INFORMATION TO USERS
This manuscript has been reproduced from the microfilm master. UMI films the text directly from the original or copy submitted. Thus, some thesis and dissertation copies are in typewriter face, while others may be from any type of computer printer.
The quality of this reproduction is dependent upon the quality of the copy submitted.Broken or indistinct print, colored or poor quality illustrations and photographs, print bleedthrough, substandard margins, and improper alignment can adversely affect reproduction.
In the unlikely event that the author did not send UMI a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion.
Oversize materials (e.g., maps, drawings, charts) are reproduced by sectioning the original, beginning at the upper left-hand comer and continuing from left to right in equal sections with small overlaps.
ProQuest Information and Learning 300 North Zeeb Road, Ann Arbor, Ml 48106-1346 USA 800-521-0600
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with with permission permission of the of copyright the copyright owner. owner. Further Further reproduction reproduction prohibited withoutprohibited permission. without permission. AN INQUIRY INTO THE POSSIBLE TRADEOFFS BETWEEN
ANTITRUST ENFORCEMENT AND EMPLOYMENT
bv
Van H. Pho
submitted to the
Faculty of the College of Arts and Sciences
of American University
in partial fulfillment of
the requirements for the degree
of Doctor o f Philosophy
in
Economics
Robert Lerman
Agapf Somwaru
Dean of the College
Date
2002 American University Washington D.C. 20016
AMERICAN UNIVERSITY LIBRARY
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. UMI Number: 3069097
___ ® UMI
UMI Microform 3069097 Copyright 2003 by ProQuest Information and Learning Company. All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code.
ProQuest Information and Learning Company 300 North Zeeb Road P.O. Box 1346 Ann Arbor, Ml 48106-1346
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ACKNOWLEDGEMENTS
To my parents, Daniel and Hong, who have sacrificed so much.
Special thanks to my family and friends for their unending support, encouragement, and
understanding of what it means to be on a graduate student budget.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. AN INQUIRY INTO THE POSSIBLE TRADEOFFS BETWEEN
ANTITRUST ENFORCEMENT AND EMPLOYMENT
BY
Yvon H. Pho
ABSTRACT
This research investigates the impact of U.S. federal antitrust enforcement
on employment opportunities and wages for U.S. manufacturing workers from 1979-
1999. This study is conducted on two levels. First, a case study approach is used to
determine direct and deterrent effects from antitrust indictments on economic activities
for six firms. Second, panel data analysis is employed to examine antitrust enforcement
effects on industry employment (2-digit SIC code level) and w'ages for seven occupation
groups.
Antitrust law enforcement potentially can result in wage and job losses if
generally efficient business practices are disrupted. While the aim of antitrust is not to
attack efficiency, antitrust indictments may lead to negative labor market effects. A
theory-based relationship between employment and antitrust enforcement is developed.
Then, a panel dataset was created from the Trade Regulation Reporter, Annual Survey of
Manufactures, Current Population Survey, and National Income and Product Accounts.
Ten models are estimated using GLS procedures to control for nonspherical disturbances
in the data.
iii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The resounding theme from these analyses is that antitrust enforcement
does affect the labor market. A pattern emerging from the case studies is an immediate
rise and subsequent decline in competitor employment once the antitrust indictment is
issued. Corresponding with the results from the industry analysis, this could be an
outcome of the competing firm maneuvering to improve its market position and
competitiveness. The studies also provided evidence of lagged adverse employment
effects occurring during and at the conclusion of the antitrust case.
The industry analyses indicate that antitrust enforcement generally
benefits employment. Employment increases persist two years beyond the initial
indictment. However, the rate of increase does not. In most cases, employment growth
surges one year after the indictment, but tapers off considerably. Additionally, high-wage
workers who tend to possess specific human capital benefit the most from antitrust
enforcement in terms of quantity and price, whereas low-wage workers who embody
general human capital benefit the least.
In summary, firms respond to antitrust enforcement in ways that
ultimately affect both labor quantity and price. This dissertation provides evidence of
employment and wage consequences derived from government intervention in the
marketplace.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. TABLE OF CONTENTS
ACKNOWLEDGEMENTS...... ii
ABSTRACT...... iii
LIST OF TABLES...... viii
LIST OF FIGURES...... ix
Chapter
1. GOALS AND OBJECTIVES...... 1 Background
The Antitrust Laws and Its Enforcers
Literature Review
Methodology
2. EMPIRICAL ANALYSIS...... 25 Level 1: Firm analysis
Level 2: Industry analysis
Extensions to the core models
Industry concentration
Worker categories
Major industry group
3. EMPIRICAL RESULTS FROM FIRM ANALYSES...... 43 Case Study 1: The Stanley Works and Black & Decker
The Stanley Works, Incorporated
V
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Case Study 2: Archer-Daniels-Midland and ConAgra
Archer-Daniels-Midland Company
ConAgra Foods, Incorporated
Case Study 3: Merck and Schering-Plough
Merck & Co, Incorporated
Schering-Plough Corporation
Summary of Case Studies
4. DATA AND ECONOMETRIC ISSUES...... 102 Trade Regulation Reporter
Annual Survey of Manufactures (ASM) and Economic Census
Current Population Survey (CPS)
National Income and Product Accounts
Limitations
5. EMPIRICAL RESULTS FROM INDUSTRY ANALYSES...... 121 Analysis of effects on employment levels
Model 1: Analysis of employment levels - Core model
Model 2: Analysis of employment levels, including industry concentration interaction effect
Model 3: Analysis of employment levels, including worker category interaction effects
Model 4: Analysis of employment levels, including major industry group interaction effects
Model 5: Analysis of employment levels, including both worker category and major industry group interaction effects
Analysis of effects on average wages vi
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Model 6: Analysis of average wages - core model (natural log specifications)
Model 7: Analysis of average wages, including industry concentration interaction effect
Model 8: Analysis of average wages, including worker category interaction effects
Model 9: Analysis of average wages, including industry interaction effects
Model 10: Analysis of average wages, including both worker category and major industry group interaction effects
6. CONCLUSION 143
APPENDICES 151
WORKS CITED 296
VII
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. LIST OF TABLES
Table Page
1. Summary of employment models ...... 41
2. Summary of average wage models ...... 42
3. Summary of The Stanley Works business activity ...... 48
4. Summary of Black & Decker business activity ...... 57
5. Summary of Archer-Daniels-Midland business activity ...... 64
6. Summary of ConAgra business activity ...... 74
7. Summary of Schering-Plough business activity ...... 90
8. Descriptive statistics for antitrust lawsuit outcomes for the years 1979 through June 2001 (N = 532) ...... 103
9. Descriptive statistics for employment analysis ...... 110
10. Descriptive statistics for wage analysis ...... 110
11. Estimates of the effect of antitrust enforcement on number of employees (in thousands, N = 2,660) ...... 122
12. Estimates of the effect of antitrust enforcement on the natural log of average wages (in 1996 constant dollars, N = 2,650) ...... 136
viii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. LIST OF FIGURES
Figure Page
Figure 1. The Stanley Works employment as a share of manufacturing sector and industry employment for the years 1979 - 1999 ...... 53
Figure 2. The Stanley Works sales as a share of manufacturing sector and industry sales for the years 1979 - 1999 ...... 54
Figure 3. The Stanley Works sales per employee for the years 1979 - 1999 ...... 56
Figure 4. Comparison of Employment Levels between The Stanley Works and Black & Decker for the Years 1979 - 1999 ...... 59
Figure 5. Black & Decker employment as a share of manufacturing sector and industry employment for the years 1979 - 1999 ...... 61
Figure 6. Black & Decker sales as a share of manufacturing sector and industry sales for the years 1979 - 1999 ...... 62
Figure 7. Black & Decker sales per employee for the years 1979 - 1999 ...... 63
Figure 8. Archer-Daniels-Midland employment as a share of manufacturing sector and industry employment for the years 1979 - 1999 ...... 70
Figure 9. Archer-Daniels-Midland sales as a share of manufacturing sector and industry sales for the years 1979 - 1999 ...... 72
Figure 10. Archer-Daniels-Midland sales per employee for the years 1979 - 1999 ...... 73
Figure 11. Comparison of employment levels between Archer-Daniels-Midland and ConAgra for the years 1979 - 1999 ...... 80
Figure 12. ConAgra employment as a share of manufacturing sector and industry employment for the years 1979 - 1999 ...... 81
Figure 13. ConAgra sales as a share of manufacturing sector and industry sales for the years 1979 - 1999 ...... 82
Figure 14. ConAgra sales per employee for the years 1979 - 1999 ...... 83
ix
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 15. Merck employment as a share of manufacturing sector and industry employment for the years 1979 - 1999 ...... 87
Figure 16. Merck sales as a share of manufacturing sector and industry sales for the years 1979- 1999 ...... 88
Figure 17. Merck sales per employee for the years 1979 - 1999 ...... 89
Figure 18. Comparison of employment levels between Merck and Schering-Plough for the years 1979 - 1999 ...... 93
Figure 19. Schering-Plough employment as a share of manufacturing sector and industry employment for the years 1979 - 1999 ...... 94
Figure 20. Schering-Plough sales as a share of manufacturing sector and industry salesfor the years 1979 - 1999 ...... 95
Figure 21. Schering-Plough sales per employee for the years 1979 - 1999 ...... 96
x
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 1
GOALS AND OBJECTIVES
The goal of this dissertation is to test the hypothesis that U.S. federal
antitrust enforcement increases employment opportunities or wages for workers. The
hypothesis will be tested for all workers, in addition to specified groups of workers.
Antitrust law. while commonly thought of as beneficial, may not provide any other
benefits extending beyond consumer welfare. The Industrial Organization field of
economics has focused primarily on consumer welfare and the factors that motivate
corporate behavior. Only in rare instances is the effect of corporate behavior on workers
taken into account. This dissertation strives to extend the analysis and bring the well
being of workers into consideration.
Although there is much debate regarding the original intent of the antitrust
laws, their purpose is by and large to protect economic freedom and opportunity by
promoting and maintaining a competitive marketplace.1 While the consensus of
economists is to promote efficiency, there is continuing debate as to the efficacy of the
antitrust laws. Antitrust enforcement is generally viewed as welfare enhancing when it
1 A symposium of varying opinions expressed earlier can be found in Keezer (1949) whereas more contemporary views can be found in McCutcheon (1997).
1
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. encourages competition. The laws provide the opportunity for firms to compete on the
basis of price and quality alone.
However, the enforcement of antitrust laws potentially can result in job
losses. This may occur if generally efficient business practices that have emerged in the
market have been disrupted. While the aim of antitrust is not to attack efficiency, an
antitrust indictment issued against a firm exhibiting a high profit margin may not
necessarily lead to positive effects on firm production and/or employment. These
potential negative effects on firm production and employment are not limited to the firm.
Changes at the level of the firm have ramifications on production and employment in the
industry and the economy as well.
If high profits or large market share are achieved through market
domination or monopolization, then antitrust enforcement may be desirable and likely
will lead to increases in industry output and employment. However, if they are achieved
through the firm incurring lower costs through greater efficiency, then a reduction in
output and employment is more likely to ensue. The firm may be operating efficiently
due to economies achieved through its larger size. There are efficiencies associated with
size, especially in manufacturing, which may be unobtainable with a large number of
small firms. As a result, antitrust may reduce employment through inefficient re
allocation of resources.
Furthermore, antitrust enforcement compels the firm to defend itself in
court. Litigation raises costs to the firm and this, in turn, has the potential to affect
employment adversely. In an effort to maximize profit, the firm may respond to litigation
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. costs by reducing output and employment. To the extent firms have monopsony power in
the labor market, there may also be an effort to restrain wages.
Finally, antitrust enforcement potentially can reduce employment through
firm structural reorganization. The government, being an outside party and not wholly
familiar with the market in which the firm is operating, may propose (and a judge may
order) a firm restructuring that would reduce overall efficiency. A possible outcome of
such an event would be a structure with inefficient allocation of resources, rendering the
firm unable to either compete or conduct business in a profitable manner.
One possible situation in which antitrust enforcement would have little to
no effect on employment is if the firm views an antitrust indictment as a one-time event,
and no structural remedy is likely. If the firm embraces this view, its long-term behavior
likely will not be altered. Thus, the firm may simply endure higher costs and lower profit
rates (resulting from litigation costs) in the near term.
Several corporate practices exist that fall under the umbrella of antitrust.
Generally, the intent of these practices is to increase market power either by a single firm
or a group of independent firms. The most commonly cited acts performed are price-
fixing conspiracies, almost universally regarded as output restricting; corporate mergers
likely to reduce the competitive vigor of particular markets; and predatory or other acts
designed to achieve or maintain monopoly power. In these instances, antitrust
enforcement is intended to prevent such practices, thereby suggesting that output and
employment would increase.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4
The scope of actions that are potentially included under the antitrust
umbrella is extensive. This dissertation will be restricted to the aforementioned types of
practices. Specifically, the cases included in this analysis deal with price fixing, restraint
of trade, resale price maintenance, bid rigging, territorial allocation, acquisitions, joint
ventures, premerger notification failures, and monopolization. Hereafter, the term
“antitrust” will refer exclusively to these types of antitrust law violations. Additionally,
the analysis will be limited to antitrust cases initiated by the U.S. government; however,
in the case studies discussion of private enforcement related to the government action
will be considered.
In summary, this analysis sheds light on effects of government
intervention in markets. It utilizes data to examine whether antitrust enforcement has
employment or wage consequences for workers. The conclusions from the analysis have
important ramifications for the workforce and the future of antitrust policy. An estimate
of the number of workers affected by antitrust enforcement will be calculated to quantify
the effects. Finally, the results will identify the types of workers most likely affected by
antitrust. Using the information from this study, antitrust policy can be structured to
minimize negative impacts on workers or, alternatively, be strategically amended to
benefit workers. It is without dispute that the intent of market regulation is to protect and
benefit citizens. However, if the final outcome experienced in the economy adversely
affects workers, perhaps a reassessment of the policy is warranted.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5
Background
The monopoly model suggests a relationship exists between employment
levels and antitrust enforcement. Economic theory suggests that a perfectly competitive
firm will charge a price equal to its marginal cost. In this case, the firm earns only normal
profit in the long run. A firm earns above normal profit in the long run if it possesses
market power, or exerts the ability to influence the price commanded in the market. This
is achieved by charging a price (P) exceeding the marginal cost (MC) of the good.
Monopoly power involves restricting output such that the firm is able to create a wedge
between price and marginal cost. Government intervention would attempt to force the
firm to price more competitively and/or undergo a reorganization of the firm’s structure.
Thus, there are two possible ways in which antitrust enforcement would lead to a positive
effect on employment.
Greater production induced by a lower price requires increased inputs into
the production process, thereby potentially increasing employment. This result, however,
assumes that monopolistic firms are operating efficiently. If the firm is operating
inefficiently as a monopolist, increasing production may not increase employment at all.
Rather, the firm will be extracting more productivity from its resources, thereby using
them more efficiently and effectively. The result, then, is increased production with no
increase in employment.
In addition, a lessening of market power would allow other competitors to
enter the market. Thus, a rise in employment is anticipated due to new entrants hiring
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. workers. For these reasons—lower price for consumers and increased employment—
antitrust enforcement always has been considered desirable and beneficial to society.
The Antitrust Laws and Its Enforcers
Three primary laws govern U.S. antitrust. The first is the Sherman
Antitrust Act of 1890. The Sherman Act contains two sections that are frequently cited
and used to regulate monopoly power in the marketplace. Section 1 prohibits “every
contract, combination in the form of trust or otherwise, or conspiracy, in restraint of trade
or commerce among the several states. . .” Section 2 declares guilty of a misdemeanor
“every person who shall monopolize, or attempt to monopolize...”
