases with
expected surplus and specific investrnents, and decreases with uncertainty. For example,
the poisson regression estimates indicate a mine-mouth plant has 41 percent longer
contract duration than the other plants, and one thousand more miles between the plant
and supplier decreases the contract length by 13 percent. The results support the
hypothesis that contract length increases with expected surplus. In addition, contract
duration decreases with uncertainty because when there is one more contract signed
between the same plant and supplier, the duration of the contract on average increases by
three percent. At last, when the agreed quantity is increased by one trillion Btu's, contract
duration extends about two percent. The effect indicates contract duration increases with
specific investments, and it diminishes as the quantity increases. The longer contract
length in the Midwest and the West relative to the East provides the other evidence

supporting the positive relationship between contract duration and specific investments.

44

Table 9: OLS and MLE (Truncated, Poisson, and Zero-Truncated Poisson
Regressions): Dependant Variable (Contract Duration)

 

 

 

 

 

 

 

Zero-
Variable OLS Truncated Poisson Truncated
, . Poisson ..-.-
Mine-Mouth Plant 10.312*** 7219*" 0.408*** 0.213“
(=1 if the plant is mine-mouth plant) (1.887) (2.744) (0.088) (0.109)
Total Distance ofA Route (1,000 miles) -O.790*** -2.514*** -0.130*** -0. 139*"
(0.249) (0.683) (0.033) (0.035)
Accumulated Number of Contracts 0.536*** 0804*" 0.031*** 0032*"
Signed between the Same Plant and
Supplier (0.049) (0.085) (0.004) (0.004)
Annual Contract Quantity (trillion Btu's) 0204*" 0406*" 0.022*** 0.021***
(0.011) (0.022) (0.001) (0.001)
The Square ofAnnual Contract Quantity -0.853*** -2.354*** -0.133*** -0.130"‘**
divided by 1,000 (trillion Btu's) (0.129) (0.229) (0.011) (0.011)
Midwest (=1 if the coal production 0.581“ 1.054 0.039 0.064*
is in the Midwest) (0.265) (0.648) (0.033) (0.033)
West (=1 ifthe coal production 1.126“ 3381*" 0.128” 0170*”
is in the West) (0.466) (1.058) (0.053) (0.054)
Initial ContractA 1376*" 2123*“ 0092*" 0.093***
(0.188) (0.537) (0.027) (0.027)
Percentage of Sulfur of the Coal Shipped -0.627*** -1 356*" -0.065*** -0.073***
(0.115) (0.298) (0.015) (0.015)
Percentage of Ash of the Coal Shipped -0.063* -0.112 -0.004 -0.004
(0.033) (0.083) (0.004) (0.004)
Percentage of Moisture of the Coal Shipped 0.013 0.097" 0.004* 0.004“
(0.023) (0.047) (0.002) (0.002)
Year Fixed Effects YES YES YES YES
Total Number of Observations 5,835 5,571 5,835 5,571
R-squared 0.671 0.582 0.675 0.677
Log-likelihood -15222. 169 —l9488.75 -18162.854
Note: A. Initial Contract=1 if the observation is among the initial contracts signed between the same parties

during the sample period.

 

The coefl'rcient estimates are not marginal effects. Robust deviations are in parentheses.
* significant at 10%; ** significant at 5%; *" significant at 1%.

45

THE CHOICE OF PRICING PROVISION

A long-term contract saves the renegotiation cost and induces more efficient
specific investments. However, the binding agreement also reduces the flexibility of the
contracting parties when facing future contingencies. It is therefore important to apply
appropriate pricing provision to reduce the costs of signing a long-term contract while
preserving the benefits of such contract. Joskow (1988) argued a mutually satisfactory
long-term coal contract should incorporate the pricing provision that can 1) equalize the
expected revenues and costs of the coal supply; 2) minimize the incentives for the
contracting parties to breach; 3) provide enough flexibility to facilitate efficient
adaptations to future contingencies; and 4) guard against the potential holdup problem. In
this section, I discuss whether the current pricing provisions used in the coal markets
have fulfill the requirements above, and then empirically test the determinants of the

provision choice.

The Pricing Provisions of the Coal Contracts

Fixed Price Contract. A long-term fixed price contract will be agreed only if the
expected revenue from the coal supply is greater than the expected cost for the supplier.
However, any future contingency can change the costs of both the buyer and seller and
drive the market value deviates from the original expectation, provides the incentives for
the contracting parties to breach. It is therefore less likely to have a long-term contract

using the fixed price provision.

46

Market Price Contract. The market price contract is an opposite pricing
provision to the fixed price contract and thus minimizes the incentives for the contracting
parties to breach when facing the future contingency. However, how market price is
defined in the contract is hardly to be accurate because the heterogeneity of the coal
contents across different mines (Joskow, 1985, 1988). Besides, there may not even be a
market price of the coal for a mine-mouth plant because it is the monopsony of the mine
that provides the coal. Consequently, the market price contract creates the possibility of
future haggling and provides less protection for the specific investments in the long-term
relationship.

Evergreen Contract. The evergreen contract functions similarly with the market
price contract but with predetermined renegotiation interval. It therefore suffers the same
problem as the market price contract with the vague definition of the market price and
creates the possibility of haggling from either the buyer or seller. For example, Carlson
(1995) illustrated the case that when different grades of coal escalate at different rates,
the buyer will bring the evidence that favors the lower price to the annual renegotiation
while the seller will evidence that favors a higher price. Therefore, the evergreen contract
may offset the benefits from a long-term contract.46

Cost Plus Contract. The difference between contract prices and market values of
coal is the major problem that induces the possibility of breach or haggling. Cost plus
contract, on the other hand, provides the incentive for the supplier to stay in the
relationship by allowing the price to vary with changes in the cost of producing coal.

However, this type of contracts fails to provide adequate incentives for the supplier to

 

‘6 Guriev and Kvasov (2005) showed that efficient investment can be induced through an evergreen
contract which is perpetual and allows unilateral termination with advance notice. However, the definition
of the evergreen contract in their context is different from here.

