:Ezpttfi
.. .53.! 12.. u.
1‘
.3... 3:6
1.:

’91:... 7.: 3
....:l..s .I: ..
2.

«1 ‘17.».
:24}:

5.1.3.5:
:3! 5...: 13.: s? i

:1. .finuflnuu:

$9..
:1 :33}.
a

31‘.
i . .
3. .1152
4.....-
.1

{.5121},
vi . r...
.

i
‘2... . ..
w . q 1:5.
.3; z : .33.. Z.
11‘. . ..
.sasa I.
...
.r 2.2:

x. :1...

. .
i.

‘

, :

iv: {1 a
1...... the... H.

20.13 .

10.5...

, 3:...

.5 :. 1:7!

‘ 1!...
9. I.

hidu.
2.. .

:3! I

z. :2. .t .x ‘

12......“ J...
.5 1
at... ‘ 1

. v i
i. 1... ‘ . :13.
‘ .2

‘24
v
3:

~ ti. ‘
L...

 

 

 

 

 

:1...
.1... :3?

i2;

 

vain
I 3:;
2:.- \

9.;
>v\. 39.1..

«:31‘tk. ,

3". x
Yr.
..

. it
:5 I. r:

. 1!
.na}.)...3
‘,:..

.

THESlS

9
2001

 

LIBRARY
Michigan State
University

 

 

 

This is to certify that the
dissertation entitled
CONTINGENT WORKERS' IMPACT ON THE PAY AND

PROMOTIONS OF TRADITIONAL EMPLOYEES IN THE
INFORMATION TECHNOLOGY PROFESSION

presented by

Jennifer Lynn Gregory

has been accepted towards fulfillment
of the requirements for

Ph. D. degree in Social Science

 

 

: /
@l W [OLE Qw all] /
Valor professfl’

Date 8/2 B’l/O/

MS U i: an Affirmative Action/Equal Opportunity Institution 0-12771

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

'

 

3, DATE DUE

DATE DUE

DATE DUE

 

37'.)
....
s..-

235? 1 92003

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 cycincmmmpes-ms

 

COATINGI-
TKADITI

 

 

CONTINGENT WORKERS’ IMPACT ON THE PAY AND PROMOTIONS OF
TRADITIONAL EMPLOYEES IN THE INFORMATION TECHNOLOGY
PROFESSION

By

Jannifer Lynn Gregory

AN ABSTRACT OF A DISSERTATION

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

DOCTOR OF PHILOSOPHY

School of Labor and Industrial Relations

2001

Edilberto Montemayor

 

corn:
nu

This
Contingent r
regular empf
market b} e1

Workers from

 

 

Potential for
Workforce cl
“Ilh all of III:
emploiffs m’
accommodak
Wotection ha
djficlences in
employees an
adIUSImentS a]

for Crafi Tami;-

ABSTRACT
CONTINGENT WORKERS’ IMPACT ON THE PAY AND PROMOTIONS OF
TRADITIONAL EMPLOYEES IN THE INFORMATION TECHNOLOGY
PROFESSION
By

J annifer Lynn Gregory

This research tests for any effects that firms’ use of Information Technology
Contingent Staff (ITCS) will have on the regular, IT employees hired by the firm. These
regular employees have traditionally been protected from the forces of the external labor
market by either craft or enterprise Internal Labor Markets (ILMs). These ILMS shielded
workers from swings in pay due to labor supply and demand and provided workers with
potential for career growth through promotions and training. By bringing ITCS into the
workforce, employers are directly exposing their employees to the external labor market
with all of the potential and pitfalls that may come with it. It is hypothesized that
employers will probably need to make adjustments in their pay policies and practices to
accommodate the information brought into their systems by the IT CS. The ILM
protection has typically varied for craft and enterprise ILMS and it is suggest here that the
differences in how employers handle the introduction of ITCS will also depend on if the
employees are part of a craft or an enterprise job family. Results indicate that no such
adjustments are made for enterprise families and that only limited adjustments are made

for craft families.

 

 

 

200 l

 

 

COPyright by
JANNIFER LYNN GREGORY

2001

To my parents, Tom, and my sister, without their support and understanding this research
would not exist.

iv

 

 

 

 

 

 

iii
lwould be
channemi
that not on';‘
thought thn
including K;

ektremel} ht

 

Each one. it:
the Way [0 m
and Particula
endeaxm b}
“Dialling-\- SI
that this dISSt

“malice I

ACKNOWLEGEMENTS

The research in this dissertation was achieved with the help of many individuals.
I would be remiss to not acknowledge their contributions. My dissertation committee
chairperson, Edilberto Montemayor, has spent countless hours working with me to ensure
that not only this project but also my entire graduate school experience was carefully
thought through and that I achieved the goals I had set. My dissertation committee
including Karen Roberts, Alison Barber, Michael Moore, and Neal Schmitt, have all been
extremely helpful and patient while reading through the multiple drafts of this document.
Each one, using their own specialties and knowledge, has asked the right questions along
the way to make the most of this research. Also, the people of William M. Mercer, Inc.
and particularly the Data and Systems Group, who have assisted me greatly in this
endeavor by supplying the data on which this research is founded. They have also been
unfailingly supportive in providing me great flexibility in my own work arrangements so
that this dissertation could be completed. Thank you all for your contributions and

assistance. They were greatly necessary and appreciated.

TABLE OF CONTENTS

LIST OF TABLES viii
LIST OF FIGURES ix
CHAPTER 1 - INTRODUCTION 1
CHAPTER 2 — LITERATURE REVIEW 6
Contingent Work Arrangements 7

Shifis in Employment Arrangements and Demographics Over Time 10

Advantages to Contingent Work Arrangements 14
Disadvantages to Contingent Work Arrangements 17

Information Technology Contingent Work 20

Internal Labor Markets 23

Basic Structure of ILMs 24

Reasons for ILMs 27
Compensation in ILMs 32

ILM Level so Administration 34

IT Professionals in Internal Labor Markets 36

Neoclassical Economic Models in ILMs 37

External Labor Market’s Impact on Internal Labor Markets 41
CHAPTER 3 — HYPOTHESES 47

Hypothesis Set 1: The pay differential between IT CS and regular IT employees. 50
Hypothesis Set 2: Regular employee pay differences between firms using ITCS

and firms not using ITCS. 51
Hypothesis Set 3: The extent of ITCS use and regular employee pay. 55
Hypothesis 4: ITCS use and regular employee pay variability. 56
Hypothesis Set 5: Difference between “core” and “periphery” industries in the
pay consequences of IT CS use. 56
Hypothesis Set 6: ITCS use and long-term incentive for regular employees. 60
Hypothesis Set 7: ITCS use and promotion for regular employees. 61
Hypothesis Set 8: Extent of IT CS use and salary structure range overlap for
regular employees. 63
CHAPTER 4 - DATA AND METHODOLOGY 66
Data 66
Methods 72

CHAPTER 5 - RESULTS 84
Hypothesis Set 1: The pay differential between ITCS and regular IT employees.88
Hypothesis Set 2: Regular employee pay differences between firms using ITCS

and firms not using ITCS. 90
Hypothesis Set 3: The extent of ITCS use and regular employee pay. 94
Hypothesis 4: ITCS use and regular employee pay variability. 97
Hypothesis Set 5: Difference between “core” and “periphery” industries in the
pay consequences of ITCS use. 100
Hypothesis Set 6: ITCS use and long-term incentive for regular employees. 103
Hypothesis Set 7: ITCS use and promotion for regular employees. 106
Hypothesis Set 8: Extent of ITCS use and salary structure range overlap for
regular employees. 109
CHAPTER 6 — CONCLUSION 1 14
Summary of Results 115
Possible Explanations for These Findings 118
Future Research 121
Conclusions 124
APPENDIX A — Survey Questions 127
APPENDIX B — Job Descriptions 131
APPENDIX C — Industry, Geographic, Unemployment, and IT Percent of Employment
Tables Distribution Tables 139
BIBLIOGRAPHY 147

vii

LIST OF TABLES.
Table 1: Descriptive Statistics and Correlations
Table 2: Paired t-test Results of Hypotheses 1a and 1b
Table 3: Paired t-test Results of Hypothesis 1c
Table 4: Moderated Regression Results for Hypothesis Set 2
Table 5: Moderated Regression Results for Hypothesis Set 3
Table 6: Moderated Regression Results for Hypothesis Set 4
Table 7: Mediated Regression Results for Hypothesis Set 5
Table 8: Moderated Regression Results for Hypothesis Set 6
Table 9: Moderated Regression Results for Hypothesis Set 7
Table 10: Moderated Regression Results for Hypothesis Set 8

Table 11: Summary of Results for all Hypotheses

viii

86

89

90

93

96

99

103

105

108

111

112

LIST OF FIGURES

Figure 1: Mediating Model of Hypotheses 5a and 5b 79, 101

 

worke .

comp:

works

experii ::

work 3

COMM: t

of Worl

researcl

employ ‘

3131mm

0! not 1h
(7001111 gs

e"1910)::

machter.
w(”Riffs in

behalf m'

SUCh as he,

emplol’ees

 

IL

9(.

h i

Y»
g.

0F

 

 

 

CHAPTER 1 - INTRODUCTION

There has been a growing concern in the United States about the use of contingent
workers. Concerns about these work arrangements stem from the apprehension that
companies provide lower levels of pay, benefits, and working conditions for contingent
workers than they do for their regular employees. This means that contingent workers
experience a lower standard of living than regular employees and studies on contingent
work arrangements tend to support these ideas. However, researchers have not yet
considered the potential impact contingent work arrangements may have on other aspects
of work (Pfeffer & Baron, 1988). To broaden our understanding of this trend, the
research presented here addresses the use of contingent workers fiom the perspective of
employees in regular/long-term employment relationships. While contingent workers
appear to directly suffer from lower working conditions, this research examines whether
or not the working conditions of regular employees are indirectly impacted by the use of
contingent workers. Further, this research examines if this impact varies for regular
employees performing different types of work

The regular work relationship in the United States is one where employers and
employees implicitly agree to continue their association over multiple time periods
(Wachter, 1990; Doeringer & Piore, 1971; Cordova, 1986). Employers provide their
workers with ongoing employment and career advancement opportunities, unless they
behave with gross negligence. In addition, employees are provided with other benefits
such as healthcare and retirement insurance. These work arrangements protect

employees from swings in external labor market conditions. Employees have a sense of

job security and are encouraged to learn firrn-specific skills which would be less valuable
to them in other firms. This regular work relationship is called an Internal Labor Market
(ILM) (Doeringer & Piore, 1971).

Contingent work relationships are those which are not explicitly or implicitly
long-standing and may take one of many forms including temporary, part-time, leased,
self-employed, contracted out or home-based work. The key defining element of
contingent work, however, is the degree of uncertainty regarding the continuance of the
work arrangements with the hiring organizations (Polivka, 1996).

The research in this paper studies the impact that the presence of these contingent
workers has on the regular employees still “protected” by traditional ILMs. Because
work outcomes are not determined by single, independent events, but rather by many
intermingled events, companies’ use of contingent workers should be expected to change
the terms of work arrangements between employers and employees. Employers using
contingent workers will have access to potential employees while incurring only minimal
recruiting expenses. They will have previewed these employees’ work habits and know
more about them than they would a typical job applicant. These employers are likely to
be more aware of the availability of workers outside their firms than employers not using
contractors. Similarly, regular employees working with contingent workers will have
more exposure to the external labor market than employees, who do not work with
contingent workers. They will be more aware of favorability or unfavorability of the
external employment opportunities for their jobs. All of these factors should effect the
demand and supply of labor, which in turn will effect employee wages and other work

arrangements.

Prior research on contingent work arrangements has discussed the myriad
implications these temporary work arrangements have for the contingent workers.
Researchers have looked at the demographics of these workers, their hours of work, and
their pay levels in comparison with regular employees. Additionally, there has been
extensive debate in the United States about the lower level of health and retirement
benefits available to the contingent labor force, because private employers through long-
terrn employment are the normal delivery mechanism for these benefits (Cordova, 1986;
Christensen, 1988; Loveman, 1988). Most of the prior research concludes that contingent
workers have lower pay, benefits, and working conditions.

This study attempts to clarify that the problems associated with contingent
workers effect not only the contingent workers, but also the regular employees working
with them. If contingent workers are used in companies, then their terms of employment
should be anticipated to have some impact on the terms of employment for regular
workers. This impact may occur despite or because of the fact that contingent workers
are only hired for a specific, limited period of time and consequently they have no long-
term commitment to the hiring organizations. It is hypothesized here that contingent
workers’ presence will change the way organizations treat their regular employees with
respect to compensation level and mix, as well as, promotion opportunities. Further,
changes in companies’ pay philosophies or internal labor markets with respect to their
regular employees may differ depending on the types of work being performed (i.e., craft
or enterprise).

This research examines the effect that contingent workers in the Information

Technology (IT) profession have on the wages and other ILM conditions of regular, 1T

workers. Information technology is one of the largest and fastest growing segments of
contingent work. Kosters (1997) reports a 300% grth in IT contracting since 1982.
Sharpe (1997) claims that 40% of contracted work in the United States during 1996 was
in IT services and that this is the largest single type of contracted work. Because of IT’s
large presence in the contingent work arena, it is interesting to study the impact
Information Technology Contingent Staff (ITCS) may have on the work relationships of
regular IT employees as these findings effect a large group of workers. Also, it is
possible that the findings for IT workers are applicable to other types of workers, which
would increase the value of these results.

In this research it is hypothesized that the effect of the use of ITCS on the wages
and other ILM conditions of regular, IT employees is different when they work in job
families that belong to two types of ILMs: craft and enterprise. Craft employees
generally, have a status of independent professionals (Doeringer & Piore, 1971;
Osterman, 1983). They are assumed to have a substantial body of knowledge before
being hired by a company and are most likely to be responsible for their own continuing
education The hypotheses predict that the presence of ITCS will be favorable for
regular, craft IT employees. The second group of hypotheses addresses the impact ITCS
have on enterprise IT employees. Enterprise IT employees are not required to have skills
when hired into an organization. They are most likely to be trained on the job and do not
require additional skills beyond those learned at the beginning of employment (Doeringer
& Piore, 1971; Osterrnan, 1983). The hypotheses predict that the presence of ITCS will

have an unfavorable impact on the work relationships of these enterprise, IT employees.

The data used in this research are from employee compensation surveys
conducted in 1998 and 1999. The data from these two surveys are combined into one
data set to increase the sample size and thereby increase the potential for finding effects
should they be present. Most of the analyses will include control variables for company
size, industry and general labor market conditions to avoid conclusions based upon
spurious relationships.

This research attempts to increase our understanding of how using contingent
work is changing our traditional Internal Labor Markets. Prior research on contingent
workers is interesting and necessary, however, we should start considering that there are
potentially many more employees, whose work lives are being changed by this
phenomena. Employees, who have done nothing to alter any part of their work
relationships, may now be finding themselves facing employers with new ideas regarding
the worth of their jobs and the security of the implicit long-term employment contract
many believed they enjoyed. These regular employees, who work side-by-side with
contingent workers, are the main focus of this research. It is hoped that this study will
increase our understanding of the workings of ILMs and draw attention to a different

perspective of the contingent work debate, that has so far been left unmentioned.

CHAPTER 2 — LITERATURE REVIEW

This chapter reviews three literature streams. The first literature reviewed is that
on contingent labor. Researchers doing studies in this area have struggled with
quantifying the magnitude of these work arrangements and who these workers are. These
efforts have been and continue to be confounded by problems of definitional clarity.
Other researchers have studied the reasons for these arrangements to exist from both the
employers’ and the employees’ perspectives, as well as, the disadvantages of these
arrangements. Research on the use of contingent workers in Information Technology
(IT) jobs is specifically addressed. The next literature is on Internal Labor Markets
(ILMS) and the role that neoclassical Economics has within this framework. Doeringer
and Piore’s (1971) idea of the Internal Labor Market (ILM) plays an important role in
many human resource systems. ILMs shield many employees from the vagaries of the
external labor market through policies and rules for pay and promotion. Economic
theories work within these systems and as employees look outside their firms for other
opporttmities. The research reviewing herein describes the advantages and disadvantages
of these systems, the level of analysis for ILMs, and how neoclassical economic theories
integrate with these systems. Finally, the third literature reviewed which looks at the
interaction of ILMS and external labor markets and the co—existence of regular employees
and contingent workers. This third section presents research most similar to this study
because it shows how aspects of the labor market internal and external to the organization

are interrelated. Pfeffer and Baron’s (1988) discussion provides interesting directions for

future research on internal changes related to the externalization of work. Some of which

are incorporated in this study.

Contingent Work Amggements

A growing number of researchers and public policy makers have become
concerned with the phenomena of “flexibility” in the workplace. Striving to be more
flexible has occurred in many countries and in almost every industry (Cohany, 1996;
Polivka, 1996b). Researchers have begun to study the wide range of practices, which
result in flexibility. At one extreme, firms may become more flexible by moving their
operations to locations with more attractive business legislation and labor markets. In a
more moderate approach, firms may strive to achieve flexibility by having a workforce
that is functionally flexible or a workforce that is numerically flexible (Pollert, 1991).
Functional flexibility refers to a workforce that can be easily reallocated to different tasks
to meet the changing skill needs of a firm. Numerical flexibility is a workforce that can
grow and shrink as workload varies. This research focuses on implications of numerical
flexibility through the use of contingent labor.

Ideally, numerical flexibility would allow firms to change their employee
headcount rapidly and with a minimum of disruption to company operations. Longer-
term work arrangements, as exemplified by Internal Labor Markets (ILMS), are discussed
in the next section, but generally are typified by employees and companies have an
implicit agreement to sustain their employment relationship for multiple time periods.

This type of work arrangement complicates firms’ usage of numerical flexibility as a

staffing technique because the continuing employment agreement conflicts with changing
headcount.

The key characteristics of contingent workers are that they are not recognized as
regular employees of the firms, who are protected by the standard work arrangements
mentioned above, and as such they do not have any agreement with the user firms for
employment in future time periods. Their future employment with user firms is
contingent upon these firms’ labor needs in the next time periods. Contingent work
arrangements afford workers very little job security and force these workers to be
continuously searching the labor market for new employment situations. Throughout this
paper the term contingent work arrangements will by synonymous with a work
relationship between a firm and a worker that is conditional upon the firm’s labor needs
in future time periods.

Contingent work arrangements may take many forms. These “new” work
arrangements may include any derivations of temporary/leased and self-
employed/contracted employees (Polivka & Nardone, 1989). Independent contractors
and self-employed workers are hired on a project basis and their employment with the
firm often ends when the project is completed. Temporary and leased employees are
typically employees of another firm and are essentially rented out to the hiring firm. All
of these work arrangements are often considered contingent, but for purposes of this
research they will only be contingent if there is substantial insecurity about future
employment.

Contingent work arrangements are difficult to study because there is no single

definition that stipulates which workers have contingent status. Polivka ( 1996a) used

BLS data and three different definitions of contingent worker to estimate the prevalence
of these work arrangements. These definitions of contingent workers Polivka used
ranged fiom people working for firms, where their work relationships are expected to last
less than a year, to any workers, whose their current employment is not considered
permanent. Using this spectrum, in 1999 contingent workers represented anywhere from
1.9% or 2.5 million workers of the American workforce to 4.3% or 5.6 million workers
(Bureau of Labor Statistics, 1999c). Further, this study found that 8.2 million workers or
6.3% of the American workforce are considered to be independent contractors.

This wide range of estimates shows how important it is that researchers clearly
define what is meant by a contingent worker. Polivka and Nardone (1989) use the
stability of a job to create a distinction between contingent and non-contingent workers.
These authors state that many contingent work arrangements (e.g., leased, self-employed,
or contracted workers) may be long lasting and steady work relationships. When these
work arrangements exist over long periods of time, they should not be considered
contingent because they are probably not jobs that employers would eliminate if faced
with decreased workloads. Only those work arrangements with little or no job security
should be counted as part of the contingent labor market. Polivka and Nardone define
contingent work as, “Any job in which an individual does not have explicit or implicit
contract for long-term employment or one in which the minimum hours worked can vary
in a nonsystematic manner” (1989, p. 11).

Other research uses more legalistic definitions of contingent/contract work.
Christensen (1988) describes how the BLS definition of contingent is different fiom the

IRS definition of contract workers. BLS defines three categories of contingent workers:

1) wage and salary workers whose jobs will not last more than two years; 2) wage and
salary earners, self-employed, and independent contractors whose jobs will not last more
than two years; and 3) workers who do not expect their current jobs to last for an
additional year (BLS, 1999c). Contractors would be included in the self-employed
category, but certainly do not constitute this entire group. The IRS has outlined several
working conditions that must be satisfied for a worker to be classified as an independent
contractor (Wells, 1987). An independent contractor, according to the IRS, must have: 1)
some degree of control over his/her work, 2) an opportunity for profit or loss, 3) an
investment in facilities, 4) a non-permanent relationship with the contracting company,
and 5) the skills required to complete the work contracted without assistance from the
hiring company. All of these conditions are meant to determine the degree of
independence a contractor has from the hiring company and ensure that the contracting
relationship is more than an accounting tool to lower the employer’s liability. These
definitions focus on different facets of work from Nardone and Polivka’s (1996a)
definition, thereby illustrating how common terms (e.g., contractor) for work
arrangements can mean very different things. Definitional consensus on terms continues

to elude researchers, making research efforts inconsistent.

Shifts in Employment Arrangements and Demographics Over Time

Researchers’ interest in contingent work arrangements has been increasing
because these work arrangements cover a significant portion of our workforce and little is
known about them. Some evidence has shown that these work arrangements have not

fostered growth in workforce productivity and that the difference in pay and productivity

10

between high and low paying jobs is growing. These findings have led some researchers
to think these contingent work arrangements have created “bad” jobs (Loveman & Tilly,
1988). While Loveman and Tilly (1988) does not provide evidence that the growth of
wage differences is the result of contingent work arrangements, they have suggested it as
one potential cause. Further many of our economic models are based on a traditional
work arrangement with one employer providing steady employment over a worker’s
lifetime. Under contingent work arrangements, the economic predictions based on
traditional employment standards will not accurately portray labor market outcomes and
social policies enacted to support traditional work arrangements will lose their
effectiveness (duRivage, 1992, Cordova, 1986). The predictions and policies necessary
to accurately portray and protect contingent workers may vary depending on the type of
work being performed (i.e., professional, managerial, clerical, etc.) as the contingent
work relationships for these different work types may vary.

Two approaches to determine if the shift away from lifetime employment is really
happening are to measure changes in employees’ tenure or to measure changes in the
prevalence of contingent work arrangements. Studying changes in both is necessary to
accurately isolate shifts in employment practices because an increase or decrease in either
lifetime employment or contingent work arrangements may occur for many reasons like
the growing or shrinking of the labor supply. Measuring both simultaneously would
allow researchers to identify actual shifts in employment trends over time.

A decline in overall employee tenure indicates a less stable labor market, where
moving from one job to another is more acceptable. This type of environment may make

the use of contingent workers more common as employers and employees become more

11

accustomed to shorter-terrn employment relationships. Estimates of changes in the level
of long-term employment in the United States have been attempted by several
researchers. Carter (1988) calculated the change in lifetime employment rates by
comparing the rates of workers in San Francisco in the 1890’s to the American labor
force in 1978. Her results indicated that lifetime employment was more common the
1970’s than it was late in the last century. More recent investigations of this phenomena
have focused on shorter time spans (Diebold, Nemnark & Polsky 1997; Swinnerton &
Wial, 1995). Both Diebold et al. (1997) and Swinnerton and Wial (1995) studied job
tenure from the 1970’s into the 1990’s. Diebold et al. (1997) found little evidence of
declining job tenure overall, but did find evidence of a decline for a certain segment of
the labor force. Specifically, workers with a high school diploma or less experienced
lower job tenure than more educated workers. Swinnerton and Wial (1995) found a
curvilinear trend in job tenure over these years with higher retention occurring between
1983-87 and lower job retention in the early and late 80’s. Given the economic
prosperity during this middle time range, the authors conclude that the job tenure of the
middle time period is a result of higher employer demand rather than employee
preferences.

The second approach to studying changes in work arrangements is to directly
measure the prevalence of contingent work arrangements. Much research has focused on
the development of the temporary help industry because it is an easily identifiable
contingent work arrangement (Gonos, 1997; Mangurn, Mayall & Nelson, 1985; Moore,
1965). This research shows that temporary help agencies have been in existence since

World War H and are not new (Gonos, 1997; Moore, 1965), but that they are one of the

12

fastest growing industries in the United States. The type of work performed by
temporary workers is changing as this segment of the work force grows. Temporary
workers have traditionally performed clerical work, but temporary help firms have
expanded into manual labor, as well as, more professional job categories such as drafters,
engineers, and information technology professionals such as computer programmers
(Mangum et al., 1985). The need for these new skills is especially apparent with respect
to computer related jobs, where evolving technology requires workers to constantly learn
new languages and skills. Kosters (1997) has noted a 300% increase in computer related
contingent employment since 1982. Also, temporary agencies tend to specialize in a
specific type of temporary worker such as clerical, industrial, medical or
engineering/technical employees (Carey & Hazelbaker, 1986) giving them a more
complete knowledge of their industries/professions. The relationships with client firms
differ for each type of temporary from the daily outsourcing of clerical employees to
engineering temporary firms where assignments may last up to two years or more. Carey
and Hazelbaker (1986) predict that the use of technical temporaries will continue to
increase because many technical skills are not firm-specific and are easily transferable to
many companies.

In addition to simply looking at changes in work arrangements, there has been
much research on the demographic makeup of employees working under these non-
standard arrangements. The demographics of contingent workers vary by the type of
work arrangement.

Self-employed workers make up a small portion of the contingent work group.

They are more likely to be men (Casey, 1991; Fain, 1980; Rothstein, 1996), but women

13

comprise the fastest growing segment of self-employed workers (Becker, 1984). Self-
employed men work longer work weeks than wage workers, and typically earn more than
their wage earning counterparts (Fain, 1980). Self-employed women tend to'work fewer
hours and earn less than their wage earning counterparts. These trends may reflect a
pattern of self-employed men working in management and professional jobs and self-
employed women working in low-paying sales and service jobs (Becker, 1984). While
trend data on the level of self-employed workers in the United States show mixed results,
it appears as though this group has grown in overall size, but not as a proportion of the
total workforce (Bregger, 1963; Bregger, 1996). There is some research suggesting that
the level of self-employment in a country may be positvely related to governmental
policies, which encourage self-employment (Staber & Bogenhold, 1993, Ray, 1975).
Independent contractors, a group which may include both part-time and self-
employed individuals, tend to work in highly skilled fields and often work alone
(Cohany, 1996). This group is comprised of more college-educated, older men. They
tend to work in service industries, which require less capital investment. The majority of
people in this group seem to prefer working in their current alternative arrangements

rather than in a traditional work arrangement.

Advantages to Contingent Work Arrangements

Most companies use contingent employees for one or more of four reasons. One
of the main reasons for using contingent workers is to lower the labor costs of the firm
(Abraham, 1990; Belous, 1989). Contingent workers rarely receive healthcare or

retirement benefits, which are a significant portion of total labor costs, therefore,

14

contingent workers save firms substantially through lower benefit costs. Also, contingent
workers are frequently paid different rates than permanent employees because they are
not paid efficiency wages, which are set higher than market clearing wages, but allow
employers to reduce supervision costs (Abraham & Taylor, 1996). Contingent workers
also allow employers, whose pay levels target paying regular employees above/below
market, the ability to use contingent workers for non-value adding jobs and pay them at
or below/above market (Abraham, 1990). These firms are not thought to suffer adverse
effects of internal equity since these contingent workers are not regular employees of the
company. For example, firms targeting their regular employee population’s
compensation level below market, using contingent workers, where it is a necessity to
pay at least market, should save these firms from having to pay all of their regular
employees at the higher market rate.

The second reason to use contingent workers is to focus the company’s efforts on
its core competencies (Belous, 1989; Deavers, 1997; Sharpe, 1997). Core competencies
are those activities that are central to the value adding operations of the firm. By shifting
the responsibility for support functions such as information technology, human resources
or marketing outside of the company, employees can focus on the finn’s core
competencies that are critical for the company to succeed such as product development or
quality assurance. Further, many companies that are doing this contingent work have
these frmctions as part of their own core competencies. Therefore, in addition to being
centered on those activities that will further grow the company, non-core activities are
being performed by firms which specializes in such actions. These activities may not be

core competencies of the user firm, but they are the core competencies of the supplier

15

firms. Therefore the supplier firms can achieve better outcomes than user firms would
because they have value adding knowledge in these areas that the user firms don’t
possess.