The second law governing antitrust is the Clayton Act of 1914. This Act
prohibits price discrimination, exclusive dealing and tying contracts, acquisitions of
competing companies, and interlocking directorates when the effect of the practice “may
be to substantially lessen competition or tend to create a monopoly.” Neither Acts declare
a monopoly illegal, rather, it is the act of monopolization that is unlawful.
Finally, there is the Federal Trade Commission (FTC) Act of 1914.
Section 5 of the FTC Act prohibits “unfair methods of competition in commerce and
unfair or deceptive acts or practices in commerce.” Several subsequent amendments have
strengthened several provisions of the statutes. One such amendment is the Robinson-
Patman Act of 1936. This statute made modifications to the Clayton Act’s price
discrimination section. Another amendment is the Cellar-Kefauver Act of 1950 that
closed an existing loophole in anti-merger law.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Two agencies within the U.S. government are responsible for enforcement
of these laws. These agencies are the Bureau of Competition of the Federal Trade
Commission (FTC) and the Antitrust Division of the Department of Justice (DOJ). Both
agencies seek to prevent business practices that restrain competition, thereby violating the
antitrust laws.
The Bureau investigates alleged antitrust law violations and, when
appropriate, makes recommendations to the FTC to take formal enforcement action. The
recommendation is then reviewed and assessed. If the FTC decides to take action, the
Bureau assists in implementing that decision through litigation in federal court or before
an administrative law judge.
Similarly, the DOJ prosecutes serious and willful violations of the antitrust
laws by filing criminal suits that can lead to large fines and jail sentences. Where criminal
prosecution is not appropriate, the Division institutes a civil action seeking a court order
forbidding future violations of the law and requiring steps to remedy the anti-competitive
effects of past violations. To prevent duplication of effort, the two agencies consult
before opening any case.
In addition to federal antitrust laws, firms are subject to state antitrust laws
as well. In fact, most states have laws in effect regulating business practices. Essentially,
these laws prohibit corporate activity that unreasonably deprives consumers of the
benefits of competition, leading to higher prices for products and services.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 8
Literature Review
Numerous studies have examined the antitrust laws, their enforcement,
and the subsequent economic effects. The studies span the academic research spectrum
from heavily empirical to purely theoretical. This review of the literature will first
address the empirical studies, and then focus on the theoretical discussions.
Of all these works, only one—to my knowledge—directly addresses the
potential employment effects of the laws. This study, conducted by Shughart and Tollison
(1991), examines the relationship between unemployment and antitrust regulation. Their
hypothesis states that an unexpected increase in antitrust activity potentially leads to an
increase in the general level of unemployment. They contend that the antitrust authorities
may bring cases against the wrong firms or impose relief measures that are ineffective in
some instances, and have unintended consequences in others. As a result, vigorous
enforcement efforts may lead to an economy characterized by lower rates of growth in
real output, higher prices, and lower employment levels than otherwise. They claim that
their results suggest the existence of an antitrust analog to the short-run Phillips curve
relation.
Shughart and Tollison’s empirical analysis supports their hypothesis.
Their econometric model controls for trend, inflation, the female labor force participation
rate, and the ratio of the average weekly unemployment insurance benefit to the average
weekly wage in the nonagricultural sector using aggregated U.S. data. They find that an
unexpected rise in antitrust enforcement between the period 1890 to 1981 tends to
increase unemployment, ceteris paribus. The authors estimate a regression model of the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 9
natural log of antitrust cases (instituted annually per thousand of real budgetary
expenditures) on the natural log of real GNP. The residuals from this estimation equation
are used as a proxy for the unexpected component of antitrust enforcement. Specifically,
an unanticipated 1.0 percent increase in Antitrust Division cases per thousand dollars of
budgetary appropriations leads roughly to a 0.17 percent increase in the economy-wide
unemployment rate.
There were a few criticisms to Shughart and Tollison’s analysis. Frey
(1991) applauded the authors’ attempt at attacking the conventional idea that antitrust
policy raises economic welfare. At the same time, he brings light to the shortcomings of
their work, of which a few will be highlighted here. First, Frey suggests a theory of
bureaucratic behavior is warranted. He states that such a formulation would reflect the
specific behavior of the antitrust agency and could possibly yield different results.
Second, Frey states that measuring antitrust activity simply by incidence of a case and not
taking into consideration case outcome is incomplete. He suggests differentiating the
cases by successful and unsuccessful charges. Third, Frey questions Shughart and
Tollison’s measure of the inefficiency of antitrust policy: short-term unemployment. He
believes that any type of government intervention, efficient or otherwise, into the free
market can be expected to produce temporary unemployment. Thus, a better measure of
the inefficiency of antitrust policy would be medium- or long-term unemployment.
Finally, Frey alludes to a possible specification problem in the authors’ model. He states
that the coefficient estimates are unstable across the models estimated, and exhibit an
unusually high level of error.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 10
Schlieper (1991) also provides a critique of Shughart and Tollison’s work.
In addition to commending the authors on their efforts in addressing the issue of the
benefits of antitrust law enforcement, he identifies a few areas for improvement. First,
Schlieper suggests clarification be made to the term “beneficial” in determining whether
antitrust measures are beneficial. He states that the authors speak loosely of an increase in
economic welfare and overall efficiency in their article, and then proceed to empirically
estimate the effect on the rate of unemployment. Moreover, he suggests including a
model that would link efficiency to the rate of unemployment or the level o f employment,
since the connection between the two is not readily apparent. Second, as mentioned by
Frey, Schlieper states that any antitrust action brought against an industry—competitive
or monopolized—will always raise costs and uncertainty in the short-run, which would
likely lead to reductions in output and employment. Third, Schlieper believes
employment is a bad indicator for efficiency. He provides an example where an industry
is forced to use smaller and inefficient plant sizes due to antitrust action. In this situation,
efficiency is reduced in that more factor inputs are used per unit of output, yet
employment rises. Consequently, Schlieper believes that the empirical evidence
presented by Shughart and Tollison is inconclusive in that it could also provide support
for the notion that antitrust works in the desired direction.
An earlier study that investigated the broader effects of antitrust was
conducted by Stigler (1966). Stigler used three types of comparison studies to determine
the effectiveness of antitrust policies in reducing industry concentration. The first was a
country case study of seven industries in the U.S. and England. The second study
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. II
compared periods pre- and post- passage of antitrust law. The final study compared
industries subject to antitrust laws to industries exempt. Each study generated a slightly
different, yet interesting result.
The seven industries analyzed in the country case study were automobile,
cement, cigarette, glass, soap, steel, and rubber tire. England was chosen because, unlike
the U.S. that enacted the Sherman Act to limit concentration, England had no existing
policy. Additionally, the two economies are similar in that they both perform under a
comparable legal environment and technological base. Through comparing concentration
in these industries over time beginning in 1900, he found that the antitrust laws only have
had a modest effect on reducing concentration in the U.S.
The second study evaluated the effect of the 1950 anti-merger amendment
to the Clayton Act. Stigler recorded the number of horizontal mergers partaken by the
200 leading companies in manufacturing and mining for three periods: 1948-1953, 1954-
1959, and 1960-1964. He discovered the Clayton Act amendment was highly effective in
discouraging horizontal mergers. Even more surprising was that there were no major
mergers since 1950 in any of the seven industries from the country case study.
The final study examined antitrust cases involving some type of
conspiracy reported in the Commerce Clearing House Bluebooks.2 Through 1963, over
three-quarters of the cases involving collusion were won by the government. Stigler
created two categories of collusion: efficient and inefficient. He defined efficient forms
of collusion as agreements that can be easily cheated, whereas inefficient forms of
2 Also known as the Trade Regulation Reporter.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 12
collusion were defined as agreements that were easily enforced. Stigler’s analysis sorted
the cases by these two categories. Examples of efficient types of collusion are joint sales
agencies and customer assignment. Again, this type of collusion is efficient because these
actions were more easily detectable and proven to exist by the DOJ. From his analysis,
Stigler concluded that the Sherman Act reduced collusive activity by reducing the
“efficient” methods of collusion. Additionally, he found that while the Sherman Act was
effective in reducing collusion, it was not as effective in reducing concentration.
Posner (1970) conducted a comprehensive statistical study of antitrust
enforcement by the DOJ, Federal Trade Commission, state agencies, and private
plaintiffs from 1890 to 1969. One of the aims of his study was to identify the weaknesses
of the resources for antitrust statistics, and suggest methods for improving them. Through
use of the Trade Regulation Reporter, he records various statistics such as the number of
cases, the length of legal proceedings, the record of success of an antitrust claimant, the
use of various civil and criminal remedies, the pattern of violations alleged; and the
industries involved. Posner also investigates the possible explanatory role of politics in
antitrust activity.
Through his study, Posner tries to explain the variation in antitrust cases
and finds that both the incidence of antitrust violations and the resources available to
combat them are pro-cyclical with the economy. That is, violations and resources
increase as the economy expands, and decrease as the economy contracts. However, this
proves true only for the period up to 1940. After that time, there is little correlation
between the two variables. Posner also finds that the current antitrust remedies do little to
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 13
deter future violations. Additionally, there was not much evidence in support of the
political party in office influencing the quantity or quality of antitrust activity.
Long, Schramm and Tollison (1973) drew upon Posner’s empirical work,
and further investigated whether certain factors influence the types of cases brought forth
by the DOJ. They assume the DOJ “rationally weighs the economic benefits and costs of
bringing individual cases,” and pursues litigation only in those situations where the
benefits outweigh the costs. The authors focus primarily on the economics aspect of the
cases the DOJ decides to prosecute. That is, they assume the benefit-cost approach is
unconstrained by legal and/or other factors. The authors refer to the basic model of dead
weight loss from a monopoly, and apply welfare-loss measures developed by Harberger
and Kamerschen to formulate their index that incorporates industry concentration into the
calculation of welfare.
Ordinary least squares (OLS) regression was conducted on the number of
cases and their derived measure of industry welfare losses mentioned earlier. Variations
to the model included regressions of antitrust cases on sales, profits and concentration.
Data was collected on all cases brought by the DOJ against two-digit SIC manufacturing
industries for the period 1945-1970, as well as average profits and sales data over the
same period. The results indicate there is a positive relationship between cases brought
and the welfare measures, although the explanatory power of these measures were
minimal. Moreover, the authors conclude that aggregate benefit measures do little to
explain the variation in antitrust cases among manufacturing sector industry groups.
Industry sales were found to have a strong influence on the antitrust cases brought forth
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 14
by the DOJ. While the profit rate and the concentration index increased the explanatory
power of their empirical models, they were not statistically significant, and played a
secondary role to sales.
Block, Nold and Sidak (1981) study deterrent effects of public and private
antitrust enforcement on the decision to collude for the bread industry. They develop a
theoretical model of the collusive pricing decision (to be discussed in greater detail later
in this chapter), and determine that the collusive markup would be reduced if the
probability of detection rises or the penalty for price fixing increases. The authors test
their theoretical conclusion by using data from the bread industry to create a markup
indicator.
Using a standard recipe for bread as their reference, Block, Nold, and
Sidak obtain annual input and output prices from the Bureau of Labor Statistics to derive
a recipe adjusted bread price. This adjusted bread price was the market price of the bread
less the ingredient cost. The authors also estimate the variation in the adjusted bread price
due to changes in energy and labor costs to determine non-ingredient bread costs.
Utilizing these components, the markup indicator was then created by subtracting the
sum of the ingredient costs and an estimate of the non-ingredient costs from the market
price for bread, and dividing by estimated unit costs.
The authors used an antitrust dummy variable to measure the effect of an
antitrust action against a bread producer. The dummy variable was set equal to one 1 year
after an action was filed by the Antitrust Division, thereby introducing a lagged effect of
antitrust enforcement. Private treble damage cases were also recorded and incorporated
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 15
into their analysis. Changes in the markup were used as a proxy for the effectiveness of
antitrust enforcement, and evidence was found in support of deterrent effects from DOJ
enforcement efforts. Specifically, DOJ enforcement deterred raising markups, but never
deterred price fixing itself. According to the authors’ estimates, the total fines as a
percentage of pretax profits of the colluding firms averaged only 7 percent.
Consequently, they found that private class action suits were the true credible threat and
deterrent to price fixing, because the penalties imposed by government for price fixing
were so trivial.
Choi and Philippatos (1983) examine the effect of antitrust enforcement
upon the financial decisions of affected firms. Their study found that antitrust challenges
have been effective in the sense that the indicted firms do, in fact, show restraint in
pricing subsequent to the prosecution. The deterrent effect, however, was stronger in the
case of inexperienced violators. Choi and Philippatos also examined other strategic
choices made by firms in their sample, such as sales growth and financial leverage. They
concluded that these decisions were not seriously affected by the antitrust challenge.
Feinberg (1984) also examined deterrent effects from antitrust
investigations. His study, however, focused primarily on firm strategic and pricing
responses. Feinberg examined five Justice Department cases filed between 1965 and
1974 that involved manufacturing price-fixing conspiracies. Using time series regression
techniques, Feinberg developed a price index for the products involved in the antitrust
investigations, and compared this index to a broader producer price index series. The
empirical model was also able to examine strategic effects by including antitrust dummy
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 16
variables that capture different aspects of the case. His findings suggest that pricing and
strategic effects from antitrust enforcement do exist. The termination of an antitrust price-
fixing case resulted in lower prices for a sustained period of time. Moreover, Feinberg
found evidence of short-term strategic price reductions occurring in cases where there
was an antitrust investigation but no subsequent indictment.
In addition to empirical studies, considerable work has been done on the
theoretical aspects of the antitrust laws. Keezer (1949) compiled a symposium of
opinions regarding the development and enforcement of the federal antitrust laws. The
contributors consist of academic scholars and those who have worked in the field. In his
article, a wide spectrum of views on the effectiveness of the laws is expressed. Some
contributors state the laws provide minimal protection from industry concentration.
Rather, the laws exemplify the illusion that Government promotes competition when, in
effect, not much is done to ensure monopolies do not emerge. Opposing views suggest
that size is necessary in certain industries to capture efficiencies. That is, a tradeoff exists
between having a few firms of efficient size that are under little pressure to compete, or
having many firms that are competitive but tuo small to be efficient. Despite the range of
opinions, the consensus was that the antitrust laws would be more effective without
“judicial bungling of economic problems, conflicts, inconsistencies, and loopholes in
administration, and inadequate support for enforcement.”
Peterson (1957) expressed the idea that the existence of monopoly power
may not be disruptive to the market process. He believes the model of perfect
competition is unattainable, and thus should not be used as a benchmark for economic
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 17
analysis. Furthermore, the possession of monopoly power can still be regarded as
“workable competition” as long as there is the threat of competition that extracts
efficiencies and productivity and progress for consolidation without the negative
consequences.
Both Williamson (1969) and Lande (1988) argue that antitrust policy
inadvertently brings about inefficiencies throughout the economy. Williamson highlights
the commonly neglected efficiency aspects resulting from conglomerate organization, the
innovation process, and product variety. In an effort to preserve the efficiencies and
minimize the inefficiencies, Williamson proposes antitrust investigations pass an
allocative efficiency standard prior to enforcement.
Jorde and Teece (1989) suggest that collaboration among firms may not
always be undesirable. High technology industries are characterized by rapid change,
more than one innovation source and global competition that leads them to believe that
collaboration in these industries may not be a hindrance to competition. In contrast,
Brodley (1990) believes innovation collaboration can create anti-competitive risks, and
asserts that the existing antitrust laws are inadequate in deterring innovation research
collaboration. In fact, he asserts that the 1984 National Cooperative Research Act
(NCRA) simplified the process of R&D joint ventures thereby fostering collaboration,
and eliminated punitive antitrust penalties. Regardless, Brodley argues that sweeping
reforms of the antitrust laws would be unwarranted because there are some instances in
which collaboration among firms is desirable, and the indirect effects of the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 18
modifications may reduce this. Basically, the potential costs of antitrust law reform
outweigh the potential benefits.