47

supply efficiently. The negotiated price can also deviate from the market price because
the production costs of a single mine may fluctuate differently from the market values.
(Joskow 1985, 1988).

Base Price Escalation Contract. The other pricing provision that tracks the
market value of coal is the base price escalation contract. This type of contracts allows
the agreed price to adjust with some appropriate price and productivity indices reflecting
in changes of the cost and productivity of supplier. It can break down the price into
different components and relates each component to different indices." Different from
the cost plus contract, base price escalation contracts provide incentives for the supplier
to minimize costs because the market price fluctuation is independent of the supplier's
production decisions (Joskow, 1985). However, this advantage also implies the indices
may not be sensitive to unanticipated changes in the market. The market price can deviate
fiom the contract price and creates the possibility for haggling. In addition, some

components of the production cost may also be difficult to index.

 

'7 For example, see Joskow (1985).

48

EMPIRICAL RESULTS

A provision that enhances the benefit of using a long-term contract should be
positively related to the determinants that extends contract duration. From the discussion
above, the fixed price contract accounts for the least flexibility for a long term contract.
Market price contract and evergreen contract can be applied for a long-term contract
when there is less haggling expected between the contracting parties over the definition
of market price. Base price escalation contract and cost plus contract, on the other hand,
provide the most flexibility and the least possibility of haggling. However, an uncertain
environment with more possible unanticipated shocks will reduce the probability for the
price escalation contract and cost plus contract being negotiated. Cost plus is also a less
likely choice when there are more difficult for the utility/plant to monitor the production
efficiency of the supplier.

Table 10 presents the empirical results. To examine the determinants that affect
the choice over different pricing provisions, I estimate a multinomial logit model and use
the same regressors as in the contract duration regression. Since the coefficient estimates
of the multinomial logit regression solely indicate the log-odds change between the base
category and each category as the regressors change, Table 11 presents the choice
between pricing provision pairs in the Appendix. Specifically, Table 11 provides the
probability change between any pair of the pricing provisions, indicates which provision
will be the more or less likely choice than the alternative after the regressors change. For
example, when there are one more contract signed between the same parties, base price

escalation contract is the more likely choice than cost plus contract, while evergreen

49

contract is a less likely choice compare to cost plus contract. In addition, only the
estimates that are statistically significant in the ten percent level are reported, and the

results for provisions that fall out of the five categories discussed above are not presented.

 

 

 

   

 

 

 

 

 

 

Table 10 Multinomial Logit Regression: Dependent Variable (Pricin - Provision)
.,-.:,:.-.-,-.» . » . gas . C
,Mariable , ._...Esc ' n , Cost Plus . Evergreen . Market Price .
Mine-Mouth Plant (=1 ifthe 1.524 3398*“ -39.349 -36.802
plant is mine-mouth plant) (1.160) (1.290) (0.000) (0.000)
Total Distance of A Route 0440"" -0.679 0724*" 0.688
(1,000 miles) (0.125) (0.434) (0.206) (0.419)
Accumulated Number of Contracts Signed 0.009 -0.164** 0.068” 0.008
between the Same Plant and Supplier (0.024) (0.087) (0.036) (0.091)
Annual Contract Quantity 0031*" 0.094‘" 0.041*** 0.054“
(trillion Btu's) (0.006) (0.014) (0.010) (0.021)
The Square of Annual Contract Quantity -0.195"'** -0.630*** -0.301** -0.570"‘
divided by 1000(trillion Btu's) (0.071) (0.151) (0.132) (0.317)
Midwest (=1 if the coal production -l.140*** -2.158*** -1.237*** -31.369
is in the Midwest) (0.149) (0.402) (0.249) (0.000)
West (=1 if the coal production -0.801*** -0.817 4589*“ -1.070"
is in the West) (0.200) (0.539) (0.573) (0.629)
Initial ContractA -0103 —0.1 11 0.288 0.144
(0.110) (0.377) (0.179) (0.386)
Percentage of Sulfur of the Coal Shipped -0.014 0364*” 0165 -0.633*"'
(0.063) (0.104) (0.101) (0.294)
Percentage of Ash of the Coal Shipped 0.006 -0.042 0103*" 0.054
(0.019) (0.043) (0.030) (0.056)
Percentage of Moisture of the Coal Shipped -0.027*** -0.072*** 0.010 0.015
(0.009) (0.024L M21) (0.029)
Year Fixed Effects YES
Observations 5,835
Log-pseudolikelihood -4979.177
Pseudo R-griared 0.268

 

Note: A. Initial Contract=1 if the observation is among the initial contracts signed between the same parties

during the sample period.
B. Base category is fixed price contract.

The coefficient estimates are not marginal effects. Robust standard errors presented in parentheses.

* significant at 10%; ** significant at 5%; *** significant at 1%.

50

The empirical results are consistent with the argument of Masten and Saussier
(2000) that specific investments and uncertainty are important to the contract design, and
also confirm the importance of expected surplus to the contract type choice. To be
specific, cost plus contract is the most likely choice for a mine-mouth site. In most cases,
cost plus contract is also the less likely choice when a plant is more far away from its
supplier. The results indicate cost plus contract is the preferred choice when there are
more expected surplus.48 For the plant and supplier who have traded more frequently thus
. with less uncertainty, evergreen contract is the more possible choice while cost plus
contract is the less likely choice. In addition, fixed price contract is the least likely choice
as one more trillion Btu's contracted. It indicates fixed price contract may not be an
appropriate provision when there are more specific investments. Cost plus contract, on
the other hand, is the most likely choice when there are more specific investments.
However, fixed price contract is a more preferred candidate in most of the cases when the
coal production is in the Midwest or the West. According to Joskow (1987), the results
may indicate that cost surplus contract is the more possible choice when there are more
dedicated asset specificity, while fixed cost contract is the choice when there are more
physical asset specificity. In conclusion, cost surplus contract is the preferred choice of
pricing provision when there are more expected surplus, uncertainty, and dedicated asset
specificity. Evergreen contract is the more possible choice when there is less uncertainty,
while fixed cost contract is the likely choice when more physical asset specific

investments are required.