The third reason for contingent workers is to acquire specialized skills. Many
firms would prefer to purchase skills unavailable in their own workforces, rather than
spend the time and training money to develop those skills in their current employees
(Abraham, 1990; Belous, 1989; Segal & Sullivan, 1995). This may be particularly true
for those skills the firm will only need for a limited time. Information technology skills
are a good example as they take time to learn, change rapidly and are difficult to attain.
If employers wanted to develop these skills in their current workforce, they would be
constantly investing resources in training them. Purchasing these skills through
contractors requires less of a time commitment and potentially may be cost saving. For
this reason information technology skills are the largest segment of contingent work
(Sharpe, 1997) and IT contingent workers are continuing to rapidly grow as a portion of
outsourced labor (Kosters, 1997).

Finally firms may use contingent work arrangements as part of a hiring strategy.
By employing a person as a contingent worker, firms may get a chance to preview a
worker’s abilities and work habits before committing a regular position to the employee
(Segal & Sullivan, 1995). Hiring a regular employee endows that worker with certain
legal rights and typically gives them access to costly benefits. Contingent workers, who
prove themselves to be good workers, may be hired into the company as regular
employees and contingent workers, who do not prove themselves to be good workers,

may be let go at the end of their assignment without entangling the employer in benefits’

continuation or other possible complications. Hiring contingent workers on as regular
employees is so common that most temporary help firms charge companies a
“liquidation” fee if a temporary employee is hired to help them recover some of their

training investment in that employee (Carey & Hazelbaker, 1986).

Disadvantages to Contingent Work Arrangements

Despite the added potential competitive advantages of contracting out work, there
are problems associated with these practices. Workers and public policy makers have
reason to be concerned about these arrangements. Research has shown that contingent
workers earn less than regular, full-time workers and are less likely to receive benefits
such as healthcare and retirement (duRivage, 1992; Loveman & Tilly, 1988; Rothstein,
1996; Tilly, 1992). As discussed earlier, contingent work is a significant portion of the
labor force. Therefore the lower wages and lack of benefits in this group effects a sizable
portion of American workers. Because healthcare and retirement are provided through
private employers in the United States, contingent employees assume much of the risk of
unemployment even while they are employed Contingent employees need to provide
themselves with health insurance which often will cost an individual much more than if it
were purchased as part of a large group of employees. Also their retirement fimds and
usually the “employer’s” share of income taxes (e.g., Social Security and Medicare) must
be paid by these employees. All of these costs make employment a more expensive
prospect for a contractor than a regular employee. Many contractors are able to set their
rates higher to offset these costs, but research does show that most contingent workers

earn less than regular employees with the exception of the self-employed (Loveman &

17

Tilly, 1988; Tilly, 1992). Additionally, since there is no implicit or explicit guarantee of
long-term employment, as is frequently present in regular employment settings,
contractors face added risk of losing their healthcare or retirement funds at the end of a
work assignment because they cannot afford the premiums and don’t have the COBRA
protection of regular employees.

The financial risks of contingent arrangements are only some of the problems
contingent workers face. Researchers have found that contingent workers are more likely
to be involved in accidents in the workplace Otebitzer, 1998; Rousseau & Libuser, 1997).
Because the user firm is not technically the contingent workers’ employer, the firm is not
legally responsible for the safety training of these workers. The contingent workers may
never receive such training because they are self-employed and do not have access to
safety training or because their employer does not provide such training. Contingent
employees typically spend less time at a job site than regular employees. Consequently,
they may be less familiar with the site and unaware of some of the safety risks present.

They also probably have less thorough communication and familiarity with the
permanent workers, which may result in lower productivity (Goodman & Leyden, 1991).
Goodman and Leyden (1991) have shown that specific knowledge with the job, the work
environment, and co—workers can lead to higher productivity. This lack of familiarity and
communication may be exacerbated by their status as contingent workers. Rogers’

( 1995) research of clerical temporaries shows that temporaries often feel alienated from
workers at the hiring firm, from the work they are doing, and fi'om their own work
ambitions. Being a temporary employee connotes to others a secondary status of being

not as important as regular employees. As such these contingent employees are not

18

afforded the same courtesy, acceptance, and information at the workplace as other
employees.

From the hiring firm’s perspective, outsourcing work may pose problems, if the
work being outsourced is part of the firm’s core competencies (Bettis, Bradley & Hamel,
1992). Researchers typically agree that outsourcing a core competency is not part of a
sound long-term strategy, but firms may still choose to do this to reap short-term benefits
from these arrangements. In these cases, firms must be careful not to let go skills that are
necessary for the production or technical development of its products. If the company
loses control over these important aspects of its business it is likely to lose control over
the future direction and abilities of the company. The company may no longer have a
good grasp on evolving products in its industry and won’t be able to react to the market
as is necessary to compete. This situation is even more sensitive if the outsourcing firm
is a potential competitor, who may eventually use the knowledge learned by contracting
to enter the product market itself and compete directly against the hiring firm.

For the firm that is outsourcing non-critical work, there are still other concerns. If
the contracting firm is small and relies upon the hiring firm for revenue, then only using
the contracting firm when there is increased product demand may mean the contracting
firm will not be working some of the time. This shortage of work puts the contracting
firm in jeopardy of going out of business. The hiring firm must be aware of this
possibility and protect itself against not having a contractor when it is needed (Irnrie,
1986). The hiring firm may have to continue using the contracting firm’s time, at least
minimally, even during low product demand to ensure it is around when product demand

rebounds.

l9

Information Technology Contingent Work

The focus of the research presented in this study is on the effects of using
Information Technology (IT) contingent workers in the workplace. As mentioned before,
IT is one of the largest and fastest growing segments of contingent labor (Kosters, 1997 ;
Sharpe, 1997). IT contractors are also one of the better paying contracting professions
(Williams, 1989). In 1987, contracted engineers earned on average $24.74 per hour,
systems analysts earned $18.17, and computer programmers earned $15.96. This high
wage rate implies that the opportunity costs for IT workers wishing to leave traditional
employment is lower than most other occupations. Also the high wages suggest that it
should be easier for computer people to find contingent work arrangements because there
is a high demand for their expertise.

There is a wide range of IT skills that may be contracted out by firms. Because
most IT skills are non-firm specific almost anything can be done with contingent labor.
Lesser-skilled IT work, such as data entry, is just as easily completed by contingent
workers as higher-skilled IT jobs, such as software systems engineers, programmers, or
analysts. Firms that use contingent workers will need to assess their own situations to
determine what types of IT work are best suited for contingent labor.

In the previous section four reasons for hiring contractors in general were
discussed: different pay rates, focusing on core competencies, acquiring specialized
skills, and as part of a hiring strategy. The relatively higher wages of professional
computer contractors (Williams, 1989) appears to confirm the first reason, of differing

pay levels between contingent and regular workers. IT contractors are often being paid

20

more than regular IT employees doing the same or similar jobs are being paid. However,
benefits, such as healthcare and retirement, represent a significant portion of total
remuneration costs, thereby making professional IT contractors with their higher pay
rates, in most cases, still more economically efficient than hiring regular employees.

The second reason for hiring contractors was to allow a firm’s regular employees
to focus on core competencies. IT services are typically part of a company’s support
function and not directly related to the main processes of the company. This being the
case, many firms may opt to contract out their IT function to firms that specialize in these
services. In fact IT functions may be one of the first areas firms look to contract out,
when trying to save money because it is considered an ancillary function and not a profit
center.

Contracting is especially beneficial for the higher level IT functions from the
perspective of the third reason of acquiring specialized skills. IT skills are very complex
and change rapidly. IT firms typically already possess the skills and capabilities that
other non-IT firms may be unwilling to invest training dollars to acquire. With the
current, fast pace of technological developments, technical obsolescence is a real concern
for IT professionals and firms (Lee, Trauth & Farwell, 1995), many firms find it
advantageous to contract their IT function out to avoid the need for extensive, on-going
training of IT workers required just to keep up-to-date. This reasoning will be more
applicable to the complex IT skills, such as programming and development (craft), than
to the lower skill IT functions such as data entry (enterprise). Further, most complex IT
skills are non-firm specific making them easily transferable from one company to another

(Applebaum, 1989). This transferability is another reason why firms may choose not to

21

invest in training for complex IT skills. IT workers are more marketable with these skills
and employers investing in IT training may be likely to lose these investment costs to
other employers once the employee has acquired these new skills.

Using contracting as part of a strategy for hiring IT workers may not be as
successful as it is for other professions. The unusually high demand and low supply of IT
professionals with complex IT skills is a partial explanation for the higher salaries IT
contractors receive. Since all IT professionals have the possibility of making these high
wages, it is less likely that they will be interested in regular employment without some
other non-pecuniary inducements. Workers in other professions, which are not facing
such extreme labor shortages, might be more interested in finding stable employment
through regular employment rather than work in the relative insecurity of a contracting
relationship. Again this reason may be more applicable to professionals with complex IT
skills rather than the simpler functions, where extensive training is not required and a
labor shortage is not as evident.

The disadvantages to contracting discussed previously tend to be less important
for IT contracting. Lower wages for these employees is often not a reality, especially if
the contractor has access to healthcare benefits through other sources, such as a spouse.
Lower training for contractors does not seem to be a concern as higher level IT
professionals are part of a craft and typically take responsibility for the majority of their
training rather than relying on a company to provide training. Training, however, may be
a larger issues for clerical II functions. Employer fears of lower productivity may not be
realized either as most IT skills are non-firm specific (Applebaum, 1989). IT contractors

may be equally productive in any firm because the computer language and general

22

approach to IT work does not change from one computer system to another. Since IT
functions are typically support functions, firms need not be concerned with outsourcing
critical functions which may endanger the health of the organization in the long-run.
Overall, contracting IT workers should in theory allow companies to benefit from
many of the positive aspects of contracting and gain much flexibility without having to
suffer through the normal consequences of contracting. Firms may not gain as much in
lower salaries for IT contractors as they would for other professions, but they don’t suffer
from as many side effects by letting go of critical function or of lower productivity from
contingent workers. Plus firms that contract IT skills save significantly on lower training
costs. The benefits of contracting, however, may be greater for the more complex IT
functions such as programming, analysis and systems engineering than for the more
clerical tasks like data entry. These more complex skills require more training and
change more rapidly than data entry jobs, therefore firms would realize greater savings in
training costs and acquire more up-to-date skills by contracting out their complex IT
functions rather than the simpler functions. For these reasons, it is anticipated that IT
contracting will continue to grow and will impact employers’ relationships with their
regular IT employees, who work under an ILM structure. It is also anticipated that this

impact will vary with the type of IT work being contracted out.

Internal Labor Mar
Compensation of regular employees is determined by many factors both internal
and external to the organizations in which they work. Internal factors are often

determined by the presence of an Internal Labor Market (ILM), while external factors are

23

controlled by neoclassical wage setting models. Internal Labor Markets were initially
described and studied by Doeringer and Piore (1971) as systems that firms use to
administer pay and promotions and protect themselves from the fluctuations of labor
supply. Neoclassical wage setting models explain how compensation is determined for
workers across all firms hiring their skills. The literature reviewed herein describes the
basic elements of ILMs, the distinction between craft and enterprise ILMs, the
advantages and disadvantages of ILMs, and how the neoclassical models interact with

ILMs to determine wages.

Basic Structure of ILMS

Internal Labor Markets were initially described and studied by Doeringer and
Piore (1971 ). They represent one of the most common work systems in the United
States. Many of the research articles reviewed this paper followed closely after
Doeringer and Prioe’s first writings. These research efforts were extensive and are still
applicable today even though ILMs as a research topic became less popular by the
1990’s.

Doeringer and Piore (1971) described the main elements of an ILM as limited
ports of entry at the bottom of defined career ladders, promotion from these ports of entry
up the career ladders over time, and an overall protection from the vagaries of the
external labor market as a result of internal promotion policies. ILMS provide more
continuity of work arrangements from one time period to the next. This Iong-terrn
approach distinguishes ILMs from the neoclassical approaches discussed later that rely

on firms and workers to renegotiate their work arrangements every time period.

24

Doeringer and Piore (1971) identified two different types of ILMs: a craft based
ILM and an enterprise-wide ILM. Craft ILMs focus mostly on general human capital,
skill set, which is not dependent upon the hiring firm. The labor supply for this craft in a
local or regional area feeds this kind of ILM Entry into the field is determined by
knowledge of a craft or skill, which may be monitored by a union or naturally through a
training/selection process that eliminates those workers not having the required skills. It
is usually the responsibility of the worker to obtain these skills. The craft or skill requires
knowledge that typically is not specific to a single firm, which facilitates the entry into a
firm at all levels. Entry into a firm may occur at almost any level, but promotion within a
firm for employees in a craft system may be based on a combination of seniority and
ability. The craft ILM model provides a good approximation of the work arrangements
for highly skilled IT professionals.

Enterprise ILMs focus on firm-specific human capital. The firrn’s current work
force feeds this kind of ILM because the skills required are mostly firm specific.
Employees working in these systems may be in any part of the company, but entry is
limited to lower level/lower skill jobs, employees are trained internally. There are few, if
any, skill requirements for hiring. Promotions are determined mostly by seniority.
Although descriptions of enterprise ILMs have been developed primarily from blue collar
systems, enterprise ILMS are good models for the work arrangements of clerical IT
employees.

Generally, craft systems rely more on the external labor market than enterprise
systems because of the external training requirements of craft workers. The distinction

between craft and enterprise ILMs shows that some ILMs are more open to external

25

influence than others. Craft ILMs’ practices fall somewhere between the completely
external labor market approach of the contingent model and the solely internal focus of
enterprise ILMs. It, however, should be stressed that while craft ILMs do tend to admit
more employees fi'om outside of the firm at all levels, internal promotion policies and
internal equity in wage determinations are the predominant forces in both types of ILMs.
Other research on labor relation strategies have resulted in similar classification
systems (Osterman, 1983 and Osterman, 1984a). Osterman (1983 & 1984a) described
three labor/employer relationships: industrial, craft, and secondary. Industrial
relationships mimic the enterprise category described above. Craft work relationships are
characterized by lower job security than the industrial category, but these workers
possess complex, non-firm specific skills and therefore have more bargaining power with
employers. This category is similar to the craft EM defined by Doeringer and Piore
(1971). Training in the craft relationship is the responsibility of the worker. Firms hiring
craft workers assume the workers’ have a minimum level of knowledge at the beginning
of the work relationship. Firms hiring craft workers seldom provide training for these
employees. The craft category may be preferable to workers, who can easily leave one
organization for another because their skills are general and very marketable. F inns may
not prefer the craft system because they have less control over the cost of labor in these
situations. Computer programmers represent a good example of jobs that fit into this
kind of craft system. Osterman’s secondary group has low job security and low skills
leaving these workers with little bargaining power against employers. In a later article,
Osterman (1987) adds a fourth category, salaried, to his typology of employment

systems. The salaried system is similar to the industrial system because it provides

26

workers with higher wages and job security, but the jobs are broader and individuals’
abilities may play a greater role in determining wages and promotions in the salaried
systems than in the industrial systems.

Because the craft and enterprise ILMS have been found by multiple researchers
(Doeringer and Piore, 1971, Osterman, 1983, 1984a, 1987), this research will focus
exclusively on them. It will not use the additional secondary and salaried categories
described by Osterrnan (1987). The craft and enterprise ILMs are easily distinguished
from each other and are conceptually appealing for classifying the IT job families

included in this research, which will be described later.

Reasons for ILA/ts

There are many reasons a firm would adopt a craft or an enterprise ILM.
Generally the less a firm wants to be dependent on market forces or the more it wants to
be responsible for training, the more likely it is to develop an enterprise ILM. Craft ILMS
anticipate that employees will have a minimum level of general skills before arriving on
the job and they are more reliant on external forces to supply skilled labor than enterprise
ILMS. Piore and Sabel (1984) suggest that craft ILMs may be more appropriate when
firms are producing specialized goods, while enterprise ILMs may be better suited for the
production of standardized goods. With respect to the relative openess of a human
resource system to external labor market forces, craft ILMs may be considered an
intermediary step between a predominantly internally focused enterprise ILM and a
predominantly externally focused contingent work arrangements. Therefore the reasons

for developing a craft or an enterprise ILM are similar. However, these reasons may play

27

a larger role in the development of an enterprise ILM than a craft ILM because the
enterprise ILM limits talent contributions from external labor sources to entry level jobs,
while the craft ILM has an option to acquire talents at all levels. This practice allows
craft ILMs to acquire talent when and where it is needed without a lengthy employer-paid
training period. It should, however, be remembered that while the craft ILM is more
open to employee entry from the external market at all levels, organizations with these
work arrangements are still interested in maintaining longer-term work relationships with
their employees than would be expected in contingent work relationships. Long-term
work relationships will allow these organizations to keep the institutional and firm
specific knowledge that come with employee tenure.

F inn specific skills are one reason that employers are motivated to develop ILMs.
If an employer’s technology is extremely specific to the firm, then the skills required to
perform tasks, such as programming computers, are unique to the organization. In the
case of computer skills, this may be applicable when legacy systems (i.e., old computer
systems that have been modified over the years to uniquely meet firms’ needs) or
exceptionally complex systems, which are highly interwoven with all aspects of the
firm’s business, need to be modified and/or maintained. The employer will have
difficulty finding these skills in the labor market and will need to invest resources to train
employees on these skills (Becker, 1993 ). Employees will need to invest time and effort
to learn these skills. Additionally, these skills will not be helpful to the employees in any
other organizations. The mutual investment of employer and employees creates an
incentive to maintain their employment arrangement over a longer period of time than if

this investment did not exist. Firms will want to recoup their training investment, which

28

may not be possible if employees quit shortly after training. Employees will probably not
be able to receive higher pay from other employers for having these skills as their firm
specific nature makes them ineffective for other firms. These employees will most likely
want to remain with the company where they will earn more because of their new skills.

Enterprise ILMs often develop when retention of employees for on-the-j ob
training is needed (Osterman, 1995). On-the-job training is especially important for a
company with many firm-specific skills because these cannot be learned outside the
organization To train new workers, firms rely heavily on experienced employees to
work with new employees and share their expertise. These long-tenured employees are
the only people with enough knowledge about the firm-specific skills to teach new
workers. It is in the firm’s best interest to retain these workers as a source of instructors.
Even for more general, non-firm specific skills, on-the-job training is often less expensive
than formal training. An ILM is one method of retaining employees to gain this
experience.

For companies with many firm-specific skills, constantly training new employees
will increase costs. To the extent that internal hiring practices are efficient, an ILM may
prove a good way for a company to effectively screen applicants. Because most
hires/promotions are from within the firm, management should know enough about the
employees’ abilities, interests, and work habits to make an informed decision about
which employees should be promoted or transferred. Further there will be a minimum
amount of training needed for these promoted employees since they are already familiar
with much of the employer’s operation. Some research has shown that employees with

firm-specific skills are more likely to be promoted (Scoones & Bernhardt, 1998).

29

The costs of hiring for a firm may also contribute to developing an ILM. The
transaction costs associated with turnover and external hiring may be substantial for a
company. Turnover may be quite expensive if a company has a high percentage of its
workforce leaving and/or the costs of finding replacement employees are high.
Termination costs associated with turnover may be significant when both legislated
continuation of benefits and company offered severance packages are considered.
Recruiting costs may be significant if labor supply is lower than demand, as is the case
with IT workers.

Higher recruiting costs may also be a problem for firms with ILMS, but this is
probably because research has shown that these firms engage in more selection
procedures than firms without ILMS (Cohen & Pfeffer, 1986). Firms with ILMS may
have lower turnover, but they spend more than firms without ILMS on selection
procedures. These multi-faceted selection processes are probably because companies
with ILMS expect to have employees stay for a longer time and therefore are more careful
in their screening techniques. Administering tests, and having more elaborate screening
and interviewing methods, however, will increase the costs of hiring.

Unexpectedly, if a finn’s technology changes rapidly there may be reason to
develop an ILM (Pfeffer & Cohen, 1984). If job requirements are changing quickly due
to changes in technology, as is perpetually the case with information technology,
employers may want to train employees internally rather than constantly searching for
new employees. The company’s current employees possess institutional knowledge that
will carry over to the new technology and would be lost if new employees were hired.

The value of this knowledge must be weighed against training costs when deciding on a

30

strategy for staffing new equipment and processes. Additionally, recruiting costs
incurred when employers continuously replenish their workforces may become
burdensome. Ifintemal training costs are less than the costs of searching for, recruiting
and socialization of new hires plus the termination costs of old employees, then internal
training would be the preferable course of action. For this reason, firms may actively
recruit and hire employees, who possess a higher ability to learn new technologies as a
means of further lowering training costs (Kalaitzidakis, 1997). If training costs for new
skills are too high frrrns may choose to buy talent in the external labor market (Piore &
Sabel, 1984). This reason for developing an ILM may be more applicable to a craft
rather than an enterprise ILM.

From the employees’ perspectives, an ILM is attractive because it should reduce
search costs and the risks associated with being unemployed. Without an ILM, most
employees need to look for new work every time their immediate job is finished or when
their skills improve enough for them to find a better job. An ILM allows employees to
continue growing within one company. Employees under these systems enjoy better job
security and the possibility to continue acquiring new skills and promotions. As part of
the strategy for long-term retention of employees, companies may choose to weight the
compensation of their workers more heavily toward the end of their careers as an
incentive for long service (Wachter & Wright, 1990). Workers may be paid less than
their marginal productivity in the early years of their careers, but be compensated more
than their marginal productivity in their later years with the company. This type of
compensation scheme is another reason for employees to remain with a company to earn

back deferred wages.

31

Research has shown some organizational factors are related to the development of
ILMs (Baron, Davis-Blake & Bielby, 1984). Larger organizations are more likely to
have an ILM. The presence of human resources departments is positively related to
having ILMs (Pfeffer & Cohen, 1984). Pfeffer and Cohen (1984) also showed that non-
manufacturing firms were more likely than manufacturing firms to have ILMs. Since
initial research on ILMs focused on the blue-collar manufacturing environment, this
finding is somewhat surprising (Pfeffer & Cohen, 1984). Pfeffer and Cohen (1984)
suggest one reason for this finding is that manufacturing firms may be better able to
control workers through manufacturing technology than non-manufacturing firms which
must rely on bureaucracy.

The use of ILMs was originally believed to be present only in “core” firms.
“Core” firms are large entities that have oligopolistic power in their industries while
“peripheral” firms are smaller and have more volatile product demand (Beck et al., 197 8;
Tolbert, Horan & Beck, 1980; Averitt, 1968). The volatility of the peripheral firms’
environments makes it difficult for these firms to supply the benefits of an ILM to their
employees because ILMs require significant resources that would need to be sustained
through difficult times. The oligopolistic nature of “core” firms allow them to more
accurately predict their product demand and their need for labor. This enables the firms

to more easily sustain an ILM over time.

Compensation in ILMS

Wages in an ILM are typically set to incorporate both internal equity and external

labor market conditions. Compensation professionals spend time allocating wage dollars

32

based on the jobs’ relative contributions to the organization and their worth in the open,
external labor market (Livemash, 1951). Jobs are ranked in order of importance to the
firm based upon criteria established in a job evaluation program (e. g., skill, supervisory
responsibility, etc.). Similar jobs are grouped into salary grades and a dollar range is
attached to that range. The value of the range may be determined through the use of
compensation surveys and through recruiting experiences. Compensation surveys are
used to estimate the going (external labor) “market” rates for jobs among the firm’s
external labor market competitors. Recruiting at entry level jobs for enterprise ILMs and
at all levels for craft ILMs will give firms another source of data regarding the current
market rate for jobs. These recruiting data are likely to be more precise in a craft ILM,
where more levels of compensation data are collected. By combining the internal equity
established with a point factor plan and the external equity established by compensation
surveys and recruiting information, a firm will be able to meet the market rate in most
cases, while maintaining internal equity (Osterman, 1984a).

Internal equity is important in an ILM environment where workers will be
together over an extended period of time and will be looking at the relative contributions
of their jobs to gauge what their compensation “should” be. Further, by lengthening the
employment relationship, employers and employees are no longer as concerned about
wages matching marginal productivity during every time period Rather they may be
more attentive to ensuring that wages match marginal productivity over the length of the
employment relationship. As such, employers will not have to adjust their wage
structures for every minute change in the labor market. These small fluctuations are

expected to be accounted for in wages over time (Groshen, 1991).

33

ILM Level of Administration

Initially, all ILM research was conducted at the company level of analysis. It was
assumed that companies used the same human resource policies for all employees and all
jobs. Some researchers, however, think that ILMs are not uniform across a company and
that other levels of analyses may provide a new perspective on traditional research
questions (Lazear, 1995). These researchers think that a company may choose to
implement such policies at the job family level (Baron et al., 1986; Osterman, 1984a;
Spilennan, 1977). Baron et al. (1986) found that organizational measures of ILMS
showed great variance that was better explained when the ILM characteristics were
measured at the job family level instead of at the firm level. They found that professional
and technical jobs were less likely than blue collar jobs to be in career ladders,
managerial jobs were diffith to enter from outside the organization, and jobs with more
incumbents were more likely to be in a ladder, as were jobs typically held by men. This
research confirms that companies have the ability to determine which job families would
benefit most fi'om the different work relationships present in craft and enterprise ILMS
and that companies can structure their human resource programs accordingly.

Why would a firm choose to treat some job families with the added protection
from the external labor market that exists in an enterprise ILM and offer a somewhat
reduced level of security that is present in a craft ILM to other job families? A firm may
want to capitalize on a better trained external supply of professionals through a craft ILM
that allows for external entry at all levels in jobs where skills are a critical part of

performance. In jobs where the skill requirements are relatively low, firms may find it

34

 

more beneficial to rely on internal training and promotions to retain as much institutional
knowledge as possible.

The simultaneous use of craft and enterprise ILMs means multiple work
relationships exist within one firm. If firms employ workers under both craft and
enterprise ILMs, then some of their employees enjoy the greater employment security
and wage stability, although not necessarily the going market wages, of an enterprise
ILM. Employees under the craft ILM may have fewer promotional opportunities, but are
more likely to be paid market clearing rates. These conflicting policies may or may not
preserve internal equity amongst all employees. Firms that choose to use multiple
employment systems must find a way to integrate two or more pay systems that have
different goals. Enterprise ILMS focus the pay relationship on internal equity with only
general pay levels dictated by market rates. Craft ILMs’ goals are more focused on
external competitiveness due to external hiring at all levels and the need to acquire a
minimum level of skill in new hires, that will demand a premium.

This research uses craft and enterprise ILMs as models of the employment
systems existing for highly skilled and less skilled computer professionals, respectively.
Highly skilled computer professionals generally fall into the craft ILM category because
they are typically responsible for their own training and have more bargaining power
with employers. Less skilled IT professions, such as data entry, are more likely to be part
of an enterprise ILM as this type of work arrangement is more common for blue

collar/clerical work.

35

I T Professionals in Internal Labor Markets

The IT profession covers a wide range of skill levels and a variety of technical
areas. When considering whether an IT job family would be part of a craft or enterprise
ILM, it is important to look at the specific duties and responsibilities of each family
separately. Some IT job families may be better suited for a Craft ILM, while other IT job
families may be more appropriate in an Enterprise ILM.

Highly skilled families such as software systems engineering or
programming/analysts are more likely to be part of Craft ILMs. These job families
require significant amounts of knowledge before employees may be productive.
Software systems engineers create programs that allow computer operating systems to
function. These professionals are responsible for ensuring the continued operation of a
computer network or telecommunications network They work with hardware changes
and multiple operating systems to provide end-users with continuous access to software
applications. Applications analysts and programmers design software applications based
upon end-user requirements. These professionals write, test, and debug the coding to
create these applications. These coding structures can quickly become complicated and
the computer languages involved in these jobs may change rapidly. The skills required to
perform in either a software systems engineering or an applications programmer/analyst
job family are complex and take time to learn.