Moschel (1991) argues that the antitrust laws may be largely ineffective in
promoting competition. He identifies the propensity for antitrust law to be manipulated
by private plaintiffs through the high costs associated with attorney fees. That is, a private
suit would be filed against a firm regardless of whether it is truly violating antitrust law.
The firm, forced to defend itself in court, incurs large expenses, thereby diverting its
resources away from production and sales to litigation. The possibility of treble damages
serves as an added incentive for plaintiffs to file court cases. The resulting outcome,
Moschel argues, is a restraint in competition.
Grandy (1993) questions the consumer welfare component of the antitrust
laws. Grandy argues that of the variety of goals expressed in the recorded congressional
debates, consumer welfare was not one of them. In fact, Congress seemed more
concerned with producer, rather than consumer, welfare. The protection of small
independent business arose repeatedly in discussions of antitrust’s goals. The language
Congress chose in defining the law directly dealt with producers and their behavior, not
with consumers. In sum, Grandy argues that the legislative history of the Sherman Act
fails to support the consumer-welfare hypothesis and suggests that Congress focused its
concern on producer behavior. Given that producer behavior was their primary concern, it
is likely that employment effects of the antitrust laws were not considered.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 19
More recent work relating to antitrust has focused primarily on investment
and stock market, rather than labor market, implications from antitrust actions/ In
summary, the economic aspects of the federal antitrust laws have been investigated at
length. The works range from assessing the motivation behind the laws and who benefits
from them to questioning the necessity for and effectiveness of the laws to ascertaining
their economic ramifications. Although much research has been done in the area, more is
needed. This dissertation draws upon and contributes to the existing literature in several
wavs.
First, a panel data set was created containing information regarding
specific variables in the manufacturing sector for the years 1979 to 1999. These specific
variables, to be discussed at length in Chapter 3. include data pertaining to employment,
wages, labor expense, capital expense, and industry concentration. Similar to Posner. data
on antitrust cases was also collected. These data were then used to conduct further
analysis on possible employment effects of the antitrust laws. Second. Shughart and
Tollison's study was expanded and modified for analysis at both the manufacturing sector
industry and worker levels. Furthermore, the extension to their analysis allows for cross
effects between antitrust enforcement and manufacturing industry, and antitrust
enforcement and worker category' to be captured. Third, as mentioned previously, the
time period in this dissertation continues beyond that of the existing literature to include
the most recent data from 1979 to 1999.4 My final contribution to the research in this area
’ A few articles relating to the effects of antitrust enforcement on the stock market can be found in Bittlingmaver (1993). and Bizjak and Coles (1995).
4 For some variables, data beyond 1999 were collected.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 20
is through case study analysis. Six case studies are conducted to determine if and how
antitrust enforcement affects the performance of an individual firm. Three of the case
studies examine firms who were indicted for antitrust, and the remaining studies attempt
to assess any impacts on their main competitors who did not undergo investigation. The
specific details concerning the case studies are explained in Chapter 2.
Methodology
A relationship between employment and antitrust enforcement can be
derived from two theoretical models. The first is through a model mentioned earlier
developed by Block. Nold and Sidak (1981). Their model examines a firm's price-fixing
decision and imposes three basic assumptions. They assume that firms produce under
conditions of constant marginal and average cost so that me = c. They also assume that
the pricing decision is undertaken jointly by all firms, finally, they assume that the "most
significant cost of any collusive device is its impact on the probability of detection." The
colluders establish an agreed-upon output price, p. that meets their objective of
maximizing the expected value of profits. The expected profit function, then, can be
expressed as:
max E(tt) = (1 - cl) tui + d Kz (1)
where d - probability of being indicted for antitrust 7i| = profit level where the colluders avoid being detected 7t: = profit level where the colluders are detected.
The first measure of profit. 7i|. can be expressed as:
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 21
71 \ = (p-c) Q{p) (2)
where Q(p) = demand for the industry’s output.
The second measure of profit, ttj, can be expressed as:
*2 = [p ~ c (1 + X/)] Q(p) (3)
where \ = (p - c) I c, and represents the markup under constant costs t = the anticipated damage multiple of the combined civil and criminal penalty for price fixing
In addition to the three basic assumptions, their model also presumes that
firms are never falsely accused of price-fixing. The authors state that if antitrust
enforcement becomes more stringent (t increasing), the firms will reduce their markup
over costs to avoid detection. Provided the assumption of constant costs, a reduction in
the markup implies that output price must fall, and ultimately output must rise.
Subsequently, employment must also increase due to higher levels of output.
Block, Nold, and Sidak’s model could be extended beyond price-fixing,
and expressed in more general terms as an intuitive argument. For instance, government
intervention via antitrust enforcement reduces the monopoly power held by a firm or a
combination of firms (a cartel). A reduction in monopoly power results in a lower price
and higher output. Higher output, in turn, is likely to increase employment. In order for
output to increase, additional quantities of inputs into the production process are
necessary. Assuming labor and capital are inputs into the production process, more of
each is required to obtain higher output.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 22
However, now consider a perfectly competitive market without collusion.
Suppose the production function is given by the following:
v=/(L,K) (4)
where y = total output L - labor K = capital
The firm’s profit equation can be expressed as:
n=pf{L, K)-wL-rK (5)
where k is profit, p is output price, w is the price of labor, and r is the price of capital.
Assuming the firm is a price taker, the only actions the firm can take to influence profits
involve the decisions of how much labor and capital to utilize. Furthermore, in the short
run capital is fixed, and thus the only variable input is labor. The amount of labor hired,
then, will directly affect the firm’s profits. Recall, this model presupposes that the firm is
a price taker. This being so, the firm would now be falsely accused of engaging in
antitrust practices—unlike Block, Nold and Sidak’s theoretical model.
Extending this model to include costs of antitrust litigation yields the cost
equation below.
TC = wL + rK + /«ATR (6)
ATR is a dummy variable triggered when the firm undergoes an indictment. The
coefficient associated with the ATR variable, m, represents the costliness of antitrust
litigation to the firm. If total costs are also a function of antitrust indictments, then a rise
in costs due to antitrust enforcement would have an effect on firm behavior.
n — p/ ( L, K) - vvL - rK - mATR (7)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Determining the amount of labor that would maximize profit requires differentiating
Equation (7) by labor.
dn=p MPi - w - m dATR = 0 (8) dL dL
Anticompetitive practices tend to be output restricting, and labor is positively correlated
with output. Thus, it is likely that labor and antitrust indictments are inversely related.
That is, an increase in labor (with a corresponding increase in output) would lead the
antitrust enforcers to be more reluctant to file an action. So,
<3ATR < 0 (9) dL
Consequently,
p MPi = w + m dATR < w (10) dL (-)
Equation (10) states that if an increase in labor leads to a lessened chance of being
indicted on antitrust charges, the effective marginal cost to the firm of hiring labor is
reduced. Thus in the short run, a firm may increase labor and quantity to minimize the
risk of an antitrust indictment. This is especially true as m, the costliness of antitrust
litigation, increases since the marginal cost of labor is even further reduced. However, the
added cost would increase the probability that a competitive firm would experience either
short run losses leading it to shut down or long run negative profit leading it to exit. In
either case, overall employment would fall.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 24
In summary, there are two plausible cases supporting the notion that
antitrust enforcement may influence firm employment behavior. The first model was
developed by Block, Nold and Sidak. Their model assumes that colluding firms are never
falsely accused, and implies antitrust enforcement would ultimately increase employment
levels. The second model incorporates antitrust litigation costs into the total cost function.
This model assumes that firms are price-takers and thus falsely accused of violating
antitrust laws. In this situation, antitrust enforcement may increase employment in the
short-run, but could ultimately decrease employment levels. The models provide
ambiguous predictions for employment. Therefore, the question of the true impact must
be answered empirically.
This study will be conducted on two distinct levels. The first uses a case
study approach to examine the effect, if any, of an antitrust indictment on economic
activities at the firm level. The case study will be a comprehensive examination of the
firm before and after the antitrust decision, and is comprised of two main elements. One
element provides a descriptive background analysis of the firm and the results from the
federal antitrust investigation. Another element analyzes graphically employment figures
and attempts to draw a correlation between antitrust litigation and the variation in
employment. The second method will be an econometric examination of industry-level
employment and wages with the focus on changes, pre- and post-indictment. This
element is similar to this research project’s first level of analysis. In summary, the
research question will be answered both at a firm and industry level.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 2
EMPIRICAL ANALYSIS
Level I: Firm analysis
Analyses of six firms are conducted for the first level of this dissertation.
The criteria used to choose a firm for study takes into consideration many factors, one of
which is type of legal outcome. While it is pertinent to examine firms found at fault of
monopolization in the courts, it is also worthwhile to analyze cases where the firm was
indicted, but not found at fault. It is of interest to determine whether the litigation
process, in and of itself, has a negative impact on employment regardless of the final
outcome obtained in the courts.
The companies chosen for case study analysis were picked based on
certain criteria; they were relatively close in range in terms of total number of employees,
and were relatively diversified. Product line diversification was used as a criterion to
avoid the possibility of the firm being too adversely affected by the indictment, and
consequently driving the results to suggest a stronger employment impact from antitrust
enforcement than one would normally obtain. The three companies that underwent
antitrust enforcement and are studied are: The Stanley Works, Inc.; Archer-Daniels-
Midland Company; and Merck & Co., Inc. Three additional case studies are conducted on
25
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 26 their main competitors. The competitors are Black & Decker Corporation, ConAgra
Foods, and Schering-Plough Corporation, respectively.
The latter three companies were chosen because they were not subject to
antitrust indictments during the relevant time period, and they are among the indicted
companies’ top competitors.5 The purpose of studying the main competing companies is
to investigate deterrent effects of antitrust enforcement on their workforce and overall
performance. In other words, these case studies may reveal employment effects on firms
whose rivals are subject to an indictment.
The case studies begin with a comprehensive company profile. This
profile consists of several items including history of the company, milestones and major
events, and a description of line of business and company products. Next, employment
and selective financial data are gathered, from various data sources to be discussed in
Chapter 3, for each firm for the period 1979 to 1999. The selective financial data
constitute firm sales per employee, in addition to firm share of both industry sales and
industry employment.6 These variables—employment and sales per employee—are
illustrated in separate graphs over the relevant time period with the dates of the antitrust
litigation highlighted. To allow comparisons to aggregate trends, the employment graphs
are plotted against both manufacturing sector and industry employment.
5 As reported by Hoover’s Online company profile http: /www.hoovers.com. Obtained on 10/1/01. While these firms may be the indicted firms’ overall top competitors, it is not necessarily the case that they are the top competitors in the product line investigated for antitrust violation.
b The variable, sales per employee, is a proxy for firm productivity.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 27
Level 2: Industry analysis
The second level of analysis consists of a time-series study using antitrust
indictment data for the period 1979 to 1999. The intent is to estimate a reduced-form
model that would capture the effect of antitrust enforcement on the labor market by
industry.7 A complete market analysis considers both quantity and price effects. Thus,
national employment levels and average wages will be examined separately.
Independent variables included in the model are those that potentially
would impact employment and average wages. In addition to examining yearly levels,
analysis on year-to-year changes was conducted. The results of both are included in this
dissertation; however, the former is discussed whereas the latter is included in Appendix
C. The core quantity model takes the following form:
E =/(TREND, RGDP, HHI, RATE, K_LEXP, ATR, LI ATR, L2ATR) (11)
where E = employment levels; RGDP = Real GDP; HHI = average Herfindahl-Hirschmann Index; RATE = 5-year average growth rate of firms; K_LEXP = capital to labor expense ratio; ATR = contemporaneous antitrust indictment; LI ATR = 1-year lagged antitrust indictment; L2ATR = 2-year lagged antitrust indictment;
Equation 11 states that employment levels (E) are a function of eight
variables. The first is a control variable. Employment in manufacturing has exhibited a
strong negative pattern over the period of study. For this reason, a trend variable is
included. The second variable, real GDP, measures the state of the economy. If the
7 For more research on inter-industry effects on wages, see Krueger and Summers (1988) and Bartel and Sicherman (1999).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 28
economy is growing, one would expect to see higher levels of output, and consequently
employment, due to the economic expansion. As a result, a positive relationship between
these two variables is anticipated
The third variable is the weighted average of the Herfindahl-Hirschmann
Index (HHI) in the 2-digit SIC major industry. This index measures concentration by
taking the sum of the squares of firm market shares within an industry, as measured by
value of shipments. The HHI is reported either ranging from 0 to 1 or 0 to 10,000
(depending on whether firm shares are treated as fractions or percents). In this analysis, it
is defined to be between 0 and 10,000. The greater the value of the HHI, the higher the
degree of concentration within the industry. To the extent that increased HHI leads to
greater monopoly power and output restriction, there may be an inverse relationship
between this variable and employment.
The fourth variable provides insight into industry employment
characteristics relating to establishment size and growth. The yearly growth rate of the
number of firms within an industry was calculated, and then a 5-year moving average
was computed. This variable reveals the overall survival rate of firms. Tracking the 5-
year average growth rate of firms within an industry could potentially explain a relatively
large share of the fluctuation in industry employment levels over time. As a result, a
positive relationship is anticipated for this variable and the level of employment.
The fifth variable, the ratio of capital-to-labor expenses, acts as a control
for variations in industry employment. Employment levels within more capital-intensive
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 29
industries should be lower, ceteris paribus. The converse is true for more labor-intensive
industries.
The next three variables are the ones of primary interest—antitrust
indictments filed by the DOJ. This variable was obtained from the Trade Regulation
Reporter (to be discussed in greater detail in Chapter 4). The Trade Regulation Reporter
records all violations of antitrust laws initiated by the DOJ. The violations are sorted by
type of violation and also by the product line in which the violation occurred. Every
record in the manufacturing sector and relevant violation category was documented by
year, and sorted into the appropriate major industry group. Sorted according to year, the
incidence of at least one indictment in a given year within an industry results in a dummy
variable value of one for that industry and year.
The effect of antitrust indictments on employment is analyzed at three
different time periods: contemporaneous, once lagged, and twice lagged. As discussed
previously, Block, Nold, and Sidak (1981) also use the lagged form of their antitrust
indictment variable in their analysis. There are several reasons why the antitrust variable
should be lagged. First, lags accommodate for a litigation process that may take several
years to complete. Second, lags permit time for a firm found at fault in the courts to
comply or modify its practices accordingly. Third, even if government price fixing cases
finish in one year, more time may be required to distinguish any employment effects due
to “piggyback” private lawsuits.8 In fact, analysis of the data indicated that the shocks
8 Piggyback lawsuits are filed on the coattails of a guilty verdict from a government lawsuit. Hence, the name "piggyback”. The parties bring their own case against the indicted company, and only need to prove they were injured to obtain monetary damages.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 30
were fully absorbed after a maximum of four years.9 The effects after the first few years
are substantially smaller. Thus, the bulk of the impact from antitrust enforcement is felt in
the first few years. For ease of interpretation, only the first two lags are included in this
analysis.
Based on the theoretical discussion, no clear relationship is anticipated for
any of these variables. However, there may be a strategic argument for a positive
contemporaneous antitrust indictment effect. There may be a short run employment
impact in response to antitrust enforcement. A firm indicted for antitrust may increase its
output, and potentially its staff, as one means to avoid being found at fault in the courts.
For safe measure, other firms may increase their output in order to avoid being indicted if
they see another firm within their industry indicted for antitrust practices. Thus, there
may also be initial positive deterrent effects from antitrust enforcement.
As mentioned earlier, careful labor market analysis requires consideration
of influences on both aspects of the market: price and quantity. Equation 12 defines the
natural log of average wages to be a function of the same explanatory variables as
Equation 11. For both models, annual changes in the dependent variables were analyzed.