 

‘8 The monitoring problem inherited in the cost plus contract can be the alternative explanation of why such
contract is the least likely choice when the plant is far away from its supplier.

51

THE SIMULTANEOUS CHOICE OF CONTRACT DURATION AND PRICING

PROVISION

Since both contract duration and the choice of pricing provision are affected by
the same determinants, a joint analysis is applied to examine whether the simultaneity has
affected the above empirical evidence. The estimation results using the seemly unrelated
regression (SUR) model are listed in Table 12. The dependent variables include contract
length and the dummy variables that indicate whether a specific pricing provision has
been used. The independent variables are the same as the previous regressions. As
Wooldridge (2002) pointed out, the coefficient estimates fi'om the SUR model are the
same as the OLS estimations equation by equation with improved estimation efficiency.
However, the contract length regression results are not significantly different from the
previous estimation with the consideration of simultaneous pricing provision choice. The
effects from expected surplus, uncertainty, and specific investments on contract duration
remain statistically significant. Moreover, the SUR model also provides comparable
results with the estimates from the multinomial logit model on the pricing provision
choice: the probability to use a cost plus contract increases with expected surplus;
evergreen contract is the more likely choice with less uncertainty while cost plus contract
is the more likely option when there are more uncertainty; and the fixed price contract is
the more likely choice when the coal production is in the Midwest or the West where
more specific investments are required. In addition, the SUR model also indicates base
price escalation contract and cost plus contract are the more possible choice with more

dedicated asset specificity, while the fixed price contract is the less possible option. The

52

model also indicates the relationship between expected surplus and the adoption of fixed

price contract are unclear, because the probability to use fixed price cost is smaller for a

mine-mouth site but the probability also decreases with the total distance of a route.

Table 12 Seemly Unrelated Regression: Dependent Variables (Contract Duration

 

 

 

 

  
    

 

 

 

 

 

‘ Contract ., Market ‘ FiXed- "
Variable Duration . Cost Plus . ‘ Evergeen Price Price
Mine-Mouth Plant 10.312*** -0.070 0.149“ 0.039 -0.019 -0.125**
(=1 if the plant is (0.982) (0.077) (0.024) (0.041) (0.018) (0.059)
mine-mouth plant)
Total Distance ofA -0.790*** 0.051*** -0.017*** 0.016 0.003 -0.059"““"'
Route (1,000 miles) (0.227) (0.018) (0.006) (0.010) (0.004) (0.014)
Accumulated 0.536‘" 0.003 -0.004**"‘ 0003* -0.000 -0.003
Number of (0.040) (0.003) (0.001) (0.002) (0.001) (0.002)
Contracts Signed
between the Same
Plant and Supplier
Annual Contract 0204*“ 0002*" 0001*“ 0.001 0.000 -0.003***
Quantity (trillion) (0.009) (0.001) (0.000) (0.000) (0.000) (0.001)
Btu's
The Square of 0 020*)”.
Annual -0.853*** -0.009*** 0007*" -0.004 -0.002* ‘
Contract Quantity (0 005)
divided by (0.087) (0.007) (0.006) (0.004) (0.001) '
1000(trillion Btu's)
Midwest (=1 ifthe 0.581" -0.052*** -0.021*** -0.008 -0.010** 0139*"
coal production is in (0.259) (0.020) (0.006) (0.011) (0.005) (0.015)
the Midwest)
West (=1 if the coal 1.126*"'* 0.043 -0.001 -0.105*** 0.002 0125*“
production is in the (0.340) (0.027) (0.008) (0.014) (0.006) (0.020)
West)
Initial ContractA 1376*" -0.017 -0.006 0.020" 0.002 0002
(0.199) (0.016) (0.005) (0.008) (0.004) (0.012)
Percentage of Sulfur -0.627""“* 0.010 0012*” -0.008"‘ -0.004* 0.003
of the Coal Shipped (0.103) (0.008) (0.002) (0.004) (0.002) (0.006)
Percentage of Ash -0.063*"' 0008*" -0.001 -0.006*"‘* 0.001 0.001
of the Coal Shipped (0.031) (0.002) (0.001) (0.001) @001) (0.002)

 

53

Table 12 (cont’d)

 

 

 

 

 

 

 

 

Percentage of 0.013 -0.004*** -0.001 -0.000 0.000 0004*"
Moisture of the Coal (0.016) (0.001) (0.000) (0.001) (0.000) (0.001)
Shipped
Year Fixed Effects YES YES YES YES YES YES
Observations 5,835 5,835 5,835 5,835 5,835 5,835
R-squared 0.67 0.22 0.18 0.23 0.10 0.32

 

Note: A. Initial Contract=l if the observation is among the initial contracts signed between the same parties

during the sample period.

To check the simultaneity between the regressions, Table 13 presents the

correlation matrix of residuals from the SUR estimation. The results show the residuals

from the contract length regression are not significantly correlated with the unexplained

variations from the pricing provision regressions. Table 13 also shows the errors from

different pricing provision regressions are not significantly correlated with each other,

except the substitutions between base price escalation contract and the rest contract types.

At last, the Breusch-Pagan test can not reject the existence of simultaneity.

Table 13 Correlation Matrix of Residuals from the Seemly Unrelated Regression

 

 

 

 

 

 

 

 

Fixed Base Price Market Cost
Duration Price Escalation Price Plus Evergreen

Duration 1.000
Fixed Price -0.147 1.000
Base Price Escalation -0.004 -0.539 1.000
Market Price 0.014 -0.058 -0. 147 1.000
Cost Plus 0.087 -0.014 -0.259 -0.011 1.000
Evergreen 0.050 -0.133 -0.349 -0.033 -0.023 1.000

 

CONCLUSION

This paper extends the seminal paper of Joskow (1987) by examining how
contract duration is affected by expected surplus and uncertainty, in additional to his
focus on specific investments. The empirical evidence shows contract duration increases
with expected surplus and specific investments, and decreases with uncertainty. The
results are consistent with Tang (2009). Moreover, this paper also looks at the choice of
pricing provision in the coal market. The empirical results indicate a cost plus contract is
the more likely choice when there are more expected surplus. The estimates also Show
that evergreen contract is the more likely choice when less uncertainty is presented while
cost plus contract is the more likely option when more uncertainty exists between the
contracting parties. The specific investment is also important for the choice of contract
type, where cost plus contract is the more possible choice when there are more dedicated
asset specificity, and fixed cost contract is the likely choice when there are more physical
asset specificity.