Employees working in these capacities are likely to have learned their trade
before going to work at an organization. Because the skills change so quickly, firms are
unlikely to want to provide internal training on them. Also, since computer languages

and hardware are not firm-specific there is no benefit to the firm for training these

36

employees (Becker, 1993 ). They can easily take these skills to other employers and the
original firm would lose the training investments. Professionals skills learned by the
individual are one characteristic of a craft H.M.

Further, internal promotion policies in these families may be less beneficial to the
firm. As the skills sets change with new technological advances, employers may wish to
find new employees with newer skills. If the company relied solely on internal
promotions, external sources of new skills would be very limited. A craft ILM allows
firms to hire higher level professionals from outside of the firm.

Other less skilled IT professions, such as data entry operations, may be more
appropriate in an Enterprise ILM. Data entry operations are responsible for keying
information into databases or other organizational systems that track and use data.
Employees seeking work in this field do not need any foreknowledge of data entry. All
of these skills may be taught shortly after beginning work through on-the-job training.
Firms may choose to include data entry in an Enterprise ILM because there is little to be
gained from purchasing specialized data entry skills in the external market. The internal
promotion policies of enterprise ILMs may be beneficial for data entry operations
because there is usually institutional knowledge acquired over time that would be helpful

to the organization to retain.

Neoclassical Economic Models in ILMS
Traditional, neoclassical economics asserts that differences in pay, across firms,
for individuals participating in the same external labor market can be explained by

considering the interplay between labor supply and demand in competitive (external)

37

labor markets. Neoclassical economics predicts workers and companies will renegotiate
their work relationships every time period. Perfect information allows both parties to
know the “market” rate for a job and the goodness of fit between a potential employee
and a job. This is in direct contrast with ILMS that are systems, which protect employees
from the vagaries of the external labor market. ILMs provide more continuity of work
arrangements from time period to time period rather than the market mediated work
systems described in neoclassical economics.

While the current study uses primarily ILM theory to discuss wages and
promotional opportunities of employees, economic theories must be considered as they
introduce the external labor market into the wage setting process for employees who
participate in the same external labor market. Specifically, three economic wage theories
that will be addressed here are compensating wage differentials, efficiency wage theory,
and human capital theory. These theories co—exist with the ILMs and will be integrated
with ILM theory as acting forces throughout all of the pay relationships discussed in this
paper.

Compensating Wage Differential theory predicts that some firms, whose terms
and/or conditions of employment are less agreeable, will need to pay their employees
more than other frrrns pay, that is, above the market clearing rate (Rosen, 1986). The
negative work characteristics related to compensating pay differentials may be anything
that would make the job unattractive to people at the regular market rate such as safety
concerns, unusual hours, or a higher risk of unemployment. The worse the working
conditions of a job the higher the compensating wage differential will need to be to create

a labor supply. Firms may choose to offset these negative conditions through purchasing

38

better technology to alleviate the problems for workers or through higher wages. The
higher the compensating wage differential is for a particular negativity, the more people
will be willing to do the job. Employers will then pay a differential high enough to create
an adequate supply of workers. In the context of this study, the theory of compensating
wage differentials predicts higher pay rates for IT CS, than for comparable regular
employees because of their higher unemployment risks and/or the lack of non-pay
benefits. Further, regular employees in ILMs would probably not receive compensating
wage differentials, but the hypotheses presented in the next chapter will suggest that the
introduction of IT CS may create a need for firms to pay such a differential.

Efficiency Wage theory predicts that some firms will pay above market clearing
rate, that is more than the rest of their (external) labor market competitors, in order to
achieve other objectives such as lower turnover, lower shirking, and/or higher
productivity (Akerlof & Yellen, 1986). Under this theory employees will recognize that
their wages are higher than they would receive in other firms. They will not want to
voluntarily leave these jobs because they couldn’t earn equal money in other firms. The
potential threat of losing a high paying job creates a self-enforcing contract, where
employees have little incentive to shirk (Carmichael, 1989). Further Akerlof ( 1982) has
developed a sociological model that predicts firms paying higher wages can induce
higher productivity in workers. Lower turnover is key in the IT profession, where labor
supply is short of demand Generally, efficiency wage theory predicts higher pay rates
for regular IT employees than for comparable IT CS because firms are interested in long-
terrn retention and motivation for regular employees but not for contingent workers. As

explained in the next chapter, this study suggests that there may be less need for an

39

efficiency wage for regular employees in enterprise ILMs when firms are using
comparable ITCS.

Human capital theory (Becker, 1993; Hutchens, 1989) predicts that peoples’
wages will differ due to differing education and work experience among individuals.
Higher wages for more complex skills should motivate individuals to expend time, effort,
and/or money to acquire skills with greater value in the labor market. Human capital
theory, however, distinguishes between general and firm-specific skills. General skills
are equally applicable in any firm and will consequently be equally valuable for
employees in any place of employment. Employers will not be inclined to provide
training for general skills unless employees’ wages can be lowered to offset the costs of
this training. In contrast, firm-specific training provides employees with skills that will
only be valuable in the firm that is providing the training. Employers will be more likely
to cover the costs of firm-specific training as these skills will not be available in the
external workforce. Further, firms may pay employees trained in firm-specific skills
more than employees without these skills as a means of retaining them and not losing
their training investments. In the context of this study, Human Capital theory helps
explain why some IT job families are in enterprise or craft ILMs. Human Capital theory
also helps explain why regular employees in all job families may be paid more than
comparable IT CS for their firm-specific and institutional knowledge.

Each of these three theories provides reasons why firms would pay some
employees more than other comparable employees/IT CS. However, they all focus on a
unique factor of the work relationship. It is important to recognize that all of these

factors are working simultaneously when work arrangements are determined. Therefore,

40

these theories by themselves may be limiting. As described in the previous sections, the
ILM framework used in the current research encompasses many factors internal to firms
and may be combined with these economic theories to provide a more complete
explanation of how work relationships are created The next section describes some

research, which combines both the internal and external labor market factors.

External Labor Market ’3 [M on Internal Labor Markets

Organizations do not operate in a vacuum and even the most sheltered employee
is aware of major changes in the employment environment. Companies that have
implemented extensive ILMs insulating employees from market forces will still be
impacted by changes in the supply of labor outside their firms and other factors effecting
the health of their firms. Several researchers have studied the effects of the external labor
markets on firms’ employment practices and employees’ wages and job security (Bills,
1987; Fay & Medoff, 1985; Grimshaw & Rubery, 1998; Moss Kanter, 1984; Pfeffer &
Baron, 1988; Rebitzer & Taylor, 1991).

This study builds on the discussions of Pfeffer and Baron (1988), who suggested
that the externalization of work through contingent workers will have multiple
implications for the regular employees of the organization. First, for those employees
managing the contingent workers there will be a new set of skills required to effectively
use these workers to the benefit of the firm. They will have to coordinate and integrate
the efforts of the contractors and the regular employees. Because the mechanisms of

control are different for these two groups the managers’ skills will need to expand and

41

adapt to meet the requirements of the additional complexity associated with multiple
work arrangements. Second, the regular employees will be continually exposed to
external labor markets. In a traditional ILM, employees are only confronted with the
external labor market at ports of entry. With employees constantly working with
contractors, they will be continuously reminded of the external labor market. Research
evidence exists to show that these reminders and contacts in the outside world make
employees more likely to leave their current employer for other opportunities
(Granovetter, 1986). This potential to leave should force employers, who are using
contingent labor, to treat their employees more carefully. Also there is potential for the
presence of these contracted “outsiders” to alter the balance of wages, power, and
privilege in the organization (Pfeffer & Baron, 1988) because they are paid on a different
scale and have different motivations than regular employees. As regular employees
become aware of these alternative arrangements they may have second thoughts about
their own work arrangements. Therefore the interaction of regular employees and
contractors may result in a reduction in regular employee satisfaction or a change in the
firms’ pay levels in acknowledgements of the dichotomous work arrangements.

Bills (1987) studied three organizations that had craft ILMs. Two of the
companies had significant commitments to training. The first, a manufacturing firm, paid
below market wages, but had an extensive training program taking entry level employees
to a journeyman level in a craft. It had significant turnover at the journeyman level
because of the lower wages. This company recognized it had high turnover, but accepted
this as the trade for lower payroll costs. The second firm, a hospital, allowed new nurses

to rotate through all of the hospital’s areas to help them decide on an area of

42

specialization. The third, an engineering firm, relied primarily on increasing the
employees’ professional status as they became more proficient in their areas of expertise.
When these companies faced economic distress the manufacturing firm and the hospital
had to reduce the amount of training offered, while the engineering firm was able to
maintain its relatively lower cost policies. This finding confirms that ILMs are expensive
commitments to employees and should be balanced with the external labor market needs
of the firm. Ifthe external labor market has a large supply of workers for a job, there
may not be as great a need for an ILM than if the external labor market supply is tight.
Firms must be able to determine for themselves if the costs of their ILM are worthwhile
given the current and probable future external labor market supply.

Another external factor impacting the operations of an ILM is the level and
variability of product demand. Lower product demand, not surprisingly, will force most
companies to lower their employment levels. Fay and Medoff (1985) found that lower
product demand does result in a lower employment level. However, the level of
employment during the economic “trough” is not as low as would be predicted by the
production needs of the organization. In fact, they found that most companies hoard
some workers through these difficult periods to retain company knowledge. If
workloads will also vary over the course of a company’s business cycle. Similar to lower
production staff during times of peak product demand, companies may lower IT staff
during periods with low demand for IT work

Rebitzer and Taylor (1991) developed an economic model relating product
demand to the simultaneous use of primary and contingent workers. Primary employees

are paid at higher wages to incent them to put forth significant work effort and not shirk.

43

Theoretically the higher wage will make it sufficiently unattractive to them to lose their
jobs that they will work harder to avoid being fired. The size of this “no-shirking”
premium will increase as the probability of being laid off in the next time period
increases. The model predicts that firms will attempt to employ enough primary workers
to cover their production needs during times of low product demand. However, firms
will be better off using contingent staff to supplement these primary workers when
product demand is higher because the premium for contingent labor is smaller than that
of primary workers facing the potential for a layoff in the next time period. If product
demand was highly variable or unpredictable companies would probably be more likely
to keep the level of primary employment lower throughout the business cycle and
increase the use of contingent labor as needed. Their conclusion is that firms facing
moderate to high product demand will employ a combination of primary and contingent
workers simultaneously. Unfortunately, Rebitzer and Taylor’s (1985) model does not
indicate which jobs will be held by the primary and contingent workers. This model may
apply to IT workers in the sense that some IT functions are more critical to the firrn’s
operations than others. Daily operations such as accounting programs, production
control, and inventory management require constant attention from IT professionals.
Other IT work such as data warehousing activities, which do not directly effect daily
operations may be expendable when human resources are in short supply. IT
professionals, who work on the critical IT functions of a company, may be hired as
primary workers, who will be carried on as employees even through times of low demand

for IT work. Contingent IT workers, who are working on special projects or during

 

 

periods

belet.

intera
emph
that v
the er
hill 1
prom
skills

are Ie
the c:
mod:
pent
C1511
that
indi

bett

periods of high demand for IT work, may be hired during times of high demand, but will
be let go when such increased demand for IT work declines.

Another pair of researchers, Grimshaw and Rubery (1998) have speculated on the
interaction of the skill specificity required for a job and the bargaining power of the
employees as determining factors in the development of an ILM. These authors conclude
that when a job group has high skill specificity and a low supply of potential workers in
the external labor market, as is the case currently with computer programmers, employers
will implement ILMs as a means of retaining employees through higher wages,
promotion opportunities and training. When a job group requires highly firm specific
skills and the external labor supply is high, both employers and employees try to develop
an ILM that will make sense to both parties. In cases of low skill specificity, employers
are less likely to develop an ILM, but if the employees have more bargaining power, as is
the case when the labor market is tight, they may force the company into an ILM. The
model created by Grimshaw and Rubery also suggests that an ILM is shaped by the
performance of the organization, through its ability to provide an ILM, and by firm
customs and practices, which may or may not support an ILM The implication here is
that firms must be aware of their environment and that these external factors will
indirectly and possibly directly bring about changes in the employment relationships
between themselves and their employees.

Pfeffer and Baron (1988) alluded to these potential effects of using contractors
simultaneously with permanent workers over ten years ago and as of yet there have been
little or no research efforts to measure these changes. Popular literature has discussed the

plights of contingent workers, but at the same time very little has been said about the

45

 

 

porch

in ter
chapt
perm

work

potential problems or benefits these contingent workers may impart on regular employees
in terms wages and career opportunities. The hypotheses and data presented in the next
chapters look specifically at possible changes in wages and promotion opportunities for
permanent employees as a result of the concurrent use of contract and permanent

workers.

46

 

 

 

 

Tech
cond

oork

«rip
anph
isexp
hater
llhls
and pr
PWSCI
incoq
theon'
dealt

them}

film i

the pf:

CHAPTER 3 - HYPOTHESES

The hypotheses presented in this chapter study the impact Information
Technology Contingent Staff (IT CS) have on the wages and internal labor market
conditions of the regular, IT workers. Since IT CS are a large portion of the contingent
workforce, they are a good group to study to gain understanding of the possible
ramifications of these contingent work arrangements.

These research questions will examine the impact using IT CS has on the
compensation, salary structures, and promotions of regular Information Technology (IT)
employees in Internal Labor Markets (ILMs). Throughout this chapter, each hypothesis
is explained from the perspective of the anticipated impact the presence of ITCS will
have on employees in craft or enterprise ILMs. The contrast between craft and enterprise
ILMs is descriptive of the differences in the rules and norms that govern compensation
and promotions for the various IT job families studied here. The first six hypotheses sets
presented here focus on compensation for IT employees. These six hypotheses sets
incorporate the perspectives of compensating wage differential and efficiency wage
theories in addition to the ideas of ILM theory. The last two sets of hypotheses, which
deal with promotions and salary structure, respectively, are based exclusively on ILM
theory.

Most researchers in the area of contingent work arrangements have discussed how
these work arrangements effect the ITCS workers themselves. The literature presented in
the previous chapter shows that IT contingent workers typically earn a higher rate of pay

than most contingent workers, but they have lower job security than regular employees

47

 

 

exp:
worl
there
mth
our 1
empl

regul

trainr
com;
data 1
Work
bane
may

knor

5.151:

teChr

1111mm

(Williams, 1989). ITCS are faced with additional costs because they rarely receive
healthcare or retirement benefits. When companies use contingent workers, however, it
is not just the status of the contingent workers that is altered. Regular, IT employees are
expected to work together with ITCS to achieve company goals. To understand how the
work place and work relationships are changing as a results of contingent arrangements,
therefore the situations of both the contingent staff and the regular employees working
with them should be studied. The following hypotheses have been developed to increase
our knowledge about the effects of using ITCS on the careers of the regular, IT
employees of the company and how the type of ILM, craft or enterprise, in which the
regular employee belongs, may cause differences in these ITCS effects.

There is a range of job families involved in Information Technology work. The
training and knowledge required for work in these job families varies widely in terms of
complexity and rate of technologically-driven change. Jobs such as those involved in
data entry represent the low end of complexity and rate of technological change in IT
work. Skills for these jobs require less training and change more slowly. Firms may
benefit fiom enterprise ILMs for these low-complexity IT jobs because enterprise ILMs
may help firms protect their investment in firm-specific human capital: the institutional
knowledge employees gain through on-the-j ob experience.

In contrast, jobs that involve developing, modifying, and maintaining computer
systems and/or applications represent the high end of complexity and rate of
technological change in IT work The skills required for these jobs represent general
human capital and firm-specific, institutional knowledge. The attainment of the general

human capital skills is probably the responsibility of the professional, as firms are not

48

 

 

lileh
than.

from

Clllfl'l

their L
Works
With e
01 do
these

ral‘idl

less p;
“1d or

”101? h

likely to invest in general (transferable) human capital (Becker, 1993). These skills also
change rapidly as a result of the myriad technological innovations in IT such as the shift
from mainframe to client/server systems and the advent of fifth generation programming
languages. Firms may benefit from craft ILMs for these hi gh-complexity IT jobs because
craft ILMs may allow firms to retain a base of regular IT employees with a blend of
general human capital (IT technological knowledge) and firm-specific human capital
(institutional knowledge), while having the ability to attract employees with current,
state-of-the-art general human capital (IT knowledge).

As explained below, the impact that using IT CS has on compensation, salary
structures and promotions of regular IT employees is predicted to differ for craft and
enterprise ILMs.

IT Contingent Staff are often singled out as a special category of workers because
their unique skills give them an advantage in the labor market that traditional temporary
workers (i.e., clerical workers) do not enjoy. ITCS often possess more bargaining power
with employers than other contingent workers because their skills are not easily learned
or developed in-house (Abraham, 1990). Firms’ decisions to “buy” rather than “make”
these skills may result from a combination of the expense to “make” them and that
rapidly changing technology may quickly render these skills obsolete (Deavers, 1997).
These conditions may cause firms, which choose to internally train their workers, to get
less payofi‘ for their training investments. Since the costs of “making” IT talent are high
and the payoff period relatively short in comparison with other types of jobs, firms are

more likely to “buy” If talent than other skills. IT CS is one source to “buy” IT talent and

49

its use should then be expected to rise at least as quickly or possibly quicker than

contingent work arrangements for other jobs.

Hmtltesir Set I: The m diflerential between ITCS and rfllar IT employees

Wages are one of the main means employers have to attract and retain employees
and it is commonly believed that ITCS are paid more than regular, IT employees. The
theory of compensating wage differentials predicts higher wages for ITCS than for
regular, IT employees because a premium must be paid to these IT professionals to offset

the greater job insecurity and fewer healthcare and retirement benefits. Most regular
employees in ILMs have healthcare and retirement benefits, a variety of employee
welfare benefits (e.g., paid time-off, child care, etc.), and a virtual guarantee of a job in
the next period As firms offer more of these non-cash benefits to their regular
employees, they may be allowed to pay their workforce at or even below market-clearing
wage levels. These factors lead to the expectation that employers will need to pay a
premium to compensate for all of the differences in the work relationships for their ITCS
or lose them to another firm.
However, the relationship between the wages of the ITCS and the regular
employees may vary depending on whether their job classification is part of a craft or
enterprise ILM. The compensating wage differential for craft ILM employees should be
greater than it is for enterprise ILM employees. Craft ITCS have a higher skills, which
provide them with more bargaining power with employees. In contrast, there is a
relatively ample supply of people, who can be readily trained on-the-job for the work in

an enterprise ILM. Additionally, efficiency wage theory predicts that firms will pay a

50

premium to the regular employees in an enterprise ILM to protect their investment in
firm-specific training. The combination of these factors suggests that the differential
between regular employees and ITCS will be greater for craft job families than enterprise
job families.
Hypotheses 1a, 1b, and 1c test for differences in the cash pay premiums for IT CS
in craft and enterprise ILMs.
H I a: The wages of I T CS in crafl job families are greater than the wages of regular, IT
job families in craft ILMs.
H] b: The wages of I T C S in enterprise job families are greater than the wages of regular,
I T job families in enterprise ILMs.
H 1 c: The premium for craft I T CS will be greater than the premium for enterprise I T CS.
The remaining hypotheses focus on how the actual ILM practices imposed on

regular, IT workers change with the use of ITCS.

H Set2: l ee di erencesbetween mnsusi ITCSand
finnsnotustmlTCS.

ILMs have been characterized by wage structures determined by internal equity
and internal job promotions with less emplmsis on external labor market forces and by
internal job promotions (Doeringer & Piore, 1971). Enterprise and craft ILM wage
policies and structures are developed to maintain internal equity among employees. The
external labor market is given consideration during the hiring, salary budgeting and
structure development processes, but it is not as important as ensuring that all employees

are paid relatively fairly for their contributions to the organization. This policy implies

51

that wages will generally reflect the open market, but will not be directly tied to it.
Furthermore employees in an ILM are afforded many non-cash benefits. The more
notable of these benefits are an implicit promise of employment during the next time
period and healthcare and retirement benefits. The amount of such non-cash offerings
may be negatively related to the cash wages firms must pay to employees. Firms that
offer rich benefits’ packages may pay their employees less than market wages as part of
the trade for increased job security and other non-pecuniary benefits.

The introduction of ITCS into the workplace may alter the relative importance of
internal equity in wage determination. Pay practices for ITC S are based more directly on
external market forces than ILMs’ pay practices. ITCS’ pay will need to start at the
market-clearing rate for the job and risk-related premiums may be added to this rate. As
stated in the previous section, these premiums would compensate the ITCS for the extra
risks associated with being contingent, such as not receiving healthcare or retirement
benefits, and the higher risk of unemployment

ILM systems rely on internal equity and the use of IT CS, who are being paid
more than the regular employees, may pose problems for maintaining equity in wages
among the regular employees. This may be especially bothersome in companies
employing ITCS for the same tasks as their regular employees. In order to maintain
equity, organizations employing ITCS may need to adjust their regular employees’ wages
to more clearly reflect market levels.

Because internal labor market policies may vary by job family (Baron, Davis-
Blake & Bielby, 1986; Osterman, 1984; Spilerman, 1977), it’s plausible that the impact

of IT CS on permanent employees’ pay will also differ by job family. Internal labor

52

market theory predicts companies will want to retain employees in craft and enterprise
job families over multiple time periods. Employees in jobs that require significant skill,
initial training to be productive, or can provide training to others are typical of craft H.M
jobs. Employees in jobs that require significant firm-specific and institutional knowledge
but little or no skills upon entry in the company are typical of enterprise ILM jobs. Both
of these job types may be covered by an ILM, where long-term work relationships are a
goal. However, because the skill sets of people entering enterprise ILMs are minimal,
there is a larger potential applicant pool, making these employees easier to find and less
costly to replace, the secrnity of these employment relationships over multiple time
periods may be diminished. Therefore, firms’ reactions to the ITCS influence may be
different for craft and enterprise ILM employees.

If a company is going to maintain a sense of equity for its craft employees, then
the pay rates of the ITCS must be factored into the wage setting processes. ILMs’ wages
are focused on internal equity, but if employees work side by side with ITCS, doing the
same job and are paid less than the ITCS, the importance of internal equity will be
diminished If ITCS in craft jobs are paid significantly higher due to compensating pay
differentials for greater unemployment risks and lower benefits, then permanent
employees may become dissatisfied with their situation and demand similar wages or
leave the organization to pursue higher paying contracting options on their own. Leaving
the organization is an easier option for most craft workers because this work is
characterized by a substantial level of general human capital, which is valuable to many
firms. Since craft employees also possess significant firm specific knowledge not held by

ITCS, companies should want to retain them over a long period of time, it is important

53

for companies using ITCS to address the inequity concerns of these craft workers. The
introduction of craft, IT CS workers has increased the exposure of regular, craft, IT
workers to external labor market conditions. Efficiency wage theory, which focuses on
paying above market to reduce turnover, supervisory costs and shirking risk for the firm,
would predict an increase in pay for regular craft workers because they are in the long-
tcnn relationship of the ILM. The expectation then for craft job classifieations would be
for employers to increase pay rates for regular employees to closer approximate, although
probably not completely matching the ITCS’ rates. This may be a more cost effective
option than paying the transaction costs of continually replacing highly skilled workers,
who have substantial levels of general human capital, and investing in the firm specific
training of these new hires.
For employees in enterprise job classifications working side by side with ITCS,
companies may choose a different approach. While enterprise employees have firm-
specific knowledge, they may not possess much professional knowledge (general human
capital), which requires exterml training. In addition, internal training costs for these
employees are minimal because of the simplicity of the job. Generally, employers are
less concerned with maintaining a continuous employment relationship for these job
classifications. Therefore increasing the pay rate for regular employees in an enterprise
ILM to simulate the IT CS’ rates may not be necessary. Many organizations may decide
for enterprise jobs to pay a wage more consistent with internal equity regardless of the
IT CS ’ wages. The cost of replacing these workers, if they decide to leave the
organization, will be less than the cost of paying them higher wages. Further, regular

enterprise employees in firms, which use ITCS for enterprise jobs, may face an added

54

disadvantage in that their replacements are readily available in the existing IT CS pool of
labor. The ready labor supply may even induce firms to lower the wages of their
enterprise employees below the wage levels required for internal equity. These enterprise
employees may not be aware of the IT CS’ wages, but should they find out about them
and decide to leave, employers will probably be able to quickly replace them with an
ITCS. As long as there exists an adequate supply of ITCS, firms have a motivation to
lower regular wages as much as possible until their supply of replacements dwindles.
H2a: The pay level of regular, crafi, IT job families will be higher in companies that use
I T CS for some of their craft work, than the pay of regular, crafi, ITjob families in
companies that don ’t use I T CS for any of these services.
H2b: The pay level of regular, enterprise, ITjob families will be lower in companies that
use I T CS for some of their enterprise work, than the pay of regular, enterprise, I T job

families in companies that don ’t use I T CS for any of these services.

H Set3:Theextento ITCSuseandr ee

Hypotheses 2a and 2b concern the differences between firms that use and firms
that do not use IT CS. This approach assumes that the mere exposure to external labor
market conditions, through the hiring of at least one contingent worker, may be enough to
motivate firms to adjust the pay rates of regular employees. It is also possible that the
extent to which firm use IT CS will be related to any change in regular employees’ pay
rates. Ifthis is true, then using ITCS can be considered a continuous variable ranging
fi'om 0% for companies that do not use any ITCS in a particular area to 99% for

companies that use ITCS for almost all of an IT work area. For companies that contract

55

out 100% of an area of IT work, the impact these contractors have on regular employee
pay is a moot point. The following hypotheses test the continuous variable approach for
IT CS on permanent employees’ wages.
H30: The extent of a company ’s crafi work completed by I T CS will be positively related
to the pay level of its regular, craft, ITjob families.
H3b: The extent of a company ’s enterprise work completed by I T CS will be negatively

related to the pay level of its regular, enterprise, I T job families.

° Hmtbesis 4: ITCS' use and ram eMgee M variabm.
One of the features of ILMs is an emphasis on internal equity. As explained in

the literature review, the presence of IT CS is likely to reduce the importance of internal
equity in pay administration. By enforcing a policy of internal equity, employees
performing the same duties and responsibilities are paid similarly. When IT CS are part
of firms’ strategy, their attention to internal equity is likely to be reduced. Therefore it is
anticipated that the variance in pay within job families will be greater in firms with ITCS
than in firms without ITCS.

H4: The variation in pay for regular 1 T job families will be greater in companies that

use I T CS than in companies that do not use I T CS.

Hypothesis Set 5: Bitter: ence between “core” and fgerlghflz” industries in the M

co es 0 ITCS use.
Another factor, which may contribute to the impact ITCS wages have on regular

employees’ pay is the industry of the employer. By changing the level of analysis from

56

the industry level to the job family level as is being done in this research, the impact of
industry membership is not explicitly considered. While the majority of ILM explanatory
power may lie at the job family level, the industry segment of a firm may still have
significant impact on ILM policies and labor relationships in general.

Initial research on ILMs assumed that these human resource systems were only
applicable in “core” industries (Baron, 1984; Beck, Horan & Tolbert, 1978; Doeringer &
Piore, 1971). “Core” industries are characterized by larger firms, more stable product
demand, and a less competitive market than “peripheral” industries. “Peripheral”
industries face more volatile competition and product demand, making it difficult for
firms to accurately predict labor needs. ILMs are costly to maintain and may be an
extraneous expense for firms in “peripheral” industries.

This “core/peripery” distinction is helpful in assessing industry impact on a
variety of outcomes. For purposes of this research it may be more useful to tailor the
core/periphery distinctions with a specific focus on the roles that IT skills have in each
industry. Industries will be defined as “core” if IT skills are central to their profit
producing operations. For example, communications, and manufacturing of control
devices and medical instruments are industries that can be considered as core from the
standpoint of IT work Industries will be defined as “periphery” industries if they use IT
skills, but they are not critical to the main operations of these industries. For example,
trade contractors and agricultural services are industries that can be considered as
“periphe ” fi'om the standpoint of IT work.