The difference analysis for wages is included in Appendix D. The core price model
parallels the core quantity model, and is as follows:
LNW =/(TREND, RGDP, HHI, RATE, K_LEXP, ATR, L1 ATR, L2ATR) (12)
where LNW = the natural log of wages; RGDP = real GDP;
9 Analysis was conducted on employment data that included additional lags to the core model (Model 1) to determine when the effect of an antitrust indictment would no longer have an impact on employment. The results of the analysis indicate that including seven lags rendered the fourth lag statistically insignificant. Up to the fourth lag, all the lagged antitrust variables were statistically significant.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. HHI = average HHI: RATE = 5-vear average growth rate in number of firms: KLEXP = capital to labor expense ratio: ATR = contemporaneous antitrust indictment: LI ATR - 1-year lagged antitrust indictment: L2ATR - 2-year lagged antitrust indictment:
The effect of the independent variables on average wages is somewhat
analogous to the employment effect. In terms of real GDP. periods of economic growth
tend to be associated with rising wages. This is primarily due to rising labor demand as
more output is being produced in the economy.
The effect of concentration on wages is not as straightforward. Machin
and Wadhwani (1991) state that "workers and firms may agree on a wage that is higher
than the competitive level as a form of rent-sharing.” firms in concentrated industries
tend to have relatively higher profit margins than firms in less concentrated industries
because of their larger share of the market.1" Continuing this notion, firms in
concentrated industries are thus likely to have more rent to share with their employees.
Provided that the workers in these firms have sufficient bargaining power, industry'
concentration would be positively correlated with wages.11
Conversely, to the extent that firms with power in the output market may
also have monopsony power in the buyers' market for labor, highly concentrated
industries may have an overall lower demand for labor and pay lower wages. Thus,
increased competition within the industry could potentially benefit workers by increasing
10 For more information, see Scherer (1990).
" Askildsen (1998) found that as workers gain greater worker participation and control in a firm's decision making process, firm employment decreases and their wages increase.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. labor demand due to new entrants into the market. The relationship between
concentration and average wages, then, is contingent upon the strength of the workers'
bargaining power.
Similar to real GDP. the 5-year average growth rate in the number of firms
within an industry should be positively correlated with wages. If an industry is growing,
its demand for labor that would increase, raising wages. An industry's capital-to-labor
expenditure ratio should also have a positive influence wage rates, firms that invest more
heavily into capital rather than labor also tend to use capital goods more intensively in
their production process. Although less labor is usually required, the labor that is hired in
these firms tends to be skilled. One reason for this is because the employees need to
know how to operate, maintain, and. to a certain extent, repair the machinery. As a result,
the non-management employees are still quite technically skilled. Generally, more skilled
workers possess these abilities, and are able to command higher wages. Higher capital-to-
labor expenditure ratios then, may be positively correlated with wages.
The relationship between antitrust indictments and wages is ambiguous.
As explained earlier, antitrust litigation raises costs to the firm. If negative employment
effects from antitrust enforcement via higher costs exist, wages could either fall or rise
depending on the type of worker displaced. If a firm releases low-wage workers, the
average wage may rise. Conversely, if high-wage workers are let go. average wages may
fall. Additionally, there may be little to no effect if w orkers at all wage levels are let go
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. proportionally or if the indictment only slows down the growth rate of wages.12
Moreover, if there are no employment effects, there may also be little to no effect on
wages. However, if positive employment effects exist, then wages should rise due to
increased labor demand. Wage effects from antitrust enforcement are again examined
contemporaneously, after one year, and after two years.
Extensions to the core models
Several variations of the core models. Models 1 and 6. will also be
analyzed. These two models simply examine whether there is a labor market effect from
antitrust enforcement. The extended analyses approach the same research question, but
from different angles. It is worthwhile to examine whether a particular market
environment, as measured by the level of industry concentration, changes the
employment effect of an antitrust indictment. Additionally, it is of value to investigate
whether all workers would be affected equally due to antitrust litigation in their industry'.
Finally, it is interesting to determine if all industries w ould be affected similarly due to an
antitrust indictment. The extensions to the core model take into consideration all of these
aspects.
In total, ten models will be estimated. Models 1 through 5 correspond to
the quantity analysis as expressed by Equation 11. Models 6 through 10 are associated
with the price analysis seen in Equation 12. Within each of these two groups of models.
'* More on temporary and permanent employment etYects on certain categories of workers in the upcoming section.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 34
the effects of antitrust enforcement on industry concentration, w orker category and major
industry' group are analyzed.
Industry concentration
It is interesting to examine if the combination of antitrust cases and certain
variables mentioned above affect employment. In order to address this issue, interaction
variables were introduced into the core models. The first extension to the core models
occurs in Models 2 and 7. and includes an interaction variable between the weighted
average HHI and the dummy for an antitrust indictment in the previous year. This
analysis examines the impact from the 1-year lagged antitrust variable interacted with the
current market environment. The higher the degree of concentration within an industry,
the greater the impact and intluence an antitrust indictment should have on employment.
The effect of industry concentration on antitrust effects on employment should be
positive as it increases the likelihood that increased competition and output expansion
would result.
The interaction variable incorporates the 1 -year antitrust lag. but maintains
the measure for industry concentration contemporaneous. The variable is meant to
capture the current market environment and how it interacts with antitrust proceedings.1'’
A 1-year lagged antitrust variable seems to be the most appropriate of the three antitrust
variables, because some of the initial shock would already have been absorbed, thus
|j Sensitivity analyses were conducted on lagging both variables in the interaction and lagging only the antitrust variable. The results of the analyses showed there were no benefits from using one specification over the other. Both variables provided similar results in terms of statistical significance. Since neither variable proved to be superior, the concentration index was kept contemporaneous for the aforementioned
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. allowing for potential employment impacts separate from the shock to be assessed. The
primary reason this and other interaction variables (to be discussed later) are included in
the model is because antitrust litigation is unlikely to have the same effect across the
manufacturing sector. The interaction variables allow these effects to vary and be
identified.
As stated previously, firms in highly concentrated industries are inclined
to have more producer surplus or profit since they control a larger share of the market.
Provided that workers in these firms have sufficient bargaining power, they would be
able to extract some of this surplus for themselves through higher w ages, as suggested by
Machin and VVadhwani (1991).
As noted above, increased concentration implies that the introduction of
antitrust enforcement would tend to increase competition in the industry thereby reducing
surplus to the industry, and ultimately reducing surplus to the worker. On the other hand,
the greater employment expansion predicted could raise wages through growth in labor
demand. Therefore, the impact of this interaction term is indeterminate.
Worker categories
The second extension to the core model incorporates interaction variables
between the once lagged antitrust variable and worker categories. Models 3 and 8 build
upon the Models 2 and 7 by including dummy variables for the interaction between each
worker category, less the omitted class, and the 1-year lagged antitrust indictment
reasons. The analysis entailed estimating the same regression equation with both variables, separately, and
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 36
dummy. These analyses separate out the employment effect on an individual category' of
workers.
If employment or wage effects from antitrust enforcement exist, then the
effects can be either temporary or permanent. The temporary component suggests that the
antitrust indictment was neither anticipated nor expected to have long-lasting
repercussions. The company views the indictment as a temporary setback towards the
achievement of its goals. If reducing the firm's workforce were necessary, the more
prudent course of action for the firm would be to maintain workers with specific human
capital rather than general. Thus, the firm would be more inclined to let go of lower
skilled workers because they typically possess general human capital.
In terms of a temporary wage effect, the firm may decide to reduce the
rate of wage increases. Due to the stickiness of w ages, it is unlikely that a worker will see
a reduction in their nominal wage.14 Rather, a lower wage increase for all workers, or
higher wage increases for higher skilled workers relative to lower skilled workers is
likely to ensue.
The permanent component suggests that the company views the antitrust
indictment will have a prolonged impact and it will be unable to return to its previous
competitive position. The firm may find it necessary to operate on a smaller scale than
previously for an indefinite period. In this situation, any reductions in the firm's
workforce or wage cuts would more likely be proportional across worker categories, not
affecting one class of worker more than another.
comparing the statistical significance of each.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 37
The same logic applies to positive, rather than negative, changes in
employment. If antitrust enforcement increases employment through market entry of
firms, then the temporary component of the effect would be to hire the type of worker
most needed first. Again, this may be a worker who possesses specific human capital who
is integral to getting the firm up and running. Once the firm has succeeded in becoming
somewhat established in the market, it would find it necessary to increase its labor force
proportionally by all types of workers in order to for it to continue its growth. This would
be the permanent component of the positive employment effect.
The inclusion of worker categories into the model enables the
determination of w hat types of workers are most affected. Certain workers may be more
frequently displaced due to antitrust proceedings compared to others. Moreover, if the
data analysis indicates that all workers are not affected equally, then it is important to
determine who these workers are. The seven worker categories are: Technicians &
Related Support. Professional Specialty: Executive. Administrative & Managerial: Sales:
Administrative Support, including Clerical: Service: and Production. The omitted
category' in these models is Production workers.
Average wage calculations for the manufacturing sector and its 20 major
industries, was performed on these seven detailed occupation categories.1' The
calculations resulted in a close range of average wage values for certain groups of
workers. This clustering of average wages enabled three broader classes of wages to
emerge naturally from the data. The low wage worker category consists of
14 For economic theories on sticky wages, see Romer (1996).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 38
Administrative Support & Clerical. Service, and Production workers. The medium wage
worker category consists of Technicians & Related Support and Sales workers. Finally.
Professional Specialty. Executive. Administrative & Managerial occupations comprise
the high wage worker category.16
Major industry group
The final extension to the core models incorporates the interaction
between the once lagged antitrust indictment dummy and each of the manufacturing
sector's major industry groups. The antitrust laws apply to nearly all sectors and level of
business, including manufacturing, transportation, distribution, and marketing.
However, this dissertation will consider only the manufacturing sector. Each major
industry group within the manufacturing sector will be examined. They are:
• Food & Kindred Products (20): • Rubber and miscellaneous plastics products (30); • Tobacco Products (21): • Leather and leather products (31); • Textile Mill Products (22); • Stone, clay, glass, and concrete products (32); • Apparel and Other Textile Products (23); • Primary metal industries (33); • Lumber and Wood Products (24); • Fabricated metal products (34); • Furniture and Fixtures (25); • Industrial machinery and equipment (35); • Paper and Allied Products (26); • Electrical and electronic equipment (36); • Printing and Publishing (27); • Transportation equipment (37); • Chemicals and Allied Products (28); • Instruments and related products (38); • Petroleum and Coal Products (29); • Miscellaneous manufacturing industries (39)
15 Current Population Survey data, described in Chapter 4. were used to perform the wage analysis.
Employment and wage analyses were also conducted by wage category (low. medium, and high) instead of worker category. These analyses were not as fruitful as the current analyses, and were thus not reported. The results are included in Appendix G.
'' Some industries and businesses are exempt from the federal Antitrust Laws. A few of them include farm marketing associations, labor unions, shipping, banking, and major league baseball.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 39
where the numbers in parentheses represent the 2-digit 1987 SIC (Standard Industrial
Classification) code equivalent.
Models 4 and 9 allow for antitrust enforcement effects to differ only by
industry. The purpose of these analyses is to determine if antitrust enforcement has a
varying effect on industry employment regardless of worker type. In this manner, any
industry effects can be separated out. Although all are within the manufacturing sector,
each of the 20 industries is unique. For example, one way in which the industries may
differ is in their relative share of factor inputs used in the production process. For these
reasons, it is likely that an employment effect will vary. Models 4 and 9 are sensitive to
these industry-specific peculiarities. Dummy interaction variables are included for 1-year
lagged antitrust indictment and industry less the omitted industry, which is the Food &
Kindred Products industry (SIC code 20).
Models 5 and 10 bring all the previous models together by examining
industry concentration, worker, and industry effects simultaneously. All the previously
interaction variables are included in these two models. The antitrust-industry
concentration interaction variable examines how the employment and wage effects of
antitrust enforcement impact the market environment. The antitrust-occupation
interaction variables capture how an antitrust case affects employment and wage for a
particular type of worker. The antitrust-industry interaction variables indicate the impact
of an antitrust case on a particular industry, and examine how this dynamic might
influence employment and wage levels.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 40
Models 3 and 5 examine for each worker category the previously
mentioned temporary component from the employment effect, while Models 8 and 10 do
the same for wages. These analyses will determine if the employment and wage effects
are greater for a certain class of workers. If so, that would suggest the presence of a
temporary component of the employment and wage effects. It is highly probable that
workers are affected by antitrust enforcement differently and, even likely, that lower-
skilled workers are more adversely affected by increases in firm costs from antitrust
litigation compared to high-skilled workers.
In summary, one large panel dataset is created to conduct two types of
analyses: labor quantity and labor price. The cross section aspect of the data is the worker
categories. The data are collected for a 20-year time period, which constitutes the time
series aspect of the data. Five models are estimated for each dataset. Models 1 and 6 are
the core models, and determine whether there is an employment effect from antitrust
enforcement. Models 2 and 7 build upon the core models by introducing the additional
effect that industry concentration may have on the relationship between antitrust
indictments and employment. Models 3 and 8 further build upon the core models by
examining the employment effect of antitrust enforcement by worker category. These two
models allow for worker-specific effects to be captured. Models 4 and 9 expand upon
Models 2 and 7 by including interactions between once lagged antitrust indictment
dummy and industry. From these two models, industry-specific effects can be identified.
Models 5 and 10 are the most comprehensive of the analyses because they include all the
aforementioned specific effects: market environment, worker, and industry. Each model
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 41
builds upon the another, enabling the isolation and identification of the gradual effect of
each additional variable’s contribution to the analysis.
The data are organized such that within each of the 20 manufacturing
major industry groups there are seven worker categories. Pooling the data for each of the
21 years in the analysis results in the final data set containing 2,940 observations. Table 1
below provides a detailed summary of each of the five quantity models.
Table 1. Summary of employment models Dependent Variable: Employee Count bv Worker Categon- Model Independent Variables: I 2 3 4 5 (a) (b ) (c) (d) (e) (0 Real GDP X XX XX Weighted Average HHI X XX X X Capital-to-Labor Expense Ratio X XX XX Average 5-year Firm Growth Rate X XX XX ATR Dummy X XX X X 1-Year Lagged ATR Dummy X X XXX 2-Year Lagged ATR Dummy X XX XX Interaction Variables: 1-Year Lagged ATR Dummy and XX XX Weighted Average HHI 1-Year Lagged ATR Dummy and X X Worker Category Dummy 1-Year Lagged ATR Dummy and XX Industry Dummy Note: Author's analysis.
A parallel analysis was conducted for the price analysis of the natural log
of average wages. The intent is first to examine if wages are influenced by antitrust
enforcement. If they are, then the next step is to determine to what extent are wages also
sensitive to the market environment, worker- and industry-specific effects. Table 2 below
illustrates the wage analyses.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 42
Aside from the dependent variable, the price analysis is identical to the
quantity analysis. Average wages for workers by worker category and by manufacturing
industry were calculated. The different models employed in this study provide a
comprehensive examination of the potential price and quantity effects from antitrust
indictments.
Table 2. Summary of average wage models Dependent Variable: Natural Log o f Average Wage by Worker Category Model Independent Variables: 6 7 8 9 10 (a) (b ) (c) (d) (e) (0 Real GDP XX XX X Weighted Average HHI XX XX X Capital-to-Labor Expense Ratio XX XX X Average 5-year Firm Growth Rate XX XX X ATR Dummy XX XX X 1-Year Lagged ATR Dummy XXX X X 2-Year Lagged ATR Dummy X XX X X Interaction Variables: 1-Year Lagged ATR Dummy and XXX X Weighted Average HHI I-Year Lagged ATR Dummy and X X Worker Category Dummy 1-Year Lagged ATR Dummy and X X Industry Dummy Note: Author’s analysis.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 3
EMPIRICAL RESULTS FROM FIRM ANALYSES
The following analyses examine the ramifications of antitrust enforcement
more narrowly by investigating possible effects on a single firm. The firm analyses
consist of case studies on six companies. As stated earlier, the companies included in the
case studies were thoughtfully chosen and met several specified criteria. The first
criterion was similarity in size and magnitude. The second criterion was product line
diversification. This criterion was established so that any effects obtained from the
analyses would be conservative estimates. The final criterion was variation in type of
charges and legal outcome. Every criterion was met successfully.