While the simultaneous estimation confirms the above findings, the empirical test
can not reject the existence of simultaneity between the choice of contract duration and
pricing provision. The result indicates the other estimation techniques may be needed to
deal with the simultaneous choice problem. There are two potential solutions to address
the concern. First, an instrumental variable regression can be applied to address the
simultaneity as Crocker and Masten (1991). However, appropriate sets of instrumental
variables are necessary for this estimation method. In addition, since contract duration is

a count variable and the choice of pricing provision is a discrete choice variable, the

55

nonlinear simultaneous estimation method is the other choice instead of the SUR model.
While this paper confirms the simultaneity between the choice of contract duration and
pricing provision choice does not significantly affect the results, it will be worthwhile to
see the future research exploring how the simultaneity will affect the coal contract design

nonlinearly.

56

REFERENCES

Bandiera, Oriana. "Contract Duration and Investment Incentives: Evidence from Land
Tenancy Agreements," Journal of the European Economic Association, 2007,
5(5), pp. 953-86.

Brickley, James A.; Misra Sanjog; and Van Horn, R. Lawrence. "Contract Duration:
Evidence fiom Franchising," The Journal of Law and Economics, 2006, 49(1), pp.
1 73-96.

Carlson, Kenneth E. "Fossil Fuels," Power Plant Engineering edited by Larry Drbal and
Kayla Westra, 1996, Springer Science+Business Media Inc., pp71-123.

Cheung, Steven N.S. "Transaction Costs, Risk Aversion, and the Choice of Contractual
Arrangements," The Journal of Law and Economics, 1969, 12(1), pp. 23-42.

Chiappori, Pierre-Andre and Salanié, Bernard. "Testing Contract Theory: a Survey of
Some Recent Work," CESifo Working Paper Series No. 738, 2002, SSRN:
http://ssm.com/abstract=3 1 8780

Crocker, Keith J. "Regulatory Issues with Vertically Disintegrated Public Utilities: A
Transaction Cost Analysis," in John Groenewegen (eds), Transaction Cost
Economics and Beyond, 1996, pp. 85-103, Kluwer Academic Publishers.

--------- . and Masten, Scott E. "Mitigating Contractual Hazards: Unilateral Options and
Contract Length," RAND Journal of Economics, 1988, 19(3), pp. 327-43.

--------- . and Masten, Scott E. "Pretia ex Machina? Prices and Process in Long-Term
Contracts," Journal of Law and Economics, 1991 , 34(1), pp. 69-99.

--------- . and Reynolds, Kenneth J. "The Efficiency of Incomplete Contracts: An
Empirical Analysis of Air Force Engine Procurement," RAND Journal of
Economics, 1993, 24(1), pp. 126--46.

Guriev, Sergei and Kvasov, Dmitriy. "Contracting on Time," American Economic
Review, 2005, 95(5), pp. 1369-85.

57

J oskow, Paul L. "Vertical Integration and Long-Terrn Contracts: The Case of Coal-

Burning Electric Generating Plants," Journal of Law, Economics, and
Organization, 1985, 1(1), pp. 33-80.

--------- . "Contract Duration and Relationship-Specific Investments: Empirical Evidence
from Coal Markets," American Economic Review, 1987, 77(1), pp. 168-85.

--------- . "Price Adjustment in Long-Term Contracts: The Case of Coal," Journal of Law
and Economics, 1988, 31(1), pp. 47-83.

Leffler Keith B. and Rucker Randal R. "Transaction Costs and the Efficient Organization
of Production: A Study of Timber-Harvesting Contracts," Journal of Political
Economy, 99(5), 1060-87.

Masten, Scott E. "Contractual Choice," in B. Boukaert and G. De Geest (eds.),
Encyclopedia of Law and Economics, Vol. III, The Regulation of Contracts, 2000,
pp. 25-45. Edward Elgar Publishing.

--------- . and Saussier, Stephane. "Econometrics of Contracts: An Assessment of
Developments in the Empirical Literature on Contracting," Revue d'Economie
Industrielle, 2000, 92, pp. 215-36.

Tang, Meng-Chi. "Contract Length: Expected Surplus, Uncertainty, and Specific
Investments," 2009, Michigan State University Working Paper.

Wallace, Frederick H. "The Effects of Shock Size and Type on Labor-Contract
Duration," Journal of Labor Economics, 2001, 19(3), pp. 658-81.

Williamson, Oliver E. "Credible Commitments: Using Hostages to Support Exchange,"
American Economic Review, 1983, 73, pp. 519-40.

Wooldridge, Jeffery. Econometric Analysis of Cross Section and Panel Data, 2002
Cambridge, Mass: MIT Press.

58

CHAPTER III

THE EFFECT OF AN AGENT’S EXPERIENCE ON NATIONAL FOOTBALL
LEAGUE CONTRACT STRUCTURE.