“Core” industries’ more stable IT demand may give companies the ability to

predict more accurately the number of IT workers needed. Because “core” industry firms

57

have more stable IT labor needs, they should be able to protect themselves from changes
in the supply of IT labor better than “peripheral” industry firms, which are unable to
predict with accuracy their future IT labor needs. Further, “core” industry firms may
offer higher job security and more of an emphasis on long-term employment than
“peripheral” industry firms. This greater stability and ability to predict future IT labor
needs should reduce the necessity of hiring ITCS, especially on short notice to meet
unexpected workload demands. Therefore, core industry firms then are less likely to use
IT CS, which means that the higher regular employees’ pay rates that are predicted in
Hypotheses 2a and 33 by compensating wage differential theory will be less necessary.
For craft, IT workers in “core” industries, higher job security means that the
compensating wage differential for the higher job risk of working with IT CS will be
reduced, but an efficiency wage premium to reduce supervisory costs and/or shirking
behaviors may be more common due to the long-term focus of the work relationship.
Craft, IT workers in “peripheral” industries may have lower job security because of
“peripheral” industries’ greater variability in IT demand. This volatility may result in a
greater use of ITCS and therefore more interaction of their employees with IT CS. The
"periphera " craft, IT employees will be more aware of ITCS opportunities and may be
more likely to take advantage of them (Granovetter, 1986). “Peripheral” industries will
probably need to address craft, 1T employees’ higher risks of unemployment and desires
to leave by paying compensating wage differentials more than “core” industries.
Efficiency wage premiums, however, may not be feasible in “peripheral” industries
because the higher wages may impose prohibitive cost on an already volatile IT demand.

Any difference in wages for craft, IT employees in “core” or “peripheral” industries

58

should indicate that the magnitude of either the compensating wage differentials in
“peripheral” industries or the efiiciency wage premium present in “core” industries is
greater.

For enterprise workers in “core” industries, there will be little need for a
compensating wage differential as the risk of unemployment is low. The use of
efficiency wage strategies may also be unnecessary because supervising this group is
simpler than craft workers, making it unnecessary to pay these enterprise employees
more to avoid shirking. In “peripheral” industries using IT CS for enterprise work, the
compensating wage differential paid the regular enterprise employees may be minimal
since long-term retention, while valued, is not critical. Efficiency wage premiums are
probably not necessary for enterprise employees in “peripheral” industries because
supervisory costs are minimal for this group. It is predicted that for enterprise workers
the compensating wage differential of the “peripheral” industries will cause the wages of
enterprise, IT employees in the “peripheral” industries to earn more than enterprise, IT
employees in “core” industries.

H5a: Peripheral industries will have a higher usage of I T CS than core industries,
causing wages for craft, 1 T job families in peripheral industries to be higher than the
wages of craft 1F job families in core industries.

H5 b: Peripheral industries will have a higher usage of I T CS then core industries causing
companies using I T CS in peripheral industries will pay their regular, enterprise, I T job

families more than companies using I T CS in core industries.

59

Hmthesis Set 6: ITCS use and log-term incentive to: r rmlar mlgzees.

Wages are not the only tool used by employers to retain and motivate employees.
Long-term incentives (LTI) may be used as a means of linking employees’ interests with
those of the firm. LTI plans pay employees for company performance over a three to five
year period. This long-term measurement period encourages employees to engage in
behaviors good for the long-term health of the firm instead of focusing activities on short-
tenn gains that may prove detrimental over a longer time period. LTI may be paid in
company stock or deferred compensation. By delivering the incentive in stock,
companies can further tie the employees’ interests to those of the company because the
value of the stock is dependent on firm performance where deferred cash payouts have no
tie to future firm performance. LTI awards frequently have a vesting period, which
requires the employee to remain with the firm for a pre-specified number of years before
receiving the award. This feature makes LTI awards a particularly effective retention
tool.

The use of IT CS may impact the firrn’s use of long-term incentives. If permanent
employees are continually interacting with ITCS, then they may be more inclined to see
the benefits of contingent work (i.e., more flexible hours, higher wages, etc.) and decide
to leave the company. As part of an efficiency wage strategy to reduce turnover, firms
using ITCS may be more likely to use long-term incentives. An employee eligible for a
LTI that is not payable until some specified time in the future will have an added reason
to stay with the company than an employee not eligible for LTIs. As explained above,
the use of ITCS may increase significantly the risks of turnover among regular employees

in craft IT jobs because of their substantial level of general human capital. Consequently,

60

it is anticipated that more craft job families in firms using ITCS will be eligible for LTIs
than craft job families in firms not using ITCS. In contrast, the risk and impact of
turnover by regular IT employees in enterprise IT job is not as great when they are
exposed to ITCS. Therefore, no increase in LTI eligibility is expected for enterprise job
families in firms using If CS. Offering LTI at the level of enterprise job families may be
the result of a firrn-wide philosophy of equality that the use of ITCS will not change.
H6a: Companies that use I T CS for crafi, IT work will be more likely to ofler long-term
incentives to regular, crafl, ITjob families than companies that do not use I T CS for craft,
1 T work.
H6b: Companies that use I T CS for enterprise, IT work will be equally likely to offer long-
term incentives to regular, enterprise, ITjob families than companies that do not use

I T CS for enterprise, IT work.

Hmthesis Set 7: I TCS use and gromoa‘on [or rfllar emigrees.
In addition to wages and LTI, which are “protected” from the external labor

market, firms with ILMs have policies that give preferential treatment to internal
promotions over hiring from external sources (Doeringer & Piore, 1971). The benefit for
employees is that promotion decisions are restricted to the possible candidates inside the
company and employees are not competing against external candidates, thereby
narrowing the field of possible applicants. Most companies promote employees up a
career ladder based upon some predetermined criteria (e.g., performance or seniority).
This internal promotion policy ensures that managerial employees have a base of firm

specific knowledge (Osterman, 1987). Therefore managers in these firms are more likely

61

to understand the impact of company-wide objectives on their areas and how their work
effects processes in other parts of the company.

When companies use ITCS, they have direct access to other IT workers capable of
performing the tasks of their regular, IT employees. They may choose to hire these IT CS
into higher level positions instead of promoting their regular employees while still
incurring a nrinimurn of transaction costs and training time. For craft jobs, there are high
costs associated with hiring unqualified, non-entry level employees, firms could be
expected to hire the most qualified workers regardless of whether they were regular
employees or IT CS. Further since the skills associated with craft IT jobs change rapidly
with new technological developments, external sources of employees may offer
companies more up-to-date skills than internal promotions. Hiring from ITCS would
allow companies to acquire up-to-date skills without investing in expensive recruiting
and training efforts. Therefore companies that use ITCS may hire IT professionals from
outside instead of promoting from within to get the newest skills. Because these firms
are hiring IT CS, there will be fewer promotion opportunities for regular employees. In
firms that do not use IT CS more of the promotion opportunities will be available for
internal hires because there is not a ready supply of replacements. Therefore the average
promotion rate of companies using ITCS for craft IT job families should be slower than
in companies not using IT CS.

Regular employees in enterprise job classifications typically are covered by the
internal promotion policies of ILMs that are even more exclusive than craft ILM policies.
However, if a firm wanted to hire an ITCS, the transaction costs would be lower than if it

had to find a worker with no leads. Companies using IT CS for a portion of their

62

ICE:

IQ

rd:

enterprise jobs may consider both regular workers and ITCS for the position because they
are easy to inform about the position and these employers already have information about
the work habits of both groups of people. Training for these jobs are minimal making the
firm indifferent to promoting an employee or IT CS into an open position The lower real
and potential costs associated with IT CS do pose a threat to the exclusivity of a
traditional enterprise ILM promotion policy. However, firm specific and institutional
knowledge will most likely be greater in the regular employees, leading companies to be
theoretically inclined to promote regular employees rather than hire IT CS. By having
ITCS working on enterprise tasks, firms will have hired fewer regular, 1T employees and
thus the pool of potential applicants will be smaller thereby increasing the chances of a
regular employee to be promoted.
H 7a: The extent of using I T CS for crafi, 1T work will be positively related to the average
number of years until promotion for employees in regular, craft, ITjob families.
H 7b: The extent of using I T CS for enterprise, I T work will be negatively related to the
average number of years until promotion for employees in regular, enterprise, I T job

families.

Hmhesis Set 8: Extent at ITCS gs; and salter structure range M2 [or mlar
memes

Employees, who are promoted, usually receive significant increases in their wages
to compensate for the higher level of responsibility that comes with their new jobs. The
wages for both the old and new jobs of the employees are based upon guidelines

established as part of salary structures. Firms measure the internal and external equity of

63

jobs and determine ranges, which indicate the minimum and maximum they are willing to
pay for jobs in order to best balance the demands of both types of equity. Jobs that differ
in responsibility and scope will belong in different ranges. Jobs that are sequential to
each other in a career progression will often belong in adjacent pay ranges for which
there will probably exist some overlap from the minimum of the higher range and the
maximum of the lower range. More overlap in adjacent ranges means that employees in
the lower range have greater potential to earn the same or more than employees in the
higher range. The higher wage rates given for promotions are supposed to act as an
incentive to take on the added responsibility of the promotions. However, if the overlap
between adjacent grades is significant, then employees in the adjacent grades may be paid
close to each other and the raise earned for promotion will not be as great, diminishing
the incentive for promotion. In situations where employees are less likely to be
promoted, it may be good strategy of firms to design their pay structures to minimize the
benefits of promotion. Conversely, in situations where employees are more likely to be
promoted, it may be good strategy of firms to design their pay structures to maximize the
benefits of promotion Because of the reasons described in the Hypotheses 7a and 7b,
craft employees may be less likely to be promoted and this means that firms should create
salary structures for these job families with significant overlap. Enterprise employees
may be more likely to be promoted, which means firms should create salary structures for
these job families with less overlap.

H80: The extent of using I T CS for craft, 1T work will be positively related to the amount

of overlap between adjacent salary ranges for craft 1 T job families.

H8b: The extent of using 1 T CS for enterprise, IT work will be negatively related to the
amount of overlap between adjacent salary ranges for enterprise ITjob families.
These hypotheses study the relationships that using contingent IT workers has on

the pay and promotions of regular IT employees in an organization They are designed to
more closely examine the relationship between contingent staff, employees, and
organizations at the job family level, which research indicates is a more appropriate level
to measure companies’ ILM policies and practices (Baron et al., 1986; Osterman, 1984;
Spilerman, 1977). Further, the theoretical ideas associated with ILMs, in combination
with Compensating Wage Differential Theory, Efficiency Wage Theory, and Human
Capital Theory suggests that ITCS may have differing impacts on craft and enterprise
jobs. The hypotheses presented here are developed to test the these theories and to assess
how they adapt to the introduction of the external labor market in the form of ITCS. The

next chapter will detail the data and methodology for testing these hypotheses.

65

CHAPTER 4 — DATA AND METHODOLOGY

M

The data for these analyses come from an employee compensation survey
administered to companies in 1998 and 1999 by William M. Mercer, Incorporated. The
1998 sample consisted of 994 firms reporting on the pay practices for 2,829 job families
with 92,163 incumbents. The 1999 sample consisted of 961 firms reporting on the pay
practices for 2,197 job families with 96,136 incumbents. 598 firms participated in both
the 1998 and 1999 surveys. Once the data had been collected and analysis begun, it was
necessary to combine the two waves of survey data to achieve the needed variance in the
variables to ensure finding an effect if one existed This combined sample consists of
1,357 firms with 3,780 witlrin-finn job families and 132, 925 employees. All of the data
was screened for accuracy by phone calls from William M. Mercer’s survey
administration team.

Human Resource professionals completed the questionnaires, which asked about
their companies’ pay practices and actual pay levels for information technology
employees. Pay practices questions asked about the use of IT contingent staff.
Specifically, companies supplied the percentage of total hours worked by IT contingent
staff and the percentage of contingent pay with respect to in-house employees for five
information technology job families: applications systems analysis and programming,
applications systems analysis, applications programming, software systems engineering
311d data entry jobs. For regular employees, data were collected, at the incumbent level,

about annual base pay and short-term incentive amounts. At the job level, data were

collected regarding the eligibility for long term incentives, such as stock options, as well
as, average promotion rates and pay ranges, which can be used to assess the internal labor
markets of these job families. (See Appendix A for actual survey questions.)

The five job families included in this study have been segmented into either craft
or enterprise labor markets, based upon their major duties, responsibilities, and
knowledge requirements. Generally, the families classified into craft ILMs require
substantial knowledge of a specific craft/profession which may take a substantial time to
acquire, training is done externally of the firm, and these skills are not specific to any one
firm. Families in enterprise ILMs do not require employees to possess any knowledge
before they begin work. Training is accomplished in-house and is largely firm-specific.
Of the five job families covered in this study job in four of them, Applications Systems
Analysis and Programming, Applications Systems Analysis, Applications Systems
Programming, and Software Systems Engineering, are classified as craft jobs and jobs in
the fifth family, Data Entry, are classified as enterprise jobs. The craft families require
incumbents to know programming languages, principles of logic and coding, and/or
information system requirements and limitations. These skills are typically acquired
through undergraduate studies in computer science and continuing training on new
languages and innovations in the information technology arena The enterprise family
requires incumbents to operate data entry devices, such as computers, and verify data
already entered into these devices. These duties may be taught easily on the first day of
employment and require no foreknowledge of the company’s data entry operations. (See

Appendix B for the complete job descriptions.)

67

To assesswages at the job family level two pay indices are calculated. The base
salary index is the average of the ratios of a firm’s average base pay to the average base
pay of the entire data sample for all of the film’s jobs in a given job family. A second
ratio is calculated similarly using total cash compensation, which is the sum of base pay
and short-term incentive payments. This ratio approach eliminates problems with units
that may occur when firms have reported different job levels within a family. Also, the
pay indices are easy to understand as values greater than 1 indicate a job family is paid

above the market average and values less than 1 indicate a job family is paid below the
market average.

The findings of this study may be applicable to only the firms and job families
captured in these samples or may generalize to a larger population of firms and job
families depending on the representativeness of the sample database. The ability to
generalize findings to a larger population depends on the external validity of the sample
(Cook & Campbell, 1979). The results of this study may be applicable to many types of
finns or job families, but because the hypotheses are only tested on a subset of firms or

10b families, the samples external validity must be tested before such generalizations
Should occur. One threat to external validity is possible selection bias in the sample.
This bias may occur when the companies in the sample are not representative of the
conlpanies in the population to which the results are intended to be generalized To test
for Selection bias, sample was compared to national statistics regarding geographical
distribution, industry distribution, and company size.

The geographic distributions of the sample was similar to the total geographic

distribution as determined by the Bureau of Labor Statistics with the exception that the

68

data sample has fewer employees on the West Coast, specifically California, than the
whole employee population (BLS, 1999a). California may be argued to have a
proportionately higher population of IT employees and ITCS because of the high
concentration of IT work in Silicon Valley. The study sample therefore may have a
lower population of employees that are being exposed to ITCS than is apparent in the
larger population. This fact should lower the possibility of finding a relationship between
IT CS and IT employee work relationships. If the hypotheses are confirmed, then it will
be despite the fact that the samples do not contain a realistic representation of Silicon
Valley firms that, at least anecdotally, are the most prolific users of IT contractors.
Industry distribution may be another threat to external validity. If the distribution
of industries present in the data sample differs from the distribution of industries in the
nation, it will make it diflicult to generalize to this larger population of employees and
not just to the subset of industries present in the sample. Generally, sample contained
employees fi'om all of the non-farm, two-digit SIC codes. The sample had
Proportionately more employees in the durable manufacturing and services industries
than is present in the national population of employees and a smaller proportion of
wl'lolesale/retail trade and public administration employees (BLS, 1999b). Industry
Seglnent is used as a control variable in these analyses and therefore the disproportionate
industry representation should not effect the results of the research, but some care should
be talren before assuming that the results from this research are applicable to the country
as a whole.
The size of the organizations in which the sample employees work may also differ

f1- Oln the national distribution of company sizes. The survey fiom which these data are

69

extracted is extremely complex. It is customary for a compensation professional to
complete the survey and therefore the set of companies adequately staffed to participate
will be generally those companies with fully formed human resource departments.
Further, the cost to companies to pmticipate in this survey is relatively high, which may
discourage smaller firms that do not have a large budget for salary surveys. It is assumed
from these two circumstances that the firms in this data sample are on average larger than
the typical company in the United States. Therefore the results of these hypotheses
should only be generalized to employees working at larger firms.

Another possible source of selection bias is in the job families being used for the
study. This study focuses on five job families in the Information Technology profession:
Applications Systems Analysis and Programming, Applications Systems Analysis,
Applications Systems Programming, Software Systems Engineering, and Data Entry.
While these families cover a broad range of duties within the Information Technology
field, there are other parts of the Information Technology that are not part of these
families. These families represent only the software portion of Information Technology

jOb families and do not include any families that are responsible for hardware
requirements, such as network operations. There does not appear to be theoretical reason
Why the results of this research would not be applicable to IT job families focusing on
hardware, however, generalization to the entire spectrum of IT or non-IT job families
should only be done with a cautionary note to the reader regarding the original sample.

In addition to selection bias, variables present in the organizations studied may

p036 threats to external validity because they may influence pay and promotion

relationships, but they are not variables of contracting use. Company size, industry, and

70

unemployment rates may all confound the study’s results. Industry effects on ITCS may
be different in high technology industries where the IT job families are more central to
the operations and profitability of the organization. This effect will be tested for in
Hypotheses 5a and 5b. Organization size has been shown to impact pay levels, regardless
of contracting practices (Baron, 1984) and unemployment should directly impact wages
through market forces (Pfeffer and Cohen, 1984). Industry segment, organizational size,
and state unemployment rates will be used as control variables in this study to isolate
their impact of these factors.

Another threat to external validity is the potential for the point in time when the
questionnaire was completed to impact the relationships found. Each of the samples in
this study was achieved through mail in questionnaires making the actual time of
completion vary by organization, although there is a common effective date on the
questionnaires to ensure all of the data represents the same point in time. The
QUCStionnaire is extremely lengthy requiring one day at a minimum for completion. It is
unlikely that there would be any one specific event that would effect all of the
Participants in such a way that would skew the study outcomes. Further, by using
Combining the samples from 1998 and 1999, any time specific events that might impact
one of the samples would not be likely to have an impact on the other sample.

The internal validity of this study also needs to be addressed. Internal validity
l'effil's to ability of the study to correctly infer that the predicted relationships between the
Studied variables does or does not exist. To eliminate possible threats to internal validity,
the Sample was examined for differences between the contracting and non-contracting

finhs on the control variables that have potential to cause similar differences in pay

71

policies to those predicted in the study’s hypotheses. Using t-tests on all of the control
variables it was found that there are proportionately fewer services firms in the
contracting group for the Applications Systems Analysis job families. The contracting
firms in the Applications Systems Analysis and Programming family has proportionately
fewer servicesrfirms, but more transportation firms. Finally, the Software Systems
Engineering job family has proportionately more transportation firms than the non-
contracting firms. Consequently, industry variables are put into the tests of these

hypotheses as control variables to ensure the effects of these differences are measured

separately from the effects of the study contracting variables.

Methods
Throughout the testing of these hypotheses the level of analysis will be the job

family. For each hypothesis set the number of cases in the analysis changes because
either the hypothesis itself resticts which cases from the sample should be included in the
analysis or because survey participants were unable to provide information on a specific
dependent variable. For example, Hypothesis Set 1 examines the differences between
regular employees’ pay rates and the pay rates of IT CS in the same firm. Therefore only
fifths that reported the pay rates of their ITCS are included in these analyses. However,
IPIYDOtheses Set 2 and 3 include all firms that reported regular employee wage rates for a
job family, so the sample sizes in these hypotheses will be much larger. More specific
detail about the samples for each hypothesis are described below. Another factor that
affects the sample sizes is missing data. For Hypotheses Set 6, 7, and 8, not all

participants were able to provide information on their LTI practices, promotion rates or

72

salary ranges. When data was not provided for a dependent variable the firm was

excluded from the analysis. These missing data will also effect sample sizes.

Hypothesis Set I .' The pay diflerential between I T CS and regular 1T employees
Hypotheses la, 1b, and 1c test the relative pay levels of ITCS versus regular IT

employees. It is hypothesized that ITCS in both crafi and enterprise job families will be

paid more than regular IT employees, but that this differential will be greater for the craft
job families. The dependent variable for these hypotheses is IT CS’ pay rates as a
percentage of the companies’ regular, IT workers base pay. Participants reported the
relative pay for ITCS in comparison with their regular employees. For example if IT CS
are paid twice as much as the regular employees, participants reported 200%. (See
Appendix for survey questions.) The comparisons are made using a t-test to estimate the
Significance of the difference in the means between the ITCS and the regular employees
pay rates. If the ITCS rates are significantly higher than 100%, which is the regular
employees’ pay rate, then Hypotheses 1a and 1b should be considered confirmed If the
t-va.lue generated by the difference between the enterprise and crafi contractor pay rates
is significant, then Hypothesis 1c will be confirmed To test Hypotheses 1a and lb only
thOse firms that reported the pay rates of their ITCS in a job family are included The pay
rates of their regular employees were defaulted to 100% because the survey question
asked for a percentage of pay based upon the regular employees’ pay rates. To test

I‘IyIJothesis lo, the sample included only those firms that reported the pay rates of their

ITCS in both the craft and enterprise job families.

73

Hypothesis Set 2: Regular employee pay diflerences between firms using ITCS and firms
not using I T CS
Employee pay can be defined by two methods. Base pay is the minimum salary

paid to an employee for completing a job. Total cash compensation is all of the cash paid
to an employee over the course of a year for all services. The difference between base
Salary and total cash compensation may include bonus payments made in addition to
regular wages, overtime payments, or any other cash given to an employee in exchange
for work In testing Hypotheses 2a and 2b indices derived from employees base pay and
total cash compensation were created. These base pay and total cash pay indices were
developed at the job family level. They represent the employee weighted average of the
ratios for all of the jobs in a job family of a firm’s average pay in a job to the market
average pay for that job. The base pay indices should be similar to the total cash indices,
except that the total cash indices should reflect any variable pay components of wages.
Difference in the results of these tests should give an indication that IT CS may be
impacting the use of variable pay.

Hypotheses 2a and 2b are tested using a moderated regression analyses. The
moderated regression approach enters all variables simultaneously including control
val‘iables, dummy variables, for each of the five job frnailies except the Software Systems
Engineering, the dichotomous variable, IT CSDICH, and the interaction terms, which are
the cross product of the job family dummy variables and ITCSDICH (Cohen & Cohen,

1 983). Both the base salary and total cash compensation indices are used as dependent
variables. The sample includes all firms that reported regular employees in a job family.

TWO regression models are performed to test both of these dependent variables.

74

Control variables in this study include industry segment (dummy coded with the
wholesale trade omitted), firm size (computed as the logarithm of total employees), and
state unemployment rates. Industry is used as a control variable because prior research
indicates that industry has a significant effect on wage levels in firms. Industry segments
are coded using the two-digit SIC codes. These codes are used because they are readily
available and cover all possible industries in the sample. Firm size is used as a control
variable because prior research indicates that larger firms typically pay ore than smaller
firms. It is necessary to pull this effect out of the variance before measuring for the

impact of IT CS on wages. The logarithm transformation is used because the scale of the
total employees responses is much greater than other scales being used in these analyses.
The logarithm transformation will reduce the amount of influence this firm size variable
has simply due to scale. State unemployment rates are also used as a control variable in
these analyses. Unemployment rates have been shown to be related to wages.
Controlling for the wage effect of unemployment seems appropriate when trying to
isolate the effects of other variables on wages. State unemployment rates are used
because the jobs in this study are probably subject to regional recruiting areas, not
national recruiting areas and state unemployment rates are a better proxy of the labor

market environment regionally than using a national unemployment rate.

I"Iodel l with Base Salary Index as Dependent Variable:

Base Salary Index = Bo + 51 (ln(Total Employment) + B; (Unemp) +
[33(Mining) + 34 (Construction) + [35 (Mfg-Durable) + [36 (Mfg-
NonDurable) + B7(Transportation) + [3. (Retail) + [3, (Finance) + [31.
(Services) + Bu (Public Administration) + [312(ITCSDICH) + [31: (ASP) +
B,.(ASB) + Bis (ASA) + Bic (DE) + Br: (IT CSDICHxASB) + Bin

75

(ITCSDICHxASA) + [3,, (ITCSDICHxASP) + [320 (ITCSDICHxDE) +
Error

Model 2 with Total Cash Compensation Index as Dependent Variable:
Total Cash Compensation Index = B0 + [31 (ln(Total Employment) +
B; (Unemp) + [33(Mining) + [34 (Construction) + BS (Mfg-Durable) + [35
(Mfg-NonDurable) + [37 (Transportation) + [38 (Retail) + [3, (Finance) + [319
(SCTVICCS) + Bu (PUbllC Administration) + BuaTCSDICH) + B13 (ASP) +

[314(ASB) + [3,5 (ASA) + 5.,(DE) + [317(ITCSDICHxASB) + a].
(ITCSDICHxASA) + [31,(ITCSDICHxASP) + flm(TTCSDICHxDE) +

Error

The same results are anticipated with both models, but should there be differences
it may indicate that firms are attempting to alleviate market pressures from ITCS through
the use of variable pay. Significant, positive regression coefficients on the ITCS variable
for the craft job families will confirm Hypothesis 2a. Significant, negative regression

coefficients on the IT CS variable for enterprise job family will confirm Hypothesis 2b.

Hypothesis Set 3: The extent of I T CS use and regular employee pay

Hypotheses 3a and 3b are tested using the same dependent variables and
mOderated regression model as is used for Hypotheses 2a and 2b. However, the ITCS
val‘iable entered is the continuous variable IT CSCONT defined as the percentage of work
completed by contractors. This continuous variable measures the extent to which IT C8
are used in the workplace. The dependent variables base salary index and total cash
compensation index are analyzed in two separate regressions. The sample includes all
fifths that reported regular employees in a job family. The results of this test delve

deeper into the ITCS effect to determine if merely the presence of ITCS will cause a

76

change in employee pay rates or if there actually needs to be some critical amount of

contracting actually occurring before employees’ pay levels are changed.

Model 1 with Base Salary Index as Dependent Variable:

Base Salary Index = Bo + 51 (ln(Total Employment) + B; (Unemp) +
BMMng) + Be (Construction) + Bs Mir-Durable) + Bs(Mfg-
NonDurable) + B7(Transportation) + [33(Retail) + B, (Finance) + 310
(Services) + [311 (Public Administration) + [312 (ITCSCONT) + [3,; (ASP)
+ Bu (ASB) + Bis (ASA) + Bu (DE) + [317(ITCSCONTxASB) + Bis

(ITCSCONTxASA) + 319 (ITCSCONTxASP) + [32. (ITCSCONTxDE) +
Error

Model 2 with Total Cash Compensation Index as Dependent Variable:

Total Cash Compensation Index = Bo + Bl (ln(Total Employment) +

B; (Unemp) + B;(Mining) + B4 (Construction) + 55 (Mfg-Durable) + B‘
(Mfg-Noanb16)+ viTmnSPOflation) + human) + B9(Finance) + Bio
(Services) + Bu (Public Administration) + 512 (IT CSCONT) + 513 (ASP)
+ [314(ASB) + 315(ASA) + 5160313) + [317(ITC3(3()1‘I'I"‘1“t313)‘L Bra

(ITCSCONTxASA) + [3,, (ITCSCONTxASP) + B; (IT CSCONTxDE) +
Error

Significant, positive regression coefficients on the contracting variable for the
craft job families will confirm Hypothesis 3a Significant, negative regression
Coefficients on the contracting variables for the enterprise job families will confirm

Hypothesis 3b.

Hypothesis 4: I T CS use and regular employee pay variability
Hypothesis 4 examines the variation in regular employees pay resulting from the
Presence of IT CS. To test this variance hypothesis, two moderated regression models are

used, where the dependent variables measure the variability (range) in base pay and total

77

cash compensation within a job family. The regression models include control variables
of industry segments, total employees (logarithmic transformation) and state
unemployment rates. The sample includes any firms that reported multiple employees in
at least one job of a job family. The continuous ITCS variable, ITCSCONT, from
Hypotheses 3a and 3b is used as the ITCS variable.
Model 1 with Base Pay Range as Dependent Variable:

Base Salary Index = Bo + B1(ln(Total Employment) + B, (Unemp) +

B3(Mining) + B, (Construction) + B, (Mfg-Durable) + B, (Mfg-

NonDurable) + B, (Transportation) + B, (Retail) + B, (Finance) + B“.

