EMPLOYEE SELECTION PRACTICES
AND PERFORMANCE PAY
By
Amy M. Swaney

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
Business Administration—Doctor of Philosophy
2014

ABSTRACT
EMPLOYEE SELECTION PRACTICES AND PERFORMANCE PAY
By
Amy M. Swaney
Moral hazard occurs when employee effort is unobservable and there is information
asymmetry regarding the degree of congruence between employee and organizational
goals. The empirical research in management control systems focuses on mitigating
moral hazard in current employees, ex post of the hire date. However, both economics
and management literature acknowledge that there is significant variation in employees’
goal congruence even prior to hire. I examine the effectiveness of employee selection
practices in resolving the information asymmetry regarding an applicant’s ex ante goal
congruence. I find that more extensive use of employee selection practices intended to
identify goal congruence is positively associated with future employee performance.
Further, I find that the association between employee selection practices and performance
is weaker in firms that use performance pay in the initial employment contract.
Interestingly, I also find a positive association between employee selection practices and
use of performance pay in initial contracts.

ACKNOWLEDGEMENTS

I gratefully acknowledge the support and guidance of my dissertation committee: Karen
Sedatole (Chair), Ranjani Krishnan, Isabel Wang, and Arijit Mukherjee. Karen
introduced me to the world of accounting academia and has been an inspiration as both a
researcher and teacher. I hope one day to be half as accomplished as she. I appreciate the
helpful comments and suggestions from Susanna Gallani, Liz Conners, Sally Widener,
Alex Woods, Eric Marinich, Tyler Thomas, Mike Shields, and workshop participants at
Michigan State University, Harvard Business School, Kent State University, University
of Illinois, Central Michigan University, and Northeastern University. I thank the
Department of Accounting and Information Systems at Michigan State University for
providing financial support for this research. Last but not least, I could not have
accomplished this goal without the support, encouragement, and patience of my family:
Steve, Josephine, and Alina Swaney, and my parents Tom and Kathy Pustell.

iii

TABLE OF CONTENTS

LIST OF TABLES ............................................................................................................. vi
LIST OF FIGURES .......................................................................................................... vii
CHAPTER 1: INTRODUCTION ....................................................................................... 1
CHAPTER 2: LITERATURE REVIEW AND HYPOTHESIS DEVELOPMENT .......... 6
2.1. Adverse Selection in the Labor Market ................................................................... 6
2.2.1. Signaling Devices ............................................................................................. 7
2.2.2. Screening Devices ............................................................................................. 8
2.3. Moral Hazard ......................................................................................................... 10
2.4. Management Control through Employee Selection ............................................... 11
2.5. Management Control through Performance Pay.................................................... 14
2.6. Hypothesis Development ....................................................................................... 16
2.6.1. The association between the firm’s employee selection effort and
performance .............................................................................................................. 16
2.6.2. The association between employee selection effort and the use of
performance pay........................................................................................................ 19
2.6.3. Interaction between employee selection practices and performance pay ....... 20
CHAPTER 3: WORKPLACE AND EMPLOYEE SURVEY ......................................... 23
CHAPTER 4: SAMPLE SELECTION, VARIABLE MEASUREMENT AND
EMPIRICAL SPECIFICATION ...................................................................................... 26
4.1. Sample Selection .................................................................................................... 26
4.2. Variable measurement ........................................................................................... 26
4.2.1. Employee performance ................................................................................... 26
4.2.2. Effort in the employee selection process ........................................................ 27
4.2.3. Performance pay ............................................................................................. 28
4.2.4. Controls ........................................................................................................... 28
4.3. Empirical Specification .......................................................................................... 30
CHAPTER 5: EMPIRICAL RESULTS ........................................................................... 32
5.1. Sensitivity analyses and alternative explanations of results .................................. 33
5.2. Alternative explanations ........................................................................................ 36
CHAPTER 6: CONCLUSION ......................................................................................... 37
CHAPTER 7: LIMITATIONS ......................................................................................... 40
APPENDICES .................................................................................................................. 43
iv

Appendix A: Figures .................................................................................................... 44
Appendix B: Tables ...................................................................................................... 47
REFERENCES ..................................................................................................................55

v

LIST OF TABLES

Table 1

Survey Sample Selection ...........................................................................48

Table 2

Variable Definitions ...................................................................................49

Table 3

Descriptive Statistics ..................................................................................50

Table 4

Pearson Correlation Coefficients ...............................................................51

Table 5

SEM Model Using Salary Increase as a Proxy for Employee
Performance ...............................................................................................52

Table 6

SEM Model Using Promotion as a Proxy for Employee Performance .....53

Table 7

OLS Models and Sensitivity Tests .............................................................54

vi

LIST OF FIGURES

Figure 1

Timeline of the hiring process, with the associated control problems
and control systems ....................................................................................45

Figure 2

Theoretical model for the effects of employee selection and
performance pay on employee performance .............................................46

vii

CHAPTER 1: INTRODUCTION
A significant challenge that has been extensively studied is how performance pay can be
used subsequent to an employee’s hire to motivate effort and reduce moral hazard. However,
moral hazard can also be reduced by hiring the ‘right’ employees in the first place: employees
who are likely to have high performance, even absent performance pay. The first step in that
process is to identify which individual(s) in an applicant pool would have high performance in a
given position. Information asymmetry in the hiring process makes this a nontrivial task:
applicants have superior information regarding their own performance potential, and low
performance applicants have an incentive to misrepresent their type. There is a wide literature in
personnel economics regarding ways to resolve this information asymmetry problem in the
hiring process, with the analytical work in this field modeling the ‘right’ employee as the one
with the greatest ability.
Employee performance, however, is a multidimensional construct (Dearman and Shields
2005; Merchant and Van der Stede 2007). In addition to having the right ability, an employee
must also be goal congruent to perform highly—a person with exactly the right skills and
technical knowledge will not be productive unless they also have the motivation to work hard
toward the goals of the firm. Information asymmetry also exists with respect to the employee’s
inherent goal congruence, the extent to which their personal goals and values align with the
interests of the firm even prior to their hire. In this study, I isolate the information asymmetry
problem regarding a new employee’s goal congruence by controlling for employee ability, and
examine the effectiveness of a management control system (employee selection) in resolving this
problem ex ante to the hire date. First, I examine the association between employee selection
practices and employee performance. Next, I examine the interaction between employee

1

selection and performance pay in their effect on employee performance. Finally, I examine if
employee selection and performance pay tend to be used together in firms.
Firms expend significant time and money on employee selection practices and tools, such
as in-person interviews and personality tests. For example, personality tests conducted by
external sources can cost upwards of $5,000 per applicant (Daniels 2001). In-person interviews
can be similarly expensive, yet are widely used across industries, job levels, and countries
(Akyeampong 2006; Huo, Huang, and Napier 2002; Wilk and Cappelli 2003).1 Since it is less
costly for an employer to mitigate the information asymmetry problem regarding applicant
ability by verifying the applicant’s education and certifications, it can be inferred that a major
role of these employee selection practices is in assessing the individual’s goal congruence (Huo
et al. 2002).2
Employee selection is not the only mechanism through which employers mitigate the
information asymmetry problem in the hiring process: performance pay is widely acknowledged
in the accounting and economics literature as separating high ability from low ability applicants
(Banker, Lee, Potter, and Srinivasan 2001; Dohmen and Falk 2011; Lazear 1986; Rothschild and
Stiglitz 1976; Wilson 1980). Employee selection and performance pay in the initial contract are
thus both costly control systems that address the problem of information asymmetry in the hiring
process, although it is unclear whether firms will use these controls together or independently.
On the one hand, firms that are rigorous in their employee selection practices have already
identified and selected goal congruent employees (employees who will work hard toward the
firm’s goals with or without extrinsic motivators such as performance pay). Therefore, the firm’s

1

Gatewood and Feild (1998) roughly estimate the cost per applicant per interview to be one and a half to two days
of interviewer and staff costs, in addition to any travel and living expenses during the stay.
2
In addition to evaluating goal congruence, employee selection practices are also used to evaluate ‘soft’ skills (such
as communication skills) that are critical for job performance but difficult to objectively measure.

2

marginal return to performance pay contracts is lower for firms that are rigorous in their
employee selection practices, so firms should trade off between investment in employee
selection and investment in performance pay and the systems should be used independently. On
the other hand, since the selection effect of performance pay is widely documented, firms may
consider performance pay as simply another tool for employee selection, consistent with
employee selection and performance pay being used together.
Employee selection practices that identify ex ante goal congruence should also identify
intrinsic motivation. Motivation crowding theory shows that performance pay can have a
negative performance effect on intrinsically motivated employees (Benabou and Tirole 2003;
Bonner and Sprinkle 2002; Ellingsen and Johannesson 2008; Gneezy, Meier, and Rey-Biel
2011). Consistent with motivation crowding theory, I predict an interaction between the use of
performance pay and employee selection effort, such that the association between employee
selection effort and employee performance is weaker in the presence of performance pay.
I conduct my study using survey data obtained through the Canadian Workplace and
Employee Survey (WES), collected annually 1999-2006 from approximately 6,300 workplaces
and 24,000 employees at those workplaces. The WES is unusual in that it links employee-level
responses on topics including employee selection practices, pay and benefits, job characteristics,
education, and promotion/termination to firm-level data regarding business strategy, firm
performance, and delegation of decision making. This structure allows me to link use of ex ante
control systems (employee selection practices and compensation structure) to employee
performance while controlling for firm-level factors that would affect the control environment.
I find results generally consistent with my predictions. Using an employee’s salary
increase in the first year of employment to proxy for performance, I find that more extensive use

3

of employee selection practices intended to identify goal congruence is associated with future
employee performance – as expected for an effective control system. Further, I find a significant
negative interaction between performance pay and employee selection practices in their effects
on employee performance. I find the association between employee selection practices and future
employee performance is weaker in firms that use performance pay in the initial employment
contract, consistent with motivation crowding theory which predicts that explicit incentives
crowd out the intrinsic motivational effects of ex ante goal congruence.
Finally, employee selection and performance pay tend to be used together: I find a
positive association between more extensive use of employee selection practices and
performance pay contracts. This is consistent with firms perceiving performance pay as an
additional selection tool, despite the negative performance effect of their combined use.
This study contributes to the literature on management control systems in several ways.
First, management control research generally assumes that employees are not goal congruent
when they join the firm, and concentrates on ways to mitigate moral hazard ex post of the hiring
decision. However, individuals vary in their level of goal congruence, even before they are hired
(Akerlof and Kranton 2005; Heinle, Hofmann, and Kunz 2012). I provide archival evidence that
employee selection practices mitigate goal congruence problems. Second, recent empirical work
on employee selection and management controls uses a single employee selection practice
(referral source) to detect goal congruence (Campbell 2012). I extend this literature by
identifying additional indicators of goal congruence: in-person interviews, personality tests,
direct recruitment by the firm, and use of headhunters. Third, I extend the literature on the
adoption and use of control systems. I examine the potential tradeoffs between investing in
employee selection practices and investing in performance pay, and I examine an interaction

4

between one-time ex ante controls (employee selection practices) and a control that likely
persists throughout the tenure of the employee (performance pay). My findings are consistent
with firms adopting control systems as a package, disregarding the costly negative
interdependence between employee selection and performance pay (Grabner and Moers 2013).
Finally, empirical evidence supporting the use of employee selection practices as a control
mechanism to align incentives is relatively sparse and focuses on practices in a single firm
(Campbell 2012). In contrast, I use a large database across industries to address this question,
which allows me to examine the robustness of these results across different firm-level
environmental conditions.
The remainder of the paper is organized as follows: Chapter 2 reviews current literature
and formalizes the research hypotheses. Chapter 3 describes the Workplace and Employee
Survey data used to conduct my analyses. Chapter 4 describes the measurement of variables and
the empirical specifications used to test the hypotheses. Chapter 5 presents the results, and
Chapters 6 and 7 conclude and discuss limitations and opportunities for future research.