“Hiring an agent who has the most experience or who has represented the
most number of players is not necessarily the best choice. There are more
than 750 certified agents with varying degrees of experience and numbers
of clients. ‘Just because an agent is less experienced or works at a smaller
firm does not necessarily mean he or she can ’t provide the same level of
service or expertise in contract negotiations, ’ Levin said. ”

http://wwwnflplayers.com/user/content.aspx?fmid=l78&lmid=443&pid=2557&type=n&weigh=443,0,2557,n
(March 16, 2009)
INTRODUCTION

The above quote on the National Football League Players Association (NF LPA)
website indicates that a player’s best choice in terms of representation may not be the
agent with the most experience. Mark Levin, the NFLPA’S Director of Salary Cap and
Agent Administration, clearly states in the quote above that an agent’s expertise in
contract negotiations is not strictly based on the agent’s experience level. He also implies
that there are other services, besides contract negotiations, that are provided by an agent
which are important for the player to consider when choosing representation. While this
paper cannot directly address the question of whether a player is better off by selecting an
agent with more experience in contract negotiations, it is the first research we are aware
of that empirically tests whether an agent’s experience affects contract structure and

contract negotiations. Specifically, this paper estimates how an agent’s experience

 

' This is a joint work with Mike Conlin and Joe Orsini. For providing the contract data, we are indebted to
Michael Duberstein and Mark Levin of the National Football League Players Association. We are also
grateful to Mark Levin for providing the agent certification information and invaluable information on the
agent’s roll in contract negotiations. Finally, we thank Jeffery Wooldridge and Stacy Dickert-Conlin for
their comments, insights and suggestions.

59

affects contract length, how long the negotiations last and whether the contract includes
an incentive clause. We use whether an agent is certified by the NFLPA and how long
ago an agent obtained the certification as proxies for agent experience.

Using contract information on 1,872 players drafted prior to the implementation
of NFL salary cap in 1993 and on 1,782 players drafted after its implementation, we find
that experienced agents are more likely to negotiate shorter contracts and negotiations are
more likely to last past the start of the NFL team’s training camp.49 One explanation for
these empirical results is that experienced agents are better informed in terms of the
tradeoffs associated with contract duration and missing the start of training camp due to
protracted negotiations. For example, experienced agents may realize that while they
would be able to get a larger signing bonus if they agree to a contract of longer duration,
a longer duration would delay a perhaps sizable increase in salary associated with the
player’s subsequent contract.50 Perhaps experienced agents also better understand the
cost to the team of a player missing the start of training camp and this cost induces the
team to make a more lucrative contract offer after training camp starts.“ (See Conlin

1999 and Conlin & Emerson 2003 for a more detailed discussion of these issues.) Our

 

‘9 Leeds & Kowalewski (2001) and Larsen, F enn & Spenner (2006) address how the salary cap and free
agency afl‘ect player compensation and the competitive balance within the NFL.

50 All else equal, drafted players prefer contracts of shorter duration because they can generally expect to
receive a much higher yearly salary in their subsequent contracts if they perform well. The increase in pay
between the last year of one contract and the first year of the next contract tends to be greater than the
increase in base salary between two years covered under the same contract. As the NFL Player's
Association puts it in the 1991 edition of On the Sidelines, “Historically, career earnings have been larger
when contract lengths have been shorter... rookies with two-year packages... have done better than
entering players with longer deals.”

5' The December 1991 edition of the National Football League Players Association’s (NF LPA’s)
publication, 0n the sidelines: an annual economic analysis of the national football league prepared for
members of the NFL Players Association, states: “[W]hen a contract is signed has a major impact on what
gets signed. For draftees especially, early deals as a rule produce numbers not only below the final
averages in a round, but in many cases also under averages from the previous season.”

60

results also suggest that, prior to the NFL salary cap, experienced agents are more likely
to negotiate a contract that includes an incentive clause.

While the empirical results presented in this paper provide evidence that
experienced agents are more likely to negotiate shorter contracts that contain an incentive
clause and reach contractual agreement after a longer negotiation period, this does not
imply that a player is better off selecting a more experienced agent or that experienced
agents negotiate superior contractual terms. As Mark Levin mentions, negotiating
contractual terms is just one service an agent provides the player. In addition, we do not
have information on how fees vary across agents.52 It may be the case that more
experienced agents charge higher fees or that their fee structure differs from

53 What this paper does demonstrate is that there are systematic

inexperienced agents.
differences in contract structure and how quickly contractual terms are reached when

players are represented by an experienced compared to an inexperienced agent.

 

52 See Sobel (1993) and Dai, Lewis & Lopomo (2006) for two principal-agent models with interesting
implications for how the structure of the agent’s compensation should vary as a function of the agent’s
growledge.

61

DATA AND SUMMARY STATISTICS

The NFL conducts a draft in late April, during which players from the collegiate
level are selected by a specific team. All teams are given a draft pick in each round and
the order of the picks is based on how the team performed in the prior season. When a
player is drafted by a particular team, that team has exclusive rights which prevent other
NFL teams from negotiating with that player — unless the other NFL team trades for the
player’s rights. In early July, drafted players who have reached contractual agreements
attend their team's training camp. At training camp, players improve their conditioning,
learn their team’s plays and compete against other NFL teams in exhibition games.
Players who have not signed a contract cannot attend training camp. Ahnost all players
Sign contracts before the regular season begins, in late August or early September.

Our data consist of contract information for drafted players from two time periods:
1986 through 1991 and 2001 through 2007. The earlier time period is prior to a salary
cap negotiated between the NFL and the NFLPA in their 1993 Collective Bargaining
Agreement (CBA). While a number of changes have occurred in the CBA post-1993, the
basic structure of the salary cap has not. The most significant changes for our analysis
are the restriction in 2006 on the maximum length of a drafted player’s contract and the
changes in what constitutes a restricted free agent, an unrestricted free agent and a

“franchised” player.54 Some minor changes have also occurred in the draft. Specifically,

 

5" Section 5 of Article XVII of the 2006 CBA states: “Tire initial Players Contract of a Rookie, including
any Club option, may not exceed four years in length, except that the initial Player Contract of a Rookie
drafied with a selection in the first half of the first round (e. g., the first sixteen of thirty-two selections in
the 2006 Draft), including any Club option, may not exceed six years in length, and the initial Player
Contract of a Rookie drafted with a selection in the second half of the first round, including any Club
option, may not exceed five years in length.” Articles XVIII and XIX of the CBA provide the details on
restricted free agents, unrestricted free agents and the “franchise” designation.