Company financial data for the case studies were obtained from Standard
& Poor’s COMPUSTAT (North America) database. There are three broad categories of
this database—North America, Global, and Specialty.18 Since all eight companies under
case study examination have headquarters in North America, the North American version
of the COMPUSTAT database was utilized. This version contains information on roughly
18 For a detailed description of the categories, see www.compustat.com.
43
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 44
10,000 actively traded U.S. companies, 11,000 inactive U.S. companies and 1,100
Canadian companies. Specifically, this database contains the following information: 55
annual data reports; 334 annual data items; 132 quarterly data items; 300 financial ratios
including dividends, growth rates, profitability and relative market performance; 1,500
indices including S&P, Dow Jones and Russell; 20 years of annual data history (optional
history available to 1950); up to 48 quarters of history (optional history available to
1962); 20 years of monthly market data (optional history available to 1962); company
name, address and officer information; operating segment information; economic
industry sector data; company business descriptions; growth and value indicators; daily
updates available on more than 200 data items (all other items updated weekly); and
Global Industry Classification Standard (GICS).
The case studies provide a historical overview of the company, product
line description, important events contributing to the firm’s development, selective
employment and financial data over the 1979 to 1999 time period, and pertinent
information relating to the antitrust case.19 Dun and Bradstreet’s Million Dollar Database
was used to select the companies used in the case studies. This database contains
information on over 1,500,000 U.S. and Canadian leading public and private businesses.
Company-level information on number of employees, annual sales, type of ownership
and industry, principal executives and biographies is provided. The Million Dollar
Database is updated every 60 days, and is available only by subscription. At the time the
information was accessed for this dissertation, the last update occurred on February 2,
19 In some cases, data for the early years were unavailable.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 45
2001 and contained January 2001 data. There were 63,991 records in the database whose
major industry was reported to be the Manufacturing sector (SIC codes 20-39). From
these records, only firms with at least one employee were kept.20 This reduced the total
number of records to 57,565. Privately-owned firms are not required to provide financial
reports to shareholders, and thus data for these companies are not readily available to the
public; deleting these resulted in the final sample containing 7,204 observations.
For each case study, the historical overview was obtained from both the
company’s website and the firm synopsis provided by FIS Online. FIS Online is a
subscription-based database offering detailed business and financial information on
approximately 11,000 U.S. public companies and 17,000 non-U.S. public companies.21
Mergent Inc., formerly known as Moody’s Financial Information Services, provides and
maintains this service. Mergent has been publishing detailed business descriptions,
corporate histories and financial statements since 1900. Case study data unavailable from
either FIS Online or the COMPUSTAT database were obtained from the individual
company annual reports. Specifically, accessing annual reports for particular years
enabled certain gaps in the data to be filled in. Detailed data on reported business
segments were obtained exclusively from the annual reports.
■° Both firm headquarters and branch locations are included in the database. In a majority of the cases where there were no employees, the firm listed was a branch with employees at the headquarters.
Specifically, this database contains the following information: company directory; summary financials; company history; line of business; properties; subsidiaries and associated businesses; officers and directors; auditor and counsel; long term debt; capital stock including price ranges, stock splits, and dividends; annual income statements for the last five years; quarterly income statements for the most recent five quarter; annual balance sheet; and cash flow statement.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 46
Details concerning the antitrust cases were obtained from the Trade
Regulation Reporter (to be discussed in greater detail in Chapter 4). Information on
private antitrust cases and legal research were obtained using Lexis-Nexis. Lexis-Nexis
also provided financial information from newspaper reports. The antitrust cases filed by
the U.S. Government against the specific companies included in the case studies were
obtained from the database and reviewed. Additionally, a search for subsequent class
action or civil lawsuits filed in response to the antitrust indictment filed by the U.S.
Government was conducted using this data source.
According to the National Bureau of Economic Research’s (NBER)
Business Cycle Dating Committee, there were three economic recessions within this
twenty-year period.22 The first lasted six months from January 1980 to July 1980. The
second continued for 16 months beginning in July 1981 and ending in November 1982.
The final recession took place from July 1990 to March 1991. An attempt will be made to
separate out firm antitrust effects from business cycle effects. The financial data have
been deflated into constant 1996 dollars using the GDP chain-type price index.23
Descriptive data and graphical analysis will be performed for each case study for
employment, and sales.
The aim of the case studies is to assess not only the employment effects
from antitrust indictments, but also the potential effects on firm performance compared to
” For a complete list of U.S. Business Cycle Expansions and Contractions, see h ttp :www.nber.oriycvcles.html. Obtained on 3/14/02.
23 This index was obtained from the Bureau of Economic Analysis website the following address: http: '■ www.bea.doc.ttov bea dn nioaweb TableViewFixcd.aspSMid. Obtained on 3/14/02.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 47
aggregate trends as measured by sales. Graphs are created to illustrate the following firm-
level attributes:
• Employment as a percentage of industry and manufacturing employment; • Sales as a percentage of industry and manufacturing sales; and • Sales per employee
As stated earlier, although six companies are studied, only half were
indicted for antitrust violations during the time period considered. The other three
companies are among their main competitors. Recall the objective of studying the
competing companies is to explore potential deterrent employment effects from an
antitrust indictment. That is, antitrust enforcement may have wider-ranging impacts that
affect firms other than the ones indicted. One additional graph is created for each of the
non-indicted companies that compares their employment levels to the indicted firm’s.
The companies that have been subject to antitrust action are The Stanley Works, Inc.,
Archer-Daniels-Midland Company, and Merck & Co., Inc. Their main competitors are
Black & Decker Corporation, ConAgra Foods, and Schering-Plough Corporation,
respectively.
Case Study 1: The Stanley Works and Black & Decker
The Stanley Works, Incorporated
Frederick T. Stanley founded the Stanley Works, Inc. in 1843. Located in
New Britain, Connecticut, the company began operating from a one-story wooden
building manufacturing hinges, bolts, and other door hardware. Mr. Stanley sought to
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. create a hardware company of excellent quality and customer service. Experiencing much
success, the small company became incorporated on July 1, 1852, and began exporting
their products by the 1870s. In an effort to expand its international operations, the Stanley
Works acquired Stanley Rule and Level in 1920. Stanley Rule and Level at that time was
the largest U.S. manufacturer of hand tools, and was founded in 1857 by a relative of
Frederick T. Stanley.
Over 150 years later, The Stanley Works is renowned for the quality of its
tools, hardware, doors and home decor products. The Stanley Works currently offers over
50,000 different products for professional, industrial, and consumer use. With sales in
excess of S2.7 billion, Stanley Works has two reportable business segments: Home
Improvement and Consumer, and Industrial and Professional. Stanley products have a
strong presence both globally and domestically in each of these segments. Table 3 is a
summary of business activities conducted by the Stanley Works that have made it into the
company it is today.
Table 3. Summary of The Stanley Works business activity ______Date Business Activity (a) (b) ______June 1926 Acquired property of American Tube & Stamping Co. at Bridgeport, Connecticut (sold in 1954) October 1959 Formed Intemational-Stanley Corp., Omaha, Neb. a joint venture with International Paper Co. to sell conugated freight car doors and car liners (sold in 1977) 1963 Formed Stanley-Brockhaus Gesellshaft fur Verpackung M.b.H. (Germany) to sell steel strapping and strapping tools (sold in 1975), and Stanley-Titan Property, Ltd organized for manufacture of hand tools in Australia, both under joint venture agreement January 1964 Stanley Chemical Division acquired assets of Wilbur & Williams Co., Inc. (sold in 1969) June 1964 Purchased for cash production assets of Arnold Artex Co. (sold on March 1970)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 49
Table 3. Cont. Date Business Activity (a) (b) 1965 Acquired Berry Industries, Inc., Birmingham, Michigan, manufacturer of garage doors. Also acquired Berry Door Ltd., Wingham, Ontario, Canada and McKinney- Skillcraft Co., Toronto April 1966 Acquired assets and business of Volkert Stampings, Inc.. New York for S 10,000,000 August 1966 Acquired Amerock Corp., Rockford, Illinois in a tax-free merger in exchange for common stock. Operations discontinued in 1974 (sold September 28, 1975) October 1966 Acquired controlling interest in four Latin American subsidiaries of the Collins Co. August 1969 Acquired Taymouth Industries, Ltd., Toronto, Canada, operated as Stanley- Taymouth Ltd. (sold in 1977) September 1970 Acquired Walter Finkeldei GmbH Wuppertal-Barmen. Germany (now Stanley Tools GmbH) December 1970 Acquired a majority interest in S.A. Quenot-Mabo, Besancon, France. Acquired S.A. Wetty & Sons, Inc., Royersford, Pa. (now Royersford Plant, Stanley Tools Division) April 1971 Acquired Prestressed Concrete of Colorado, Inc. (now Stanley Structures Inc.) 1971 Acquired Bumeda Steel Products, Ltd (sold in 1979) April 1972 Acquired Ackley Manufacturing Co. and Ackley Sales Co., Clackamas, Oregon for 92.650 common shares (now Stanley Hydraulic Tools Division) May 1972 Acquired Prestressed Concrete Products Inc., Albuquerque, NM for 12,300 common shares September 1972 Acquired Jed Products Co. for 65,000 common shares (now Stanley Automatic Openers Division) 1973 Acquired Armbro Materials and Construction Ltd., Antichoc, S.A., William Mills, Ltd., and Pennsylvania Saw Corp. for 58,020,000 January 1973 Acquired Wilson Concrete Products, Ltd. for cash (sold in 1985) September 1973 Acquired remaining minority interest in S.A. Quenot-Mabo, Besancon, France 1974 Acquired certain assets of Williams Mills & Co. Ltd., and Rapner Holdings Ltd. (became Stanley Garden Tools, Ltd., sold in 1980). Acquired Multi-Elmac Co. 1976 Acquired Wolco Corp. for 85,000 common shares July 1976 Acquired remaining 50% outstanding common stock of Stanley Works Pty. Ltd. for cash April 1980 Acquired Mac Tools, Inc. through an exchange of shares June 1980 Acquired C.E.S. Inc. through an exchange of shares. Sold electric tool business to Robert Bosch, GmbH April 1981 Sold drapery hardware business to Newell Companies, Inc. June 1983 Acquired Taylor Rental Corp. for 511,948,000 including 474,217 common shares April 1984 Acquired Proto Industrial Tools for 534,958,000 June 1984 Acquired HED Corp. for 54,269,000 February 1986 Acquired Textron’s Bostitch division for approximately 5201,043,000 1986 Acquired Chiro Tool Manufacturing Corp., Taiwan March 1986 Acquired the hand tools business of the Peugeot group in France June 1986 Acquired Electronic Sortation Systems, Inc. November 1986 Acquired Hartco Co. for 359,409 common shares. Acquired Sutton-Landis of St. Louis, Missouri December 1986 Acquired Halstead Enterprises, Inc. December 1986 Acquired operating assets of National Hand Tool Corp. for approximately 578,000,000 1987 Acquired Acme Holding Corp., a leading manufacturer of sliding and folding door hardware, for 899,999 shares of common stock February 1987 Merged Acme Holding Corp. for 899,999 common shares
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 50
Table 3. Cont. Date Business Activity (a) (b) March 1987 Acquired Plan-A-FIex Designer Co. of New Orleans, La. August 1987 Sold Stanley Steel Division, Stanley Strapping Systems Division, and Stanley Precision, Inc. to Cold Metal Products Company, Inc. for $36,306,000 September 1987 Acquired for cash selected assets of Beach Industries, Ltd. (Canada) October 1987 Sold Stanley Structures Inc. divisions in San Antonio, Tx and Albuquerque, NM 1988 Sold remaining prestressed concrete operation for $9,290,000 July 1989 Acquired Rabone Chesterman, U.K. Acquired Windor Manufacturing Ltd. of Vancouver, B.C. Canada August 1989 Acquired The Parker Group Inc., of Northboro, Mass. Acquired Propoint Inc. of Cary. Illinois October 1989 Acquired American Pneumatic Technologies, Inc. of O’Fallon, Missouri January 1990 Sold drapery hardware business of Stanley Works Italia of Figino, Italy to Metallyttans, A.B. of Anderstorp, Sweden 1990 Acquired assets of three businesses for $15,600,000 1991 Acquired Sidchrome Tools in Australia; Monarch Mirror Door Co.; J.B. Supplies, Inc.; and Nirva in France May 1991 Acquired Mosley-Stone, Ltd., a U.K. manufacturer of paint brushes, rollers and decorators’ tool 1992 Acquired Goldblatt Tool Company. Acquired controlling interest in Tona a.s. Pecky, a Czech manufacturer of mechanics tools January 1992 Acquired LaBounty Manufacturing Inc., a Minnesota-based manufacturer of large hydraulic tools, in exchange for 642,940 shares of common stock April 1992 Acquired Mail Media (Jensen Tool, Inc. and Direct Safety) July 1992 Acquired American Brush Company, Inc., a U.S. manufacturer of paint brushes and decorator tools 1993 Acquired Rikkoh-Sha Co. Ltd. a mechanical tools distributor in Japan June 1993 Sold the franchise operations of its subsidiary, Taylor Rental Corp., to SERVISTAR Corp. July 1993 Acquired Friess & Co. KG, a German manufacturer and marketer of paint rollers and brushes June 1994 Sold Taylor Rental Corp. company-owned stores to General Rental L.P. November 1997 Acquired assets of Atro Industriale, a manufacturer and distributor of pneumatic fastening tools, collated nails, and staples for $46.3 million August 1998 Completed acquisition of ZAG Industries, Ltd. for $ 129,300,000. The purchase price included a cash payment of $114,400,000, contingent payments based on ZAG’s estimated earnings over a five-year period and acquisition related costs April 2001 Acquired Contact East, Inc. Source: Company data report from FIS Online (http:/ www.fisonIine.com). Obtained on 10/9/01.
Along with three other companies and five individuals, The Stanley
Works was criminally indicted on May 22, 1990 for conspiring to fix prices charged for
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 51
architectural hinges sold in the U.S.24 Architectural hinges are used to hang heavy doors,
and are commonly found in large buildings. The defendants were charged for
participating in a price-fixing conspiracy that began as early as April 1986 and continued
for at least two years. The firms involved in the price-fixing conspiracy colluded to
reduce list prices of the architectural hinges, but restricted their discounts to customers.
The net effect of these actions would be an overall increase in prices. A federal grand
jury in St. Louis, Missouri returned a one-count indictment charging that their acts
violated Section 1 of the Sherman Act.
Richard H. Ayers, the chairman and chief executive officer of The Stanley
Works stated “In July 1988, we reported a grand jury investigation into possible price-
fixing activities in the sale of architectural hinges used for non-residential applications,
which constitute a small portion of the company’s overall business. The investigation has
proceeded and we have set aside reserves in the first quarter which, in addition to
reserves previously made, we believe are adequate to cover the potential consequences of
this investigation and any litigation resulting from it. The amount reserved for this matter
in the first quarter approximates the gain the company realized in the first quarter from
the sale of its drapery hardware business in Europe.”25 This comment was revealing in
that it acknowledges that Stanley management recognizes that there are costs associated
with an antitrust investigation, and are making preparations to finance the expense. In this
24 The other defendants were The Hager Hinge Co., McKinney Products Co., Lawrence Brothers, Inc., Richard G. Martin, John F. Hollfelder, Robert A. Haversat, David B. Gibson, and John A. Lawrence. Architectural hinges are also known as contract-grade or commercial-grade hinges.