5

CHAPTER 2: LITERATURE REVIEW AND HYPOTHESIS DEVELOPMENT
The main focus of prior literature on management control systems has been on mitigating
the moral hazard problem: implementing control systems to induce the alignment of managerial
preferences with the organization’s goals, with the assumption that employees’ individual goals
are contrary to those of the firm when they are hired. Agency theory, however, predicts another
source of agency problems: adverse selection. In this chapter, I review the literature on adverse
selection and moral hazard, as well as literature on employee selection and performance pay as
mechanisms to alleviate these problems. I then develop hypotheses regarding the effects of
employee selection and performance pay. Figure 1 shows a timeline of the hiring process, the
potential information asymmetry problems associated with time periods prior to and following
hire, as well as some control systems used to address these problems and their potential
unintended consequences.
2.1. Adverse Selection in the Labor Market
Adverse selection arises in the labor market when there is information uncertainty
regarding applicant quality prior to the employment contract, where low quality individuals are
disproportionally attracted to a firm. Without the ability to observe an applicant’s productivity,
the firm then has a greater likelihood of hiring a low quality individual (see Figure 1 period 1). It
should be noted that the concept of quality in the labor market has many definitions across the
different literature streams. The economics literature generally defines employee quality in terms
of overall productivity outcomes (Bedard 2001; Borjas and Goldberg 1978; Horner 2006;
Kubler, Muller, and Normann 2008). The management literature often defines employee quality
as the worker’s ‘fit’ for a job (Autor 2001). Consistent with the management control literature, I
define a high quality employee as one who is both high ability (has the knowledge and skills

6

necessary to complete the job tasks) and goal congruent (motivated to work hard toward the
goals of the organization) (Alles and Datar 2002, Campbell 2012).
Two types of devices exist to address the adverse selection problem, signaling devices
and screening devices, distinguished by the party (firm or individual) that undertakes an action to
uncover individual quality (Riley 2001; Horner 2006). Next, I define and discuss these two
types of devices in turn.
2.2.1. Signaling Devices
Signaling devices are characterized as costly actions taken by the individual to credibly
communicate their own quality to the employer. For these actions to effectively separate high
from low quality individuals, the action must be sufficiently more costly for low quality
individuals than high so that low quality individuals are not tempted to mimic high quality by
also investing in the action. Education has been often cited as a signal of high quality in the
labor market (Spence 1973; Verrecchia 1983; Trueman 1986; Bedard 2001; Kim 2007; Kubler et
al. 2008). Education is costly to both high and low quality individuals, in terms of time and effort
as well as in monetary costs. Both types will only invest in education, then, if the returns to
education outweigh the costs. However, the cost of education is negatively associated with
quality: although tuition and financial costs of education are constant across different quality
individuals, it is easier (less costly in terms of time and effort) for high quality individuals to
succeed. A high quality individual can thus signal their type to the prospective employer by
investing in education. When the employer observes the education signal, they will then infer the
individual’s type and offer a higher wage (relative to the less educated applicant). In addition to
investing in education, individuals can signal their quality by accepting a low initial wage with
opportunity for advancement (and the associated wage increases) over time (Tirole 1988;

7

Lindbeck and Snower 2001). As with investment in education, low initial wages are
differentially more costly to low quality than to high quality workers.
Experimental work has shown that signaling devices (e.g., investment in education) do
not form a perfect separating equilibrium. That is, real people do not always conform to the
signaling investment level decisions predicted by the analytic models. High (and low) quality
individuals tend to choose investment levels significantly below (above) the levels predicted by
economic models, and separating equilibriums do not always occur (Kubler et al. 2008, Miller
and Plott 1985). In a series of experiments, Miller and Plott (1985) show that separating
equilibriums only occur when the signaling cost differential across quality types is very high.
In addition, there may be aspects of a given job for which there is no credible signal of
individual quality. For example, creative problem solving (or multitasking) may be essential for
high productivity in a certain job. In this environment, a credible signal of quality would be an
action the individual could take that is both positively associated with problem solving creativity
and more costly for individuals who are less able to think outside the box. It is difficult to
imagine a credible signal for such an attribute. Similarly, it is difficult to imagine a credible
signal of goal congruence (when a individual’s goals are aligned with those of the organization).
Therefore, signaling devices alone are unlikely to completely alleviate the information
asymmetry problem in the employee selection decision. Not surprisingly, employers often also
employ screening solutions to the adverse selection problem.
2.2.2. Screening Devices
In contrast to signaling devices where costly actions are undertaken by the individuals,
screening devices are characterized as costly actions taken by the firm with the intent to uncover

8

individual quality, either directly or indirectly (by encouraging individuals of one type to selfselect into the organization).
Performance pay is a screening device well noted in the economics and accounting
literature (Banker et al. 2001; Banker et al. 2003; Bonner and Sprinkle 2002; Lazear 1986;
Lazear and Shaw 2007; Milgrom and Roberts 1992; Prendergast 1999). Lazear (1986) notes that
for any given occupation, high quality individuals will be attracted to firms that offer
performance pay, since they are more likely to achieve the high pay associated with strong
performance. In contrast, low quality individuals will self-select into firms where their pay is not
affected by their effort or ability. In one empirical study, Banker et al. (2003) use proprietary
data from a retail establishment that eliminated performance pay for sales employees, while
adjusting hourly wages to reflect prior productivity. In effect, the firm removed compensation
risk from these employees, while maintaining the individual’s expected compensation. They find
a strong negative productivity differential of individuals who joined the firm after termination of
the performance pay plan relative to the productivity of employees who left the firm in the same
period. They infer that the largest impact of this performance pay plan was indirect, through the
self-selection of highly productive, high quality individuals into the firm. Indeed, Prendergast
(1999) suggests that up to one third of the impact of performance pay on individual performance
is due to self-selection. However, there are limits to the effectiveness of performance pay as a
screening device. Lazear (1986) finds that the self-selection value of performance pay is reduced
when the measured output is noisy or costly, when individuals are not aware of their own quality,
and when high quality individuals are risk averse.
Employee selection practices are direct examples of screening devices. Firms spend time
and effort in the employee selection process in order to determine (or at least get a better

9

indication of) individual quality. Firms use a wide variety of employee selection practices,
including interviews, and pre-employment tests for cognitive ability and personality. I discuss
these practices in more detail in a separate section. It should also be noted that these employee
selection practices can also have an indirect, self-selection effect in the labor market. Once the
potential applicant pool is aware of more intense employee selection practices used by a firm,
low quality individuals (who deem themselves less likely to ‘survive the gauntlet’) will selfselect away from the firm.
In addition to performance pay and employee selection practices, the economics literature
has identified a number of other screening devices, including seniority-based pay, and
probationary contracts (Milgrom and Roberts 1992; Salop and Salop 1976; Coyte 1984; Autor
2001; Gibbons and Katz 1991). Although it would be interesting to address all these devices, in
this study I focus on employee selection practices and performance pay.
2.3. Moral Hazard
Information asymmetry regarding employee actions can also adversely affect firm
performance. Moral hazard exists when the firm cannot monitor employee actions, allowing
employee decisions to diverge from the best interests of the firm. Prior empirical literature on
management control systems has focused on Figure 1 period 4, mitigating the moral hazard
problem after employees have been hired (Baiman 1990; Holmstrom 1979; Indjejikian 1999;
Lambert 2001).
However, firms can also use management control systems to manage their risk for moral
hazard in the hiring process (the second and third periods in Figure 1) by selecting goal
congruent employees (Merchant 1982). Both economics and management literature acknowledge
that there is significant variation in individuals’ goal congruence prior to hire (Akerlof and

10

Kranton 2005; Gatewood and Feild 1998; Murphy 2012; Oyer and Schaefer 2011; Sekiguchi and
Huber 2011).
The focus of the current study is on use of management control systems prior to the date
of hire, where the firm is identifying to which applicants they will offer contracts (period 2), and
how to structure those contracts (period 3). The control problem in these two periods is to reduce
the potential for moral hazard by hiring only employees with high overall performance potential,
including both high ability and goal congruence.
2.4. Management Control through Employee Selection
Employee selection practices, along with other personnel controls, have long been
regarded as an important management control tool (Brody 2010; Cohen and Pfeffer 1986;
Merchant 1982; Merchant 1985; Ouchi 1979; Terpstra and Rozell 1993; Widener 2004; Wilk
and Capelli 2003). Since firms invest in employee selection in hope of identifying the ‘right’
individual for a given position, employee selection practices are by definition screening solutions
to the adverse selection problem. They are “designed to make it more likely that employees will
perform the desired tasks satisfactorily on their own because, for example, the employees are
experienced, honest, and hard working” (Merchant and Van der Stede 2007, p. 76, emphasis in
original) and are often used to gain information regarding qualitative aspects of the selection
process, such as how well the individual fits with the culture and values of the organization.
Effort and expense invested in employee selection can thus reduce the risk of moral hazard by
identifying individuals with goals that are innately aligned with those of the firm.
Despite the importance and potential benefits of employee selection, there is limited
empirical research on how firms address the hiring problem (Campbell 2012; Oyer and Schaefer
2011). Even in the management literature, employee selection research traditionally addresses

11

the problem of applicant ability rather than goal congruence (Sekiguchi and Huber 2011).
Certainly, finding people with the right technical and cognitive ability is important. In fact,
research shows that cognitive ability tests are a significant predictor of employee performance
(Murphy 2012; Ones, Dilchert, and Viswesvaran 2012). Knowing that employers will pay more
for an employee with certain skills and high ability, a high ability applicant has incentive to
credibly signal their type through education or certifications (Spence 1973). It is then relatively
costless to resolve any information asymmetry regarding applicant ability through a basic
background check to verify the employment, education, and certifications listed on an
application or resume: a simple web search identifies numerous companies that will conduct
basic background checks and resume verification for $7.50 - $15 per previous employer,
certification, or educational degree listed on a resume.3
Many of the more costly commonly used employee selection practices have little value in
assessing applicant technical ability, at least not incremental to the information obtained through
a basic background check (Huo et al. 2002; Gatewood and Feild 1998). Huo et al. (2002) infer
that the major role of employee selection is in alleviating information asymmetry regarding
social skills and goal congruence. There is a wide range of commonly used employee selection
practices, including in-person interviews, personality tests, and employee referral source.
First, in-person interviews are the most widely used employee selection practice, across
all industries and types of jobs. Studies report that between 70% and 95% of firms use in-person
interviews for employee selection, despite the fact that interviews are not seen as particularly
reliable or valid predictors of performance (Akyeampong 2006; Ekuma 2012; Gatewood and
Feild 1998; Graves and Karren 1996; Huo et al. 2002). Prior research extensively examines how

3

See http://www.studentclearinghouse.org/verifiers/fees.php/ for verification of education and certification, or
http://www.hireright.com/express/Background-Check-Services/ for employment verification.