62

the number of NFL teams (each of which has drafting rights) has increased from 28 in
1985 to 32 in 2007, the number of rounds in a draft has decreased fi'om twelve in the
earlier time period to seven in the later time period and 32 “compensatory picks” were
awarded (each year from 2001 thru 2007) to teams that lost players due to free agency.
Most players hire an agent or multiple agents prior to the draft to negotiate their
contract. While some players represent themselves or have a family member/friend
negotiate the contract, most players hire professional agents who specialize in
representing professional athletes. In 2003, the NFLPA restricted the agent’s
compensation to be a maximum of three percent of the contract’s value but this rate is

55 In addition to this compensation, agents often receive compensation for

negotiable.
providing additional services to the player such as negotiating promotional deals. Once a
player is drafted, a representative of the drafting team negotiates contract terms with the
player’s agent/representative. These negotiations ahnost always result in a contractual
agreement where the player signs what is often termed a “Standard Form Contract”. (See
Appendix C of the 2006 CBA for a sample of this contract.) Besides the amount of the
signing bonus, base salaries, contract length, and (on occasion) incentive clauses, the
contracts signed by players drafted prior to the salary cap were largely standardized.
Those contracts signed in the 2001 through 2007 drafts were less standard because the

structure of the contract influenced how much the player’s compensation counted against

the salary cap. For example, these contracts often contained a wide variety of incentive

 

55 A player can also compensate his representative on a fixed fee or an hourly basis. However, most
agents’ compensation is a function of the contract terms (i.e., a contingent fee).

63

clauses because the “value” of an incentive clause for salary cap purposes depends on
whether the incentive clause was “likely to be earned”.56

Our dataset contains contract information, the player’s college and the date of
agent certification provided by the NFLPA. The NFL provided training camp starting
dates for the different teams while team information (including win-loss records,
attendance, stadium capacity and head coach tenure) was collected from the NFL Record
and Fact Books. Finally, information on whether the drafi picks attended a Division IA
college football program was collected from the Official National Collegiate Athletic
Association College Football Records Books.

Table 14: Descriptive Statistics (Means and Standard Deviations)

 

 

1986-1991 2001-2007
. Drafts :2 : Drafts i
Contract Duration (years) 2. 736 3.809
(0.805) (1.024)
Rookie Training Camp (=1 if sign after start of 0.504 0.409
Training Camp, 0 otherwise) (0.500) (0.492)
Incentive Clause (=1 if incentive clause, 0 0.033 0.262
otherwise) (0.179) (0.440)
Agent Certified Prior (=1 if agent certified prior to 0.334 0.937
Draft, 0 otherwise) (0.472) (0.244)
Number of Years Agent Certified prior to Draft 4.408 13.258
(1.945) (5.656)
Agent Ever Certified (=1 if agent was ever 0.426 0.941
certified, 0 otherwise) (0.495) (0.237)
Selection Number in Draft 161 128
(95) (74)

 

 

 

5‘ See Section 7 of Article XXIV in the 2006 CBA for a detailed discussion of how incentive clauses are
accounted for in salary cap calculations.

64

Table 14 (cont’d)

 

 

Fraction of Empty Seats in Stadium for Team in 0.182 0.055
Prior Year (0.143) (0.081)
Nmnber of Team Wins in Prior Year 7.769 7.877
(2.965) (3.054)
College Division (=1 if player competed in Div IA 0.791 0.914
college program, 0 otherwise) (0.407) (0.280)
Stadium Capacity (10,000) 6.802 6.971
(0.919) (0.611)
Increase in Stadium Capacity from Prior Year 0.067 0.079
(0.251) (0.269)
Decrease in Stadium Capacity from Prior Year 0.025 0.042
(0.155) (0.200)
Population of Team’s MSA (in 1,000,000) 4.782 5.008
(4.935) (4.955)
Tenure of Team’s Head Coach (years) 5.723 3.960
(6.381) (3.045)
Total Number of Observations 1,872 1,782

 

 

2007. 57

Table 14 contains the variables’ means and standard deviations for the 1,872 draft

 

65

selections from 1986 through 1991 and for the 1,782 draft selections from 2001 through
In terms of contract length, the average for the 1986-91 selections is
approximately a year less than for the 2001-07 selections (2.736 years compared to 3.809
years). The primary reason for this difference is that early round draft choices usually
sign longer contracts than late round choices and there were five more rounds in the early

time period. This also explains why the fraction of draftees from Division IA schools is

57 We have information on all contracts reported to the National Football League Players Association
(NF LPA). Approximately 95% of all drafted payers in 1986 thru 1991 reported the contractual terms to the
NFLPA and almost 100% reported from the 2001 thru 2007 drafts.

much less for the early time period - players from smaller football programs are more
likely to be drafted in the later rounds. Draftees in the early time period are also less
likely to Sign a contract before the start of training camp and much less likely to have an
incentive clause in their contract. The main reason why incentive clauses are more
prevalent in the 2001-2007 drafts is because providing compensation in the form of
incentive clauses allows teams to better manage their salary cap. In terms of agent
certification, Table 1 indicates that a third of the draftees selected in the 1986-1991 drafts
were represented by a certified agent while 93.7 percent of the 2001-2007 contracts were
negotiated by agents who were certified. For those represented by a certified agent, the
average number of years between certification and contract negotiation is 4.41 years for
1986—91 and 13.26 years for 2001-07.58 For the different number of years between
certification and contract negotiation, Figure 8 presents the relevant number of 2001-07
contracts. As the figure depicts, the number of years range between 0 and 25 with a
concentration between 13 and 19 years.

Figure 8 Number of Years Agent Certified Prior to Draft for the 2001-2007

Contracts
| l I l I .

58 For players represented by more than one agent, we define the number of years between certification and
contract negotiation based on the agent who first became certified.