(Services) + Bu (Public Administration) + Bu(ITCSCONT) + B13 (ASP)

+ Bu (ASB) + [315 (ASA) + Bis (DE) + Bu (IT CSCONTxASB) + Bis

(ITCSCONTxASA) + B1, (ITCSCONTxASP) + B, (ITCSCONTxDE) +
Error

Model 2 with Total Cash Compensation Pay Range as Dependent Variable:

T CC Pay Range = Bo + B1(ln(Total Employment) + B; (Unemp) +
B:(Mining) + [34 (Construction) + BsMg-Durable) + Bc(Mfg-
NonDurable) + B7 (Transportation) + B, (Retail) + B, (Finance) + B1.
(Services) + Bu (Public Administration) + Bu(ITCSCONT) + B13(ASP)
+ Bu (ASB) + B15 (ASA) + B“ (DE) + Br] (ITCSCONTXASB) + B];

(IT CSCONTxASA) + B1, (ITCSCONTxASP) + [3,. (IT CSCONTxDE) +
Error

A significant, positive regression coefficient on the contracting variable would

confirm Hypothesis 4.

Hypothesis Set 5: Difi'erence between “core ” and “periphery " industries in the pay
consequences of I T CS use

Hypotheses 5a and 5b look more closely at the possible industry effects on
employee pay with contractors. The dependent variables for testing these hypotheses are

the base pay and total cash compensation indices used in Hypotheses Sets 2 and 3. To

78

test the core/periphery arguments presented in the previous chapter, subsets of the
databases including only those job family/firm combinations hiring contingent staff are
used in these analyses. Firms will be classified on a core/peripheral continuum based
upon the percentage of IT employees in their industries as determine by national
Occupational Employment Statistics (BLS, 1998). These OES statistics measure the
percent of employees within an industry that work in IT jobs. It is assumed for the
purposes of this study that a higher percentage of IT employees in an industry correlates
to a more central role for IT tasks within that industry. If firms need more IT workers to
complete their work, then it should mean that IT workers are more important or central to
the operation of the firms.

These hypotheses are tested using the mediating regression model illustrated

 

 

 

 

 

 

 

below:
Figure 1: Mediating Model of Hypotheses 5a and 5b
IT CS
/ \
Core/Periphery $ Job Family Wage
Industry 1 Indices

 

 

 

 

 

 

Using the methods described by Baron and Kenny (1986), this model is tested by
examining the regression weights of four regression equations. First, the
independent variable, core/periphery, is regressed directly on the base salary/1' CC
index used in hypotheses 2 and 3 (linkl). Second the core/periphery variable is
regression on the continuous ITCS variable used in Hypotheses 3a and 3b (link 2).

The third regression estimates the relationship of the continuous ITCS variable on

79

the base salary/T CC index (link 3). If all of these regression relationships exist,
then the core/periphery variable is regressed on the base salary/T CC index
holding the IT CS variable constant (links 2 and 3). For the mediating model to
hold, the relationship between the core/periphery and the base salary/T CC index
in the last regression model should be smaller than the relationship in the first
regression model. Throughout all of the these regression models the effects of
unemployment and organizational size (ln(total employment)) are controlled for.
Significant, regression coefficients on the core/peripheral variable for the
craft job families would indicate the one of the compensating wage differential or
the efficiency wage premium is greater than the other. Significant, positive
regression coefficients on the core/periphery variable for the enterprise job

families will confirm Hypothesis 5b.

Hypothesis Set 6: I T CS use and long-term incentives for regular employees

Hypotheses 6a and 6b predict the impact ITCS has on the retention practice of
long-term incentives. The dependent variable, eligibility for long-term incentives, is
calculated as the percentage of employees in a job family eligible for LTI. Again, a
moderated regression model is used with total employment (logarithmic transformation)
as a control variable. The sample used to test these hypotheses consists of job families
from firms that were able to report on the eligibility of its employees in a job family for
LTI. Firms that were unable to report on LTI eligibility were excluded from the analysis.
Model with Percent Eligible for LTI as Dependent Variable:

Percent Eligible for LT I = B0 + B1 (ln(Total Employment) + [32 (Unemp) +
B3(Mining) + B4 (Construction) + B, (Mfg-Durable) + B, (Mfg-

80

NonDurable) + B; (Transportation) + B, (Retail) + B, (Finance) + B1,

(Services) + Bu (Public Administration) + Bu (IT CSCONT) + B13 (ASP)

+ B“ (ASB) + B15 (ASA) + B15 (DE) + B17 (ITCSCONTxASB) + B1,

(IT CSCONTxASA) + B1, (IT CSCONTxASP) + B], (ITCSCONTxDE) +

Error

Significant, positive regression coefficients on the contracting variable will
confirm Hypothesis 6a. Non-significant regression coefficients on the contracting

variable will lend support to Hypothesis 6b.

Hypothesis Set 7: I T CS use and promotion for regular employees

Hypotheses 7a and 7b predict the impact of ITCS on the ILM practice of intemal
promotions. The moderated regression model presented for Hypotheses 6a and 6b is
used, but the dependent variable is the average number of years until promotion within a
job family. Finns were asked to report the average number of years it takes employees to
be promoted for each job in a job family. The employee weighted average of these
numbers of years for all of the jobs in a family is the dependent variable in these
analyses. The sample used for these hypotheses is those firms reporting the average
number of years until promotion for at least one job in a job family. Firms that were
unable to provide information on promotion rates were excluded from these analyses.

Model with Average Years for Promotion as Dependent Variable:

Avg Years to Promotion = Bo + B, (111(Total Employment) + B; (Unemp) +
B3(Mining) + B. (Construction) + B, (Mfg-Durable) + B‘ (Mfg-
NonDurable) + B; (Transportation) + B, (Retail) + B, (Finance) + B1,
(Services) + Bu (Public Administration) + B 12(ITCSCONT) + B], (ASP)
+ 9140433) + 515(ASA) + 516 (DE) + 517 (UCSCONTXASB) + Bra
(ITCSCONTxASA) + 919 (ITCSCONTxASP) + B, (ITCSCONTxDE) +
Error

81

Significant, positive regression coefficients on the contracting variable for the
craft job families would confirm Hypothesis 7a. Significant, negative regression
coefficients on the contracting variable for the enterprise job families would confirm

Hypothesis 7b.

Hypothesis Set 8: Extent of I TCS use and salary structure range overlap for regular
employees

Hypotheses 8a and 8b predict the differences exists in the amount of salary range
overlap that exists if a company employs ITCS or not. Salary range overlap is defined as
the percent of the salary structure range that overlaps with the next higher salary structure
range. The dependent variable in these analyses is the average percent of overlap for all
jobs in a job family. The sample includes those firms that provided information
regarding their salary structures. Firms that were unable to provide salary structure range
information were excluded from the analysis. Again, the moderated regression model
presented for Hypotheses 6a and 6b is used, but the dependent variable is the average
percent of overlap for a job family.
Model with Average Percent of Overlap as Dependent Variable:

Avg Percent of Overlap = Bo + B1(ln(Total Employment) + B; (Unemp) +

BdMining) + [34 (Construction) + Bs(Mfg-Durable) + Bc (Mfg-

NonDurable) + B, (Transportation) + B, (Retail) + B, (Finance) + B1,

(Services) + Bu (Public Administration) + BuOTCSCONT) + B13 (ASP)

+ [514(ASB) + 315 (ASA) + Bic (DE) + Bu (IT CSCONTxASB) + Bis

(ITCSCONTxASA) + B1, (ITCSCONTxASP) + B; (ITCSCONTxDE) +

Error

Significant, positive regression coefficients on the contracting variable for the

craft job families will confirm Hypothesis 8a. Significant, negative regression

82

coefficients on the contracting variables for the enterprise job families will confirm

Hypothesis 8b.

83

CHAPTER 5 - RESULTS

The data used to test the hypotheses presented in the previous chapter come from
two years of the Compensation Survey of Information Technology Professionals
conducted by William M Mercer, Inc. The data were collected for this survey in April
1998 and April 1999. These data were submitted by human resources professionals and
covered the wages paid to their companies’ regular information technology employees,
salary structure ranges, long-term incentive eligibility, the average number of years until
promotion, and the use of ITCS in a variety of information technology areas.

The survey collects data about five job families. The first family, Data Entry
jobs, is an example of IT work organized in an enterprise ILM. The other four job
families, Software Systems Engineering, Applications Systems Programming,
Applications Systems Analysis, and Applications Systems Analysis and Programming,
are examples of IT work organized in craft ILMs.

The data sets from the 1998 and 1999 surveys were combined into one sample to
ensure there is an adequate sample size to capture the effects of all the variables in
question and any effect of IT CS that may be present will be apparent in the results. The
combined data set contains data on 132,925 employees in 3,780 job families from 1,357
organizations. As all of the hypotheses have been stated at the job family level, we will
focus on the 3,780 job families as the unit of analysis. The term “job family” will be
used to designate a set of positions, from one of the five job families listed above, within

a particular firm.

84

The sample analyzed here is generally representative of the US population of
organizations. The geographic distribution is relatively similar to that of the US
employee population (BLS, 1999a). The industry representation of the sample
companies is also similar to the US industry distribution, however, it does have a slightly
larger percentage of durable manufacturing and services firms and a smaller percentage
of public administration and retail/wholesale trade firms than are present in the US (BLS,
1999b). (See Appendix C for more detail on the industry and geographic distributions.)
One difference from the US population is that the companies submitting data for these
job families are assumed to be larger than the US population of companies. Given these
demographics, the results of these analyses may be applicable to a larger population than
the firms in this sample. However, caution should be employed before applying these
results to jobs other than those in the Information Technology field or to smaller
organizations.

The survey questions used to test these hypotheses included policy data on the
amount of work completed by ITCS and the relative pay rates of ITCS for each type of
job family, job level questions about salary structure ranges, average promotion rates, and
the use of long-term incentives. For regular IT employees, data were submitted for their
base pay and any short-term incentives paid in the last 12 months to them. Descriptive

statistics for all of the variables used in these analyses are presented on the following

pages-

85

 

 

Table 1: Descriptive Statistics and Correlations

E a a a a
a? E S 'c o 'c g g c I l— e
_ 3 3 "3 *3 3s 8 (D T: e ,. a g B E g
= 8 3 .52“ E” 3 “E E “E :5 E e .2 E .8 2 1E a a a a
a e g g E 3 a g a E E E :3 a a r: g e o o ‘g a a g
Variable '2 '8 E LL} 3 2 8 2 z 2 o 3 E: :52 u: «8 i < i L" E: 6 < < <
In (Total Employees) 8.101 1.454
Unemployment 4.040 1.052 -.009
Mining .015 .120 —.046** .047**
Construction .001 .028 -.019 —.026 —.003
Manufacturing—Nondurable .097 .296 .070*** -.015 —.040* -.009
Manufacturing — Durable .055 .228 -.009 -.070*** -.029 —.007 -.079***
Wholesale Trade .023 .149 —.029 .061*** -.019 —.004 -.050** —.037**
Retail Trade .059 .236 .230*** .006 -.030* -.0074.082***-.060*** -.038**
Finance .238 .426-.135*** .015-068*“ -.016-.183***—.135***-.085***-.140***
Services .393 .488 .004 .009 .098*** —.023-.264***».194***-.123***».201***e.449***
Public Administration .040 .200-.098*** -.030* -.025 -.006-.067*** -.049** -.031* —.051**-.114***-.163***
Transportation .080 .272 .044** —.001 -.036** -.008-.097***-.071*** -.045**-.074***—.165***—.238***-.060***
ITCSDICH .176 .381 —.044** —.020 .025 -.013 .0058 .019 -.005 029* .007-.092*** .027* .080***
ITCSCONT 3.778 11.934 —.032** -.021 027* -.009 .001 .006 .027 .020 —.031-.062*** .018 .098*** .685***
Contract Pay 30.497 81.060 —.032 —.041** -.004 -.011 .006 .025 -.007 .023 .002 -.054** .023 .034** .814*** .553***
ASP .145 .352 -.011 -.014 -.019 —.012 .002 .006 .018 .012 .022 -.034** .020 -.006 .023 .013 .015
A513 .403 .490 -.045** .026 .004 ~.004 .022 .015 .001 -.007 —.008 —.0034 —.012 -.005 .171*** .135*** .147***-.338***
ASA .159 .366 .011 .011 .014 .039** .016 —.007 —.028 -.011 —.012 .004 .000 .015-.055*** -.047** -.056**-.179***-.358***
SSE .216 .412 .046** ~.031* .017 —.015 —.018 —.011 .006 .002 .025 -.022 -.018 .029*-.103*** -.080**-.080***-.216***-.43l***-.228***
DE .077 .267 .010 .005 ~.027* —.008 -.038** —.009 .003 .008 -.036** .079*** .023 -.049**-.110*** -.076**-.091***-.119***-.237***~.126***
ITCSxASP .601 4.884 -.044 .002 -.007 —.003 .001 -.012 .011 .008 .004 -.026 .014 031* .266*** .377*** .195*** .299***-.101*** -.054**
ITCSxASB 2.310 9.542 —.055** -.019 .033** —.007 —.010 .020 .032** .012 -.014 -.055** .012 .066*** .524*** .770*** .459***-.100*** .295***-.105***
ITCSxASA .396 3.968 .013 -.010 .026 -.003 .028* .012 -.016 —.009 —.027* -.012 .016 .018 .222*** .295*** .122***-.042***-.084*** .235***
ITCSxSSE 0.422 4.366 0.025 -.007 -.012 -.003 .002 —.022 .006 .010 -.027 —.011 -.007 .077*** .209*** .339*** .167*** —.040**-.080*** -3042**
ITCSxDE 0.048 1.312 —0.003 -.008 -.004 -.001 —.012 —.009 -.006 .059*** -.020 .008 .003 -.011 .079*** .09*** .018** -.015 ~.030* -.016
BaseRatio 0.975 0.194 0.029 .129*** .043** —.040** .045** .007 .011 .027 -.001 -.054** —.023 .016 -.010 .013 -.019 -.026 .018-.187***
TCC Ratio 0.968 0.204 .037** .128*** .052** -.037** .049** .004 .006 .037** .022-.085*** —.040** .037** —.010 .015 -.020 —.012 .001-.174***
Base Range 34.164 29.667 .244*** .001 -.030 —- -.039 .054** -.049* -.058** .014 .069** —.038 —.037 -.021 —.012 -.032 -.075** .194*** -.055**
TCC Range 35.55 31.826 .247*** .007 —.030 -- —.029 .052** -.052** -.061** .044* .042 —.044* -.036 -.022 —.012 —.032 -.076** .189*** -.057**
Core/Peri 4.592 4.295 -.168*** —.036**-.103*** -.029*-.298*** .034** -.036**-.247*** .289*** .180***~.115***—.116***-.125***-.097***-.116*** -.017 .056** -.033**
LTI Percent 0.123 0.311 .111*** .024 .005 -.009 —.006 .068*** .010 .098*** -.036** -.022-.072*** .001 .004 .015 -.015 -.020 .035** —.024
YrsProm 2.6 2.134 -.022 .075** .052* —.042 -.039 -.083** -.074** -.062** .025 .077** .036 .002-.099*** -.065** -.079**<.0166** -.045 048*
Overlap 0.618 0.227 .021 -.037 .029 .056** -.042* .032 .000 -.010 .071** —.058** -.004 —.004 -.021 -.026 .000 .030 —.057** .052**

N=3,780 *p<.05 **p<.01 ***p<.001

 

86

 

 

Table 1: Descriptive Statistics and Correlations (Cont'd).

a 53
< <
m a 5:
Variable 33 E E E
1n (Total Employees)
Unemployment
Mining
Construction
Manufacturing - Nondurable
Manufacturing - Durable
Wholesale Trade
Retail Trade
Finance
Services
Public Administration
Transportation
ITCSDKHl
ITCSCONT
Contract Pay
ASP
ASB
ASA
SSE
DE a152***
ITCSxASP 2065*** =026**
ITCSxASB -J27***2070*** 2030*
ITCSxASA 2054** 2030* :013 2025
ITCSxSSE .185*** 2028* 1012 a023
ITCSxDE a019.127*** 2005 a009
BayzRafio .014.238*** .021 .019
TCCIRafio .010 236*** .024 .018
BaseRange a078**—3058** a021 .008
TCIIRange 2062** =067** 2020 .006
ConflPen =034** .017 2045** <052**
LTI Percent .054** -.088*** -.011 .025
Yrflhonr .060** .018 .000 2089**
Oveflap .028 4043* 2036 2020
N = 3,780 * p < .05 ** p < .01 *** p < .001

87

at) L” to
< U) C
>< >< ><
m m m
U U U
5 L1 t
2010
a004 2004
2061*** .013 .037**
2059*** .016 .036**
.003 2021 —”025
.001 —:015 <024
2058*** =040** 2032**
2009 .013 a014
a002 .027 2030
—.013 .018 .007

o
.9 .9 i“
E E 5
0: Oi Di
3 U 3’.
:3 U c:
CD 1— CO
.977***
.047** .053**
.059** .075**.976***
.036** .045**.195***
.026 .056**.l6l***
.121***.109*** .096**
.009 .013 :051

O
on
c 'C
e o.)
‘3‘ E:
U 9'3
L) o
l— U
.191***
.159*** .002
085* .094**
-.039 .008

LTl Percent

-.070**
—.022

YrsProm
Overl p

.047

 

 

The correlation matrix reveals some interesting relationships. The use of ITCS as
measured by IT CSDICH and IT CSCONT is significantly, positively related to ITCS pay
rate. This correlation implies that as firms use more IT CS, they tend to pay them higher
rates. The Core/Periphery variable is negatively, significantly related to IT CS pay rate,
suggesting that as firms move toward the core they pay less to their ITCS. The
Core/Periphery variable is also negatively related to ITCSDICH and ITCSCONT, which
implies that core firms do not use ITCS as much as periphery firms. This finding
corroborates part of the rationale behind Hypotheses 53 and 5b. Total employment is
positively related to the total cash compensation range index and the LTI percent
variable. This finding would imply that as firms grow they have less internal consistency
in total cash compensation larger firms would typically have more employees in a job
and administering pay among a larger number of people would make it more difficult to
be consistent, thereby increasing the range of pay being offered to employees. Also
larger firms may be more stable providing them an opportunity to create long-term
performance plans that would be presumptuous in smaller, less stable firms.
Unemployment is positively related to the base pay and total cash pay indices. This
finding is counterintuitive and is consistent throughout the analyses, but may reflect the
non-IT specific nature of these unemployment rates. Finally, many of the industry
variables are related to the base pay and total cash compensation indices as is predicted
by much prior research These relationships between industry and pay have been well
documented and explain why industry segment variables are used as control variables
throughout these analyses. Although some of the correlations between the interaction

terms and their components, ITCSCONT and the dummy job family variables are fairly

88

high, none of the correlations between the independent variables is close to approaching
1.00, therefore problems with multi-collinearity will be minimal.

As discussed in the Data and Methodology chapter, the samples used to test each
hypothesis vary in size because of the focus of the hypothesis and the number of valid
responses to the survey questions. The focus of the hypotheses means that the
hypotheses are stated in such a way that in for some of them (Hla, Hlb, ch, H4, and
H5) it is appropriate to test the hypotheses on a subset of the whole sample. These
hypotheses are predicated on companies using contractors (Hla, Hlb, ch, H5) or having
multiple employees in a job classification (H4). As such, the sample was segmented to
include only those companies that fit these criteria. The second reason for the changing
numbers of cases is related to the level of detail required to complete the complex
questionnaire. Some companies were unable to provide all of the information regarding
their policies with respect to their contractors or for the salary structure, long-term
incentive eligibility, and promotion policies associated with their employees (H6, H7,
H8). In each hypothesis, the subset of companies that were able to provide these data
were included and companies tlmt did not provide information on the required variable
were excluded. This ensures that the hypotheses are tested on cleaner data than would be

used if missing data were replaced with the mean of the sample.

H t SetI:The d1 er betweenITCSand arIT lo ees.
Hypotheses 1a and 1b predict that the wages of ITCS in both craft and enterprise
IT job families will be higher than the wages of the regular employees in these job

families. IT CS relative wages were reported by companies using ITCS in each of the five

89

1

job families, as a percentage the regular, IT employees’ wages in the same family. For
example, if IT CS were paid twice the hourly rate of regular employees in that job family,
then a value of 200% was reported as the ITCS relative wages. Hypotheses la and lb
were (tested using paired t-tests to compare the ITCS relative wage to 100%, which is the
corresponding firm rate for the regular employees in a job family. Sample sizes for these
paired t-tests are smaller because the sample is limited to those companies that reported
IT CS relative wages and regular employees’ wages for the same job family. The table
below shows the results of these tests, which confirm the predictions of Hypotheses la
and 1b. As predicted, the IT CS relative pay rates are significantly higher than 100%.
This means that both craft and enterprise ITCS are paid significantly higher than regular,
IT employees within the same job families.

Table 2: Paired t-test Results of Hypotheses la and 1b

 

Job Family Mean ITCS Pay Rate N T

 

Applications Systems Analysis & 196.46 349 17.354"
_Programming (Craft) ,
Applications Systems Analysis 198.87 61 ~ 7.600“

agirafi) . ‘
Applications Systems 187.40 98 10.405"

Wining (Crafi)
Software Systems Engineering 199.58 74 . 8.533“

ELQrafi)

 

 

 

 

 

 

 

 

 

, Data Entry (Enterprise) 181.25 8 pl 3.3038
"'"' p < .001
"' p< .05

 

Hypothesis 1c predicts that the wage premiums above the regular, IT employees’
Wages paid to ITCS in craft job families will be higher than the premiums paid to ITCS in
enterprise job families. To test this hypothesis a paired t-test was again used to compare

the premiums for craft ITCS to the premium for the enterprise ITCS. The sarnples for

these tests were limited to companies reporting pay rates for the craft and enterprise ITC S
being compared These companies may not have reported regular employees in these job
families. Using a paired t-test controls for the general wage levels within the firms. The
table below shows the results of this analysis. The results are only significant for the

Applications Systems Analysis & Programming giving only limited support for

 

 

 

 

 

 

 

 

 

 

 

Hypothesis 1c.
Table 3: Paired t-test Results of Hypothesis 1c
Job Family Mean Craft Mean Enterprise N T
ITCS Pay Rate ITCS Pay Rate
Applications Systems Analysis 236.33 159.54 24 4.023"
& Programming (Craft) »
Applications Systems Analysis -- -- -- -
(Craft)
Applications Systems 156.25 124.25 4 1.355
Programming (Craft)
Software Systems Engineering 207.75 179.00 8 1.565
(Craft)
‘ p < .001
- insufficient data
H lrhestsS' 2:R'l..n “Jr/ll! eem (martian w een .14; CS ,
rnrs not

Hypotheses 23 and 2b examine the differences in pay for regular, IT employees’
between firms with and without 1T CS. A dummy variable, labeled ITCSDICH is used to
represent the presence (IT CSDICH=1) or absence (IT CSDICH==0) of ITCS for a
particular job family. It is predicted that regular, IT employees in craft families will have
higher wages when ITCS are present than when they are not present in their job family
and organization. However, it is predicted that the wages of regular, IT employees in

enterprise families will be lower when IT CS are present in their job family. All of the

91

 

companies that submitted data for a job family, whether or not they use ITCS, are
included in these analyses.

These hypotheses are tested using moderated regression model to isolate the
effect of the presence of IT CS (represented by ITCSDICH) by job family while
controlling for company size, industry, and unemployment conditions in the state.
Company size is measured as the natural log of total employees. The natural log is used
to eliminate the impact that the larger scale of the total employee variable may have the
dependent variables. Industry is coded using dummy variables for mining, construction,
manufacturing-durable, manufacturing-nondurable, transportation and utilities, retail
trade, wholesale trade, financial, insurance, and real estate services, services, and public
administration with wholesale trade omitted from the analysis. Unemployment is
measured by the state unemployment rate from BLS statistics during the month the
compensation survey data is effective (BLS, 1999a). The ITCS variable (ITCSDICH) is
dichotomous, equal to 1 if a company uses ITCS in a job family or equal to 0 if the
company does not use ITCS in that job family. Four of the job families are coded using
dummy codes with the Software Systems Engineering family omitted for the analysis.
The interaction term is calculated as the product of the ITCSDICH variable and the
dummy code for each job family.

Two regression models are tested. Model 1 utilizes a base salary index as the
dependent variable. Model 2 utilizes a total cash compensation (T CC) index as the
dependent variable. These two models are studied to determine whether variable pay

plays a role in creating these pay differentials for regular employees when IT CS are used.

92

In both the base salary and TCC models, a pay index for each company, including
those that use IT CS and those that do not use ITCS, reporting data in a job family was
calculated. For each job in a family, the company’s average pay is divided by the
average pay for all employees reported in that job for the whole sample. Then an
employee weighted average of all these ratios for the job family is calculated to create the
job family pay index. An employee weighted average is used instead of a straight
average to ensure the index reflects the actual base pay practices of the companies. In
cases where a company has many employees in one job and only a few at other levels in a
job family, an employee weighted index will more accurately portray how companies are
paying their employees. This index indicates how companies pay in relation to the
market. A value greater than one means companies pay that particular family above the
market rate and a value less than one means companies pay that family below the market.

The model includes the company size, industry, unemployment rates, the
dichotomous ITCSDICH variable, the dummy variables for job families and the
interaction terms of the ITCSDICH x the dummy job family variables as independent
variables and the base salary index as the dependent variable. The regression model is as

follows:

Base Salary Index = [30 + B1(1n(Total Employment) + B; (Unemp) +
83(Mining) + 84 (Construction) + [35 (Mfg-Durable) + B‘ (Mfg-
NonDurable) + B7 (Transportation) + B; (Retail) + B, (Finance) + Bu
(Services) + [311 (Public Administration) + Bu (IT CSDICH) + [313 (ASP) +
914053) + 515(ASA) + BIS (DE) + 517 (ITCSDICHXASB) + Bra

(IT CSDICHxASA) + [3,, (ITCSDICHxASP) + B, (ITCSDICHxDE) +
Error

93

Table 4: Moderated Regression Results for Hypothesis Set 2

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Variable Base Salary Index TCC Index
Constant 361"" ' 334MB:
(.030) (.031)
ln(Total Ees) .003 004*
(.002) (.002)
Unem 10 cut 024*" 324*":
p m (.003) (.003)
Mining .073" .100"
‘ (.032) (.033)
Construction -. 197* -. 183
(.108) (.113)
Mfg-NonDurable .035 .045‘
(.022) (.023)
Mfg-Durable .016 .020
(.024) (.025)
Transportation .020 .041‘
(.022) (.024)
Retail .016 .030
(.024) (.025)
Finance .003 .013
(.021) (.022)
Services -.017 -.020
(.020) (.021)
Public -020 -032
Administration (025) (.026)
ITCSDICH .01 1 .013
(.021) (.022)
ASP -.024'* -.015
(.011) (.012)
ASB .001 -.001
(.009) (.009)
ASA «089” -.081'" ,
(.011) (.011)
DE .l60"* , .173":
(.013) (.014)
lTCSDICHxASP .029 .024
(.029) (.031)
ITCSDICHxASB -.018 -.020
(.024) (.025)
ITCSDICHxASA -.034 -.044
(.031) (.032)
ITCSDICHxDE -.017 -.019
(.066) (.069)
R2 .112 .114
F .. ratio 23.73am 24.282‘"
N = 3,769
*p<JO
" p < .05
"* p < .001

94

 

Model 2 follows the same regression process but uses the TCC index as the
dependent variable. Table 4 shows the unstandardized weights with the standard error in
parentheses below it for each independent variable. The results are report no significant
regression weights for the interaction terms. Since the interaction terms and the
ITCSDICH regression weights are non-significant, the interpretation has to be that there
is no support for Hypotheses 2a and 2b. Some of the control variables are significant in
both models including unemployment, mining, ASA, and DE. Unemployment is
positively and significantly related to wages in both models. This is inconsistent with
conventional beliefs that higher unemployment would have a dampening effect on wages.
However, this unemployment rate is for the whole state at the times of the surveys and is
not specific to IT employment. IT employment constitutes a relatively small portion of a
total statewide labor force. The level of IT unemployment may not be reflected in the
statewide numbers. Being in the mining industry is significantly, positively related to
wages. The Applications Systems Analysis Family is paid relatively lower than Software
Systems Engineering family and the Data Entry family is paid higher than the Software

Systems Engineering family, but these relationship are not relawd to IT CS use.