25 This quote was reported to the business desk in New Britain, Connecticut on April 3, 1990 from the PR News wire.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 52
instance, the CEO of Stanley asserts that they are in a financially stable position, and able
to make the expenditure without much disturbance to the firm.
The Assistant Attorney General of the Antitrust Division, James F. Rill,
stated that a federal antitrust grand jury investigation into alleged antitrust violations
involving architectural hinges was what initially brought about the charges. Litigation
continued for three years until March 31, 1993 when The Stanley Works filed pleas of no
contest. As a result, the firm was fined S8 million. The sentence was later modified on
July 13, 1993 to a $5 million fine and community service which required The Stanley
Works to pay SI million to the courts for offering ethics courses at universities (given
sales o f S2.7 billion, this is essentially only 0.2 percent of sales).
Figures 1 through 3 graphically illustrate specific firm data against macro
data over time. From these figures, insight into whether the antitrust litigation had an
effect on firm performance is attained. The shaded area represents the period during
which the antitrust indictment against the firm was initiated and concluded.
Figure 1 reveals the share of Stanley employment as a share of the primary
industry in which the firm operates—Fabricated Metal Products—and manufacturing
sector employment. From the graph, it appears that firm employment tends to follow
generally the same trend as the industry. During the initial period of antitrust litigation,
from 1990 to 1991, Stanley Works employment falls by slightly over 2.0 percent. This
fall could be the firm’s knee-jerk reaction to an antitrust indictment and the rise in costs
due to litigation. While this reduction in employment could be partly attributed to the
antitrust case, it is likely that the economic recession and decline in industry employment
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 53
also played a role. The decline in firm employment and subsequent recovery beginning in
1990 and ending in 1993 also correspond with that which took place in the overall
manufacturing sector.
Figure 1. The Stanley Works employment as a share of manufacturing sector and industry employment for the years 1979 - 1999
1.40% - 0.10%
1.20% 0.08% 1.00%
0.80% 0.06%
0.60% 0.04%
0.40% - 0.02% 0 .20%
0.00% 0.00% 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 ■ ■ S h a re of Fabricated Metal Products Industry Enployment (SIC code 34) —♦— Share of Manufacturing Sector Employment
During the period of interest, 1990 to 1993, there is a slight inconsistency.
Between 1990 and 1991, Stanley employment as a share of manufacturing sector shows a
modest increase whereas Stanley’s share of industry employment shows a modest
decrease. This implies that sector employment fell more steeply than at the industry level.
Generally speaking, the share of Stanley employment follows the same pattern across
both industry and sector. The number of employees at The Stanley Works has fluctuated
greatly over the 20-year period examined with an average growth rate of negative 0.2
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 54
percent. By 1999, Stanley’s employee base shrank by nearly 1,500 employees compared
to 1979.
Employment at Stanley fell considerably beginning in 1995. The decline
begins two years after the firm lost the antitrust case and persists until the end of the
period under study. In 1996, while industry employment was still rising, Stanley
employment decreased by nearly 1,000 workers. The decline in firm employment
suggests that an employment lag might possibly exist from the closing of the antitrust
case. This finding suggests that the adverse employment effects do not emerge
immediately, but after some time—in this case two years. This supports the inclusion of
lags in the previous empirical analysis.
Figure 2. The Stanley Works sales as a share of manufacturing sector and industry sales for the years 1979 - 1999
- 0.08%
- 0.05‘ 0.80% r 0.O»% 0.60% f 0.03% 0.40% 4 4 0.02%
0.20% * r 0.01%
0.00% - * 0 .00 % 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 ■ ■ I Share of Fabricated Metal Products Industry Sales (SIC code 34) ♦ Share o f Manufacturing Sector Sales
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 55
Figure 2 is similar to Figure 1 except the comparison variable is no longer
employees, but rather sales. Overall sales declined considerably during the period 1988 to
1991 at the firm, industry, and sector levels. In fact, sales fell over SI90 million over this
period at The Stanley Works. The largest annual sales decline occurred between 1990 and
1991 when sales decreased by nearly S96 million. In addition to this time frame
corresponding with the recession, it was also when the antitrust litigation began.
Architectural hinges are a component of the Home Improvement and Consumer business
segment. In 1990, this segment brought in S998.6 million in sales, which constituted for
approximately 50.5 percent of total sales.26 Stanley sales rebounded after 1991, aside
from a one-year stagnant period between 1992 and 1993. Industry and sector sales
continued to rise until the end of the period. Similar to what occurred with employment,
Stanley sales declined for two years beginning in 1995. Despite the increase experienced
in both industry and sector sales, the decline suggests the possibility of a lagged adverse
effect on firm performance from antitrust litigation.
Stanley’s sales as a share of industry and manufacturing sector sales
follow a consistent pattern with gains in both shares by the end of the twenty-year period.
During the antitrust litigation period, firm shares rose in the early stages and fell in the
latter stages. This implies that industry and sector sales experienced both greater declines
and recoveries compared to the firm. The final measure for firm performance is seen in
Figure 3, which depicts sales per employee as a proxy for employee productivity. Pre-
and post-indictment period, The Stanley Works experienced growth in employee
"6 Data obtained from Stanley Works’ 1990 Annual Report.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 56
productivity. This graph is the most revealing of the figures because it clearly illustrates
the stagnation in growth that lasted for precisely the duration of the antitrust case. While
it is less clear whether the antitrust litigation affected the other measures, it seems more
likely that employee productivity was impacted.
Figure 3. The Stanley Works sales per employee for the years 1979 - 1999
* 'j
i 8
;■>
« i I** 4 I* 1 I** 1 ■t 4
Black & Decker Corporation
On September 27, 1910, Black & Decker came into being as The Black
and Decker Manufacturing Company. Incorporated in Maryland, the firm operated for 75
years under that name. It was not until the year 1985 that The Black and Decker
Manufacturing Company became Black & Decker Corporation. The firm has operations
in three primary business segments: Power Tools and Accessories, Hardware and Home
Improvement, and Fastening and Assembly Systems. Black & Decker and The Stanley
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 57
Works overlap in the Tools and Hardware reportable business segments. According to
Hoover’s, Black & Decker is considered to be one of Stanley Work’s top competitors.27
With sales in excess of S4.3 billion. Black & Decker is one of the leading
producers of electric power tools, power tool accessories, electric lawn and garden tools,
residential security hardware and plumbing products. Adding to their reputation, Black &
Decker’s plumbing products business is among the three largest faucet manufacturers in
North America. Its products are marketed and manufactured for both domestic and
commercial use.
Black & Decker is recognized for many of its products and has been a
market leader in several aspects. For instance, in 1915 Black & Decker was the first to
develop the power drill. Additionally, Black & Decker is renowned for designing the
Lunar Surface Drill. This drill was later used by astronauts to remove core samples from
the moon. Headquartered in Towson, Maryland, Black & Decker’s products and services
are sold in over 100 countries and has manufacturing operations in ten percent of the
countries. Table 4 is a summary of the firm’s business activities.
Table 4. Summary of Black & Decker business activity Date Business Activity (a) (b) 1926 Acquired Marschke Manufacturing (sold in 1982) 1928 Acquired Van Dom Electric Tool Co. for about 51,000,000 1929 Acquired Domestic Electric Co. (dissolved in 1930), Fleming Machine Co. (dissolved in 1936) and Loadometer Co. (dissolved in 1938) 1944 Renamed Domestic Electric Co. to Black & Decker Electric Co. and sold for $560,000 in cash and 5240,000 in par value o f preferred stock
27 Hoover’s Online company profile http://www.hoovers.com. Obtained on 3/14/02. The two companies have different primary SIC codes.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 58
Table 4. Cont. Date Business Activity (a) (b) 1959 Issued 35,789 common shares for assets of Master Pneumatic Tool Co., Inc., Bedford Ohio, maker of air-powered tools, and 1,215 common shares for assets of Bedford Properties, Inc., owner of property in Bedford, Ohio. Assets were transferred to Master Power Corp., which was merged in 1965. 1960 Issued 120,000 common shares to American Machine & Foundry Co. for all stock of DeWalt. Inc., Lancaster (liquidated in 1964) 1961 Acquired a 52% interest in Societe Constructions Mecaniques du Dauphine Socomeda, Lyons, France (now Black & Decker (France) S.A.R.L.-100% owned) maker of portable electric tools. Issued 59,014 common shares for all capital stock of Star Utensili Elettrici, S.p. A., Milan, Italy, maker o f portable electric tools 1969 Acquired all shares of Tatry Officina Meccanica S.r.L., Perugia, Italy, a radial arm saw maker, for $150,000 cash and 22,200 common shares 1970 Acquired assets of Carbide Router Co., Inc. Operations continued as a wholly- owned subsidiary of same name 1971 Acquired for cash certain assets and property of Wisconsin Knife Works, Inc., which made woodworking knives and tools. Operations continued as a wholly- owned subsidiary of same name 1974 Acquired McCulloch Corp 1983 Sold assets of the gasoline chain saw business 1984 Acquired substantially all of the assets and businesses o f General Electric Company’s Housewares operations 1985 Acquired EIu Machines, S.A. Switzerland January 1987 Acquired American Emblem Tool Co. 1989 Acquired all of the outstanding shares of common stock of Emhart Corporation at a price of S40 per share in cash, with the aggregate purchase price for such shares and related costs totaling approximately S2,675,000.000 January 1990 Sold its Bostik chemical adhesives and sealants business to Orkem, S.A., French chemical company, for 5345,000,000 March 1990 Sold its Texon Footwear Materials business to United Shoe Machinery Group Limited for approximately 5125.000,000 April 1990 Sold its Arcotronics business to Nissei Electric of Japan for approximately 580,000,000 1990-1991 Sold five additional Emhart businesses as well as certain other non-strategic assets including the Emhart headquarters, U.S. Capacitors, True Temper Hardware, Medic Div. Of PRC, GardenAmerica, the North American Mallory Controls and the Brazilian Mallory Control Division and the Brazilian Footwear Materials Division. The aggregate value of these sales was approximately 5300,000,000 January 1994 Sold its Corbin Russwin Archectural Hardware business to Williams Holdings PLC for 580,000,000 March 1995 Sold PRC Realty Systems, Inc. for approximately 560,000,000 September 1995 Sold PRC Environmental Management, Inc. for 535,500,000 February 1996 Sold PRC, Inc., the remaining business in its information technology and services segment, for 5425,000,000 June 1998 Sold its household products business in North America and Latin America, excluding Brazil, to Windmere-Durable Holdings, Inc. for S315,000,000. Transfer of the household products manufacturing operations in Mexico will occur upon receipt of final regulatory approval in Mexico
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 59
Table 4. Cont. Date Business Activity (a) (b) September 1998 Sold its glass container-making and inspection equipment business, Emhart Glass, to Bucher Holding AG of Switzerland for $178,700,000 2000 Sold its remaining interest in True Temper for $25,000,000. Acquired Emglo Products, L.P. Acquired Momentum Laser Source: Company data report from FIS Online (http:, www.fisonline.com). Obtained on 10/9/01.
Unlike The Stanley Works, Black & Decker was not indicted for violating
antitrust laws between 1979 and 1999. Figure 4 illustrates employment growth over time
for both The Stanley Works and Black & Decker. From 1990 to 1993, the two
companies’ employment levels move in practically opposite directions. While The
Figure 4. Comparison of Employment Levels between The Stanley Works and Black & Decker for the Years 1979 - 1999
19 5 - 43
18.5 - 38
17 5
16.5
15.5
14 5
I 3.5 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999
■— Stanley W orks Employment — B lac k & Dec ker Emp lo y men t
Stanley Works initially experiences negative employment growth and then positive,
Black & Decker shows precisely the reverse. This implies that as The Stanley Works is
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 60
shedding its employees, their competition is expanding its employee base. It is possible
that Black & Decker saw this as an opportunity to gain market share or surge ahead of its
competitor.
Figure 5 contrasts Black & Decker employment as a share of industry and
sector employment. Employment at both the industry and sector levels shows a relatively
steady and continual decline whereas Black & Decker’s employment experiences more
volatility. In fact, from 1987 to 1990 employment at Black & Decker increased by nearly
24,000 employees. According to table 4, Black & Decker acquired American Emblem
Tool Co. during this time, which may have contributed to the surge in employment. The
fall in employment in the latter half of the time series beginning with the recession in
1990 is somewhat consistent with the industry and sector. Despite the continual
downward trend. Black & Decker still exhibited employment growth between 1979 and
1999. Black & Decker’s share of industry employment follows steadily along with its
share of sector employment. During the litigation period, Black & Decker’s share of
industry employment rises modestly from 1991 to 1992, while its share of manufacturing
employment continues to fall. It is also during this time that employment levels recover
slightly from a steep decline.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 61
Figure 5. Black & Decker employment as a share of manufacturing sector and industry employment for the years 1979 - 1999
0.25%
0.20%
0.00% 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 ■ ■ 1 ShareofIndustrial&Machinery Equipment Industry Enployment (SICcode 35) —•— Share of Manufacturing Sector Employment
Figure 6 compares Black & Decker share of sales to industry and
manufacturing sales. Sales in the Industrial Machinery & Equipment industry remain
relatively steady up until 1991 when it begins a steep ascent. Sales for the firm, on the
other hand, rise from 1983 to 1990, and then begin to level off. It is interesting to note
that even when the firm’s employment declines, sales are not greatly affected. Black &
Decker sales trend upward, which happens to follow along more consistently with sector
sales. Despite this overall pattern, sales for Black & Decker and the manufacturing sector
are adversely affected by the recession of 1990 to 1991. As seen in Figure 6, Black &
Decker’s industry share follows the same pattern as its sector share. This was also the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 62
case with The Stanley Works. Figure 7 shows sales per employee, which is another proxy
for firm productivity over the period. For the majority of the period, sales per employee
are on the rise with a clear upward trend. During The Stanley Works’ antitrust litigation
period, productivity at Black & Decker is rising.
Figure 6. Black & Decker sales as a share of manufacturing sector and industry sales for the years 1979 - 1999
0.16%
0.12%*
0.06%*
0.(W% ■
0.02%*
0.00% 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 Share of Industrial Machinery & Equipment Industry Sales (SIC code 35) Share o f Mmufactiring Sector Sales
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 63
Figure 7. Black & Decker sales per employee for the years 1979 - 1999 : 10
r>o
•in
W ’ n MS'. V»s; V‘S ' m ’ WS'I 1-1 «|N VI ■* I’l'i'*
Case Study 2: Archer-Daniels-Midland and ConAgra
Archer-Daniels-Midland Company
Archer-Daniels-Midland (ADM) started with John Daniels in 1902 when
he started Daniels Linseed Company in Minneapolis. The company he Founded created
linseed oil through the crushing of flaxseed. In 1903. another flaxseed crusher. George
Archer, joined the company. Twenty years later, the company bought Midland Linseed
Products and became incorporated as Archer-Daniels-Midland in Delaware on May 2.
1923.
Currently. ADM has over 300 processing plants worldwide, and has
become the world's largest linseed oil producer. ADM procures, transports, stores.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 64
processes, and merchandises agricultural commodities and products such as oilseeds,
com, wheat, cocoa beans, oats, barley and peanuts. The product generated is used
primarily for food or feed ingredients. ADM has four reportable business segments:
oilseed products; com products; wheat and other milled products; and other products.
ADM has also ventured into research and development activities. Some examples of the
products developed and in use are textured vegetable protein (TVP®), fuel ethanol, and
55 percent high fructose com syrup. Headquartered in Decatur, Illinois, ADM has
undergone so much growth that it is now a multi-billion dollar company. In 2001, ADM
sales exceeded S20 billion. Table 5 provides a summary of their business activities.