120

0 . .
0 1n 0
,_

Years Mont Cortlflod

8

Number of Contact:
A a:
o o

8

 

 

 

 

 

 

 

1n

 

66

EMPIRICAL SPECIFICATION AND ESTIMATION

We test how an agent's experience/certification affects contract length, whether
contractual terms are reached after the start of training camp and whether the contract
includes an incentive clause. Because contract length is a count variable that ranges from
1 to 6 years in the early period and from 1 to 7 years in the later period, we estimate a
binomial regression model when considering contract length. Specifically, we assume
that the conditional mean of the length to be estimated equals np(XiB), where X; contains
the set of regressors, p(.) is a probability function estimated by the logit model and n is
the maximum contract length”. When considering how the agent’s experience affects
whether contractual terms are reached after the start of training camp and whether the
contract includes an incentive clause, we estimate probit models since both are binary
variables. The binomial regression and probit models are estimated separately for the
1986-91 and 2001-07 data.

The set of regressors are the same across models and include team, player, draft
and agent characteristics. The team characteristics consist of the fiaction of empty
stadium seats in the prior season, number of wins in the prior season, stadium capacity,
whether stadium capacity increased or decreased by over 1,000 seats from the prior year,
MSA population and head coach tenure.60 The set of regressors also includes team
specific indicator variables to control for unobserved, time-invariant team characteristics.

The player characteristics include whether the player competed in a Division IA college

 

59 Since contract length in the sample is a count variable ranging from 1 to 6 or 1 to 7, multiplying the
estimated probability function by 6 or 7 guarantees the fitted values for contract length fall within this
range. If we use ordinary least squares, it is possible to obtain fitted values of contract length that are
negative. Moreover, although count data is often assumed to have Poisson distribution, Poisson regression
may produce some fitted values exceeding the upper bound of the sample.

60 Annual MSA population measures are constructed by interpolating and extrapolating population counts
from the decennial census.

67

football program and a full complement of player position indicator variables. The
player’s selection number in the draft, indicator variables for the round the player was
selected in the draft and indicator variables for the year of the draft are also included as
regressors. These team, player and draft characteristics control for non-agent factors that
are likely to influence contract negotiations and contract structure."l Finally, to test
whether an agent’s experience affects contract negotiations/structure, the set of regressors
includes whether the agent was certified prior to the contract negotiations. For the 2001-
07 data, we also include the number of years the agent was certified prior to the contract
negotiations.62

Table 15 Binomial Regression and Probit Models

 

 

1986-1991 Drafts '
Contract Training Incentive _ _
. Duration .Camp Clause _- I
Agent
Certified Prior
(=1 ifagent -0.035** 0.240" 0.368“ 0.059 -0.023 -0.004
certified prior (0.017) (0.068) (0.129) (0.040) (0.181) (0.171)
to Drafi)
Years Agent is -0.004** 0.013" 0.004
Certified Prior (0.002) (0.006) (0.006)
to Draft
Selection —0.004** -0.001 -0.017* -0.002* 0.005 -0.003
Number in (0.001) (0.005) (0.010) (0.001) (0.003) (0.004)
Draft

 

 

 

 

6' Tang (2008) uses the same datasets as this paper to test whether expected surplus, uncertainty and
specificity of investment affects contract length. He proxies for these three characteristics of the
negotiations using many of the same team, player and draft characteristics.
62 From 1986 thru 1991, the benefit an agent derived from being certified by the NFLPA was much less
than in the later years. In recent years, the NFLPA has provided much more assistance to players deciding
which agent to select and this has increased the agent’s benefit of being certified by the NFLPA. As
NFLPA Director of Salary Cap and Agent Administration Mark Levin states: “We have files on every
agent. We can tell a player what contracts an agent has negotiated, who they represent, what fees they
charge and most importantly, whether the Committee on Agent Regulation and Discipline has ever taken
action against the agent.”

h

 

This results in far fewer agents being certified and makes years since certification a much less reliable
proxy for agent experience in the early time period. Therefore, we did not include years since certification
as a regressor when estimating the models using this early time period.

68

 

 

 

 

Table 15 (cont’d)
Fraction of
Empty Seats in
Stadium for 0.102 0.995" 0.389 -0.217 -0.444 0.945
Team in Prior (0.131) (0.505) (1.046) (0.184) (0.855) (0.905)
Year
Number of 0.001 0.026 0.021 -0.007* 0.020 0.001
Team Wins in (0.005) (0.020) (0.041) (0.004) (0.015) (0.016)
Prior Year
College
Division (=1 if
player
competed in -0.055** 0.029 0.051 -0.026 0.183 0.092
Div IA college (0.019) (0.079) (0.160) (0.031) (0.127) (0.125)
Prograln)
Stadium -0.069 -0.272 -0.712 0.017 0.197 0.226
Capacity (0.044) (0.176) (0.641) (0.037) (0.176) (0.181)
(10,000)
Increase in
Stadium -0.037 0.285 -0.100** 0.106 -0.039
Capacity from (0.049) (0.185) (0.032) (0.157) (0.166)
Prior Year
Decrease in
Stadium -0.104* -0.320 6.387 -0.050 -0.l44 -0.026
Capacity from (0.057) (0.235) (4.013) (0.039) (0.195) (0.192)
Prior Year
Population of -0.162** 0.047 -0.803 0.029 -0.246 0.159
Team’s MSA (0.053) (0.176) (0.675) (0.033) (0.173) (0.194)
(in 1,000,000)
Tenure of 0.013" 0.026" -0.030 0.007 0.005 0.006
Team’s Head (0.003) (0.012) (0.021) (0.004) (0.021) (0.021)
Coach (years)
Fixed EffectsA YES YES YES YES YES YES
R-squared 0.167 0.140 0.191 0.263 0.265 0.184
Log-
Likelihood -2,255 -1,115 -200 -2,292 -881 -831
N'm'be’ff 1,872 1,872 1,261 1,770 1,770 1,770
Observatrons

 

Notes: A. Indicator variables include the position, round, team, and year fixed effects.
The standard errors in parentheses are robust to arbitrary heteroskedasticity. * Statistically
significant at .10 level; ** Statistically significant at .05 level. The SS column denotes whether we
can reject the null that the corresponding coefficients in the related and unrelated accident probits
are equal.