12

to improve interview effectiveness in predicting performance. For example, research examines
the style of interview question (Gatewood and Feild 1998; Janz 1982; Maurer and Fay 1988;
Wiesner and Cronshaw 1988), the use of a predetermined set of questions, (Campion, Campion,
and Hudson 1994; Gatewood and Feild 1998; Graves and Karren 1996; Maurer and Fay 1988;
Searacy, Woods, Gatewood, and Lance 1993), and interviewer training (Ekuma 2012; Gatewood
and Feild 1998; Maurer and Fay 1988).
Second, nearly 20% of U.S. firms have applicants take a personality test prior to hire
(Shellenbarger 2013), and there is anecdotal evidence that firms experience significant firm level
benefits from using external vendors to conduct personality interviews. For example, Stryker
Corporation attributes its persistently strong organic growth rate largely to their employee
selection practices, relying heavily on results from personality interviews conducted through the
Gallup Organization (Suff 2012). In addition, Bristol-Meyers reports a 25% increase in retention
rate following implementation of personality tests in the employee selection process (Daniels
2001). These internally validated tests are designed to identify applicants with personality traits
similar to those of highly performing (and assumedly goal congruent) current employees.
A third widely used employee selection practice involves the referral source: giving
preference in the hiring process to applicants who are referred by current employees. Current
employees (who are assumed to be a better fit with the firm than the general population due to
their continued employment and willingness to recommend the firm) are more likely to refer
individuals who are, like themselves, goal congruent. Consistent with these findings, prior
literature shows that employees sourced through employee referrals have better post-hire career
outcomes (Autor 2001; Bidwell 2011; Campbell 2012; Montgomery 1991). Kim (2007) models
employee quality as including both technical ability and social skills, and shows that high quality

13

individuals will underinvest in education if firms highly value social skills which can be less
expensively signaled through employee referrals. Although prior literature has shown that
individuals sourced through employee referrals do tend to have better posthire career outcomes,
relying too heavily on internal referrals could lead to nepotism, groupthink and lower
performance in the long run (Campbell 2012).
In a recent related study, Campbell (2012) identifies employee selection as an
understudied element of management control systems. He examines the use of employee
selection to increase goal congruence in a financial services organization, and demonstrates an
association between individuals referred by existing employees and the effective use of decisionmaking authority. Campbell considers a single employee selection device (employee referrals)
to indicate alignment of individual goals with the organization. Other referral sources, such as
use of headhunters or direct recruitment of employees from other firms, are similarly utilized
with the expectation of raising the caliber of new employees.
Kennedy and Widener (2013) find a positive association between employees’ beliefs
regarding the long-term security and stability of their position with the firm and employees’ goal
congruence. To the extent that a high investment in the employee selection process signals the
firm’s commitment to the worker (high selection costs would make the firm less inclined to incur
further turnover), effort in the employee selection process should be positively associated with
goal congruence, and thus performance.
2.5. Management Control through Performance Pay
Once the right applicant has been identified from the applicant pool, the firm sets the
contract and makes an offer to the applicant, including the base pay and any performance pay in
the form of bonuses, piece rates, etc. (period 3 from Figure 1). As discussed in section 2.2.2.,

14

performance pay addresses the risk of moral hazard ex ante to the actual hire date by
encouraging high ability individuals to self-select into the firm (Banker et al. 2001; Dohmen and
Falk 2011; Lazear 1986; Rothschild and Stiglitz 1976; Wilson 1980). In addition, there is
extensive literature in both economics and accounting on the use of performance pay to mitigate
moral hazard in existing employees, see Bonner and Sprinkle (2002); Dutta (2007); Indjejikian
(1999); Ittner and Larcker (1998); Lambert (2001); Murphy (1999); and Prendergast (1999) for
literature reviews. Analytical studies show that the effectiveness of performance pay is reduced
when performance measures are noisy (Banker and Datar 1989; Feltham and Xie 1994),
misaligned with the firm’s goals (Feltham and Xie 1994) and insensitive to managerial effort
(Antle and Demski 1988; Banker and Datar 1989; Holmstrom 1979).
However, performance pay is a costly solution to information asymmetry in the hiring
process for two reasons. First, although performance pay provides an on-going solution to the
moral hazard problem, this effect is transitory: the well documented short term positive response
to performance pay reverses once the incentives are taken away (Banker et al. 2003; Benabou
and Tirole 2003; Gneezy et al. 2011). Therefore, the firm is likely committed to paying the
compensation risk premium associated with performance pay not just in the employee’s first year
(as agreed in the contract), but throughout the employee’s entire tenure.4 Performance pay acts
like an annuity of payments, likely increasing over time as the employee rises in the hierarchy
and lasting throughout the tenure of the employee at the firm.
Second, performance pay discourages self-selection into the firm based on personal
characteristics other than ability. Dohmen and Falk (2011) show that self-selection into jobs that

4

This makes it difficult to disentangle the ex post incentive effects from the ex ante self-selection effects of
performance pay. Prendergast (1999) note that “studies that allowed the effects of incentives to be separated from
worker selection issues suggest that perhaps one-third of the increase in performance arises from attracting better
workers” (p. 17).

15

offer performance pay is driven by risk attitude and gender, as well as ability. Performance pay
can thus result in an employee population that is predominantly male and more risk seeking than
the general population, which may or may not be beneficial to the firm.

2.6. Hypothesis Development
In this section, I present the mechanism by which more extensive use of employee
selection practices affects employee performance. First, I develop a baseline hypothesis linking
effort in employee selection and subsequent employee performance. I then pose as an open
research question whether employee selection and performance pay tend to be used together or
separately. Finally, I hypothesize a negative interaction between effort in employee selection and
performance pay in their effects on subsequent employee performance. The theoretical model is
illustrated in Figure 2.
2.6.1. The association between the firm’s employee selection effort and performance
Prior literature in the management field shows that employee selection practices are a
useful control mechanism for identifying goal congruence. Cable and Judge (1997) directly
examine interviewers’ ability to assess goal congruence by surveying a set of interviewers and
applicants through a university career office. Applicant data was collected regarding their
personal values, and data regarding organizational values was collected through the interviewers.
In addition, interviewers were asked the degree to which each applicant fit with the firm and its
current employees. Cable and Judge (1997) then measure the applicant’s inherent goal
congruence as the correlation between each applicant’s personal values and the firm values as
identified by the interviewer, and find inherent goal congruence to be significantly and positively
associated with the interviewer’s perceived goal congruence.

16

Prior research also shows that goal congruence is positively associated with employee
performance. Chatman (1991) examines the initial alignment between the personal values of
entry-level auditors and the values and culture of their hiring public accounting firms. This study
was unique in that it surveyed auditors as they entered the firm to assess the employee’s fit with
the firm values and culture, and then tracked the performance of these auditors over the
following year. She finds that auditors who are a good fit with the firm at entry adjust more
quickly to the new environment, are more satisfied, and stay with the firm longer. Stewardship
theory in the management literature “defines situations in which managers are not motivated by
individual goals, but rather are stewards whose motives are aligned with the objectives of their
principals” (Davis, Schoorman, and Donaldson 1997, p. 21). In this view, goal congruent
employees (stewards) are intrinsically motivated to work hard. Indeed, Wasserman (2006) finds
that founder CEOs, who create the firm and are thus more likely to identify with the firm and be
more goal aligned, accept lower compensation than non-founder CEOs. This is consistent with
goal congruent employees requiring less extrinsic rewards, because they are intrinsically
motivated to work.
In the accounting literature, empirical work on employee selection is limited. Campbell
(2012) identifies employee selection as an understudied element of management control systems.
He examines the use of employee selection in a financial services firm after a major shift in
competitive strategy to be more customer oriented. Success of this new strategy depended on the
decentralization of loan decisions to loan managers, rather than relying on strict action controls.
He demonstrates an association between employees referred by existing employees and use of
decision-making authority: new employees referred by current loan managers (especially those
hired after the strategy change) were more likely to override the traditionally recommended loan

17

decision, and that these loans were profitable for the firm in the long term.
One limitation of Campbell (2012) is its narrow focus on a single employee selection
device (employee referrals) to indicate alignment of employee goals with the firm. Although
employee referrals are not an uncommon employee selection device, they are by no means the
most widely used, and relying too heavily on employee referrals can lead to nepotism,
groupthink and lower performance in the long run (Autor 2001; Bidwell 2011; Campbell 2012;
Montgomery 1991).
Other research avoids this limitation by acknowledging that firms often use multiple
selection practices simultaneously, examining the firm’s effort in the employee selection process
(Cohen and Pfeffer 1986; Huo et al. 2002; Rosse, Miller, and Stecher 1994; Terpstra and Rozell
1993; Widener 2004). Terpstra and Rozell (1993) survey the heads of human resource
departments from 201 firms to collect data on five employee selection practices. They find a
significant positive association between the use of more of these practices and firm performance.
In addition, firm effort in the employee selection process can affect employee attitudes
and behavior. Kennedy and Widener (2013) measure goal congruence literally, comparing the
employee’s goals to those of the firm. They show that personnel controls (belief systems,
mentoring and peer pressure) are positively associated with goal congruence, and that this
association is fully mediated by the effect of personnel controls on employees’ beliefs regarding
the long-term security and stability of their position. To the extent that investment in employee
selection indicates the firm’s commitment to the employee (high selection costs would make the
firm less inclined to incur further turnover), effort in the employee selection process should be
positively associated with goal congruence, and thus employee performance.
Based on this research, I state my baseline hypothesis as follows:

18

H1: Higher effort in the employee selection process is positively associated with employee
performance.
2.6.2. The association between employee selection effort and the use of performance pay
In addition to employee selection practices, performance pay (in the form of bonuses,
piece rates, and other compensation directly linked to individual performance) has been widely
acknowledged in the accounting and economics literature to resolve information asymmetry
problems by encouraging high ability individuals to self-select into the firm (Banker et al. 2001;
Dohmen and Falk 2011; Lazear 1986; Rothschild and Stiglitz 1976; Wilson 1980). However,
empirical research has yet to examine if firms tend to use these two systems together, or if they
trade off between investment in employee selection and investment in performance pay, as they
are both costly solutions to the same control problem.
Performance pay is a costly solution to information asymmetry in the hiring process for
two reasons: the performance effect of performance pay reverses once the incentives are taken
away, and performance pay discourages self-selection into the firm based on personal
characteristics not necessarily related to job performance. Therefore, it would benefit the
employer to offer performance pay contracts only when they expect to have the greatest marginal
return to those contracts. Firms that are rigorous in their employee selection practices have
already identified goal congruent applicants: those who will work hard toward the firm’s goals
with or without extrinsic motivators such as performance pay. Therefore, the marginal return to
performance pay contracts is lower for firms that are rigorous in their employee selection
practices.
Recent analytical work in economics and accounting show that investment in employee
selection and performance pay are optimally used separately. Akerlof and Kranton (2005)
develop an analytical model in which the goal congruent employee identifies with the firm. As
19

such, she gains utility by working toward the interests of the firm, and loses utility when her
actions conflict with the firm goals. Therefore, firms who hire goal congruent employees (as
identified in the employee selection process) will have less need for performance pay structures:
their employees will exert high-level effort even without extrinsic rewards. Similarly, Heinle et
al. (2012) find that in a single task setting, the optimal incentive weights are decreasing in the
extent to which the employee identifies with the firm. These arguments would indicate that firms
should trade off between investment in employee selection and use of performance pay.
On the other hand, performance pay may be considered as an additional employee
selection tool, so firms that invest heavily in employee selection would also use performance
pay. The self-selection effect of performance pay systems has been widely acknowledged in the
accounting and economics literature (Banker et al. 2001; Dohmen and Falk 2011; Lazear 1986;
Rothschild and Stiglitz 1976; Wilson 1980). Indeed, in the same work where Akerlof and
Kranton (2005) propose an analytic model where firms trade off between employee selection and
performance pay, they suggest that the overwhelming success of a performance pay plan at
Lincoln Electric is due to the high goal congruence of its employees.
Therefore, I make a nondirectional hypothesis regarding the association between
employee selection and performance pay:
H2: Higher effort in the employee selection process is associated with use of performance pay
structures.
2.6.3. Interaction between employee selection practices and performance pay
Finally, if employee selection practices are used to identify ex ante goal congruence, they
also identify intrinsic motivation. When an employee’s own goals are aligned with those of the
firm, they will be intrinsically motivated to work hard toward the firm’s (a.k.a., their own) goals.
Motivation crowding theory examines the interaction between performance pay (extrinsic
20

rewards) and intrinsic motivation, providing evidence that performance pay has a negative effect
on the performance of intrinsically motivated employees, calling this a ‘crowding out’ effect
(Benabou and Tirole 2003; Bonner and Sprinkle 2002; Ellingsen and Johannesson 2008; Gneezy
et al. 2011). The crowding out effect can occur through two mechanisms. Performance pay can
be viewed as 1) a negative signal regarding the desirability of the job task itself, or 2) as a signal
regarding the firm’s lack of trust or confidence in the employee.
First, performance pay can affect the employee’s intrinsic motivation toward their
specific job if the provision of performance pay is viewed as a signal that the job is undesirable.
Benabou and Tirole (2003) develop an analytical model of intrinsic and extrinsic rewards. They
find that when the firm has private information about the job, performance pay will signal to the
employee that the job is particularly onerous, and intrinsic motivation to exert effort will be lost.
Second, performance pay has a crowding out effect on intrinsic motivation in that it
affects the employee’s perception of the firm itself (Cardinaels and Yin 2013; Christ, Sedatole,
and Towry 2012; Ellingsen and Johanesson 2008, Gneezy et al. 2011). A performance pay
contract signals a lack of trust by the firm, where the firm assumes an employee will not be
productive without extrinsic rewards.5 While purely self-interested employees are less likely to
react negatively to this signal (since the lack of trust was, in fact, warranted), performance pay
will hinder motivation of employees that value trust and other social influences. Therefore, to the
extent that effort in the employee selection process can identify goal congruent and intrinsically
motivated employees, the association between effort in employee selection and performance
should be lower when performance pay contracts are used. That is, while employee selection
practices are positively associated with employee performance because they allow the firm to

5

In contrast, Christ et al. 2012 find that relative to penalty contracts, performance pay contracts signal that the firm
has more trust in the employee’s integrity and competence.

21

identify and hire intrinsically motivated employees, the association between employee selection
and performance should then be weaker in the presence of performance pay contracts.
I therefore examine if use of performance pay interacts with employee selection effort in
their effect on future performance, such that the association between employee selection
practices and employee performance is weaker in the presence of performance pay.
H3: High effort in the employee selection process and performance pay structures interact
negatively in predicting employee performance, such that the positive effect of effort on
performance in the employee selection process is lessened when performance pay structures are
used.
The three hypotheses are presented in graphical form in Figure 2.

22

CHAPTER 3: WORKPLACE AND EMPLOYEE SURVEY
I conduct my study using data from the Workplace and Employee Survey (WES), a large
scale survey of matched employer-employee information developed and administered by a
Canadian federal government agency, Statistics Canada. A multi-level survey structure means
that employee-level responses on topics including employee selection practices, job
characteristics, pay and benefits, education, promotion, and termination are linked to firm-level
data regarding business strategy, competition, firm performance, delegation of decision making
authority, and use of technology and innovation. This uniquely allows me to link use of control
systems (e.g., employee selection practices and compensation structure) to both employee- and
firm-level outcomes, as well as to control for firm-level environmental factors that would affect
the control environment. It also alleviates some of the standard concerns in survey-based
research regarding common method bias. Although employee level responses are self-reported
by a single individual, these data are supplemented with workplace level data, as reported by a
different individual.
In addition, the WES has a longitudinal component: the sample of firms is followed
1999-2006 (with new locations added periodically to maintain a representative cross section as
firms exit the market). Employees sampled from within these firms were then followed for two
years, with a fresh sample of employees drawn every other year. That is, a sample of employees
from a given firm was surveyed in 1999, and that same sample was resurveyed in 2000. A new
set of employees from that firm were then selected to be surveyed in 2001 and 2002. Workers are
only followed for two years due to the cost and difficulty of integrating new companies into the
workplace sample as workers change companies over time. The survey was phased out in 2006
with only firm level responses collected in the final year. Thus, the sample of workers drawn for

23

the 2005 survey was not resurveyed in 2006.
Each year, the WES collected data from approximately 6,200 workplaces and 20,000
employees at those workplaces. The survey sample was selected in two stages. In the first stage,
the WES randomly drew a sample of workplaces from the Business Register, a list of all
businesses in Canada. Workplaces in Yukon, Nunavut and the Northwest Territories were
excluded from the target population, as were workplaces operating in crop or animal production,
private households, religious organizations, and public administration. The sample was then
stratified by industry, region, and size (based on estimated employment), and each selected
business unit was assigned a sampling weight proportional to the population of business units in
the stratum. By proportionally weighting the data in all estimation procedures, inferences can be
made about the population frame of firms from which the sample was drawn (Sarndal,
Swensson, and Wretman 1992). The selected firms then provided Statistics Canada with a full
list of employees. In the second stage, a sample of employees was selected from each of these
lists. Employees were selected using a probability mechanism, with a maximum of 24 employees
per business unit selected. In extremely small business units (only 3-4 employees), all
employees were selected. Statistics Canada interviewers then collected the WES data through
computer-assisted telephone interviews. As with the workplace component, each employee-level
response was assigned a sampling weight equal to the inverse of the original probability of
selection.
Annual response rates for the WES were very high (ranging from 82% to 95% at the firm
level, and from 83% to 91% at the employee level), since compliance with the WES was legally
required by Canada’s Statistics Act of 1985.6 High response rates may also be due to the
commitment to respondent confidentiality formalized in the Statistics Act, which only allows
6

http://laws-lois.justice.gc.ca/eng/acts/S-19/FullText.html

24

access to survey data to “deemed employees” and expressly prohibits the disclosure of any
survey information that could be identifiable at the individual, business or organization level.
It is worthwhile to note some differences between the U.S. labor market more commonly
used in empirical research and the Canadian labor market used in this study. First, while most
countries (including Canada) require firms to show cause prior to dismissing an employee,
employment relations are generally presumed to be “at-will” in the U.S. That is, U.S. firms have
the right to terminate an employee at any time, without notice or severance pay, for any reason or
for no reason at all.7 In contrast, Canadian firms are subject to a default assumption of implied
contract in every hiring agreement. Canadian firms can only terminate employment without
notice or severance pay for ‘just cause’, requiring documentation of multiple warnings for poor
performance or evidence of culpable behavior such as theft, fraud, or sexual harassment.
Therefore, Canadian firms face much higher penalties to poor hiring decisions, as it is much
more costly to terminate poorly performing employees.
Second, differences in labor law have led to a much more conducive environment to form
unions in Canada relative to the U.S. This is reflected in Canada’s much higher unionization rate
(28.7% in 2011) relative to the U.S. (11.8%).8 Higher unionization rates also imply a greater cost
of poor hiring decisions in Canada, as union law imposes even greater restrictions on terminating
employees. For both of these reasons, it is vital for Canadian firms to mitigate the potential for
moral hazard in the hiring decision, rather than relying on post-hire controls.

7

U.S. employment relationships default to an “at-will” status in the absence of a written contract. Montana is the
only U.S. state that has enacted legal protection from being terminated without cause after completing an initial
probationary period. In addition, 43 states have enacted protection for employees complying with (or refusing to
violate) public policy, 37 states have enacted protection for employees subject to an implied contract, and 11 states
have enacted a “covenant of good faith” prohibiting terminations motivated by malice or unfairness.
8
Center for Economic and Policy Research, 2012. http://www.cepr.net/documents/publications/canada-2012-08.pdf

25

CHAPTER 4: SAMPLE SELECTION, VARIABLE MEASUREMENT AND
EMPIRICAL SPECIFICATION
4.1. Sample Selection
In this study, I am interested in differences in future outcomes based on the control
systems used as the employee enters the firm (e.g., employee selection practices and the initial
performance pay). Therefore, I restrict the available sample from the WES to employees who
were surveyed within one year of their initial hire date. This maximizes the probability that the
respondent’s current pay structure (as of the survey response date) is the same as when they were
originally hired. It also minimizes recall bias in the respondent’s recollections of steps taken in
the employee selection process. In addition, as I am interested in the associations between these
control systems and future outcomes, I restrict my sample to employees for whom longitudinal
data is available. For instance, I eliminated from the 1999 WES sample individuals who were not
resurveyed in 2000. In addition, I restrict my sample to only full-time, regular employees of
firms that have been in existence for at least one fiscal year and have at least 25 employees. As a
result, my final working sample is composed of 906 individual observations from 746 firms.
Table 1 shows the details of the sample selection process for this study.
4.2. Variable measurement
Next, I discuss the measurement of each of the constructs of interest and controls,
beginning with the dependent variable: employee performance.
4.2.1. Employee performance
Measuring employee level performance in a multi-firm structure is problematic, since
performance as measured within each firm is not comparable across firms. Firms often use
multiple, different performance measures across different employees within the same firm, much
less maintain consistency across firms. Also, firms commonly measure performance according to
26

different scales, and may even assign different meanings to levels on the same scale. Instead, I
measure employee performance using a proxy more likely to be consistent indicators of
performance across firms: the percent increase in base salary from the employee’s first and
second year responses to the WES.
Visual inspection of data suggests that there are some coding errors in the survey. For
instance, extremely large positive or negative changes in annualized salary are sometimes seen,
especially when the respondent also indicates a change in their pay frequency (i.e., hourly,
weekly, semi-monthly, etc.). Although the potential for these errors exists, it is unlikely that
coding errors are associated with the other independent variables in the model, and so would
simply add noise and bias against finding results. Nevertheless, I reduce the potential impact of
these outliers by winsorizing at the 5% level. In addition, it is possible that variation in the
independent variables (employee selection effort and use of performance pay) exists primarily at
the firm level, while variation in salary increases is at the individual level. To reduce concerns
regarding a mismatch in the unit of analysis, I center this variable by industry mean salary
increase (%Increaset+1).9
4.2.2. Effort in the employee selection process
I follow Ryan, McFarland, Baron, and Page (1999) and Wilk and Cappelli (2003) by
measuring the extent of practices used in the employee selection process (Selectiont). The WES
includes items regarding a variety of selection practices, including both specific tests and tools
used in the selection process and measures of how the employee learned about the job opening.
Since my hypotheses rely on selection practices intended to identify goal congruence, I limit this
measure to the WES items that could potentially indicate employee type. For example, taking a
9

It would be ideal to mean center the dependent variable at the firm level rather than the industry level. Data
availability precludes this specification, however, as the vast majority (83%) of observations are the only respondent
from that firm-year.