Hmhesis Set 3: The current at ITCS use and rggflar er_n2(_o_zee m.

Hypotheses 3a and 3b are similar to Hypotheses 2a and 2b, except they take a
finer measure of IT CS use in companies. Instead of using a dichotomous variable that
firms either use ITCS or they do not, Hypotheses 3a and 3b use a continuous ITCS
variable (ITCSCONT), which measures the percentage of work completed by ITCS. This

variable measures the extent to which a firm uses ITCS. The dependent variables are the

95

Base Salary index and the Total Cash Compensation index created for Hypotheses 2a and
2b. The sample is also identical to those used in Hypotheses 2a and 2b. The results of
the regression models should indicate whether the degree to which ITCS are used effects
these pay outcomes. Hypotheses 3a and 3b predict that wages will be higher for regular,
IT employees in craft job families as firms use more ITCS in these job families and the
wages will be lower for regular, IT employees in enterprise job families as firms use
more ITCS in these job families. Table 5 shows the results of the regression analysis for
the base salary index and the TCC index The unstandardized regression weights are
shown as well as the standard error in parentheses under the regression weights.

As with Hypotheses 2a and 2b, the results do not support these predictions. There
are no significant interaction regression weights in either model. As with Hypotheses 2a
and 2b the mining industry control variable is significant. Unemployment continues to
contribute significantly and positively to wages and the Applications Systems Armlysis
and Data Entry families are still paid significantly differently from the other job families
in this study. These findings suggest that employers have really done very little with pay
to acknowledge the external labor market forces that they are bringing into their

workplaces by hiring IT CS.

Table 5: Moderated Regression Results for Hypothesis Set 3

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Variable Base Salary Index TCC Index
Constant .858‘“ 331“"
(.030) (.031)
ln(Total Ees) 003 004"
(.002) (.002)
Unem lo ent 024*” 025*”
p m (.003) (.003)
Mining .074" .101"
(.032) (.033)
Construction -.195' -.179
(.108) (.113)
Mfg-NonDurable 036 046"
(.022) (.023)
-Durable .016 .021
Mfg (.024) (.025)
Transportation 020 040"
(. 022) (. 024) -
Retail .015 .029
(.024) (.025)
Finance .004 .019
(.021) (.022)
Services -.015 -.018
(.020) (.021)
Public -.014 -.030
Administration (025) (026)
IT CSCONT .000 .001
(001) (.001)
ASP -.021“'I -.013
g (.011) (.011)
ASB -.003 -.005
(008) (009)
ASA «.086"* -.082“* .‘
(.010) (.011)
DE .159‘“ .171‘”
(.013) , (.013)
IT CSCONTxASP .001 .001
(.001) (.001L
IT CSCONTxASB .000 .000
@01) (.OOIL
IT CSCONTxASA «001 -.002
(.001) (.001)
IT CSCONTxDE .001 .001
(.002) (.003)
R2 .112 .115
F _ ratio 23.160“* 24.33?“
N = 3,769
‘p<JO
"‘ p < .05
on: p < 001

97

 

H hais4:IICSuseaud far I e if

Hypothesis 4 looked at the variability of pay resulting from using ITCS. The
prediction is that using ITCS will lessen the importance of internal equity such that the
range of pay within job families for companies using ITCS will be greater than the range
of pay within job families of firms not using ITCS. To test this hypothesis a pay range
index was created for each job family. This index equals the employee weighted average
of the percent range of pay for each job in a family. For each job in a job family with
multiple employees, the percent range of pay was calculated by subtracting the minimum
wage from the maximum wage and dividing the difference by the minimum wage paid
for the job. These percents for all the jobs in each job family were then used to create an
employee weighted average that is the range index. Given the nature of the range
calculation only those firms that had multiple employees in at least one job category in a
family are included in this analysis. Using an employee weighted average gives a better
approximation of the pay practices for an organization than a straight average, especially
if the distribution of employees is uneven across jobs in a family. A larger index means
that there is more variability in pay. This index was calculated for both the base salary
and TCC in a job family. The use of both indices is important for this analysis because
typically the range of variable pay awarded to employees is greater than the range of pay
offered in base salaries. It is likely that a difference in the range of pay will be found for
the TCC index, but not for the base salary index because delivering anomalies in pay
through variable pay easier than adding them into base pay.

These range indices were the dependent variable in a moderated regression model.

Company size (ln(total employees», unemployment, industry, and the dummy job family

98

variables were entered as control variables. The continuous ITCSCONT variable,
representing the percent of work completed by ITCS, was entered to measure the effects
of ITCS on these pay ranges indices. ITCSCONT is used for this hypothesis and all
subsequent hypotheses because the continuous nature of the variable provides better
information about IT CS usage than the dichotomous variable. Additionally, the
interaction terms, used in Hypotheses 3a and 3b, were entered to capture any differences
in variability due to the combination of ITCS and job family.

The sample for this hypothesis set consists of job families, where companies
employed multiple regular employees in at least one level of the job family. The sample
is smaller than that used in Hypotheses 2 and 3 because some smaller companies
reporting data only employed one worker in a job level, making it irrelevant to determine
the range of pay for that job.

Table 6 summarizes the results of these moderated regressions. The
unstandardized regression weights are shown with the standard error in parentheses
below them. None of the regression weights on the IT CSCONT variable or the
interaction variables are significant, giving no support to Hypothesis 4. Some of the
industry control variables, manufacturing durable, finance and services, are significantly
related to an increase of variability in wages. The Applications Systems Analysis and

Programming family has more variability of wages than the other job families.

Table 6: Moderated Regression Results for Hypothesis Set 4

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Variable Base Salary Ramge TCC Range
Constant -29.503“* -34.032***
(7.107) (7.632)
ln(Total Ees) 5924"" 6459*"
(.511) (.549)
Unem lo ent .058 .264
p m (.662) (.711)
Mining 80% -2.852
(8.985) (9.638)
Construction - -
Mfg-NonDurablc 2.097 4.087
(5.201) (5.578)
Mfg-Durable 15.648" 17.401“
(5.419) (5813)
Transportation 3.305 4.233
(5.149) (5.497)
Retail -7.668 -7.908
(5.427) (5.822)
Finance 10.615" , 13.785"
(4.787) * (5.134)
Services 11.136“ 11.625“
(4.743) (5.087)
Public 4.609 4.920
Administration (5937) (6369)
ITCSCONT -.088 -.075
(.205) (.220)
ASP 1.875 .557
(2.529 (2.703)
ASB 13.4w" 13.04am
(2.128) (2.282)
ASA 2.348 1.246
(2.350) (2.521)
DE -1.077 -3.121
(3.160) (3.399
ITCSCONTxASP .115 .112
(.248L (.266)
ITCSCONTxASB -.015 -.040
(.221) (.237)
ITCSCONTxASA .198 .173
(.271) 4.291)
ITCSCONTxDE -.087 -073
(.588) (.631)
R2 .132 .132
F — ratio 12.63"" 12.656***
N = 3,769
*p<JO
"“" p < .05
”* p < .001

100

 

The size of the companies, as measured with the ln(total employment), was positively
significant to more variability in pay. Company size may be related to variability in pay
because larger firms will have more employees in a job category and this will make it

more difficult to administer pay similarly for all employees.

 

W50 e.

Hypotheses 5a and 5b investigate the role industry plays in the use of IT CS and
consequently in the wages of regular, 1T employees. Specifically, core companies, those
that use IT services as a central part of their operations, are expected to use fewer IT CS
and consequently pay a smaller premium to their regular, craft and enterprise, IT job
families than peripheral industries, which are predicted to use more ITCS and will need
to pay their regular, craft and enterprise employees a greater premium. Because these
hypotheses focus on regular, employee wages when IT CS are employed under different
levels of IT centrality to companies’ operations, the sample is restricted to only those
companies that reported using IT CS in a job family. The number of cases included
because of this restriction is smaller than those in Hypotheses 2 and 3.

The test of these hypotheses is through a mediated regression model as shown
below. The ITCSCONT variable has a mediating role in this model because it is
anticipated that companies’ position towards the core or periphery will have an impact on

their use of IT CS, which in turn will impact their regular, employees’ wages.

101

Figure 1: Mediating Model of Hypotheses 5a and 5b

ITC SCONT \
/ 3

Core/Periphery 1 4 Job Family Wage
Indices

 

 

 

 

 

 

 

 

 

 

 

 

 

Using the methods described by Baron and Kenny (1986), the above model is
tested by examining the regression weights of four regression equations. First, the
independent variable, core/periphery, is regressed directly on the Base Salary/T CC index
used in hypotheses 2 and 3 (linkl). Second the core/periphery variable is regressed on
the continuous IT CS variable used in Hypotheses 3a and 3b (link 2). The third regression
estimates the relationship of the continuous IT CS variable on the Base Salary/1‘ CC index
(link 3). If all of these regression relationships exist, then the core/periphery variable is
regressed on the base salary/T CC index holding the ITCS variable constant (links 2 and
3). For the mediating model to hold, the relationship between the core/periphery and the
base salary/T CC index in the last regression model should be smaller than the
relationship in the first regression model.

Throughout all of the these regression models the effects of unemployment and
company size (ln(total employment)) are controlled Industry variables are not entered
into these models because their high correlation with the core/periphery variable would
make their entry redundant. The core/periphery variable is a continuous variable that is
the percent of IT employment by industry as measured by the Occupational Employment
Statistics of 1998 (Bureau of Labor Statistics, 1998; See Appendix for details). This

variable can range from 0—100% and represents the total employment in an industry that

102

consists of IT professionals. It is assumed that higher percentages of IT workers put an
industry closer to the core because more of this industry’s work is accomplished by IT
workers. Lower percentages of IT professionals put an industry closer to the periphery
because a smaller amount of the work needed to for companies to operate is done by IT
workers. The table on the following page has the results of these regression models. The
table shows the unstandardized regression weights with the standard error in parentheses
below them.
Table 7 does show some evidence suggesting support for Hypotheses 5a and 5b.
The Core/periphery variable is significant in all of the regression models. Although the
regression weight does not decrease in the complete model the amount of variance
explained for the Base and TCC models is the greatest in the full model regression This
implies that some relationship exists among the core/periphery, the ITCS variables and
regular, employees’ pay. Link 2 shows that firms in the Core do tend to use fewer ITCS.
This confirms part of reasoning behind Hypothesis Set 5 that Core industries will have a
smaller need for ITCS because they can better predict their IT workload and employ an
adequate number of regular employees to complete all of their work. Further Link 2
shows that unemployment is negatively related to the use of ITCS. This implies that
firms use ITCS when regular employees are not available in the labor market. While
high unemployment was not cited as one of the reasons to use IT CS, it is understandable
that firms would seek to use IT CS to cover needed work, when regular employees were
not available. None of the interaction terms were significant, which means the limiwd
results found here are not differentiated by job family. Despite some evidence in favor of

these hypotheses the required relationship for the mediating model are not present.

103

Therefore, no support is found for Hypotheses 5a and 5b. More research is needed to

determine the exact relationship between core/periphery industries, ITCS use and regular

employees’ wages.

Table 7: Mediating Regression Results for - Hypothesis Set 5

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Base Pa Index TCC Index
Variable Link 1 Link 2 Link 3 Full Link 1 Link 2 Link 3 Full
Model Model
Dependent Base Fl‘ C S Base Base TCC IT CS TCC TCC
Variable Pay Pay Pay Index Index Index
Index Index Index
Constant .838m 7526‘“ .857’" 834°" .809’" 7-526‘" .827m .804m .
4.022) (1.426) (.021) (.022) (.024) (1.426) (.023) (.024)
ln(Total Ees) .005" -.357" .004" .005" .006" -.357" .005" .007"
(.002) (.133) (.002) (.002) (.002) (.133) (.002) (.002)
Unemployment .0243” 4325' .024." .024." .025'“ -.325" .025": _ .026’"
' (.003) (.182) (.003) (.003) (.003) (.182) (.003) (.003)
Core/Periphery .002“ -.31 1‘“ .002“ .002“ -.31 1m .002“
(.001) (.045) (.001) (.001) (.045) (.001)
ITCSCONT .000 .001 .001 .001
(.001) (.001) (.001) (.001)
ASP 4017' 2.170" -021" -022" -.010 . 2.170" -.013 -.014
(.010) (648) (.011) (.011) (.011) ‘ (.648) (.011) (.011
ASB -.003 3.910." -.003 -005 -.006 3.910.” -.007 -.008
(.008) (.510) (.008) (.008) (.008) ‘ (.510) (.009) (.009)
ASA -.090"' .510 -.087‘" -.087”‘ -.087“‘ .510 -.084‘” -.084”‘
(.010) (.630) (.010) (.010) (.010) (.630) (.011) (.011)
DE .152m -l.l68 .153m .152": 160‘“ -l.l68 .152m .l60‘“
(.013) (.801) (.013) (.013) (.013) (.801) (.013) . (.013)
ITCSCONTxASP .001 .001 .001 .001
(.001) (.001) (.001) (.001)
ITCSCONTxASB .000 -000 -000 -000
(.001) (.001) (.001) (.001)
ITCSCONTxASA -.001 -.001 -002 -002
(.001) (.001) (001) (.001L
ITCSCONTxDE .001 .001 .000 .001
(.002) (.002) (.003) (.003)
R2 .102 .037 .102 .103 .097 .037 .097 .099
F _ ratio 6087‘“ 20.43.“ 38.81‘“ 36.18m 57.88‘“ 20.43.“ 36.63‘“ 3458‘”
N=3,756 *p<.10 "p<.05 ***p<.001

Hmtlreses Set 6: ITCS use and log-Jenn incentives [or Mr fllflees.
Hypotheses 6a and 6b predict that companies using ITCS will offer more long-

term incentives to their Craft job families, but make no changes in their LTI offerings for

Enterprise job families These hypotheses are tested using a moderawd regression model.

The LTI variable is measured as the percentage of employees in a job family eligible to

receive long-term incentives, regardless of whether or not they were granted these

104

 

incentives. Participants responded whether each job in a family was eligible to receive
LTI grants (0=No, 1=Yes) and this was multiplied by the number of employees in each
job. The sum of these products was divided by the total number of employees in the
family to determine the percent of employees in the family eligible to receive these
benefits, which is the dependent variable, LTI Percentage.

The control variables are organizational size (ln(total employment)), industry,
unemployment, and job family. The size of the organization as measured by the natural
log of the total employment of the firm is used as a control variable because larger firms
are more likely to be public and therefore have stock that could be awarded in LTI grants.
Further, offering LTI in a larger, more established firm is not as risky as it may be with
smaller companies, where LTI grants of stock or cash may be too risky for employees to

accept. Industry is used as a control variable because some industries may be more or
less likely to offer LTI grants. For example, the public administration industry is
probably less likely to offer LTI because these organizations are typically not—for—profit
and not publicly traded. Unemployment may be related as low unemployment may
induce companies to implement more retention techniques such as LTI grants.

The moderated model is as follows:

LT 1 Percentage = [30 + Bl (ln(Total Employment) + B; (Unemp) +
BMMng) + B4 (Constrtwfion) + 85 (Mfg-Dumb) + Bs(Mfg-
NonDurable) + B1 (Transportation) + [33 (Retail) + [3, (Finance) + Bio
(Services) + [311 (Public Administration) + [312 (IT CSCONT) + 513 (ASP)
+ [314(ASB) + Bis (ASA) + [316(DE)+ [317 (IT CSCONTxASB) + Bu
(ITCSCONTxASA) + [3,, (ITCSCONTxASP) + [3, (ITCSCONTxDE) +
Error

105

Table 8: Moderated Regression Results for Hypothesis Set 6

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Variable LTI Percent
Constant -.015
(.055)
ln(Total Ees) 020‘“
(.004)
Unem lo ent .008
p m (.005)
mm -.009
ng (.059)
Construction -.125
(.233)
Mfg-NonDurable --039
(.041 )
Mfg-Durable .072
(.044)
Trans rtation --034
p0 (.041)
Retail .075**
(044) .
Finance -.041
(.03Q
Services -.031
(.038)
Public , 4-127"
Administration (047)
ITCSCONT .000
(.001)
ASP «046"
(.0201
ASB -.021
(.016)
ASA ~.OSO"
(.019)
DE ,. l 30""! .
1024)
IT CSCONTxASP .000
(.002)
IT CSCONTxASB .000
(.001)
IT CSCONTxASA -.001
(.002)
IT CSCONTxDE -.002
(.004)
R2 .039
F — ratio 6733‘"
N = 3,370
*p<JO
” p < .05
’1. p < 001

106

Table 8 summarizes the results of this regression model. The results show no
support for Hypotheses 6a and 6b, which predicted higher LTI for Craft job families.
The ITCSCONT variable’s regression weight is not significant nor are any of the
interaction terms significant. The control variables showed some significant relationships
with LTI eligibility. Organizational size was positively related to LTI eligibility, which
should be expected as larger firms are more likely to survive and have an opportunity for
LTI grants to be meaningful to employees. The public administration industry variable
was negative, which confirms the thought that public employers do not offer LTI grants,
probably because they do not have stock as a possible payout mechanism. Applications
Systems Analysis, Applications Systems Programming, and Data Entry families were less

likely to be eligible for LTI grants than the other families.

Hmthesis Set 7: ITCS‘ use and promotion to: r Legu_la_r awn

Hypotheses 7a and 7b predict a relationship between the use of ITCS and years
until promotion in Craft and Enterprise job families. Hypothesis 7a predicts that for
companies using IT CS in Craft families, the number of years to promotion will be longer
than for Craft families in companies that do not use ITCS. The reasoning for this
prediction is that companies hiring for Craft jobs will be more interested in obtaining the
most knowledge from their workers on up-to-date skills that may be more readily
available in the external marketplace. This means that external hires from available
IT CS may be more likely therefore making the pool of potential applicants for a job
greater. In the case of Enterprise families, Hypothesis 7b predicted that the number of

years to promotion will be shorter in companies using ITCS than in companies that do

107

not use ITCS. In the Enterprise families, it is reasoned that the most valuable skills of
these employees are their firm-specific knowledge that will limit the pool of applicants to
existing employees. In companies that use ITCS, the headcount of Enterprise employees
should be smaller, therefore making the possibility of promotion greater.

These hypotheses are tested using moderated regression similar to Hypotheses 6a
and 6b. The dependent variable is the average number of years to promotion for
employees in these job families as reported by the human resources professionals, who
completed the survey. Participants reported the average number of years to be promoted
fiom each job in a job family. The employee weighted average of this figure for all jobs
in a family is the calculated Average Years to Promotion variable. The weighted average
is used instead of a straight average to account for the possibility that uneven
distributions of employees in certain job categories might be under represented in a
straight average. The main independent variables are the continuous ITCSCONT
variable and the interaction variables of job family x ITCSCONT. Company size
(ln(total employment)). unemployment, industry, and job family variables are controlled
for in the first step. Company size is used as a control variable because larger firms will
have more employees to promote and possible more opportunities to promote than
smaller companies. Unemployment is used as a measure of the labor market conditions,
which may impact companies’ decisions to promote from within or hire a new employee

from outside.

108

 

Table 9: Moderated Regression Results for Hypothesis Set 7

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Variable Years to Promotion
Constant 1.479“I
(.528)
ln(Total Ees) --009
(.040)
Unem lo ent .134"
p m (.056)
Mining 1.6l4"
(.517)
Construction -1.331
(1.527)
Mfg-NonDurable .647”
(.382)
Mfg-Durable .252
(.409)
Transportation .955"
(.383)
Retail .398
(.426)
Finance 962"
(.357)
Services 1.133M
(355)
Public 1.349"
Administration 6457)
ITCSCONT -006
(.015)
ASP -.585**
(.201)
ASB -.215
(.166)
ASA -.017
(.196)
DE -.026
(.284)
IT CSCONTxASP .000
(.018)
IT CSCONTxASB -.024
(.016)
ITCSCONTxASA -.015
(.020)
ITCSCONTxDE -.067
(.061)
R2 .049
F—ratio 3313*"
N = 1,325
*p<JO
" p < .05
it! p < 00]

109

Years to Promotion = Bo + B! (ln(Total Employment) + B; (Unemp) + B3(Mining)
+ [34 (Construction) + [35 (Mfg-Durable) + B‘ (Mfg-NonDurable) + B7
(Transportation) + B. (Retail) + [39 (Finance) + 510 (Services) + Bu (Public
AdminiSmItion) + Bu (1T CSCONT) + Bis (ASP) + 1314(ASB) + firs (ASA) + Bic
(DE) + [317 (IT CSCONTxASB) + B13 (IT CSCONTxASA) + 01,
(ITCSCONTxASP) + [320 (ITCSCONTxDE) + Error
Tables 9 summarizes the results of this regressions. The tables include
unstandardized regression weights and the standard errors associated with them in
parentheses below. The results do not show any support for Hypothesis Set 7. None of
the interaction terms were significant nor did the ITCSCONT variable contribute
significantly to the Years to Promotion variable. In the control variables, Unemployment
was positively related to Years to Promotion suggesting that higher unemployment
causes employers to slow their promotion rates, perhaps because they will not be as likely
to find a replacement for the promoted employee. Also the mining, manufacturing non-
durable, finance, services, and public administration industries have significant positive

regression weights suggesting that it takes longer to be promoted in these industries than

other industries.

H Set 8: Extent 0 ITCS use and sat stnacture ra or far
Mm

Hypotheses 8a and 8b relate to the use of salary structures as part of employees’
compensation package. These hypotheses predict that characteristics of a salary structure
such as range overlap may be motivators for employee behaviors. Specifically the
amount of range overlap or lack of overlap between adjacent salary grades may

encourage employees to perform their jobs in such a way that they could be promoted. If

110

 

promotion is a reasonable goal, companies will want to encourage this type of behavior
and design their salary structure with little overlap between grades. If promotion is not a
reasonable goal, companies will not want to encourage these behaviors and will design
salary structures with much overlap between adjacent grades. In the case of Hypothesis
8a, more overlap between grades is predicted for Craft families when companies employ
ITCS. Hypothesis 8b predicts that less overlap will exist in Enterprise families, when
companies employ ITCS.

These hypotheses were tested using moderated regression models. The dependent
variable was the average overlap between adjacent grades for a job family, where overlap
is defined as:

Overlap = (Maximum of lower range — Minimum of higher range) / (Maximum of
higher range — Minimum of higher range)

The independent variables include dummy variables for each job family, the continuous
ITCSCONT variable, and interaction variables of job family x IT CSCONT. Company
size (ln(total employment)), unemployment, and industry variables are used as control
variables. Company size is included because larger companies may be more likely to
have human resource departments (Cohen & Pfeffer, 1986), that may have designed their
salary structures to match their compensation strategy. Unemployment is a general
indicator of the status of the external labor market, which may have and impact on

company policies. Table 10 shows the results of this regression.

111

Table 10:

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Moderated Regression Results for Hypothesis Set 8
Variable Range Overlap
Constant 603“" *
(.056)
ln(Total Ees) .053“
(.004)
Unem lo ent -.007*
P Ym (005)
Mining .059
1060)
Construction .365"
(. 164)
Mfg-NonDurable -.030
(.041)
Mfg-Durable .028
(.043)
Trans rtation «002
p0 (.043)
Retail -.017
(.043)
Finance .030
(.039)
Services -.016
(.038)
Public .003
Administration (046)
ITCSCONT .001
(.001)
ASP .017
(.020)
ASB -.020
(015)
ASA .024
(.019)
DE -.049*
(.026)
IT CSCONTxASP (4004*
(.002)
IT CSCONTxASB -.001
(.001)
ITCSCONTxASA -.003
(.002)
ITCSCONTxDE .005
(or 1)
RT .024
F _ ratio 2207*"
N = 1,325
‘p<J0
** p < .05

ttt p < 001

112

 

This regression shows no support for Hypotheses 8a and 8b. In contrast, the
interaction term for the Applications & Systems is negatively significant. This finding
reflects less overlap for a craft family when IT CS are being used in a firm. Possible firms
using ITCS for craft families are choosing to promote from within to keep institutional
knowledge. There is no support for Hypothesis 8b. The unemployment variable has a
significantly negative weight.

Overall, the results of the hypotheses proposed in this research do not provide
overwhelming evidence of any change in compensation practice or policies due to the
introduction of ITCS. While the confirmatory outcomes of Hypotheses la, lb, and 1c
indicate that companies certainly pay IT CS more than their regular employees, the rest of
the results show no subsequent change in behavior towards their regular employees. The
following table summarizes the results of all of the Hypotheses:

Table 11: Summary of Results for all Hypotheses

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Hypothesis 1a SImported
Hypothesis 1b Slipppl‘ted
Hypothesis 1c Partially SUPPCmed
Hypothesis 2a Not Supp<med
Hypothesis 2b Not Slipported
Hypothesis 3a Not Slippoftpd
Hypothesis 3b Not Supported
Hypothesis 4 Not Supported
Hypothesis 5a Not SUppoited
Hypothesis 5b Not sllppofled
Hypothesis 6a Not Supppfled
Hypothesis 6b Not Stlpported
Hypothesis 7a Not Slipported
Hypothesis 7b Not Slipppflpd
Hypothesis 88 Not Supported
Hypothesis 8b Not Slipported

 

 

 

 

113

 

Table 11 summarizes the results of all the hypotheses tested. It shows that most of the
hypotheses received no support for the relationships predicted. The next chapter will

discuss possible reasons for these findings.

114

CHAPTER 6 — CONCLUSION

The hypotheses, data, and methods outlined in this paper are intended to
quantitatively test for changes in the employment relationship between companies and
regular, IT workers as a result of IT CS use. Previous research in the areas of contingent
workers and ILMs are combined to develop a variety of potential organizational reactions
with respect to their IT employees’ wages and pay practices in the presence of ITCS.
Researchers-to—date have focused on the work relationships of the IT CS and have not
closely examined the impact IT CS may have on the regular employees. This research
attempts to fill that gap.

Workers do not negotiate work relationships in a vacuum. They are at least
minimally aware of labor market conditions when discussing the terms and conditions of
work with their employers. This may be especially true for IT workers because the
shortage of workers in this field has been a popular news topic for many years. Few if
any IT professionals can be ignorant of their advantageous bargaining positions due to
the current labor shortage. The relative ease with which most IT employees could
become contingent staff must be taken into account by both employees and companies
when agreeing to terms and conditions of work. This research was developed to answer
some of the questions about how this factor, ITCS, in the labor market is effecting the
work relationships of non-contingent IT workers.

The data were taken from the 1998 and 1999 Compensation Survey of IT

Professionals conducted by William M. Mercer, Inc. These surveys collected data on the

prevalence of ITCS, pay policies for IT employees in these organizations and the actual

115

base and total cash compensation for all of the organizations’ IT employees. Data fi'om
five job families in the survey were used: Applications Systems Analysis and
Programming, Applications Systems Analysis, Applications Systems Programming,
Software Systems Engineering, and Data Entry.

Each of these job families were classified as part of a craft (Applications Systems
Programming, Applications Systems Analysis, Applications Systems Analysis and

Programming, and Software Systems Engineering) or an enterprise (Data Entry) ILM.

 

The hypotheses were posed and tested at the job family level because previous research
suggests this is the most appropriate level of analysis (Osterman, 1983). Different
outcomes were predicted for job families based on inclusion in a craft or enterprise ILM.
The results, however, do not show support for any differentiation based on this craft or

enterprise distinction.

M52 otZResulm
The first group of hypotheses (1a, 1b, 1c, 2a, 2b, 3a, 3b, and 4) focused on the

actual wages paid to ITCS and regular, IT employees. The results confirmed that ITCS
are generally paid more than regular employees in the same job families (1a and 1b).
With the ITCS premium established, the focus turned to how organizations with IT CS
compensate their regular employees to address potential equity issues generated by the
presence of IT CS. Hypotheses sets 2 and 3 looked at the wages of the regular employees
in firms with ITCS in comparison to firms without ITCS. None of the ITCS variables in
H2 or H3 had any significant findings. This suggests that employers have not recognized

the impact that ITCS may have on their work relationships with their regular employees.

116

Further, Hypothesis 4 looked at the variability of pay in the presence of ITCS and found
no evidence of increasing ranges in pay when ITCS are present in a job family. It was
predicted that a larger range of pay would be present when ITCS were hired because
there would be less emphasis on internal equity. This appears not to be the case.
Together Hypotheses sets 2, 3 and 4 suggest that companies have made few changes in
their pay practices when they hire ITCS.