Table 5. Summary of Archer-Daniels-Midland business activity Date Business Activity (a) (b) July 1923 Acquired the properties of Midland Linseed Products Co. for S3,175.000. This company formed in 1898 operated linseed oil mills adjoining those of Archer- Daniels Linseed Co. at Minneapolis, Toledo and Edgewater February 1928 Acquired the entire property and assets of William O. Goodrich Co. of Milwaukee. manufacturers of highly specialized and refined qualities of linseed oil, for 13.712 shares of common stock July 1928 Purchased the linseed oil properties and business of Fredonia Linseed Oil Works Co., located at Fredonia, Kansas October 1929 Acquired the Vemer G. Smith Co. July 1933 Submitted an offer to the bondholders and noteholders protective committee of Commander-Larabee Corp. to purchase all bonds and notes of such Co. at a price of 60% flat for bonds and 30% flat for notes, provided at least 80% of bonds and notes assented to such purchases. Over 80% wras then acquired and in subsequent years the balance, together with all the outstanding preferred and common stocks was acquired, now operated as ADM Milling Co. July 1954 Acquired resin division of the U.S. Industrial Chemicals 1957 Acquired Federal Foundry Supply Co., and certain assets, including plant, of Price Vamish Co. August 1960 Sold Commander Elevator Division October 1960 Acquired J&O Grain Co. January 1962 Acquired Atkinson Milling Co. April 1962 Acquired 12 Terminal Elevators from Norris Grain Co. April 1963 Acquired Hoover Grain Co. September 1965 Acquired Galesburg Soy Products Co. for cash November 1965 _____ Acquired Synco Resins, Inc. ______
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 65
Table 5. Cont. Date Business Activity (a) (b) 1967 Co.'s chemical operations were sold to Ashland Oil & Refining Co. for approximately S65,000.000 total sale price. 535,000,000 in cash and balance in interest bearing notes payable over 3 years; purchased 79.9% interest in Fleischmann Malting Co., Inc. for cash December 1967 Sold alfalfa dehydrating operations to Kansas-Hebraska Natural Gas Co., Inc. 1968 Acquired Ross & Rowe, Inc. for 6,000 shares November 1969 Acquired First Interoceanic Corp. and about 88% ownership of National City Bank, Minneapolis April 1970 Acquired certain assets of companies owned by John Vanier of Salina, Kansas, which manufacture and market a variety of retail consumer packaged food products and animal foods April 1971 Acquired 50% interest in Com Sweetners Inc. January 1973 Acquired 50% of British Arkady Holdings Ltd., which subsequently acquired British Arkady Co., Ltd., England April 1973 Acquired Salina Terminal Elevator Co.. Smoot Grain Co. and Central Kansas Mill and Elevator Co. for 434,045 common shares July 1973 Acquired Supreme Sugar Co., Inc. for 100,000 common shares January 1974 Purchased a soybean processing plant and edible oil refinery at Araraquara, State of Sao Paulo, Brazil: purchased Oliefabriek De Plocg B.V. and its subsidiary Oliefabriek De Merwede B.V.; purchased ADM do Brasil Produtos Agricolas, Ltda, Agriproducts, Inc. (Cayman Islands) and Ardanco S.A. (Spain) October 1974 Sold Flax Fibre division 1975 Acquired the remaining 50% interest in Com Sweetners Inc. and merged into Co. June 1975 Agrinational Ltd. of Cayman Islands was organized as an offshore insurance company for domestic coverage August 1975 Acquired Tabor & Company, Decatur, Illinois for 1,500,000 common shares April 1976 Co. and Nestle Alimentana S.A. signed a joint venture pact pursuant to the terms of which Nestle acquires a 50% interest in the Brazilian soybean processing and vegetable oil refinery operation currently operated by Co. at Araraquara December 1976 Acquired remaining 20.1% minority interest in Fleischmann Malting Co., Inc. for the issuance of 78,669 common shares March 1977 Acquired all the outstanding shares of New Era Milling Company for the issuance of 148,582 common shares June 1978 Merged with Ross & Rowe, Inc.; Tabor Milling Company & Wenona Elevator Company August 1980 Acquired the fixed assets of Central Soya Company’s Chicago plant June 1981 Acquired Herbert Saliba, Inc., Lubbock Cotton Oil Company, Quanah Cotton Oil Company, and Sweetwater Cotton Oil Company December 1982 Acquired a substantial part of Toepfer International Shareholdings 1986 Acquired Krause Milling Company November 1986 Acquired Kurth Malting Company December 1986 Acquired Hickory Point Bank & Trust June 1987 Acquired the Gold Kist soybean processing plant at Valdosta, GA from Gold Kist, Inc. 1989 Acquired Monarch Feed Mills and Alabama Feed Mills August 1997 Acquired an additional 1,100,000 shares of IBP Inc., Dakota City, NE, boosting its stake to 7.23% September 1997 Acquired an additional 1,040,000 shares of IBP Inc., boosting its stake to 8.36% November 1997 Acquired a 30% interest in Minnesota Com Processors
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 66
Table 5. Cont. Date Business Activity (a) (b) July 1998 Co. and Lasaffre et Compagnie announced the combination of their malting operations. Co. has a 40% ownership interest in International Malting Company. Headquartered in Milwaukee. WI. International Malting will operate in the U.S. under the name Froedtert Malt. The European unit will continue to use the name Grandes Malteries Modemes, and the Canadian division will continue to operate under the name Dominion Malt Source: Company data report from FIS Online (http: www.tisonline.com). Obtained on 10/9/01.
In two separate cases, ADM was indicted for violation of antitrust laws—
once civilly and once criminally. The first occurred on December 14, 1982 when the
Department of Justice filed a suit in the U.S. District Court in Des Moines, Iowa against
ADM and Nabisco Brands, Inc. for monopolization, restraint of trade, and acquisition.
The government challenged the lease held between Nabisco of New York City and ADM
of Decatur, Illinois. ADM agreed to lease, with options to buy, two com products
facilities owned by Nabisco for 13 years.'8 The government alleged that this arrangement
would severely lessen competition and unreasonably restrain trade in the manufacture
and sale of high fructose com syrup (HFCS). Moreover, these acts violated two antitrust
laws: Section 7 of the Clayton Act and Section 1 of the Sherman Act.
High fructose com syrup is used mainly in soft drinks and other food
products to sweeten the taste. Roughly 7.33 billion pounds of this sweetening agent was
sold in the U.S. alone in 1981. Prior to the lease agreement between the two companies,
the HFCS industry was already highly concentrated with an HHI of approximately 1,610.
Specifically, the four largest domestic manufacturers of HFCS accounted for 70 percent
of the manufacturing capacity, and the eight largest domestic producers were responsible
28 Specifically, ADM would lease com wet milling plants in Montezuma, New York, and Clinton, Iowa.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 67
for 90 percent of manufacturing capacity. According to the Los Angeles Times, ADM
controlled about a third of the S2.6 billion market for HFCS, or approximately S870
million.29
The suit claimed that ADM was the second largest manufacturer of HFCS
in the U.S. and generated roughly 25 percent of domestic production and about 21
percent of domestic finishing capacity in 1981. The suit also claimed that Nabisco was
the third largest manufacturer of HFCS and accounted for 16 percent of domestic
production and 12 percent of domestic finishing capacity in 1981. The government
believed that allowing the lease to take place would lead to an even more highly
concentrated industry. In fact, their calculations suggest that the HHI would increase 530
points to 2,140. The lease agreement enabled Nabisco to stop manufacturing HFCS,
because ADM would acquire its com products business. As a result, ADM’s control of
total domestic production and finishing capacity would increase 16 percentage points
from 25 to 41 percent and 12 percentage points from 21 to 33 percent, respectively. ADM
would then become the largest domestic manufacturer in terms of both production and
finishing capacity. Whereas previously the four largest domestic manufacturers
accounted for approximately 70 percent of manufacturing capacity, they would now
account for 80 percent. The lease agreement would allow the eight largest domestic
manufacturers to account for virtually all of HFCS manufacturing capacity whereas
previously they held 90 percent.
29 July 11, 1995, Tuesday, Home Edition.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 68
Given these estimates, the government sought to rescind the lease and the
accompanying agreements, restoring Nabisco’s com products business to a competitive
position, and injunctive relief. In response, ADM and Nabisco put forth a motion to
dismiss the case. The motion to dismiss was denied on March 29, 1984. The U.S. Court
of Appeals in St. Louis later affirmed the denial on February 24, 1986. However, on May
29, 1988 the case was dismissed. Seven months later, on December 15, 1988, the U.S.
Court of Appeals for the Eighth Circuit reversed the dismissal and declared that the
relevant market for which the government alleged that trade would be restrained could be
limited to only HFCS. Nearly nine years after the antitrust indictment was initially filed,
the case ended on December 10, 1991 when a federal district court ruled in favor of
ADM and Nabisco. According to the Trade Regulation Reporter and The New York
Times, the Court stated that the lease of com products facilities ultimately did not lessen
competition. Furthermore, the major customers of HFCS had enough influence to force
ADM and others to compete.30
Five years later in 1996, ADM was indicted again. This time the firm was
faced with two-count felony charges for allegedly having a role in two international price
fixing conspiracies that would eliminate competition and allocate sales in the lysine and
citric acid markets worldwide. Lysine is an amino acid applied by farmers to livestock
feed as an additive to assure proper growth. Citric acid is a flavor additive and
preservative created from different types of sugars. It is added to drinks, processed food,
30 July 10, 1995, Monday, Late Edition.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 69
detergents, pharmaceutical and cosmetic products. The lysine and citric acid markets are
global industries valued at S600 million and SI.2 billion, respectively.
Count 1 charges were for colluding with Ajinomoto Co. Inc., Kyowa
Hakko Kogyo Co. Ltd., Sewon America Inc. and other unnamed individuals and
corporations to fix prices in the lysine market during the period June 1992 to June 27,
1995. The identified corporations were also indicted, but in a separate case. Count 2
charges were for conspiring with others to limit and eradicate competition in the citric
acid market beginning in January 1993 and ending in June 27, 1995. The Omaha World
Herald reports that “according to company regulatory filings, court rulings, government
studies and Wall Street analysts. Archer Daniels gets about S1.3 billion of its SI 1.4
billion of annual revenue from the three products [HFCS, lysine, and citric acid] under
federal scrutiny.”31 This amounts to approximately 11.4 percent of their total revenue.
Lasting less than one year, the government quickly won the case on October 15, 1996
when ADM pleaded guilty and agreed to pay a penalty of S100 million. The firm was
fined S70 million for Count 1 and S30 million for Count 2. The punishment imposed in
this case was the largest criminal antitrust fine imposed up to that date.
On November 18, 1997 Galavan Supplements, Ltd. (Galavan) filed a class
action lawsuit against ADM stating that ADM’s involvement in the citric acid price-
fixing conspiracy caused “an unreasonable restraint of foreign trade and commerce in
violation of antitrust laws”.32 Galavan is a foreign company, and had previously bought
31 July 14, 1995 Friday, Metro Edition.
3‘ The specific case is: Galavan Supplements, Ltd. v. Archer-Daniels-Midland Co., et al. No. C 97-3259 FMS U.S. District Court for the Northern District of California 1997 U.S. Dist. Lexis 18585.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 70
citric acid from ADM. ADM filed a motion to dismiss, and on November 19, 1997 the
court granted ADM’s request for dismissal. The court held that the 1982 Foreign Trade
Antitrust Improvements Act gave it jurisdiction over the claim.33 Moreover, the court
held that Galavan lacked standing to prosecute its claim of a violation of the Sherman
Act. In sum, the nature of Galavan’s injury extended beyond the reach of the antitrust
laws, and the more appropriate plaintiffs would be U.S. consumers injured by ADM’s
actions.
Figure 8. Archer-Daniels-Midland employment as a share of manufacturing sector and industry employment for the years 1979 - 1999
1.60% - 0.14%
0. 12%
1.20% 0. 10%
0.08% 0.80% 0.06% 0.60%
0.04% 0.40%
r 0.02%
0.00% ■■IIM M M IM IM .W II" !! Mil ■ ■ ■ M-JM. ■ . ■ ■ Q.00% 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 Share of Food & Kindred Products Industry Employment (SIC code 20) —♦—Share of Manufacturing Sector Employment
33 1982 Foreign Trade Antitrust Improvements Act, 15 U.S.C.S. § 6(a).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 71
Figure 8 reveals that ADM’s share of both industry and sector
employment has risen rather consistently over the period. Comparing ADM employment
to the Food & Kindred Products industry, ADM employment has risen steadily, aside
from a considerable dip in 1989, while industry employment has remained relatively
stable over time. As stated previously, the first antitrust indictment was initiated in 1982,
and lasted for nine years. Aside from two small downward shifts in 1988 and another in
1995, ADM employment was, by and large, unaffected by the antitrust case. Unlike the
Food & Kindred Products industry, manufacturing sector employment has been trending
downwards. Whereas only the first two recessions adversely affected industry
employment, all three had negative impacts on manufacturing employment. ADM
employment, on the other hand, was harmed by neither the recessions nor the decline in
manufacturing sector employment.
Unlike employment, firm sales generally did follow a pattern similar to the
industry and sector. Akin to employment, ADM sales were unaffected by antitrust
litigation. Aside from three minor declines in sales during the litigation period that lasted
one year each beginning in the years 1981—which may have been a byproduct of the
second recession of the period— 1984, and 1989, sales have consistently risen. In 1996,
sales for the oilseed products business segment were $8,125 million. Sales in the com
products segment were $2,561 million. The wheat and other milled products business
segment reported $1,644 million in sales. The final business segment, other products, had
1996 sales of $984 million. The share of total sales for each of the aforementioned
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 72
Figure 9. Archer-Daniels-Midland sales as a share of manufacturing sector and industry sales for the years 1979 - 1999
1983 1985 1987 1989 1991 1993 1995 1997 Share of Food & Kindred Products Industry Sales (SIC code 20) Share of Manufacturing Sector Sales
business segments are approximately 61.0, 19.2, 12.3, and 7.4 percent, respectively. The
products for which ADM was indicted for antitrust in 1996, lysine and citric acid, fall
into the Other Products business segment.34 Since this business segment comprised only
7.4 percent of total sales, it is not surprising that antitrust litigation did not have much of
an impact on the company’s overall performance. Sales peaked in 1998 at $15.6 billion
only to fall nearly $2.0 billion the following year to near 1997 levels. Manufacturing
sector and industry sales fluctuated in the first half of the period under study, and trended
upwards in the latter half. Figure 9 illustrates the increasing share of both industry and
sector sales held by ADM throughout the period.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 73
The remaining ADM figure is more revealing. Despite the steady growth
in employment and sales, net sales per employee did not follow this mold. As seen from
Figure 10, sales per employee drops dramatically beginning in 1996, the year in which
the second antitrust indictment was issued and resolved. No discemable trend can be
obtained from this graph. While employment and sales are not noticeably affected by
antitrust litigation, productivity seems to be more sensitive.
Figure 10. Archer-Daniels-Midland sales per employee for the years 1979 - 1999
1.000 • ^547077030^7887954873434
ooo •
800
700 -
600 -
500 1970 1081 1983 1985 1087 1989 1991 1993 1995 1997 1999
34 Data obtained from ADM’s 1996 Annual Report. Detailed business segment sales for 1982 was not reported in the 1982 Annual Report.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 74
ConAgra Foods, Incorporated
ConAgra became incorporated on December 5, 1975 as successor to a
company originally incorporated in Nebraska on September 29, 1919. The name of this
prior company was Nebraska Consolidated Mills Co. The name ConAgra, Inc. came into
being on February 25, 1971, and the company’s present name was adopted on September
28, 2000.
ConAgra is North America’s leading foodservice manufacturer and the
second largest retail food supplier. ConAgra operates in three business segments:
Packaged Foods, Refrigerated Foods and Agricultural Products. ConAgra and ADM
compete on many levels, although primarily in the Agricultural Products line of business.