27

pre-hire medical exam is not included in my measure of employee selection effort since it is
unlikely to be correlated with goal congruence. I include the following employee selection
practices in my employee selection effort construct: in-person interviews, personality tests,
referral by family or friend, direct recruitment by the firm, and use of headhunters. Selectiont is
then coded as the additive compilation of these five employee selection practices.
4.2.3. Performance pay
WES respondents are asked each year if their pay or benefits are directly affected by the
results of their formal performance evaluation. I use this item directly from the WES, with PFPt
coded as 1 if the respondent indicates use of performance based pay in their first survey
response. Pay structure is not necessarily fixed over the employee’s tenure at the firm, so there is
the chance that performance pay reported in the first survey is was not included in the initial
employment contract. I mitigate this risk to the best of my ability by limiting the sample
selection to only employees first surveyed within their first year at the firm.
4.2.4. Controls
Although general cognitive ability is not theoretically associated with goal congruence,
there is considerable research linking ability to overall performance. High ability employees
may also self-select into firms that use performance pay, making it important to disentangle
employee selection tools that identify general ability from tools that identify goal congruence (as
developed in my hypotheses). I control for cognitive ability in two ways. First, I control for the
use of employee selection tools to identify skills and ability (Abil_testt). The WES asks
respondents if they were required to take tests on general knowledge or literacy, tests for specific
skills (e.g., typing or manual dexterity tests), or tests on job-related knowledge. Abil_testt is then
coded as 1 if any of these ability tools are used, and 0 otherwise.

28

Second, an employee’s level of education has long been acknowledged as a signal of
ability (Spence 1974). Therefore, I control for the extent to which the employee is overeducated
for the position. The WES collects data on the minimum education requirements for the
respondent’s position as well as the respondent’s current education level. I scale these responses
from 0 (no requirements/education) to 9 (doctoral degree required/attained). I then code
Educationt as the difference between respondent’s education level and the minimum education
required for the position. Thus, Educationt is positive when the employee is more highly
educated than technically necessary for the position.
Next, it is also important to control for the employee’s job level when hired, since
employees hired in at lower levels of the firm may be expected to receive larger percentage
salary increases and promotions in their early years with the firm. I therefore control for the
employee’s years of full-time work experience (Yrs_expt). In addition, the WES provides
information on the employee’s job category (i.e., management, professional, technical trades,
marketing or sales, clerical or administration, and production employees), although the
respondents’ specific job titles are not available. Within each job category, employees hired at a
low initial salary are likely in entry level positions with the firm. Therefore, I also calculate and
control for the employee’s first year salary percentile within their job category (Sal%ilet).
Performance pay is also more effective in jobs with easily measured outputs. Work in
certain job categories, like sales and production, are likely to have more easily measured outputs,
so I include a set of indicators for the different job categories, with the production job category
taken as the base case (JobCategoryt). Similarly, firm strategy may be associated with greater
reliance on specific control system, so I include a firm-level control for the overall strategy of the
firm. A factor analysis of fifteen business strategy questions from the firm level survey identified

29

three types of firm strategy. Based on these results, I categorize each firm as a differentiator
(Differentiatort), cost leader (Costt), or quality leader (Qualityt). In addition, I control for
Firm_sizet (proxied by the number of employees at the location) since larger firms may have the
additional resources necessary to manage use of multiple employee selection practices.
Finally, I control for the respondent’s age (Aget), gender (Gendert), and ethnicity
(Ethnicityt), as well as the survey year.10 These attributes are not easily manipulated by the
employee, but they still may be taken as a signal of employee quality if they affect the
employer’s perception of the employee quality probability distribution (Spence 1973). That is, an
employer may believe that an employee of a given age is more or less likely to be productive.
The employer will then structure the compensation contract according to their beliefs.
4.3. Empirical Specification
The WES is based on a stratified sample design, with employers selected randomly from
strata identified by region, industry and firm size, and employees selected randomly from each
sampled firm. Therefore, the sample is not self representing, and survey weights should be used
so that estimates for both descriptive statistics and regression coefficients will generalize to the
target population (Thompson 2012). Since my unit of analysis is the employee, I use employee
level weights in my empirical analyses. This weight (provided by Statistics Canada) represents
the inverse of the probability of the unit being selected given the sampling design, adjusted for
workplace and employee nonresponse.
In addition, the WES is collected based on samples coming from a finite population (i.e.,
the finite population of Canadian employees, excluding (a) locations in the Yukon, Nunavut and
Northwest Territories and (b) farms, hunting, private households, religious organizations, and

10

Confidentiality requirements preclude the reporting of descriptive statistics and coefficients for the job category,
firm nonprofit status, and survey year indicators.

30

public administration). In these types of samples, one should account for the stratified sampling
design to avoid underestimation of variance (Thompson 2012). Bootstrapping techniques can be
used to compute survey design consistent variances. Bootstrapping is based on resampling, with
a simple random sample of as many units as in the working sample is selected (with
replacement) from the working sample, and coefficients are estimated for this new sample. This
procedure is replicated 100 times, and the variance of the coefficient is then computed as the
variability among these resampled coefficients.
I use STATA 13 software to conduct a structural equations-based path analysis of the
hypothesized theoretical model (as depicted in Figure 2). Structural equations modeling (SEM)
is an appropriate tool for this model, because it allows one or more variables act as both
independent and dependent variables. In my hypothesized structural model, performance pay has
this characteristic: it acts as an independent variable in predicting employee performance and as
a dependent variable (with employee selection effort as an independent variable).11 STATA 13
supports the use of SEM (as well as more conventional regression models like OLS and logistic
regression) when using complex survey data sets like the WES. This allows me to use employee
level weights to obtain consistent coefficient estimates, and to use bootstrapping techniques to
obtain more accurate variance estimates.

11

SEM offers the additional complexity of creating a measurement model, in which relations between unobservable,
latent variables and multiple observable indicators are measured. In this study, the variables of interest each have a
single indicator. Therefore, I treat the items as manifest variables and estimate only the structural model component
of SEM.

31

CHAPTER 5: EMPIRICAL RESULTS
From the employee level data collected through the WES, I conduct my analysis on a
working sample of 906 full-time, regular employees hired within one year of the initial survey
date, from 746 business units that have been in existence for at least one fiscal year and have at
least 25 employees. Descriptive statistics and correlations are presented in Tables 3 and 4,
respectively.12 Employees reported an average salary increase of 4.2%, and 19.4% of employees
received a performance-based promotion within their first two years of hire. Compensation was
directly linked to evaluated performance for slightly less than half (46%) of employees in their
first year at the firm, and firms used on average 1.41 different employee selection practices to
reveal goal congruence in the hiring process.
I present the SEM results using salary increases to proxy for employee performance in
Table 5. The goodness of fit is acceptable, with a standardized root mean squared residual of
0.03, which is smaller than the recommended 0.10 (Kline 2005). Using salary increases to proxy
for employee performance, I hypothesize and find a positive association between use of
employee selection effort and performance (Table 5, Model (1)), in support of H1.
In my second hypothesis, I predict a nondirectional association between employee
selection effort and use of performance pay. Results are shown in Model (2) of Table 5. I find a
positive association between employee selection effort and use of performance pay. This
indicates that firms that employ more extensive use of employee selection practices also tend to
use performance pay in the initial contract, using the two control systems together to resolve
information asymmetry in the hiring process. This could be consistent with firms considering
performance pay as an additional employee selection tool, using employee selection and

12

Minimum and maximum values can result in a respondent’s identity being revealed, so Statistics Canada’s
confidentiality requirements prohibit reporting these values.

32

performance pay to resolve different types of information asymmetry, so that firms benefit from
using both control systems where the potential for moral hazard in employees is high. In
addition, this is consistent with firms implementing these management control practices as a
package rather than as a system, ignoring the lack of complementarity in implementing the two
systems together (Grabner and Moers 2013).
Hypothesis 3 predicts a negative interaction between employee selection effort and
performance pay on employee performance. The SEM results for H3 using %Increaset+1 to proxy
for employee performance are presented in Model (1) of Table 5. In support of H3, the model
shows a significant and negative interaction between employee selection effort and performance
pay (p<0.01). Specifically, at low (25th percentile) level usage of employee selection devices, use
of a performance pay contract is associated with a 4.68% salary increase while at high (75th
percentile) level usage of employee selection devices use of a performance pay contract is
associated with only a 1.25% salary increase (Table 5, Model (1)). This is consistent with the
extrinsic motivation provided through performance pay crowding out the intrinsic motivational
preferences of an employee whose personal goals are already aligned with those of the firm at
the hiring decision.
5.1. Sensitivity analyses and alternative explanations of results
I present sensitivity tests of my analysis in Tables 6 and 7. I begin by checking if the
results reported above are robust to an alternative specification for employee performance: an
indicator for whether the respondent has received a performance-based promotion (Promot+1).
The WES asks employee level respondents each year if they received a promotion in the year
leading up to the survey date and if so, if past performance is a primary reason for this
promotion. Promot+1 is then coded as 1 when the employee has received a performance based

33

promotion within their first two years with the firm, and 0 otherwise.
I present the SEM results using performance-based promotion to proxy for employee
performance in Table 6. Contrary to H1 and in contrast to the primary results in Table 5, the
association between employee selection effort and employee performance is negative when
performance is measured with the indicator for a performance based promotion (Table 6, Model
(1)). That is, the more tools a firm used in the employee selection process, the less likely that
employee is to receive a performance-based promotion within their first two years with the firm.
One possible explanation of this somewhat counterintuitive result relates to the firm’s
desire to hide employee performance from the external labor market. Waldman (1984) notes that
information asymmetry regarding an employee’s type is revealed to their current employer, as
they observe the employee’s effort (or outcomes therefrom). Although the employee’s type is
still hidden from other potential employers, job title is known to all parties. Internal promotion
can then be taken as a signal of employee quality, making the employee more attractive to other
employers and more difficult for the current employer to retain. Firms may then prefer to reward
high quality employees with larger salary increases, which are less visible to the external labor
market, and limit the use of internal promotion.
Results for H2 and H3 are consistent with the primary analysis. I find a positive
association between employee selection effort and use of performance pay (Table 6, Model (2))
and a negative interaction between employee selection effort and use of performance pay in their
effects on performance-based promotions (Table 6, Model (1)). Specifically, at mean level effort
in the employee selection process, use of a performance pay contract is associated with 3.4%
lower likelihood of receiving a performance based promotion.
As another (third) specification of the dependent variable, the management literature has