Hypotheses 5a and 5b looked for potential effects on ITCS use and employee
wages due to the centrality of IT services to an industry. The predictions were that
industries with high concentrations of IT workers (i.e., core industries) would hire fewer
ITCS than industries with lower concentrations of IT workers (i.e., peripheral industries).
For craft workers in core industries, this lower level of ITCS and higher job security
would translate into an efficiency wage premium, but not a compensating wage
differential premium. For craft workers in peripheral industries, higher levels of ITCS
and less job security would translate into a compensating wage differential, but not an
efficiency wage premium. For enterprise workers in core industries, higher ITCS will not
translate into an efficiency wage strategy or a compensating wage differential premimn.
In peripheral industries, the enterprise workers should see a compensating wage
differential premium. The results do not show any support for these wage predictions.
However, there is support for the idea that core industries hire fewer ITCS then
peripheral industries. This evidence suggests that further research should investigate this
relationship.

Besides pay, organizations use many approaches to incent and retain their

employees. The remaining hypotheses (H6a, H6b, H7a, H7b, H8a, and H8b) look at

117

alternatives to direct compensation that organizations may use as part of their total
remuneration package and how ITCS may change the structure of these programs. The
first alternative to direct compensation is long-term incentives (LTI). Companies may
use LTI as a means of retaining talent through company stock or other payouts tied to
long-term firm performance. The presence of IT CS was predicted to increase the LTI
offerings to craft workers, but have no impact on enterprise workers. The results indicate
no relationship between offering LTI grants and the presence of ITCS in a job family.

Hypotheses sets 7 and 8 address the use of promotions as a means of motivating
employees. Hiring ITCS is anticipated to lower the possibility of promotions in craft
families, but increase this possibility in enterprise families. The dependent variable in
hypotheses 7a and 7b is the average length of time before promotion in each job family.
The results show no changes to this average when IT CS are present Companies using
ITCS do not seem to be changing their promotion rates to increase or retard promotion
opportunities in their organizations.

Additionally, linked to the rate of promotions should be the design of the salary
structure. Salary structure design may play an important role in motivating employees to
want promotions (118a and H8b). Specifically, range overlap, which is an indicator of the
salary increase associated with promotions, is predicted to be greater in enterprise job
families where ITCS are present, but smaller in craft job families where ITCS are present.
Again, there is no evidence in these data to support these predictions.

Throughout the analyses some of the control variables have been significant.
Higher 1memployment has significantly, positively contributed to wages in Hypotheses

sets 2 and 3. This finding in counterintuitive, but may be because of the non-IT nature of

118

statewide unemployment rates. IT employment represent only a fraction of total
employment and therefore, it’s low levels may not be reflected in the statewide numbers.
Organizational size, as measured by total employees, significantly, positively
contributed to the variability of wages in Hypothesis 4. One explanation for larger
organizations having greater difficulty maintaining consistent wages is because their
larger headcormts will increase the complexity of their pay practices. More employees in
a job family is likely to increase the differences in formal education, skills and abilities
within the job family. These differences will in most companies be reflected in varying
pay rates to acknowledge different levels of contribution to the firms.
Finally, organizations in public administration seemed less likely to offer LTI
grants, which is not surprising since these organizations are not publicly traded and don’t
have stock that can be awarded. Industry variables are significant in some models, but
there is no consistent theme to which industries contribute to the dependent variables
presented here. No other control variables were consistently significant in the other

hypotheses. .

Possible £521an (a: r Them Findings
Overall, the evidence found in these analyses indicate that IT CS are paid

significantly more than regular employees. Further and more to the point of this research,
organizations appear to have done little for regular, IT employees to acknowledge these
inequities. While the introduction of ITCS should have an impact on how both
employers and employees view their working relationships, it seems that this impact has

not resulted in actions by companies. The theories described here suggest that IT CS in

119

the workplace should bring the impact of the open labor market more closely to bear on
the work relationships companies have with their regular employees. The data collected
do not show any evidence of this impact with regard to cash compensation, non-cash
programs, or salary structures.

The introduction of IT CS may not be enough of a reason to make changes to
organizational practices and policies because of several reasons. First, its possible tlmt
organizations have failed to recognize the impact ITCS may have on their regular
employees. These IT employees working alongside ITCS are exposed to the external
labor market in a very direct manner. ITCS are doing essentially the same work as the
employees, but they typically enjoy higher pay and greater flexibility in work hours and
conditions. The employees should realize that they could become ITCS if they wanted

and they could reap these same benefits. Organizations using ITCS would then be at a
higher risk of losing employees. that may be critical to their operations. The costs of
turnover, however, are rarely computed in organizations and the causes of turnover are
diffith to measure. Companies may not be able to concretely identify ITCS as a cause
of tumover nor may they be able to capture the costs due solely to ITCS related turnover.
For employees that decide to stay, there is a possibility that they would lower their
productivity level that makes them feel fairly compensated for their efforts in comparison
with the ITCS. Productivity levels are also difficult for organizations to measure making
it nearly impossible for a company to justify changing policies to address these concerns.

Another reason that organizations may not have changed their practices or

policies may be internal equity concerns across the organization. While employers may

recognize the problems that hiring IT CS may create within their IT departments, they

120

may choose to leave their policies unchanged because of concerns for equity with other
departments. The presence of ITCS may effect the work relationships of not only the IT
employees, but also employees in other departments. Special treatment of IT employees
because of ITCS may cause problems with employees in other departments feeling
unfairly treated not because ITCS have different pay rates, but because other employees
in the organization do. The potential exists for these non-IT employees to leave the
organization or lower their productivity to alleviate their feelings of unfairness. This may
effect all other departments in the organization, leaving the company open to the threat of
even larger turnover costs.

Companies may not have changed the practices and policies for their IT
employees because they have simply not had time. Changes to salary structures,
incentive programs or career progressions take some time to develop and implement.
Additionally, once these changes are made they do not have an immediate effect on
employees’ base pay and incentives. It is possible that the organizations in this sample
have not been using ITCS long enough to realize these problems and to make the
appropriate changes to their policies and practices.

Another explanation may be that because ITCS are not really employees of the
organizations, regular employees do not compare themselves to this group. This has been
an argument for using contingent staff when employers wanted to treat a group of
workers different from the majority of workers (Abraham, 1990). Contingent staff pay
' policies and practices are typically administered separately fi'om the pay policies of
regular employees and often are controlled outside of human resource departments.

These policies are so different from those of the regular employees that there may be a

121

distance created in employees’ minds between ITCS and themselves. This distance may
cause employees to not compare themselves to the contingent group as they are not really
employees of the organization and therefore are not part of the equity comparisons that
most employees make. Ifthis is truly the case with most employees, then organizations
will not nwd to adjust their pay practices and policies to acknowledge any inequities.

Research is needed in this area before this assumption can be safely made.

Future Research

Although the results of this research were not as anticipated, they do raise more
questions about how ITCS may be affecting regular, IT employees. It is unlikely that the
presence of IT CS has no impact on regular employees’ work relationships. Employees
working in these organizations must realize the differences in their work relationships
and those of the IT CS. After realizing these differences, it shouldn’t take long for these
employees to assess their own work relationships and, if necessary, make adj ustrnents to
ensure they feel fairly treated by their employers.

To understand the nature of this dynamic some additional research will be needed.
A few areas of fixture research are discussed here, but there are undoubtedly many more.
For example, do regular IT employees feel unfairly compensated in the presence of
ITCS? Since the work of regular employees and ITCS is essentially the same, it is
reasonable to expect that these groups would anticipate being paid comparably. Regular
employees, however, are often being paid less than the II‘ CS in cash compensation as
validated by the current research. Much of this difference is attributed to the cost of

benefits, which is not being paid for the ITCS and a premium to ITCS for higher

122

unemployment risks. Researchers should ask if in fact employees recognize the value of
their benefits and the additional job security that they enjoy over and above their cash
compensation? And do they believe that these benefits make up for the difference in
pay? What are the implications for organizations if employees do feel their benefits
compensate them for these differences in pay or not?

Does the productivity of regular employees change when ITCS are employed? If
regular employees feel that they are being under compensated in relation to ITCS, one
possible reaction would be to lower their productivity to equalize the ratio of input to
output with what they believe that ratio to be for ITCS. In this case, employees’ will take
longer to complete projects, to increase the wages required for the same level of output.
Also, could possible animosity between regular, IT employees and ITCS inhibit the
productivity of the organization? Such animosity could exist if the regular, IT employees
feel that ITCS are taking away their opportunities to be promoted or learn new skills or if
the regular employees think that high contractor rates are decreasing the amount of
money available for regular employees’ wages. Animosity may lead to poor
communication between regular employees and ITCS, which may slow the progress of
projects and productivity where their interaction is part of the process.

How are ITCS integrated with established work groups? Cooperation in the work
environment is critical to the success of most businesses. Even under the best
circumstances, where no animosity exists, the role of ITCS within organizations is
usually unclear because they are outside the established work relationships and
communication networks. How do companies integrate the skills of ITCS with the skills

of regular, IT employees? Would companies benefit from putting their ITCS through a

123

short socialization period to help them interact with their regular, IT employees? What
additional skills are required of managers in organizations with IT CS? Would it be
beneficial in organizations with ITCS to put managers through some team building
training to help facilitate the interaction between their regular employees and IT CS?

One reason for using ITCS is as a screening tool in the selection process. Once an
ITCS is hired as a regular IT employee, it would be interesting to know how their change
from contingent to regular worker differs from other new hires’ entering organizations.
Are ITCS, who are hired on as regular employees of organizations, accepted as members
of the organization? Is their socialization period longer or shorter than employees, who
did not have previous relationships with the hiring organizations? If there is animosity
between regular employees and ITCS, then a longer socialization period might be
anticipated. However, because the new employees, ex-ITCS, may be more
knowledgeable about the organization’s business because of their prior involvement, it
might be anticipated to take less time for them to be socialized In the long-run, is the
practice of hiring ITCS cost effective for organizations?

What changes are there in turnover rates of regular employees? Ifthe inequities
between regular employees and IT CS are great, regular employees may be motivated to
leave the organization rather then work under these conditions. Employers may be
interested to know how great the level of inequity can become before there are increases
in turnover rates. Further, it would be interesting to know where employees, who
voluntarily leave these organizations, go? Do they become ITCS or do they go to other
organizations as regular employees? Are there personality traits that would help

organizations predict which employees would leave the organization under these

124

conditions of inequity? Could these traits also predict where they would go? For
example, some people are more risk averse than other people. It’s possible that this trait
would give employers an idea of which employees are more likely to leave and which are
more likely to stay. If employers are administering personality tests during the selection
process, then they might have some foreknowledge of this situation. For companies that
employ a great number of ITCS as part of their strategy, it might be helpful in the

selection process to weed out applicants, who are not likely to thrive in this environment.

Conclusions

While the results of this research did not confirm the hypotheses, they are still
interesting. The findings presented show that organizations in this sample have not
addressed the issue of inequities when bringing in IT CS. ILM theory would predict that
organimtions would need to address these problems in order to preserve the internal
equity necessary for ILMs to be effective. The organizations in this sample, however,
have generally not changed their approaches to pay or promotion for their regular, IT
employees when they use If CS. There are many possible reasons why organizations
have not made some of the predicted changes. Organizations may not clearly understand
how using ITCS could effect their regular, IT employees. Companies may place a
priority on preserving internal equity across departments within their organizations.
Organizations may have only recently begun using ITC S, and therefore may not have had
the time to alter their policies and have these new policies impact pay and promotions.
Or companies may believe that by using contingent staff they are sufficiently removing

these ITCS from the mental comparisons made by employees regarding the equity of the

125

work relationships in their organizations. Future research may begin to shed more light
on these issues.

It is important to understand how the use on contingent workers affects all of the
parties involved with this staffing choice. In the past researchers have looked at the
ramifications of these temporary work relationship for the contingent staff, but few have
acknowledge that these contingent workers bring an additional factor into the normal
working relationship most employers have with their employees. This research is
intended to investigate how the addition of a direct link to the external labor market
through ITCS impacts the work relationships of regular employees. Contingent staff
models are one of the few work relationships that truly imitate the spot market of
neoclassical economics. Organizations should be interested in how this contingent
staffing model interacts with the more traditional ILM approach because this knowledge
could help them create staffing strategies that can take advantage of the best aspects of
both staffing models. It is hoped that firture research will continue to explore these
relationships to create to better picture of how these models may be used to

organizations’ and employees’ benefits.

126

APPENDICES

127

AM A - Sur_v_e£ Qgesdons
Survey Participants were asked to supply information regarding their companies use of
ITCS, policies and practices for job families, and actual pay rates for all employees in the
job families included in this study. The following pages show the survey questions

completed by survey participants.

Contract Stafl
How many IT contractors did you employ on average over the last 12 months?_

’_—-.——

Indicate those job families or areas of expertise for which you contract outside staff. For
each area indicated, calculate the percent of overall staff which is contracted (number of
hours worked by contract personnel expressed as a percent of total hours worked). Also

indicate the contract base salary as a percent of the in-house base salary. For example, if

the per hour rate for contract personnel is twice the rate for in-house personnel, enter

 

200%.
Hours Worked by Contract Percent Contract Pay
Penanclasl’ereeatof RatetoAverage Ill-HM“
Wilmette mm mm;

Applications Systems Analysis and
Programming _ _ 96 __ _ _ 96
(Job Codes 009-017)
Applications SystemsAnslysis (Job __% _____%
Codes 021-027)
Applicationsl’rogrammingaob ____% _____%
Codes 031-037)
SofiwareSystemsEngineeringUob ___% __.__%
Codes 040.047)
DataEnn'yOobCodesO7l-074) __% ___%
Other (specify): _ __ 96 __ __ __ 96

Job Family Practices

128

Salary Range: If a traditional salary structure is used to manage pay for this job (that is,
a structure with a minimum, midpoint, and maximum and range spreads of
approximately 50% - 80%), then enter the annual band minimum and maximum
of the range for this job. Leave the “Bromanding” fields blank.

Long-term Incentive : Eligibility - Enter a “Y” or “N” to indicate whether this job is
eligible for a long-term incentive. This refers to an incentive plan with awards
related to performance against selected criteria over a period of three years or
more.

Stock Option Eligibility - If the job is eligible for a long-term incentive, enter a
“Y” or “N” to indicate whether this job is eligible for stock options as a long-term
incentive (e.g., not stock awards that can be immediately transferred to cash). This
refers to the right to purchase a fixed number of shares of company stock at a
fixed price over a specific period of time. There may be a time requirement
(vesting period).

Cash Eligibility - If the job is eligible for a long-term incentive, enter a “Y” or
“N” to indicate whether this job is eligible for cash as a long-term incentive.

Average Number of Years in Job Prior to Promotion to Next Level: Indicate the
average time an employee spends in a job from the date of entry into the job, to
the time promoted out of the job. Although promotional timing may be based on
individual performance, please try to establish an average or typical time. Do not
simply report the number of years that the current employee has held the
job. This data item is collected for selected jobs only.

Employee Salaries

Base Salary: Enter the base pay of each employee you have matched to this job efl‘ective
as of April 1, 1999. Report the amount prior to any voluntary pay reductions. Do
not report average salaries. Do not include part-time employees or contracted staff
from outside sources.

Report the data in annual, whole dollars (for example, $32,500).

Short-term Incentive: Enter the amount of incentive earned (but not necessarily paid
out) in the most recently completed fiscal year. This refers to the amount earned
through participation in a short-term incentive plan (e.g., an annual incentive plan
or a project milestone bonus program) with awards related to performance
against selected criteria over a period of typically a year or less. Exclude any non-
perforrnance based incentives such as holiday bonuses. Do not include sign-on
bonuses, retention bonuses, profit sharing, or mandatory deferred compensation

payments.

If the employee was eligible for an incentive, but no incentive was given, enter
“0”. If the employee is not incentive eligible, leave this field blank.

Report the incentive amount in annual, whole dollars.

129

Organizational Information

Individual Supplying Information

 

 

 

 

Name:

Title:

Telephone Number: ( ) Extension:

Fax Number: ( ) e-mail Address:

 

 

Effective Date of Data: April 1, 1999 Participation Deadline: May 14, 1999

1. Major Type of Industry (See page 59 of the survey guide for a detailed listing.)
Circle the number indicating your primary industry area.

01 Agriculture 13 Government 23 Mfg - Paper & Allied
02 Business/Information 14 Healthcare Products
Services 15 Hospitality/Restaurant 24 Mining/MillingISmchina
03 Chemical 16 Insurance 25 Nonprofit, Miscellaneous
04 Computer 17 Manufacturing-Aerospace 26 Pharmaceutical
Software/Services 18 Mfg - Computer/0mm 27 Real Estate
05 Construction Equipment 28 Research & Development
06 Corraultingflegal/Account l9 Mfg - Durable, 29 Retail
i118 Miscellaneous 30 Service, Miscellaneous
07 Diversified 20 Manufactrning - 31 Telecommrmications
08 Education Electronics 32 Transportation
09 Energy 21 Manufacturing - Food 33 Utilities
10 Engineering/Architecunal 22 Mfg - Nondurable, 34 Wholesale Distribution
11 Entertainment and Media mum 35 om
12 Finance/Banking

that is applicable to your organization, enter the four-digit code.

2.

If you know the Standard Industrial Classification (SIC) code

Type of Organization (check only one)

D

D

Parent/Stand-Alone (The highest reporting entity with no parent organization
above it. Typically management and board of directors are responsible to
shareholders for overall company performance.)

Subsidiary (An independent entity with majority interest held by a parent; often
has multiple business units and is responsible for all organization functions.
Although a subsidiary may ultimately roll up into a parent entity, management

performs all functions of a stand-alone organization and is accountable to
parent company for operations and financial performance.)

Division (An independent entity accountable for all organization functions
typically for a select group of product lines. Generally, a division consists of a
single profit center or operating unit which is fully accountable to a parent or
subsidiary.)

130

If your organization is a subsidiary or division, enter the name and location of your
parent organization.

 

 

 

Parent Name:
Parent Location: BUS. ( , )
city state
CI Foreign ( , )

city country
Is your organization or parent organization included in the 1998 Fortune 1000
making? ............................................................... ClYes ......................... D No
Ownership (check only one)

D Publicly traded company (a for-profit entity whose stock is publicly traded on a
US. or foreign stock exchange)

CI Privately held company (a for-profit entity that is privately owned)
CI Government, education, or nonprofit organization

Annual Dollar Volume

Check the appropriate volume measure for the annual dollar volume of the reporting
entity (corporation, division, or subsidiary). Please report only one volume type per
participation package.

El Total Revenues/ Sales Cl Premiums (for Insurance)

D Assets (for Banking and Financial) Cl Operating Budget (for Nonprofit,

Government, and Education)

Enter the Annual Dollar Volume (domestic and international)
fortheorganization ........................................... $ ,____,____,000

Total Employment

Enter the total number of US. full-time equivalents (FTEs) in your reporting entity
The F'I'Es reported should be based on the same entity represented in question 4.

——_’_-—

131

AM B - Job mamas

Applications Systems Analysis and Programming F amiy

009

010

011

Applications Systems Analysis and Programming Director - Responsible for
applications systems analysis and programming activities. Reviews systems
development project requests and coordinates schedules and related departmental
activities. Provides overall direction and guidance to assigned project managers.
Reviews and evaluates work of subordinate staff and prepares performance
reports. Prepares activity and progress reports regarding applications systems
analysis and programming sections. Normally reports to the Corporate
Information Systems Executive (001, 002) or Divisional Information Systems
Executive (003).

Applications Systems Analysis and Programming Manager - Responsible for
applications systems analysis and programming activities for a group or section
Responsible for feasibility studies, time and cost estimates, and the establishment
and implementation of new or revised applications systems and programs. Assists
in projecting software and hardware requirements. Assigns personnel to various
projects and directs their activities; reviews and evaluates their work and prepares
performance reports. Confers with and advises subordinates on administrative
policies and procedures, technical problems, priorities and methods. Consults with
personnel in other information systems sections to coordinate activities. Prepares
activity and progress reports regarding the activities of the applications systems
and programming section Typically reports to the Corporate Information Systems
Executive (002), Divisional Information Systems Manager (003 ), Systems
Analysis and Programming Director (008, 009).

Applications Systems Analysis and Programming Supervisor - Supervises
activities of all applications systems analysis and programming personnel for a
major project, several smaller projects, or a small department. Responsible for
quality assurance. Makes decisions on personnel actions (hiring, terminations,
promotions, etc. ). Controls revenues and/or expenses within an operating unit and
responsible for meeting budget goals and objectives. Provides input to policy
level direction regarding standards, budget constraints, etc. Typically reports to
Applications Systems Analysis and Programming Manager (010), Applications
Systems Analysis and Programming Director (009) or Systems Analysis and
Programming Director (008).

132

012

013

014

016

Applications Systems Analyst/Programmer - Lead - F ormulates/defines system
scope and objectives for assigned projects. Devises or modifies procedures to
solve complex problems considering computer equipment capacity and
limitations, operating time and form of desired results. Prepares detailed
specifications from which programs will be written. Responsible for program
design, coding, testing, debugging and documentation Has full technical
knowledge of all phases of applications systems analysis and programming. Has
good understanding of the business or function for which applications is designed
Also has duties instructing, directing and checking the work of other systems
analysis and programming personnel. Responsible for quality assurance review.
Responsible for directing and monitoring the work of team members. May be
responsible for project completion and user satisfaction.

Applications Systems Analyst/Programmer - Staff Specialist - Top level
technical expert in one or more highly specialized areas of applications systems
analysis and programming. Acts independently under general direction. Provides
technical leadership on complex projects. May act as expert in business or
functional area. Formulates/defines system scope and objectives. Devises or
modifies procedures to solve complex problems considering computer equipment
capacity and limitations, operating time and form of desired results. Prepares
detailed specifications from which programs will be written. Responsible for
program design, coding, testing, debugging and documentation. Has full technical
knowledge of all phases of applications systems analysis and programming. May
be responsrble for multiple phases of a project May have duties instructing,
directing and checking the work of other applications systems analysis and
programming personnel. May have quality assurance review responsibilities.

Applications Systems Analyst/Programmer - Senior - Under general direction,
formulates/defines system scope and objectives. Devises or modifies procedures
to solve complex problems considering computer equipment capacity and
limitations, operating time and form of desired results. Prepares detailed
Specifications from which programs will be written Designs, codes, tests, debugs
and documents those programs. Competent to work at the highest technical level
of all phases of applications systems analysis and programming activities. May be
responsible for completion of a phase of a project. Regularly provides guidance
and training to less experienced analyst/programmers.

Applications Systems Analyst/Programmer - Intermediate - Under general
supervision, formulates and defines system scope and objectives through research
and fact-finding to develop or modify moderately complex information systems.
Prepares detailed specifications from which programs will be written Designs,
codes, tests, debugs, documents and maintains those programs. Competent to
work on most phases of applications systems analysis and programming activities,
but requires instruction and guidance in other phases.

133

017

Applications Systems Analyst/Programmer - Associate - Under immediate
supervision, assists in research and fact-finding to develop or modify information
systems. Assists in preparing detailed specifications from which programs will be
written Designs, codes, tests, debugs, documents and maintains those programs.
Staffed by beginners who have had sufficient educational background and/or
experience to qualify them to start in application systems analysis and
programming.

Applications Systems Analysis Family

021

022

023

Applications Systems Analysis Supervisor - Supervises activities of
applications systems analysis personnel for a major project, several smaller
projects, or a small department. Responsible for quality assurance. Makes
decisions on personnel actions (hiring, terminations, promotions, etc.). Controls
revenues and/or expenses within operating unit and responsible for meeting
budget goals and objectives. Provides input to policy level direction regarding
standards, budget constraints, etc.

Applications Systems Analyst - Lead - Formulates/defines system scope and
objectives based on user defined needs. Devises or modifies procedures to solve
complex problems considering computer equipment capacity and limitations,
operating time and form of desired results. Prepares detailed specifications from
which programs will be written. Analyzes and revises existing system logic
difficulties and documentation as necessary. Has full technical knowledge of all
phases of applications systems analysis. May use CASE tools. Also has duties
instructing, directing and checking the work of other systems analysis personnel.
Responsible for quality assurance review. Acts as project leader for projects with
small budgets or limited duration.

Applications Systems Analyst - Staff Specialist - Top level technical expert in
one or more highly specialized phases of applications systems analysis. Acts
independently under general direction. Provides technical consulting on complex
projects. F ormulates/defines system scope and obj ectives. Often acts as business
expert and assists users in defining needs. Devises or modifies procedures to solve
complex problems considering computer equipment capacity and limitations,
operating time and form of desired results. Prepares detailed specifications from
which programs will be written. Analyzes and revises existing system logic
difficulties and documentation as necessary. Has full technical knowledge of all
phases of applications systems analysis. May use CASE tools. May have duties
instructing, directing and checking the work of other applications systems
analysis personnel. May have quality assurance review responsibilities.

134

024

026

027

Applications Systems Analyst - Senior - Under general direction,
formulates/defines system scope and objectives based on user needs. Devises or
modifies procedures to solve complex problems considering computer equipment
capacity and limitations, operating time and form of desired results. Prepares
detailed specifications from which programs will be written. Analyzes and revises
existing system logic difficulties and documentation as necessary. Competent to
work at the highest technical level of all phases of applications systems analysis
activities. May use CASE tools.

Applications Systems Analyst - Intermediate - Under general supervision,
formulates and defines system scope and objectives through research and fact-
finding to deveIOp or modify moderately complex information systems. Prepares
detailed specifications from which programs will be written Analyzes and revises
existing system logic difficulties and documentation as necessary. Competent to
work on most phases of applications systems analysis activities, but requires
instruction and guidance in other phases. May use CASE tools.

Applications Systems Analyst - Associate - Under immediate supervision,
assists in research and fact-finding to develop or modify information systems.
Assists in preparing detailed specifications from which programs will be written.
Analyzes and revises existing system logic difficulties and documentation as

. necessary. This level is staffed by beginners who have had sufficient educational

background and/or experience to qualify them to start in applications systems
analysis. May use CASE tools.

Applications Programming Family

031

032

Applications Programming Supervisor - Supervises activities of applications
programming personnel for a major project, several smaller projects, or a small
department. Responsible for quality assurance. Makes decisions on personnel
actions (hiring, terminations, promotions, etc). Controls revenues and/or
expenses within operating unit and responsible for meeting budget goals and
objectives. Provides input to policy level direction regarding standards, budget
constraints, etc.

Applications Programmer - Lead - Devises or modifies procedures to solve
complex problems considering computer equipment capacity and limitations,
operating time and form of desired results. Responsible for program design,
coding, testing, debugging and documentation. Has full technical knowledge of
all phases of applications programming. Also has duties instructing, directing and
checking the work of other programming personnel. Responsible for quality
assurance review. Acts as project leader for projects with small budgets or limited
duration Responsible for directing and monitoring the work of team members.
Note: This position does not perform systems analysis functions.

135

033

034

037

Applications Programmer - Staff Specialist - Top level technical expert in one
or more highly specialized phases of applications programming. Acts
independently under general direction Provides technical consulting on complex
projects. Devises or modifies procedures to solve complex problems considering
computer equipment capacity and limitations, operating time and form of desired
results. Responsible for program design, coding, testing, debugging and
documentation. Has full technical knowledge of all phases of applications
programming. May have duties instructing, directing and checking the work of
other applications programming personnel. May have quality assurance review
responsibilities. Note: This position does not perform systems analysis functions.

Applications Programmer - Senior - Under general direction, devises or
modifies procedures to solve complex problems considering computer equipment
capacity and limitations, operating time and form of desired results. Designs,
codes, tests, debugs and documents those programs. Competent to work at the
highest technical level of all phases of applications programming activities. Note:
This position does not perform systems analysis functions.

Applications Programmer - Intermediate - Under general supervision, modifies
moderately complex applications programs from detailed specifications. Codes,
tests, debugs, documents and maintains those programs. Competent to work on
most phases of applications programming activities, but requires instruction and
guidance in phases. Note: This position does not perform systems analysis
functions.