69

The coefficient estimates associated with the binomial regression and probit
models are shown in Table 15. Both before and after the salary cap, these estimates
indicate that more experienced agents are likely to agree to shorter contracts and be
involved in more protracted contract negotiations.63 For the 1986-91 drafts, experienced
agents are also more likely to negotiate contracts that contain incentive clauses.64 For
this early period, the marginal effects corresponding to the coefficients associated with
having a certified agent indicate that an agent being certified decreases the contract
duration by 0.05 years (coefficient of -0.035), increases the probability of signing after
the start of training camp by ten percentage points (coefficient of 0.240), and increases
the probability of an incentive clause by one percentage point (coefficient of 0.368). As
for the 2001-07 drafts, the marginal effects of having an agent certified ten years earlier
are to decrease the expected contract duration by 0.07 years and to increase the
probability of signing after the start of training camp by five percentage points. However,
it has a negligible affect on the probability the contract contains an incentive clause. The
coefficient estimates for this later time period also suggest that whether the agent is
certified, of which 93.7% are, does not appear to influence when the contract is signed or
its structure in a statistically significant manner.65

As for the coefficients associated with the other regressors in the contract duration
specification, these are consistent with the results of Tang (2008) who finds that players

drafted in the early rounds of the draft, quarterbacks and players from smaller college

 

63 Because the Houston Texans entered the NFL in 2002 and therefore the prior year variables are missing,
the twelve observations of 2002 Houston Texans draft choices are not included in Table 2 regressions
(resulting in the number of observations being 1,770 instead of 1,782).

The number of observations is less when the dependent variable is incentive clause (1,261 compared to
1,872) because certain team and round indicators are perfect predictors of whether contracts have an
incentive clause.

65 For both before and after the salary cap, the qualitative results do not change appreciably if number of
drafted players an agent represents in our dataset is used to proxy for agent experience.

70

football programs (Non-Division IA) are likely to Sign longer contracts. Finally, the
estimates in Table 15 indicate that while certain team, player and draft characteristics
included as regressors influence contract length, they do not appreciably affect when the

contract is signed and whether the contract contains an incentive clause.

71

CONCLUSION

This paper uses contract information for football players drafted into the NFL to
test whether the experience level of the player’s agent influences the contract structure
and negotiation process. We find that experienced agents are more likely to negotiate
shorter contracts that contain incentive clauses and be involved in more protracted
contract negotiations. While our results indicate that the contract structure varies
systematically with the experience level of the player’s agent, we are unable to infer from
these results whether it is beneficial for the player to hire an experienced agent. One
factor that directly influences the player’s optimal agent choice is the agent’s
compensation. While we do not have information on this compensation, we believe
research focusing on how this compensation varies with agent and player characteristics
would provide valuable insight into this principal-agent relationship. In addition, this
research could test the theoretical predictions of the principal-agent models where the

knowledge of the agent varies.

72

REFERENCES

Collective Bargaining Agreement (CBA): between The NFL Management Council and
The NFL Players Association, March 8, 2006.
http://www.nflplayers.com/user/template.aspx?frnid=l 81 &lmid=622&pid=0&typ
g=_l

Dai, Chifeng, Tracy R. Lewis and Giuseppe Lopomo. “Delegating Management to
Experts”, Rand Journal of Economics, 37(3): 503-520, Autumn 2006.

Conlin, Michael. “Empirical Test of a Separating Equilibrium in National Football
League Contract Negotiations”, Rand Journal of Economics, 30(2): 289-304,
Summer 1999.

Conlin, Michael and Patrick Emerson. “Multi-Dimensional Separating Equilibrium and
Moral Hazard: An Empirical Study of National Football League Contract
Negotiations”, The Review of Economics and Statistics, 85(3): 760-765, August
2003.

Larsen, Andrew, Aju J. Penn and Erin Learme Spenner. “The Impact of Free Agency and
the Salary Cap on Competitive Balance in the National Football League”, Journal
of Sports Economics, 7(4): 374-390, November 2006.

Leeds, Michael A. and Sandra Kowalewski. “Winner Take All in the NFL: The Effect of
the Salary Cap and Free Agency on the Compensation of Skill Position Players”,
Journal of Sports Economics, 2(3): 244-256, August 2001.

Sobel, J oel. “Information Control in the Principal-Agent Problem”, International
Economic Review, 34(2): 259-269, May 1993

Tang, Meng-Chi. “Contract Length: Expected Surplus, Uncertainty, and Specific
Investments”, Michigan State University working paper, 2008.

Wooldridge, Jeffery. "Econometric Analysis of Cross Section and Panel Data, ”
Cambridge, Mass: MIT Press, 2002.

73

 

+

+

+

+

+

vommEm
:80 05.8 Rams—2 .«o owfiaoocom

3&5
=80 ofimo :3. mo owsgouom

883m
:80 05 no .525 we owfieoonom

9.63 2: 5 mm
853.95 Eco 05.x TV ~83

@8322 on: E a
eouozuoa 38 05.: TV “8322

9.3m 85::

3.3530 82:00 3:53

Saddam

was “SE 25% 2: e853 Rama
£8550 mo 55:52 poise—33‘
30:8 25.:

83M < no 8535 :38.

O53 5.85-qu

a can on; .uv ago 582852

 

out; out;
8va Sofia—2

 

succumbing

SE
8x32

 

:anmozm 2.5

 

280

 

 

5999m—

 

2:.”—
$00

82.5 nuocmco>m

 

 

ocean—2

 

secs—mumm— BEL 8am

 

35
~80

 

norm—50mm
035 Sam

 

ea... eon...—

 

233$

 

$3358? 05 82: 86:0 383 $2 2: mm 5630a moh N. 63882? 2: ~85 oomoao box: 208 2: mm comma/ca no... IL
Tue: awed 3.52532 .3 185858 FEE aegis—h ”53.5 5952. 33:0 2:. 2 03.;

EGZm—mmdw

74

   

will [laillllllll iii! lilillliliiiililiiillleS
3 1293 03063 0804