34

identified a strong association between job satisfaction and employee performance, so I examine
the effects of employee selection effort and use of performance pay on employee job satisfaction.
Results for all three hypotheses (untabulated) are consistent with the primary analysis.
Next, I verify that the results in the primary analysis are robust to the use of more
simplistic empirical specifications. Since my primary proxy for employee performance
(increases in salary) is continuous, I model the associations between salary increases, effort in
employee selection, and incentive structure using linear regression and present the results in
Table 7. Unfortunately, this simplistic model cannot accommodate the effects of a mediator, as
performance pay is hypothesized in H2 and in the graphical presentation (Figure 2). Therefore, I
limit my discussion of these analyses to H1 and H3.
Model (1) of Table 7 presents the direct effects of performance pay and employee
selection effort on employee performance, without considering an interaction (as hypothesized in
H3). Results are directionally consistent with the SEM analysis in regards to H1. Employee
selection effort is positively associated with salary increases as a proxy for employee
performance (p<0.01). Use of one additional employee selection tool (in-person interviews,
aptitude or personality interviews, referral by family or friend, direct recruitment by the firm, or
use of headhunters) is associated with a 2.0% larger salary increase after the first year of
employment. This provides further support of H1.
Having demonstrated that the baseline model delivers results consistent with the path
analysis model, I include the interactive effect of employee selection and incentive structure in
Model (2) of Table 7. Inclusion of the interaction does not qualitatively change the inferences
from the baseline model. In support of H3, the linear regression model shows a significant and
negative interaction between employee selection and performance pay (p<0.01, both

35

specifications). Specifically, at low level usage (25th percentile) of employee selection devices,
use of a performance pay contract is associated with 5.41% salary increase, while at high level
usage (75th percentile) of employee selection devices use of a performance pay contract is
associated with a 2.65% increase in salary (Table 7, Model (2)). This is consistent with the
extrinsic motivation provided through performance pay crowding out the intrinsic motivational
preferences of an employee whose goals are already aligned with those of the firm at the hiring
decision.
5.2. Alternative explanations
I consider firm strategy as an alternative explanation for these results. Firms that sustain a
competitive advantage by producing high quality, differentiated goods or services tend to operate
in more fluid environments, where employees have more decision making authority and less
measurable performance. In these types of firms, employee selection and/or performance pay
may be more valuable to the firm. Therefore, I include an interaction between firm strategy and
performance pay in Model (3) of Tables 7, and an interaction between firm strategy and
employee selection effort in Model (4) of Tables 7. Results for H1 and H3 are not qualitatively
different with the inclusion of these interactions.

36

CHAPTER 6: CONCLUSION
In this study, I examine the use and effectiveness of management control systems
(employee selection and performance pay) in addressing the information asymmetry problem
regarding an individual’s goal congruence ex ante to the hire date. I find results generally
consistent with my predictions. Using an employee’s salary increase in the first year of
employment to proxy for performance, I find that more extensive use of employee selection
practices intended to identify goal congruence is associated with salary increases in the second
year of employment. Further, I find the association between employee selection practices and
future employee performance to be weaker in firms that use performance pay in the initial
employment contract. This is consistent with motivation crowding theory, which predicts that
explicit incentives crowd out the intrinsic motivational effects of ex ante goal congruence. I also
find a positive association between more extensive use of employee selection practices and
performance pay contracts. This curious result suggests that firms use employee selection
practices and performance pay together, despite the fact that my results also show that their
combined use reduces the performance effect of each separate control.
This study contributes to the broader literature on employee selection and performance
pay in several ways. First, performance pay has been widely examined and acknowledged as a
screening device to reduce information asymmetry in the hiring process. I contribute to that
literature by providing empirical evidence regarding a drawback to the screening function of
performance pay: an adverse motivational effect on internally motivated employees. I provide
archival evidence supporting motivation crowding theory, which until now (to the best of my
knowledge) has only been tested experimentally.
Second, although there is extensive research on employee selection and on performance
pay, this is the first identified study on how the two systems work together. I find that employee
37

selection and performance pay are interdependent (the partial effect of each is reduced in the
presence of the other). However, firms do not seem to take these interdependencies into account
in their management control design choices: employee selection practices are positively
associated with use of performance pay. My findings are consistent with firms adopting these
control systems as a ‘package’, implementing them separately without regard for the costly
interdependence between employee selection and performance pay, rather than as a ‘system’,
taking these interdependencies into account in their design choices (Grabner and Moers 2013).
Third, the extant literature on mitigating information asymmetry in the hiring process
tends to focus on either a screening device, e.g., performance pay (Banker et al. 2001; Banker et
al. 2003; Bonner and Sprinkle 2002; Lazear 1986; Lazear and Shaw 2007; Milgrom and Roberts
1992; Prendergast 1999), interviews (Akyeampong 2006; Ekuma 2012; Gatewood and Feild
1998; Graves and Karren 1996; Huo et al. 2002), or referral source (Campbell 2012); or on a
signaling device, e.g. education. I extend this literature by acknowledging that firms often use
signaling devices and screening devices concurrently in practice. I also identify additional
screening practices that can uncover goal congruence (personality tests, direct recruitment by the
firm, and use of headhunters) and examine the combined use of these systems.
This study also contributes specifically to the literature on management control. First,
management control research generally assumes that employees are not goal congruent when
they join the firm, and concentrates on ways to mitigate moral hazard ex post of the hiring
decision. However, individuals vary in their level of goal congruence, even before they are hired
(Akerlof and Kranton 2005; Heinle et al. 2012). I provide archival evidence that employee
selection practices reducing information asymmetry regarding goal congruence in the hiring
process, and thus reduce the risk of moral hazard. Second, I extend the literature on the adoption

38

and use of control systems. I examine the potential tradeoffs between investing in employee
selection practices and investing in performance pay, and I examine an interaction between onetime ex ante controls (employee selection practices) and a control that likely persists throughout
the tenure of the employee (performance pay). Finally, empirical evidence supporting the use of
employee selection practices as a control mechanism to align incentives is relatively sparse and
focuses on practices in a single firm (Campbell 2012). In contrast, I use a large database across
industries to address this question, which allows me to examine the robustness of these results
across different firm-level environmental conditions.

39

CHAPTER 7: LIMITATIONS
This study is subject to certain limitations. First, the WES only captures pay structure as
of the survey date (rather than the hire date). This may be a problem, as the intent of this study is
to examine control systems used to mitigate information asymmetry in the hiring process. To the
extent that the pay structure as of the survey date differs from the pay structure in the initial
contract, I may instead be measuring the well documented post-hire use of performance pay to
reduce moral hazard. In this case, the observed negative interaction is still interesting: an
individual’s effort is bounded, so if employee selection practices result in hiring only goal
congruent individuals who are predisposed to work hard (exert high effort), adding performance
pay post-hire should have no effect on the individual’s performance. In any case, I mitigate this
risk to the best of my ability by restricting the working sample to only respondents surveyed
within one year of the initial hire.
Second, the data items in the WES do not allow me to directly measure goal congruence
prior to (or following) hire, so I depend on prior literature to infer that the intent of employee
selection practices is to determine an individual’s innate goal congruence (and so, the risk of
post-hire moral hazard) in the hiring process. In future versions of this study, I will validate the
intent of employee selection practices to reveal applicant goal congruence by determining the
correlations between employee selection effort and the likelihood of a respondent remaining in
the survey sample through the second year. If employee selection effort is correlated with goal
congruence, I would expect to see a positive association between selection effort and survival in
the sample.
Third, direct performance measures are not available in the WES, so there may be
construct validity concerns with the identified proxies for employee performance. I measure
employee performance using three proxies likely to be consistent indicators of performance
40

across firms: the percent increase in base salary from the employee’s first and second year
response to the WES, an indicator for whether the respondent has received a performance-based
promotion, and the respondent’s reported level of satisfaction with their job.
Finally, it should be noted that employee selection practices and performance pay are
often implemented prior to an employee’s hire date to mitigate both adverse selection and the
risk of moral hazard by identifying and promoting contract acceptance of high quality
individuals. Data limitations make it impossible for me to draw any conclusions regarding which
control problem (adverse selection or moral hazard) these control systems are intended to
address. Indeed, Hagerty and Siegel (1988) show analytically that a contract written to address
the moral hazard problem is indistinguishable from one intended to address adverse selection,
suggesting that this distinction could only be identified experimentally. I leave this distinction to
future research.
There is substantial opportunity for further research in the use of employee selection as a
control system. It would be interesting to apply contingency theory to this setting, as the effects
of performance pay and employee selection may be sensitive to firm environmental issues. In
this study, I control for the firm’s size, age, industry, and strategy, but there are other factors,
such as the firm’s competitive climate, that have not yet been examined. In particular, the
relation between employee selection effort and firm-level delegation of decision making
authority would be an interesting topic.
Also, goal congruence is not a stable trait in an individual, leaving it an empirical
question as to whether further socialization processes post-hire affect the significance or
direction of my results. The results of this study are consistent with a negative effect of
performance pay on employee performance through the removal of an employee’s intrinsic

41

motivation. It is possible that socialization processes post-hire can reaffirm the employee’s goal
congruence and intrinsic motivation, lessening the unintended consequence of performance pay.

42

APPENDICES

43

APPENDIX A

Figures

44

Figure 1: Timeline of the hiring process, with the associated control problems and control systems

CONTROL
PROBLEM:

Adverse
selection

1

2
Applicant
pool set

CONTROL Employee selection
SYSTEM: & performance pay
Intended
result:

Unintended
consequences:

Ex post
moral hazard

Ex ante moral hazard
(focus of the current study)

3
Contract
offered

Employee selection

Identify (or promote Identify high goal
self-selection of)
congruent and high
goal congruent, high ability applicants
ability applicants

4
Contract
accepted
(hire date)

Performance pay

Performance pay

Discourage contract
acceptance by low
ability applicants

Motivate effort,
induce goal
congruence

Discourages contract
Low effort from
acceptance by high high goal congruent
goal congruent
applicants
applicants

45

Figure 2: Theoretical model for the effects of employee selection and performance
pay on employee performance

Employee
selection effort

H2
(?)