Applications Programmer - Associate - Under immediate supervision, modifies
applications programs from detailed specifications. Codes, tests, debugs,
documents and maintains those programs. This level is staffed by beginners who
have had sufficient educational background and/or experience to qualify them to
start in applications programming. Note: This position does not perform systems
analysis functions.

136

Software Systems Engineering Family

040

041

042

Software Systems Engineering Manager - Responsible for all software systems
programming activities. Applications generally affect the overall operating
system, such as SOphisticated file maintenance routines, advanced scientific
software, large telecommunications networks and computer accounting.
Responsible for meeting budget goals and objectives. Provides input to policy
level direction regarding standards, budget constraints, etc. Makes personnel
decisions. Assigns personnel to projects and directs their activities. Projects
software and hardware requirements in conjunction with other information
systems managers. Develops standards for all software system applications and
provides technical guidance to the information systems staff. Directs the interface
of software systems with the hardware configuration and the applications systems.
Additional areas of responsibility include: configuration/capacity planning,
software products evaluation, systems performance analysis and optimization
Prepares activity and progress reports for software systems programming
activities. May report to Computer Operations Manager (061) or Director of IS
Operations (049).

Software Systems Engineering Supervisor - Supervises activities of all software
systems programming personnel for a major project, several smaller projects, or a
small department. Responsible for quality assurance. Makes decisions on
personnel actions (hiring, terminations, promotions, etc.). Controls revenues
and/or expenses within operating unit and responsible for meeting budget goals
and objectives. Provides input to policy level direction regarding standards,
budget constraints, etc.

Software Systems Engineer - Lead - Formulates/defines specifications for
complex software programming applications or modifies/maintains complex
existing applications using engineering releases and utilities from the
manufacturer. Responsible for program design, coding, testing, debugging and
documentation Usually responsible for applications dealing with the overall
operating system, such as sophisticated file maintenance routines, large
telecommunications networks, computer accounting and advanced
mathematical/scientific software packages. Has full technical knowledge of all
phases of software systems programming applications. Also has duties
instructing, directing and checking the work of other operating systems
programming personnel. Responsible for quality assurance review and the
evaluation of new and existing software products. Acts as project leader for
projects with small budgets or limited duration.

137

047

Software Systems Engineer - Staff Specialist - Top level technical expert in one
or more highly specialized phases of software systems programming. Acts
independently under general direction Provides technical consulting on complex
projects. F ormulates/defines specifications for complex operating software
programming applications or modifies/maintains complex existing applications
using engineering releases and utilities from the manufacturer. Responsible for
program design, coding, testing, debugging and documentation Usually
responsible for applications dealing with the overall operating system, such as
sophisticated file maintenance routines, large telecommunications networks,
computer accounting and advanced mathematical/scientific software packages.
Has full technical knowledge of all phases of software systems programming
applications. May have duties instructing, directing and checking the work of
other software systems programming personnel. May have quality assurance
review and/or new and existing software product evaluation responsibilities.

Software Systems Engineer - Senior - Under general direction,
formulates/defines specifications for complex operating software programming
applications or modifies/maintains complex existing applications using
engineering releases and utilities from the manufacturer. Designs, codes, tests,
debugs and documents those programs. Usually responsible for applications
dealing with the overall operating system, such as sophisticated file maintenance
routines, large telecommunications networks, computer accounting and advanced
mathematical/scientific software packages. Competent to work at the highest
technical level on all phases of software systems programming applications. May
have responsibility for the evaluation of new and existing software products. May
assist other systems programmers to effectively utilize the system's technical
software.

Software Systems Engineer - Intermediate - Under general supervision, works
from specifications to develop or modify moderately complex software
programming applications. Assists with design, coding, benchmark testing,
debugging and documentation of programs. Applications generally deal with
utility programs, job control language, macros, subroutines and other control
modules. Competent to work on most phases of software systems programming
applications, but requires instruction and guidance in other phases.

Software Systems Engineer - Associate - Under immediate supervision, works
from specifications to assist in developing and modifying operating software
programming applications. Assists with design, coding, benchmark testing,
debugging and documentation of programs. Applications generally deal with
utility programs, job control language, macros, subroutines and other control
modules. May customize purchased applications and assist in maintaining
program libraries and technical manuals. Staffed by beginners who have had
sufficient educational background and/or experience to qualify them to start in

operating systems programming.

138

Data Entry Family

071

072

073

074

Data Entry Supervisor - Supervises all data entry activities. Assigns work to
personnel and directs their activities; reviews and evaluates their work and
prepares performance reports. Normally reports to the Computer Operations
Manager (061) or Production Control Supervisor (054). NOTE: If incumbent also
supervises data control activities, match to Production Control Supervisor (054).

Data Entry Operator - Lead - Under general direction, has full technical
knowledge of data entry devices, but has duties of instructing, directing and
checking the work of other data entry operators. Assists in scheduling data entry
functions.

Data Entry Operator - Senior - Under general supervision, operates data entry
devices in recording a variety of data; verifies data entered; performs related
clerical duties.

Data Entry Operator - Under direct supervision, operates data entry devices in
recording a variety of data; verifies data entered; performs related clerical duties.

139

ndth—Ge hie!

Une In
W

ITPercento E 1 meta

Geographical Distribution of Employees

State

Alaska
Alabama
Arkansas
Arizona
California
Colorado
Connecticut
Washington, DC
Delaware
Florida
Georgia
Hawaii

Iowa

Idaho

Illinois
Indiana
Kansas
Kentucky
Louisiana
Massachusetts
Maryland
Maine
Michigan
Minnesota
Missouri
Mississippi
Montana
North Carolina
North Dakota
Nebraska
New Hampshire
New Jersey
New Mexico
Nevada

New York
Ohio
Oklahoma
Oregon
Pennsylvania
Rhode Island

Data % of Employees

0.02%
1.26%
1.68%
0.75%
5.43%
1.30%
2.61%
1.11%
0.19%
4.32%
4.21%
0.18%
1.32%
0.54%
7.71%
0.82%
0.45%
0.51%
0.10%
3.67%
1.88%
0.11%
3.00%
5.85%
2.83%
0.05%
0.09%
2.82%
0.02%
0.25%
0.26%
4.68%
0.15%
0.02%
4.43%
5.99%
0.47%
0.97%
4.47%
0.73%

140

National % of

Employees
0.23%

1.58%
0.88%
1.63%
11.87%
1.61%
1.24%
0.20%
0.28%
5.23%
2.94%
0.44%
1.15%
0.47%
4.52%
2.26%
1.02%
1.40%
1.51%
2.38%
1.99%
0.47%
3.63%
1.95%
2.08%
0.92%
0.34%
2.75%
0.25%
0.66%
0.47%
3.02%
0.61%
0.66%
6.46%
4.08%
1.19%
1.28%
4.32%
0.36%

Geographical Distribution of Employees (cont ’d).

South Carolina
South Dakota
Tennessee
Texas

Utah

Virginia
Vermont
Washington
Wisconsin
West Virginia
Wyoming
Total

141

0.37%
0.01%
1.86%
9.88%
0.88%
4.83%
0.10%
1.58%
3.10%
0.10%
0.00%

100.00%

1.43%
0.29%
1.99%
7.33%
0.76%
2.52%
0.24%
2.19%
2.14%
0.58%
0.19%

100.00%

Industry Distribution of Employees

Industry Segment

Mining

Construction

Manufacturing - Durable
Manufacturing - Non-Durable
Transportation & Utilities
Wholesale Trade

Retail Trade

Finance, Insurance, and Real Estate
Services

Public Administration

Total

142

Data % of Employees

0.08%
0.68%
24.05%
5.12%
6.93%
0.35%
3.73%
12.19%
42.46%
4.45%
100.00%

National % of
Employees
0.47%
4.62%
8.97%
6.06%
5.22%
5.42%
17.53%
5.85%
29.78%
16.06%
100.00%

Unemployment Distribution of Employees

State 1998 1999
Alaska 5.8% 6.0%
Alabama 3.9% 4.2%
Arkansas 5.4% 4.1%
Arizona 3.8% 3.9%
California 5.8% 5.6%
Colorado 3.7% 2.8%
Connecticut 3.2% 3.1%
Washington, DC 8.6% 6.3%
Delaware 3.7% 3.0%
Florida 4.1% 4.0%
Georgia 3.9% 3.6%
Hawaii 6.1% 5.5%
Iowa 2.8% 2.7%
Idaho 5.4% 5.0%
Illinois 4.2% 3.8%
Indiana 3.0% 2.5%
Kansas 3.8% 3.3%
Kentucky 4.6% 4.1%
Louisiana 5.6% 4.7%
Massachusetts 3 .0% 2.7%
Maryland 4.4% 3.5%
Maine 4.7% 4.0%
Michigan 3.5% 3.8%
Minnesota 2.6% 2.2%
Missouri 4.4% 3.1%
Mississippi 4.9% 4.0%
Montana 5.6% 5.4%
North Carolina 3.3% 2.6%
North Dakota 3.3% 2.7%
Nebraska 2.2% 2.0%
New Hampshire 3 .2% 2.6%
New Jersey 4.5% 4.2%
New Mexico 5.8% 6.0%
Nevada 4.7% 4.1%
New York 5.6% 4.9%
Ohio 3.8% 4.0%
Oklahoma 4.4% 4.0%
Oregon 5.7% 5.4%
Pennsylvania 4.6% 4. 1%
Rhode Island 4.9% 2.8%
South Carolina 3.1% 3.6%
South Dakota 2.9% 2.4%
Tennessee 4.0% 3.8%
Texas 4.4% 4.3%
Utah 3.9% 3.0%
Virginia 2.4% 2.4%

143

Unemployment Distribution of Employees (cont ’d).

Vermont
Washington
Wisconsin
West Virginia
Wyoming

144

3.5%
4.6%
3.1%
7.0%
5.1%

2.8%
4.4%
3.3%
6.9%
4.8%

Information Technology Percent of Employees by Industry

Two-digit SIC

07: Agricultural Services

10: Metal Mining

12: Coal Mining

13: Oil and Gas Extraction

14: Mining and Quarrying of Nonmetallic Minerals

15: Building Construction

16: Heavy Construction Other than Building Construction

17: Construction Special Trade Contractors

20: Manufacturing — Food and Kindred Products

21: Manufacturing - Tobacco Products

22: Manufacturing - Textile Mill Products

23: Manufacturing — Apparel and Other Finished Products Made From Fabric
24: Manufacturing - Lumber and Wood Products, Except Furniture

25: Manufacturing - Furniture and Fixtures

26: Manufacturing - Paper and Allied Products

27: Manufacturing - Printing, Publishing, and Allied Industries

28: Manufacturing - Chemicals and Allied Products

29: Manufacturing — Petroleum Refining and Related Industries

30: Manufacturing — Rubber and Misc Plastics Products

31 : Manufacturing - Leather and Leather Products

32: Manufacturing - Stone, Clay, Glass, and Concrete Products

33: Manufacturing —- Primary Metal Industries

34: Manufacturing - Fabricated Metal Products, Except Machinery and
Transportation Equipment

35: Manufacturing - Industrial and Commercial Machinery and Computer
Equipment

36: Manufacturing -— Electronic and other Electrical Equipment and
Components, Except Computer Equipment

37: Manufacturing - Transportation Equipment

38: Manufacturing - Measuring, Analyzing, and Controlling Instruments;
Photographic, Medical and Optical Goods

39: Manufacturing — Miscellaneous Manufacturing Industries

40: Railroad Transportation

41: Local and Suburban Transit and Interurban Highway Passenger
Transportation

42: Motor Freight Transportation and Warehousing

44: Water Transportation

45: Transportation by Air

46: Pipelines, Except Natural Gas

47: Transportation Services

48: Communications

49: Electric, Gas, and Sanitary Services

50: Wholesale Trade — Durable Goods

5]: Wholesale Trade — Non Durable Goods

52: Retail Trade - Building Materials, Hardware, Garden Supply , and Mobile
Home Dealers

53: Retail Trade — General Merchandise Stores

54: Retail Trade - Food Stores

55: Retail Trade - Automotive Dealers and Gasoline Service Stations
56: Retail Trade - Apparel and Accessory Stores

57: Retail Trade - Home Furniture, Furnishings, and Equipment Stores
58: Retail Trade —- Eating and Drinking Places

59: Retail Trade — Miscellaneous Retail

145

% IT Employees

0.1 1%
0.96%
0.27%
2.00%
0.32%
0.15%
0.18%
0.06%
0.51%
5.65%
0.56%
0.62%
0.41%
0.75%
0.80%
2.89%
1.65%
1.21%
0.60%
0.73%
0.59%
0.95%
0.78%

4.27%
3.83%

1.70%
4.46%

0.98%
0.64%
0.24%

0.54%
0. 73%
1 09%
2.42%
2.02%
4.88%
2.24%
3.59%
1 .46%
0.52%

0.35%
0.16%

0.085
0.28%
2.95%
0.01%
0.74%

60:
61:
62:
63:
: Insurance Agents, Brokers, and Services
65:
67:
70:
72:
73:
75:
76:
78:
79:
80:
81:
82:
83:
: Museums, Art Galleries, and Botanical and Zoological Gardens
86:
87:

88
90

Information Technology Percent of Employees by Industry (cont ’d).

Depository Institutions

Non-depository Credit Institutions

Security and Commodity Brokers, Dealers, Exchanges, and Services
Insurance Carriers

Real Estate

Holding and Other Investments Oflices
Hotels, Roaming Houses, Camps, and Other Lodging Places
Personal Services

Business Services

Automotive Report, Services, and Parking
Miscellaneous Repair Services

Motion Pictures

Amusement and Recreation Services
Health Services

Legal Services

Educational Services

Social Services

Membership Organizations

Engineering, Accounting, Research, Management, and Related Services
Private Households

Public Administration

146

4.32%
3.54%
6.01%
8.31%
3.55%
0.46%
7.34%
0.18%
0.21%

12.15%

0.26%
0.68%
1.03%
0.29%
0.82%
1.03%
0.91%
0.39%
1.04%
1.27%

6.295
4.60%
2.16%

BIBLIOGRAPHY

147

BIBLIOGRAPHY

Abraham, K. G. (1990). Restructuring the Employment Relationship: The Growth and
Market-Mediated Work Arrangements. In K. G. Abraham, R B. McKersie, (Eds), Navy
Developments in the Labor market: Toward a New Institutional Paradigm (pp. 85-120).
Cambridge: MIT Press.

Abraham, K G., Taylor, S. K. (1996). Firms' Use of Outside Contractors: Theory and
Evidence. J oumal of Labor Economics. 14(3 ), 394-424.

Akerlof, G. A. (1986) Labor Contracts as Partial Gifi Exchange. In G. A. Akerlof, J. L.
Yellen, (Eds), Efficiency Wage Models of .tlLC Labor Market (66-92). Cambridge:
Cambridge University Press.

Akerlof, G. A., Yellen, J. L. (1986). Introduction. In G. A. Akerlof, J. L. Yellen, (Eds),
Efficiency Wagg Mogels of the Labor Market (1-21). Cambridge: Cambridge University
Press.

Applebaum, E. (1989). The Growth of the US. Contingent Labor Force. In R. Drago,
Perlman, Richard (Ed), Microeconomic Issues in Labor Economics (pp. 62-82). New
York: Harvester Wheatsheaf.

Baron, J. N. (1984). Organizational Perspectives on Stratification. Annual Review of
Sociology, 37-69.

Baron, J. N., Davis-Blake, A, Bielby, W. T. (1986). The Structure of Opportunity: How
Promotion Ladders Vary Within and Among Organzations. Administrative Science

Quarterly, 31, 248-273.
Beck, E. M., Horan, P. M., Tolbert, C. (1978). Stratification in a Dual Economy: A

Sectoral Model of Earnings Determination. American Sociologiaal ReviewL43(5), 704-
720.

Becker, E. H. (1984). Self-Employed Workers: An Update to 1983. Monthly Labor
Review. 107(7), 14-18.

Becker, G. (1993) Human Capital. Chicago: University of Chicago Press.

Belous, R S. (1989). How Human Resource Systems Adjust to the Shifi Toward
Contingent Workers. MonthlLLabor Review. 112(3), 7-12.

Bettis, R., Bradley, S. P., Hamel, G. (1992). Outsourcing and Industrial Decline.
Academy of M__a_nggement Executive, 6, 7-22.

148

Bills, D. B. (1987). Costs, Commitment, and Rewards: Factors Influencing the Design
and Implementation of Internal Labor Markets. Admini We Schience Merlin 32,
202-221.

Bregger, J. E. (1963). Self-employment in the United States 1948-1962. Monthly Labor
Review 86(1), 37-43.

 

Bregger, J. E. (1996). Measuring Self-Employment in the United States. Monthly Labor
Review. 119(1-2), 3-9.

Carey, M. L., Hazelbaker, K. L. (1986). Employment Growth in the Temporary Help
Industry. Monthly Labor Review, 109(4), 37-44.

Carre, F. J. (1992). Temporary Employment in the Eighties. In V. L. duRivage (Ed),
New Policies for the Part-Time aad Contingent Workforce . Armonk, NY: M E. Sharpe.

Carter, S. (1988). The Changing Importance of Lifetime Jobs. Industrial Relations 27(3),
287-300.

 

Casey, B. (1991). Survey Evidence on Trends in ”Non-Standard" Employment. In A.
Pollert (Ed), F_arewe11 to Flexibility? (pp. 179-199). Oxford: Oxford University Press.

Cohany, S. R. (1996). Workers in Alternative Employment Arrangements. Monthly
Labor Review, 119(10), 31-45.

Cohen, Y., Pfeffer, J. (1986). Organizational Hiring Standards. Administrative Science
Merly, 31, 1-24.

Cook, T. D., Campbell, D. T. (1979). Quasi-Experimentation: Deaign & Analysis Issues
for Fieldeettiaga. Boston: Houghton Mifflin Company.

Cooper, S. (1995). The Expanding Use of Contingent Workforce in the American

Economy: New Opportunities and Dangers for Employers. Employee Relations Law
Journal 20(4), 525-539.

 

Cordova, E. (1986). From F ull-Time Wage Employment to Atypical Employment: A
Major Shift in the Evolution of [about Relations? International Labour Review 125(6),
641-657.

 

Deavers, K. L. (1997). Outsourcing: A Corporate Competitiveness Strategy, Not a Search
for Low Wages. Journal of Labor Research 18(4), 504-517.

 

Dickens, W. T., Lang, Kevin (1985). A Test of Dual Labor Market Theory. Ih_e
American Economic Review 75(4), 792-805.

 

149

Diebold, F. X., Neumark, D., Polsky, D. (1997). Job Stability in the United States.
Journal of Labor Economics. 15(2), 206-233.

Doeringer, P. B., Piore, Michael J. (1971). Internal Labor Markets & Manpower
Analysis. Lexington, MA: Heath.

Doeringer, P. B., Christensen, K, Flynn, P. M., Hall, D. J., Kapp, H. C., Keefe, J. H.,
Ruhm, C. J ., Sum, A. M., Useem, M. (1991). The Two-Tiered Workforce in U. S.
Corporations. In P. B. Doeringer (Ed. ), Lurbulence in the American Workplace (pp. 140-
155 ). New York: Oxford University Press.

duRivage, V. L. (1992). New Policies for the Part-Time and Contingent Workforce. In V.
L. duRivage (Ed. ), New Policies for the Part-Time and Contingent Workforce (pp. 89-
121). Armonk, NY: M. E. Sharpe.

Fain, T. S. (1980). Self-employed Americans: Their Number has Increased. Monthly
L_abor Review. 103(11), 3-8.

Fay, J. A., Medoff, J. L. (1985). Labor and Output over the Business Cycle: Some Direct
Evidence. American Economic Review 75(4), 638-55.

 

Golden, L., Applebaum, E. (1992). What was Driving the 1982-1988 Boom in
Temporary Employment? American Journal of Economics app Sociology. 51(4), 473-
493.

Gonos, G. (1997). The Contest over "Employer" Status in the Postwar United States: The
Case of Temporary Help Firms. _L_aw & Societv Review. 31(1), 81-110.

Goodman, P. S., Leyden, D. P. (1991). Familiarity and Group Productivity. Journal of
Applied Psychology, 76(4), 578-586.

Granovetter, M. (1986). Labor Mobility, Internal Markets, and Job Matching: A
Comparison of the Sociological and Economic Approaches. Research in Social
Stratification and Mobilitv. 5. 3-39.

Grimshaw, D., Rubery, J. (1998). Integrating the Internal and External Labour Markets.
Cambridge Journal of Economics, 22, 199-220.

Groshen, E. L. (1991). Five Reasons Why Wages Vary Among Employers. Industrial
Relations 30(3), 350-381.

 

Irnrie, R. F. (1986). Work Decentralization from Large to Small Firms: A Preliminary
Analysis of Subcontracting. Environment and Planning A, 18(2), 949-965.

Kalaitzidakis, P. (1997). On-the-Job Training under Firm-Specific Innovations and
Worker Heterogeneity. Industrial Relations 36(3 ), 371-390.

 

150

Kosters, M. H. (1997). New Employment Relationships and the Labor Market. Journal of
Labor Research. 28(4), 551-559.

Lazear, E. P. (1995). A Jobs-Based Analysis of Labor Markets. Paper presented at the
ABA

Lee, D. M. S., Trauth, E. M., Farwell, D. (1995). Critical Skills and Knowledge
Requirements of IS Professionals: A Joint Academic/Industry Investigation. @
Quarterly(September), 313-340.

Livernash, E. R. (1957). The Internal Wage Structure. In G. W. Taylor, Pierson, F. C.
(Ed), New Concefi in Wage Determination (pp. 140-172). New York: McGraw Hill.

Loveman, G. W., Tilly, C. (1988). Good Jobs or Bad Jobs: What Does the Evidence Say?
New England Economic Review. J mm/F ebmy, 46-65.

Mangum, G., Mayall, D., Nelson, K. (1985). The Temporary Help Industry: A Response
to the Dual Internal Labor Market. Industrial and Labor Relations Review. 38(4), 599-
611.

McNabb, R., Psacharopoulos, G. (1979). Further Evidence of the Relevance of the Dual
Labor Market Hypothesis for the UK. The Journal of Human Resources. 16(3), 442-448.

Moore, M. A. (1965). The Temporary Help Service Industry: Historical Development,
Operation and Scope. Industrial and Labor Relations Review 18, 554-569.

 

Moss Kanter, R. (1984). Variations in Managerial Career Structures in High-Technology
Firms: The Impact of Organizational Characteristics on Internal Labor Market Patterns.
In P. Osterman (Ed), Internal Labor Markets (pp. 109-131). Cambridge, MA: MIT Press.

Osterman, P. (1975). An Empirical Study of Labor Market Segmentation. Industrial and
Qbor Relationa Review, 28(4), 508-523.

Osterman, P. (1983). Employment Structures Within Firms. British Journal of Industrial
Relations 20(3), 349-361.

 

Osterman, P. (1984a). Introduction: The Nature and Importance of Internal Labor
Markets. In P. Osterman (Ed), Internal Labor Markets (pp. 1-22). Cambridge, MA: MIT
Press. '

Osterman, P. (1984b). White Collar Internal Labor Markets. In P. Osterman (Ed),
Internal Labor Markets (pp. 163-189). Cambridge, MA: MIT Press.

Osterman, P. (1987). Choice of Employment Systems in Internal Labor Markets.
Industrial Relations 26(1), 46-67.

 

151

Osterman, P. (1995). Skill, Training, and Work Organization in American
Establishments. Industrial Relations 34(2), 125-146.

 

Pfeffer, J., Cohen, Y. (1984). Determinants of Internal Labor Markets in Organizations.
Administrative Science QquteLm 29. 550-572.

Pfeffer, J ., Baron, J. (1988). Taking the Workers Back Out: Recent Trends in the
Structuring of Employment. In B. Staw, Cummings, L. L. (Ed), Research in
Organizational Behavior (Vol. 10, pp. 257-304). Greenwich, CT: JAI Press.

Piore, M J. (1975). Notes for a Theory of Labor Market Stratification. In R. C. Edwards,
Reich, Michael, Gordon, Daivd M (Ed), Labor Market Segmentation (pp. 125-150).
Lexington, MA: D.C. Heath.

Polivka, A. E., Nardone, T. (1989). On the definition of 'contingent work'. Monthly Labor
Review. 112(12), 9-16.

Polivka, A. E. (1996a). Contingent and Alternative Work Arrangements, defined.
Monthly Labor Review. 119, 3-9.

Polivka, A. E. (1996b). A Profile of Contingent Workers. Monthly Labor Review.
1_1_9_(10), 10-21.

Pollert, A. (1991). The Orthodoxy of Flexibility. In A. Pollert (Ed), Farewell to
F lexibiliyy? (pp. 3-31). Oxford: Blackwell Business.

Ray, R. N. (1975). A report on Self-Employed Americans in 1973. Monthly Labor
Review 98(1), 49-54.

 

Rebitzer, J. B., Taylor, L. J. (1991). A Model of Dual Labor Markets When Product
Demand is Uncertain Mrly Journal of Economics, 16(4), 1373-1383.

Rebitzer, J. B. (1998). Job Safety and Contract Workers in the Petrochemical Industry. In
K. Barker, Christensen, Kathleen (Ed), Contingent Work (pp. 243-259). Ithaca, NY: ILR
Press.

Rogers, J. K. (1995). Just a Temp: Experience and Structure of Alienation in Temporary
Clerical Employment. Work & Occumtions, 22(2), 137-166.

Rosen, S. (1986). The Theory of Equalizing Differences. In 0. Ashenfelter, R. Layard
(Eds. ), Handbook of Labor Economics Volume I (641-692). Amsterdam: Elsevier
Science Publishers.

 

Rothstein, D. (1996). Entry Into and Consequences of Nonstandard Work Arrangements.
Monthly Labor Review. 119(10), 75-82.

152

Rousseau, D. M., Libuser, C. (1997). Contingent Workers in High Risk Environments.
California _Ma__n_agemenit ReviewJ9(2), 103-123.

Scoones, D., Bernhardt, D. (1998). Promotion, Turnover, and Discretionary Human
Capital Acquisition Journal of_Labor Economics- 16(1), 122-141.

Segal, L., Sullivan, D. (1995). The Temporary Labor Force. Economic Persocctives.
12(2), 2-19.

Seiler, E. (1984). Piece Rate vs. Time Rate: The Effect of Incentives on Earnings. The;
Review of Economics and Statistics 66(3). 363-376.

 

Sharpe, M. (1997). Outsourcing, Organizational Competitiveness, and Work. Journal of
Labor Research, 18(4), 535-549.

Spilerman, S. (1977). Careers, Labor Market Structure, and Socioecomonic
Achievement. Amenm Journal of Sociolpgy, 83(3), 551-593.

Staber, U., Bogenhold, D. (1993). Self-Employment: A Study of Seventeen OECD
Countries. Industrial Relations Journal 24(2), 126-137.

 

Statistics, Bureau of. Labor. (1996). Civilian Labor Force and Unemploment .
Washington, DC: Bureau of Labor Statistics.

Statistics, Bureau of Labor. (1999a). Civilian Labor Force and Unemployment .
Washington, DC.

Statistics, Bureau of Labor. (1999b). National Emplomeng, Hours, and Earnings.
Washington, DC.

Statistics, Bureau of Labor. (1999c). Contingent and Alternative Emplomant
Arrangements. Washington, DC.

Swinnerton, K. A., Wial, H. (1995). Is Job Stability Declining in the US Economy?
Industrial and Labor Relations Review.48(2), 293-304.

Tilly, C. (1992). Short Hours, Short Shrifi: The Causes and Consequences of Part-Time

Employment. In V. L. duRivage (Ed), New Policies for the Part-Time and Contr_nge_p' t
Workforce (pp. 15-44). Armonk, NY: M E. Sharpe.

Tolbert, C., Horan, P. M., Beck, E. M. (1980). The Structure of Economic Segmentation:
A Dual Economy Approach American Journal of Sociology, 85, 1095-1116.

Wachter, M., Wright, R. (1990). The Economics of Internal Labor Markets. Industrial
Relations, 29, 240-262.

153

Wells, M. J. (1987). Legal Conflict and Class Structure: The Independent Contractor-
Employee Controversy in California Agriculture. Law & Society Review. 21. 49.

Williams, H. B. (1989). What Temporary Workers Earn: Findings from New BLS
Survey. Monthll Labor Review-109. 3-6.

 

154

lilyIliljllflliliil‘lflliyll