H1 (+)

Employee
Performance

H3
(-)

Performance
pay

46

APPENDIX B

Tables

47

Table 1: Survey sample selection
1999

2001

2003

Total

23,210

19,450

20,382

63,042

6,285

6,563

6,452

19,300

889

644

653

2,186

15,020

11,575

12,625

39,220

388

233

256

877

less no second year response

210

187

156

553

Final sample size

418

248

240

906

Employee-level survey responses
less part-time employees
less temporary/contract employees
less employees with more than one year
tenure
less business units with <25 employees

48

Table 2: Variable definitions
Variable

%Increaset+1
Promot+1

Interviewt
Personalityt
Referralt
Directt
Headt
Selectiont
PFPt
Literacyt
Skillst
JobKnowt

Abil_testt
Educationt

Description
Percent increase in respondent base salary between the first and second
survey.
WES item: Which of the following factors were important in earning
that promotion: Past performance evaluations (only asked if respondent
indicates Yes to the item, Have you ever been promoted while working
for this employer?)
WES item: When you were first hired, were you required to take a
personal interview?
WES item: When you were first hired, were you required to take
aptitude or other personality testing
WES item: When you were first hired how did you learn about the job
opening: Family or friend?
WES item: When you were first hired how did you learn about the job
opening: Directly recruited by employer?
WES item: When you were first hired how did you learn about the job
opening: Recruitment agency (headhunter)?
Sum of Interviewt, Aptitudet, Referralt, Directt, and Headt
WES item: Do the results of your job evaluation directly affect your
level of pay or benefits?
WES item: When you were first hired, were you required to take
aptitude or other personality testing
WES item: When you were first hired, were you required to take
aptitude or other personality testing
WES item: When you were first hired, were you required to take
aptitude or other personality testing
Coded as 1 if Literacyt, Skillst, or JobKnowt = 1, 0 otherwise
Coded as the difference between the respondent’s education level and
the minimum education level required for the position, scaled from 0

Scale

0-1

0-1
0-1
0-1
0-1
0-1
0-5
0-1
0-1
0-1
0-1
0-1
-9-9

(no education /requirements) to 9 (doctoral degree
attained/required).
Yrs_expt
Sal%ilet

WES item: Considering all jobs you have held, how many years of fulltime working experience do you have?
Salary percentile of the respondent within their job category (i.e.,

management, professional, technical trades, marketing or sales,
clerical or administration, or production )
Differentiatort
Costt
Firm_sizet
Aget
Gendert
Ethnicityt
JobCategoryt
Nonprofitt
SurveyYr

Factor score variable based on 15 firm level strategy survey items
Factor score variable based on 15 firm level strategy survey items
WES firm level: number of employees at the location.
Age of respondent as of survey date
Gender of survey respondent, 1 = Male
Coded as 0 for Caucasian, 1 otherwise
Categorical variables for jobs classified as Manager, Professional,
Technical/trades, or Clerical, with Production as the base case.
Coded as 1 for nonprofit firms, 0 otherwise
Categorical variables for the primary survey year

49

0-1
0-1
0-1
0-1
0-1

Table 3: Descriptive statistics
Variable
%Increaset+1
Promot+1

Mean

sd

p25

p50

p75

4.2%
19.4%

0.25
0.4

-3.8%
0

3.1%
0

10.8%
0

Selectiont
PFPt

1.41
0.46

0.73
0.5

1
0

1
0

2
1

Abil_testt
Educationt

0.14
-0.27

0.35
2.5

0
-1

0
0

0
1

Yrs_expt
Sal%ilet

12.2
46.8

9.24
27.4

4
24.7

9
46.6

18
70.4

Job_satt
Aget

4.28
34

0.82
10.25

4
26

4
32

5
40

0.52
0.5
0
1
0.17
0.38
0
0
Firm_sizet
306
509
42
100
Differentiatort
0.32
0.47
0
0
Costt
0.11
0.31
0
0
Uniont
0.14
0.35
0
0
Mfgt
0.51
0.50
0
1
Servicet
0.42
0.49
0
0
Survey_1999
0.36
0.48
0
0
Survey_2001
0.31
0.46
0
0
Survey_2003
0.33
0.47
0
0
N = 906 for all variables. Variables are as described in Table 2.

1
0
257
1
0
0
1
1
1
1
1

Gendert
Ethnicityt

50

Table 4: Pearson correlation coefficients
(1)
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
(11)
(12)
(13)
(14)
(15)
(16)
(17)
(18)

%Increaset+1
Promot+1
Selectiont
PFPt
Abil_testt
Educationt
Yrs_expt
Sal%ilet
Job_satt
Aget
Gendert
Ethnicityt

0.11*
0.06
0.10*
0.07*
0.02
0.09*
0.04
0.12*
0.05
0.09*
0.01
Firm_sizet
0.04
Differentiatort 0.00
Costt
0.03
Uniont
-0.05
Mfgt
0.05
Servicet
-0.10*

(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

(10)

(11)

(12)

(13)

-0.04
0.05
-0.04
0.14*
-0.04
-0.01
0.02
-0.04
-0.01
-0.04
-0.04
0.04
0.08*
-0.07*
0.06
-0.07*

0.13*
0.10*
0.11*
-0.05
0.09*
0.08*
-0.07*
-0.04
0.02
0.19*
0.10*
0.01
0.06
-0.05
0.00

0.04
0.16*
0.09*
0.14*
0.11*
0.07*
-0.13*
0.06
-0.01
0.17*
0.02
-0.18*
-0.02
0.02

0.04
-0.01
0.05
-0.04
0.00
0.02
0.03
0.00
-0.00
-0.09*
0.07*
-0.05
0.03

-0.01
0.11*
-0.08*
-0.06
-0.08*
0.01
0.00
-0.03
-0.07*
-0.01
-0.12*
0.08*

0.23*
0.05
0.85*
0.05
-0.03
-0.13*
0.04
-0.05
-0.03
-0.07*
0.04

0.06
0.28*
0.22*
0.04
0.11*
0.06
-0.02
0.07*
0.01
-0.06

-0.03
-0.08*
-0.21*
0.09*
0.08*
-0.04
-0.04
0.05
-0.07*

0.11*
0.10*
-0.06
0.06
-0.07*
-0.01
-0.05
0.03

0.08*
-0.08*
0.02
0.03
0.01
0.16*
-0.16*

-0.12*
-0.02
0.06
-0.08*
0.10*
-0.11*

0.20*
-0.05
0.26*
-0.12*
0.14*

* represents significance at the p< 0.05 level.

51

(14)

(15)

(16)

(17)

-0.13*
0.10* 0.01
-0.09* 0.03 -0.12*
0.12* -0.02 0.14* -0.87*

Table 5: SEM model using salary increase as a proxy for employee performance

JobCategoryt
Nonprofitt
SurveyYr

(1)
DV = %Increaset+1
2.17***
(0.6)
8.11***
(1.07)
-3.43***
(0.81)
5.00***
(0.42)
0.46***
(0.06)
0.37***
(0.03)
-0.05***
(0.01)
3.72***
(0.19)
-0.19***
(0.02)
6.22***
(0.45)
1.66***
(0.35)
0.00***
(0.00)
-2.45***
(0.31)
2.86***
(0.48)
-2.63***
(0.38)
-5.84***
(0.99)
-14.43***
(1.1)
Y
Y
Y

Observations
R-squared

906
0.18

(1)

Selectiont

(2)

PFPt

(1)*(2)

PFP*Selectiont

Controls Abil_testt
Educationt
Yrs_expt
Sal%ilet
Job_satt
Aget
Gendert
Ethnicityt
Firm_sizet
Differentiatort
Costt
Uniont
Mfgt
Servicet

Prediction
H1 (+)

H3 (-)

Prediction
H2 (-)

(2)
DV = PFPt
3.55***
(0.55)

10.73***
(1.06)
2.44***
(0.17)
0.05
(0.08)
0.19***
(0.02)
4.91***
(0.38)
0.08
(0.07)
-21.34***
(0.9)
6.43***
(1.08)
0***
(0)
15.07***
(0.93)
5.35***
(1.38)
-21.53***
(1.23)
-4.64***
(1.61)
2.6
(1.61)
Y
Y
Y
906
0.24

Note: All variables are defined as in Table 2. Standard errors were estimated with a bootstrapping procedure using 100 iterations, and are
presented in parentheses below the coefficient estimates. ***, **, * represent p<0.01, p<0.05, p<0.1 respectively.

52

Table 6: SEM model using promotion as a proxy for employee performance

JobCategoryt
Nonprofitt
SurveyYr

(1)
DV = Promot+1
-1.83***
(0.67)
5.73***
(1.39)
-3.82***
(0.85)
-3.12***
(0.90)
3.15***
(0.15)
-0.59***
(0.09)
-0.04**
(0.02)
1.89***
(0.36)
0.37***
(0.07)
0.00
(0.73)
-7.04***
(0.77)
0.00*
(0.00)
6.81***
(0.82)
13.60***
(1.20)
-5.16***
(0.90)
-1.15
(1.02)
-5.72***
(0.92)
Y
Y
Y

Observations
R-squared

906
0.18

(1)

Selectiont

(2)

PFPt

(1)*(2)

PFP*Selectiont

Controls Abil_testt
Educationt
Yrs_expt
Sal%ilet
Job_satt
Aget
Gendert
Ethnicityt
Firm_sizet
Differentiatort
Costt
Uniont
Mfgt
Servicet

Prediction
H1 (+)

H3 (-)

Prediction
H2 (-)

(2)
DV = PFPt
3.55***
(0.55)

10.73***
(1.06)
2.44***
(0.17)
0.05
(0.08)
0.19***
(0.02)
4.91***
(0.38)
0.08
(0.07)
-21.34***
(0.90)
6.43***
(1.08)
0.00***
(0.00)
15.07***
(0.93)
5.35***
(1.38)
-21.53***
(1.23)
-4.64***
(1.61)
2.60
(1.61)
Y
Y
Y
906
0.24

Note: All variables are defined as in Table 2. Standard errors were estimated with a bootstrapping procedure using 100 iterations, and are
presented in parentheses below the coefficient estimates. ***, **, * represent p<0.01, p<0.05, p<0.1 respectively.

53

Table 7: OLS Models and sensitivity tests
(1)
1.68***
(10.52)
4.42***
(17.88)

Selectiont
PFPt
PFP*Selectiont
Abil_testt

4.26***
(14.67)
0.45***
(10.61)
0.30***
(13.74)
-0.06***
(-10.29)
3.55***
(25.95)
-0.07***
(-4.35)
3.25***
(12.11)
2.17***
(8.42)
0.00***
(3.78)
-0.17
(-0.71)
-0.36
(-0.89)
-1.50***
(-4.90)
-8.08***
(-18.10)
-7.90***
(-17.37)

Educationt
Yrs_expt
Sal%ilet
Job_satt
Aget
Gendert
Ethnicityt
Firm_sizet
Differentiatort
Costt
Uniont
Mfgt
Servicet

(2)
2.72***
(10.10)
8.17***
(14.05)
-2.76***
(-6.89)
4.45***
(14.83)
0.44***
(10.20)
0.31***
(13.98)
-0.05***
(-9.73)
3.38***
(23.88)
-0.08***
(-4.39)
3.35***
(12.32)
2.24***
(8.59)
0.00***
(3.87)
-0.29
(-1.15)
-0.49
(-1.19)
-1.61***
(-5.22)
-7.76***
(-18.83)
-7.69***
(-17.81)

Diff*PFPt
Cost*PFPt

(3)
0.03***
(10.17)
0.08***
(13.77)
-0.03***
(-6.96)
0.04***
(15.10)
0.42***
(9.27)
0.31***
(14.15)
-0.05***
(-9.70)
3.39***
(23.77)
-0.07***
(-4.25)
3.30***
(12.01)
2.04***
(7.34)
0.00***
(3.43)
-1.13***
(-2.90)
-0.62
(-1.11)
-1.64***
(-5.61)
-7.79***
(-18.33)
-7.77***
(-17.55)
1.70***
(3.07)
0.38
(0.37)

Diff*Selectiont
Cost* Selectiont
Constant
Observations
R-squared

37.91***
(25.81)
906
0.209

36.85***
(24.89)
906
0.212

36.61***
(24.98)
906
0.213

(4)
1.79***
(5.87)
9.07***
(15.22)
-3.25***
(-8.08)
4.44***
(14.76)
0.45***
(10.53)
0.34***
(15.59)
-0.05***
(-10.01)
3.33***
(23.65)
-0.10***
(-6.08)
3.71***
(14.42)
1.92***
(7.44)
0.00***
(3.55)
-3.47***
(-5.53)
-6.32***
(-5.91)
-1.97***
(-6.86)
-7.67***
(-17.94)
-7.72***
(-17.27)

2.38***
(6.30)
4.36***
(5.82)
38.61***
(27.25)
906
0.216

Models are linear regressions using %Increaset+1 as the dependent variable. Variables are defined in Table 2. Z-statistics presented in
parentheses. ***, **, * represent p<0.01, p<0.05, p<0.1 respectively. Controls for JobCategoryt, Nonprofitt, SurveyYr are included in
each model.

54

REFERENCES

55

REFERENCES

Akerlof, G. A., and R. E. Kranton. 2005. Identity and the economics of organizations. The
Journal of Economic Perspectives 19 (1): 9–32.
Akyeampong, E. B. 2006. Screening job applicants. Persepctives on Labour and Income
(Statistics Canada, catalogue no. 75-001-XIE). May 2006 online edition.
Autor, D. H. 2001. Wiring the Labor Market. The Journal of Economic Perspectives 15(1): 2540.
Baiman, S. 1990. Agency research in managerial accounting: a second look. Accounting,
Organizations and Society 15(4): 341-371.
Banker, R. D., S.-Y. Lee, G. Potter, and D. Srinivasan. 2001. An Empirical Analysis of
Continuing Improvements Following the Implementation of a Performance-Based
Compensation Plan. Journal of Accounting and Economics 30 (3): 315-350.
----------. 2003. An Empirical Examination of the Impacts from Termination of a PerformanceBased Incentive Plan. Working Paper, University of Pittsburgh.
Bedard, Kelly. 2001. Human capital versus signaling models: University access and high school
dropouts. Journal of Political Economy 109(4): 749-775.
Benabou, R., and J. Tirole. 2003. Intrinsic and extrinsic motivation. Review of Economic Studies
70: 489-520.
Bidwell, M. 2011. Paying more to get less: the effects of external hiring versus internal mobility.
Administrative Science Quarterly 56(3): 369-407.
Bonner, S. E. and G. B. Sprinkle. 2002. The effects of monetary incentives on effort and task
performance: theories, evidence, and a framework for research. Accounting,
Organizations and Society 27: 303-345.
Cable, D. M. and T. A. Judge. 1997. Interviewers’ perceptions of person-organization fit and
organizational selection decisions. Journal of Applied Psychology 82(4): 546-561.
Campbell, D. 2012. Employee selection as a control system. Journal of Accounting Research
50(4): 931-966.
Campion, M. A., J. E. Campion, and P. J. Hudson. 1994. Structured interviewing: a note on
incremental validity and alternative question types. Journal of Applied Psychology 79:
998-1102.
Cardinaels, E. and H. Yin. 2013. Think twice before going for incentives: social norms and the
principal’s decision on compensation contracts. Working Paper, Tilburg University.
56

Chatman, J. A. 1991. Matching people and organizations: selection and socialization in public
accounting firms. Administrative Science Quarterly 36(3): 459-484.
Christ, M. H., K. L. Sedatole, and K. L. Towry. 2012. Sticks and carrots: the effect of contract
frame on effort in incomplete contracts. The Accounting Review 87(6): 1913-1938.
Cohen, Y. and J. Pfeffer. 1986. Organizational hiring standards. Administrative Science
Quarterly 31(1): 1-24.
Daniels, C. 2001. Does this man need a shrink? Fortune Magazine. Retrieved from
http://money.cnn.com/magazines/fortune/fortune_archive/2001/02/05/296149/.
Davis, J. H., F. D. Schoorman, and L. Donaldson. 1997. Toward a stewardship theory of
management. Academy of Management Review 22(1): 20-47.
Dearman, D. T. and M. D. Shields. 2005. Avoiding accounting fixation: determinants of
cognitive adaptation to differences in accounting method. Contemporary Accounting
Research 22(2): 351-384.
Dohmen, T. and A. Falk. 2011. Performance pay and multidimensional sorting: productivity,
preferences and gender. American Economic Review 101: 556-590.
Dutta, S. 2007. Dynamic performance measurement. Foundations and Trends in Accounting,
2(3): 175-240.
Ekuma, K. J. 2012. The importance of predictive and face validity in employee selection and
ways of maximizing them: an assessment of three selection methods. International
Journal of Business and Management 7(22): 115-122.
Ellingsen, T. and M. Johannesson. 2008. Pride and prejudice: the human side of incentive theory.
American Economic Association 98(3): 990-1008.
Gatewood, R. D. and H. S. Feild. 1998. Human Resource Selection. Fort Worth, TX: The
Dryden Press.
Gneezy, U., S. Meier, and P. Rey-Biel. 2011. When and why incentives (don’t) work to modify
behavior. American Economic Association 25(4): 191-209.
Grabner, I. and F. Moers. 2013. Management control as a system or a package? Conceptual and
empirical issues. Accounting, Organizations and Society, (forthcoming).
Graves, L. M. and R. J. Karren. 1996. The employee selection interview: a fresh look at an old
problem. Human Resource Management 35(2): 163-180.
Heinle, M. S., C. Hofmann, and A. H. Kunz. 2012. Identity, incentives, and the value of
information. The Accounting Review 87(4): 1309-1334.

57

Holmstrom, B. 1979. Moral hazard and observability. The Bell Journal of Economics 10(1): 7491.
Huo, Y. P., H. J. Huang, and N. K. Napier. 2002. Divergence or convergence: a cross-national
comparison of personnel selection practices. Human Resource Management 41(1): 3144.
Indjejikian, R. J. 1999. Performance evaluation and compensation research: an agency
perspective. Accounting Horizons 13: 147-157.
Ittner, C. D. and D. F. Larcker. 1998. Innovations in performance measurement: Trends and
research implications. Journal of Management Accounting Research 10: 205-238.
Janz, T. 1982. Initial comparisons of patterned behavior description interviews versus
unstructured interviews. Journal of Applied Psychology 67: 577-580.
Kennedy, F. A. and S. K. Widener. 2013. Socialization mechanisms and goal congruence.
Working paper, Clemson University.
Kim, Jeong-Yoo. 2007. Multidimensional signaling in the labor market. The Manchester School:
64-87.
Kline, R. B. 2005. Principles and Practice of Structural Equation Modeling (2nd ed.). New York,
NY: The Guilford Press.
Kubler, D., W. Muller, and H-T Normann. 2008. Job-market signaling and screening: An
experimental comparison. Games and Economic Behavior 64: 219-236.
Lambert, R. A. 2001. Contracting theory and accounting. Journal of Accounting and Economics
32: 3-87.
Lazear, E. P. 1986. Salaries and piece rates. The Journal of Business 59(3): 405-431.
Maurer, S. D. and C. Fay. 1988. Effect of situational interviews, conventional structured
interviews and training on interview rating agreement: an experimental analysis.
Personnel Psychology 41: 329-344.
Merchant, K. A. 1982. The control function of management. Sloan Management Review
(Summer): 43-55.
Merchant, K. A. 1985. Organizational controls and discretionary program decision making: a
field study. Accounting, Organizations and Society 10(1): 67-85.
Merchant, K. A., and W. A. Van der Stede. 2007. Management Control Systems. Performance
Measurement, Evaluation and Incentives. Harlow, U.K.: Pearson Education Ltd.
Montgomery, J. D. 1991. Social networks and labor-market outcomes: towards an economic
analysis. American Economic Review 81(5): 1408-1418.
58

Murphy, K. J. 1999. Executive compensation. In Handbook of Labor Economics, Vol. 3B, edited by
O. Ashenfelter and D. Card, 2485-2563. Amsterdam: Elsevier/North-Holland.

Murphy, K. R. 2012. Individual differences. In The Oxford Handbook of Personnel Assessment
and Selection, edited by Schmitt, N., 31-47.
Ones, D. S., S. Dilchert, and C. Viswesvaran. 2012. Cognitive abilities. In The Oxford Handbook
of Personnel Assessment and Selection, edited by Schmitt, N., 179-224.
Ouchi, W. G. 1979. A conceptual framework for the design of organizational control
mechanisms. Management Science 25(9): 833-848.
Oyer, P. and S. Schaefer. 2011. Personnel economics: hiring and incentives. In The Handbook of
Labor Economics, Volume 4, edited by Ashenfelter, O. and D. Card, 1769-1823.
Amsterdam, The Netherlands: Elsevier.
Prendergast, C. 1999. The provision of incentives in firms. Journal of Economic Literature 37 (1):
57-64.

Rosse, J. G., J. L. Miller, and M. D. Stecher. 1994. A field study of job applicants’ reactions to
personality and cognitive ability testing. Journal of Applied Psychology 79(6): 987-992.
Rothschild, M. and J. Stiglitz. 1976. Equilibrium in competitive insurance markets: an essay on
the economics of imperfect information. The Quarterly Journal of Economics 90(4): 629249.
Ryan, A. M., L. McFarland, H. Baron, and R. Page. 1999. An international look at selection
practices: nation and culture as explanations for variability in practice. Personnel
Psychology 52: 359-391.
Sarndal, C.E., Swensson, B., Wretman, J.H. 1992. Model Assisted Survey Sampling. SpringerVerlag, New York.
Searacy, C., P. N. Woods, R. D. Gatewood, and C. E. Lance. 1993. The structured interview: a
meta-analytic search for moderators. Paper presented at the Society of Industrial and
Organizational Psychology.
Sekiguchi, T. and V. L. Huber. 2011. The use of person-organization fit and person-job fit
information in making selection decisions. Organizational Behavior and Human
Decision Processes 116: 203-216.
Shellenbarger, S. 2013. Why hiring managers use personality tests. The Wall Street Journal.
Retrieved from
http://online.wsj.com/article/SB10001424127887324549004579067102747248232.html
Spence, M. 1973. Job market signaling. The Quarterly Journal of Economics 87(3): 355-374.

59

Suff, R. 2012. Stryker selects top talent. Gallup Business Journal. Retrieved from
businessjournal.gallup.com/content/19090/stryker-selects-top-talent.aspx.
Terpstra, D. E. and E. J. Rozell. 1993. The relationship of staffing practices to organizational
levels measures of performance. Personnel Psychology 46: 27-48.
Thompson, S. K. 2012. Sampling (3rd ed.). Hoboken, NJ: John Wiley & Sons.
Tirole, J. 1988. The Theory of Industrial Organization. Cambridge, MA: The MIT Press.
Waldman, M. 1984. Job assignments, signalling, and efficiency. The RAND Journal of
Economics 15(2): 255-267.
Wasserman, N. 2006. Stewards, agents, and the founder discount: executive compensation in
new ventures. Academy of Management Journal 49(5): 960-976.
Wiesner, W. H. and S. F. Cronshaw. 1988. A meta-analytic investigation of the impact of
interview format and degree of structure on the validity of the employment interview.
Journal of Occupational Psychology 79: 599-616.
Widener, S. K. 2004. An empirical investigation of the relation between the use of strategic
human capital and the design of the management control system. Accounting,
Organizations and Society 29: 377-399.
Wilk, S. L. and P. Cappelli. 2003. Understanding the determinants of employer use of selection
methods. Personnel Psychology 56: 103-124.
Wilson, C. 1980. The nature of equilibrium in markets with adverse selection. The Bell Journal
of Economics 11(1): 108-130.

60