In this segment, ConAgra carries out activities such as flour, oat and com milling, dry
edible bean processing and merchandising and barley malting. ConAgra also provides
crop protection chemicals, fertilizers, seeds and information systems to wholesalers and
retailers. Some of the major brands it distributes include Clean Crop, ACA, Awaken
mPower3 (e-merge), Savage, Shotgun, Saber, Signature and Loveland Industries. Table 6
provides a detailed summary of ConAgra’s business activities.
Table 6. Summary of ConAgra business activity Date Business Activity (a) (b) September 1963 Acquired Sheridan Flouring Mills, Inc., Sheridan, Wyo. For 14,595 shares of 5% preferred stock March 1964 Acquired Nixon & Co. for 63,106 common, and 37,862 B preferred shares June 1964 Acquired Fant Milling Co., Sherman, Texas December 1965 Acquired Maplecrest Farms, of Welman, Iowa February 1966 Merged with Nixon & Co., a subsidiary June 1966 Acquired Mauser Milling Co. of Treichlers, Pa., for 33,912 common shares 1967 Acquired Harris Milling Co
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 75
Table 6. Cont. Date Business Activity (a) JbL September 1968 Acquired J.F. Imbs Milling Co., St. Louis. Mo.; Birdsey Flour Mills, Macon. Ga.; and Red Lion Milling Co., Red Lion, Pa. 1969 Acquired Dixie Lily Milling Co. and Stral Processing Co. February 1969 Acquired 98% of capital stock of Montana Flour Mills Co., Great Falls, Montana April 1970 Acquired McGehee Poultry, Inc., Arcadia, La. January 1971 Acquired Fruen Milling Co., Minneapolis, Minn. June 1971 Acquired Harrell Poultry Cos. For 75,000 common shares January 1972 Acquired H.C. Cole Milling Co. for $2,828,461 April 1972 Acquired Feedright Milling Co. for 32,717 common shares June 1972 Acquired Security Mills, Inc. for $4,499,250 December 1972 Acquired Plum Brook Farms, Inc., Gainesville, Ga. March 1973 Acquired A.P. Merrill & Co., Inc., Birmingham, Al. 1973 Acquired Geisler Pet Products, Omaha, Nebraska and Kasco Dog Food, National City, 111. 1976 ConAgra-Albion, Inc; Feedright Milling Co.; ConAgra-Georgia, Inc.; Harrel- Poultry Co., Inc.; Security Mills, Inc.; and Fruen Inc. were merged into company February 1976 Acquired Norso Distributors, Inc. and Biota International Inc. for $1,412,500 February 1977 Acquired Pet Dealers Supply Co. a wholly-owned subsidiary of Bergen Brunswig Corp. May 1977 Acquired The McMillan Co., York, Pa. flour mill and J.P. King & Sons, Inc. of Alabama; also acquired York Flour Mills, Inc. for $ 1,022,000 in cash July 1977 Acquired Burdick Grain Co. for 22,500 class C preferred shares August 1977 Acquired Bow Wow Co., Inc. for 100,000 shares of common stock November 1977 Acquired Mr. Beef Restaurants, Inc. (now Taco Plaza, Inc.), Fort Worth, Texas for approximately $3,000,000 in Co. common stock May 1978 Acquired a 49% interest in a group of domestic companies which are engaged in distribution of agricultural chemicals for $6,038,000 in cash. December 1978 Acquired Atwood-Larson Co., for $7,700,400 in cash January 1979 Acquired Port ConAgra, Inc. (formerly Bolander-Conlan Terminal Corp.) for $250,000 in cash July 1979 Acquired remaining 51% of United Agri Products Group for $7,190,000 in cash. An additional $6,000,000 in cash was paid in fiscal year 1983 to complete the terms of the original contract June 1980 Acquired Oklahoma-Kansas Grain Corp. July 1980 Acquired Hess and Clark, Inc. for approximately $9,000,000 in cash November 1980 Acquired Banquet Foods Corp. for approximately $45,000,000 in cash and $7,000,000 in non-voting, non-convertible preferred stock 1981 Acquired 50% interest in Bioter-Biona, S.A., Spain February 1981 Acquired Singleton Packing Corp. July 1981 Acquired Sea Alaska Products for 750,000 shares of common stock 1982 Acquired Grower Services, Inc., a distributor of agriculture chemicals, for $250,000; the operating assets of the Alaska Division of Alaska Packers Association, Inc., a processor of frozen and canned salmon and herring products, for $2,000,000 and a five her operating lease July 1982 Acquired Peavey Co., a grain merchandiser, food processor and operator of specialty retail stores, for 4,998,000 Co. common shares, 708,000 Co. $2.50 cum. conv. preferred shares and $52,887,000 cash
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 76
Table 6. Cont. Date Business Activity (a) (b) September 1982 Completed organization of a poultry company which combines the U.S. poultry operations of ConAgra and Imperial Foods Limited. ConAgra invested S 18,000,000 for a 50% equity interest in the new company which ConAgra manages 1983 Acquired the operating assets of: ACLI Seafood Co., a processor, distributor, and broker of commodity shrimp and other seafood for 55,727,000; Georgia Ag Chem, Inc.. an agricultural chemical distributor for 5717,000; and Occidental Chemical Co., an agricultural chemical distributor for 52,086,000 January 1983 Sold the net operating assets and leased the property, plant and equipment of its Country Skillet Poultry operations to the newly formed poultry company for 528.306.000 October 1983 Acquired SEECO, Inc., a manufacturer of vitamin and other consumer products, for 53.303,030 December 1983 Purchased Agrichem, Inc., an agricultural chemical distributor for 51,224,000; acquired assets of Armour Food Co., a subsidiary of Greyhound Corp., for 3.400.000 shares of common stock and other considerations totalling SI 66,000,000. The acquisition does not include Armour-Dial, Inc. or any of the assets, business or trademarks of the Armour Consumer Products Group 1984 Final purchase price of Banquet Foods Corp. was determined. Banquet issued 5 10,000,000 of its preferred stock to complete the transaction based on conditions of the original purchase agreement January 1984 Purchased the operating assets of Geldermann & Co., Inc. a commodity futures brokerage, for 53,580,000; acquired Florida Feed Mills, Inc. for 52,456,000 February 1984 Sold the assets of its Pet Foods Division to Farmers Energy Corp. in a book value transaction; acquired Pacifex, Inc., a fertilizer marketing company for 5400,000 April 1984 Acquired 49% of Glendon Corp. whose principal operating company is Woodward 6 Dickerson, Inc., a worldwide trading Firm for 59,657,000 June 1984 Acquired remaining 50% interest in Country Poultry, Inc. for 518,000,000 October 1984 Acquired Saprogal, S.A. 1985 Acquired Hopkins Agricultural Chemical Co. & Security Chemical and Ag-Chem, Inc. January 1985 Acquired remaining 51% of Glendon Corp for cash March 1985 Acquired Berger & Co.; invested in a new joint venture fertilizer merchandising and marketing business AgriBasics Fertilizer Company May 1985 Acquired 50% o f the U.S. Tire, Inc. October 1985 Acquired remaining 50% of the U.S. Tire, Inc.; acquired the stock of The Cropmate Company and purchased the remaining 50% of AgriBasics Fertilizer Company November 1985 Acquired the Edible Protein Division of Pillsbury company 1986 Acquired Dale Alley Co., Boyer Valley Fertilizer Co., Agchem Service; Automatic Flagman Co., Pest Select, Inc., Camerican International Maguire Associates, Inc. February 1986 Acquired Webber Farms, Inc. June 1986 Acquired the frozen food business of RJR Nabisco, Inc. for approximately 564,000,000 in cash December 1986 Acquired three Swiss companies with grain trading and marketing operations in Canada and the U.S.; acquired Dyno Merchandise Corp. and Unique Packaging a sewing notions distributor and Howe, Inc., a fertilizer and agricultural chemical business
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 77
Table 6. Cont. Date Business Activity (a) (b) January 1987 Acquired E.J. Miller Enterprises, Inc., a producer of premium boxed beef for 1,040,000 shares of common stock; merged the net assets of its Alaskan Crab and salmon business into Trident Sea Foods Corporation in exchange for 45% equity interest; acquired O’Donnell-Usen Fisheries Corp. with fishing, fish processing and marketing operations in Canada and the U.S. for approximately 516,000,000 in cash and the assumption of approximately S 19,000,000 in long-term debt May 1987 Acquired Monfort of Colorado, Inc., a vertically integrated agri-business engaged in the production, transportation, distribution and sale of beef and lamb products, for 10,749,814 shares of common stock September 1987 Acquired a 50% interest in a corporation which controls Swift Independent Packing Co. (SIPCO) and Val-Agri, Inc. for $51,500,000 in cash; SIPCO/Val-Agri is a meat processor and food distributor October 1987 Acquired assets of Natural Food Commodities, which distributes and merchandises a wide variety of dried fruits, nuts and specialty agriculture commodities; acquired Longmont Foods, a turkey processing business May 1988 Purchased the assets of International Multifoods’ U.S. flour milling business for approximately 576,000,000 in cash and the assumption of approximately 515.000.000 in long-term debt June 1988 Acquired Blue Star Foods in a cash transaction. Blue Star Foods became a division of Co.'s Consumer Frozen Food Company; ConAgra Poultry Co. acquired Mott’s Inc., a poultry processing business based in Water Valley, Miss. July 1988 Co. merged with Fernando’s Foods Corporation (Femando’s) through an exchange of shares. Co. issued approximately 1,300,000 shares of common stock for all outstanding shares of Femando’s; completed merger with Goodmark Foods Inc. for 1.08108 shares of Co. October 1988 Acquired all of the outstanding capital stock of Cook Family Foods, Ltd. in exchange for 2,178,000 shares of ConAgra common stock February 1989 Acquired the grain merchandising division of The Pillsbury Company for approximately 5153,000,000 in cash plus the assumption of approximately $4,000,000 in long-term debt May 1989 Purchased the Sergeant’s Pet Care Division of A.H. Robins for approximately 518.000.000 July 1989 Purchased the remaining 50% interest in the parent of SIPCO/Vol-Agri from Elkhom Enterprises, Inc. (effective May 1992) for the assumption ofS51,500,000 of long-term debt and preferred stock of SIPCO totalling 5160,036,000 August 1990 Acquired Beatrice Co. from Kohlberg Kravis Roberts & Co. partnerships and management investors for a combination of cash and stock valued at 51.362.000.000 May 1991 Acquired Hyman Foods Limited, a processor and distributor of frozen meat products based in Stockport, England; acquired from Foster’s Brewing Group Limited (formerly Elders IXL Limited) certain assets of Elders Brewing Materials and Elders International Wool Division and a 50% interest in Elders Meat Division July 1991 Acquired Golden Valley Microwave Foods, Inc. in exchange for 0.8514 share of common stock for each share of Golden Valley common stock January 1992 Exchanged 5,250,000 shares of common stock for all the outstanding stock of Arrow Industries, Inc. October 1992 Klein Bros., Ltd. merged with ConAgra for a purchase price of 912,104 shares of ConAgra common stock
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 78
Table 6. Cont. Date Business Activity (a) (b) February 1993 Acquired all the outstanding stock o f United Plastic Films, Inc. in exchange for 820,000 shares of common. United Plastic became part of Arrow Industries, Inc. June 1993 Co. increased its interest in Australian Meat Holding Pty. Ltd. from 50% to approximately 90% December 1994 Sold Dyno Merchandise, Inc. to Dyno Corp , a newly formed corporation owned by BT Capital Corporation, a wholly-owned subsidiary of Bankers Trust New York Corporation, and Dyno’s management November 1995 Sold certain assets of Mott’s-Blue Coach Foods, based in Water Valley, Mississippi, to Valley Fresh, Inc., based in Turlock. California. Mott’s-Blue Coach processes poultry products for industrial, retail and foodservice markets. The principal assets sold include processing plants in Water Valley; Glasgow, Kentucky; and Talmo, Georgia March 1996 Co.’s subsidiary, ConAgra Shrimp Companies, acquired the assets of Meridian Products, Inc. in a cash transaction 1997 Acquired Wilmot Pertwee; United Specialty Food Ingredients acquired certain assets of Gilroy Foods; Co.’s subsidiary, ConAgra Shrimp Companies, acquired certain assets of the Florida Sea, Inc.; Co.’s subsidiary, ConAgra Poultry Company, acquired certain assets of Texas Smokehouse Foods, Inc., now called Texas Signature Foods; ConAgra Trading and Processing Companies acquired a 50% interest in Verde Valle, S. A., a leading packager and distributor of grocery products in Mexico December 1997 Acquired Hester Industries, Inc., Moorefield, WV, in a stock transaction that was accounted for as a pooling o f interests 1998 Sold Country General Stores to Central Tractor Farm & Country January 1998 Acquired Zoli Foods Corp. in a stock transaction February 1998 Acquired Gilardi Foods Inc. in a stock transaction April 1998 Acquired Original Italian Pasta Products, Co. Inc. The first and final liquidation of S1.28 per share was payable upon the surrender of certificates 1999 Acquired the table-spreads and Egg Beaters business from Nabisco, Inc. for $400,000,000 June 1999 Acquired Holly Ridge Foods Inc. Terms were not discosed. Holly Ridge operates part of Co.’s Lamb-Weston Inc. division September 1999 Atlanta Corporation purchased J.F. Braun & Sons from Co. The J.F. Braun executive and administrative staff members remains in place; acquired Choice One Foods, Los Angeles, CA. Choice One Foods is now operating as a subsidiary of ConAgra Inc. December 1999 Co. integrated its U.S. beef companies— Armour Fresh Meats, E.A. Miller, Monfort (Greeley CO), Northern States Beef (Omaha NE), and Signature Ground Beef—into one organization under the name ConAgra Beef Company January 2000 Acquired Seaboard Farms, the poultry division of Seaboard Corporation, for approximately $360,000,000 March 2000 Acquired Shawnee Mission’s poultry division; purchased the Emerge business of Litton Industries, Inc. Emerge provides agricultural and land-use information services and is located in Billerica, MA August 2000 Acquired International Home Foods for approximately 41,000,000 shares of common stock and $875,000,000 in cash and the assumption of approximately $1,100,000,000 in international Home Foods’ debt November 2000 Acquired Marburger Foods, a producer of pre-cooked bacon products January 2001 Acquired Artel, Inc., a manufacturer of frozen prepared foods Source; Company data report from FIS Online (http: •"www.fisonline.com). Obtained on 10/9/01.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 79
Even though ConAgra did not undergo an antitrust indictment, it is
interesting to explore whether they were affected by ADM’s antitrust case. Figure 11
plots ADM employment against ConAgra employment. Both firms exhibit positive
employment growth throughout the period, except for some minor corrections every so
often. During the nine years of ADM’s first antitrust case, the average rate of
employment growth at ConAgra was 23.5 percent. This contrasts markedly from the 15.5
percent over the entire period. ADM’s average rate of growth during the time it was
being indicted for antitrust is 5.3 percent, compared to 8.3 percent for the twenty-year
time frame. It is clear that employment surged for ConAgra while it slowed down for
ADM during the antitrust indictment period. ConAgra employment did fall by
approximately 7.8 thousand workers in 1996. This decline occurred during ADM’s
second antitrust case, and could be construed as a deterrent effect from the indictment.
That is, ConAgra is deterred from becoming an even larger firm with higher levels of
output due to the greater number of employees, because of the greater scrutiny and
market share that would likely ensue. On the coattails of steep employment growth,
ConAgra may have thought it would be sensible to modify its business practices to
prevent itself from being indicted. From 1996 to 1998, ConAgra employment stabilized,
and fell slightly again in 1999.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 80
Figure 11. Comparison of employment levels between Archer-Daniels-Midland and ConAgra for the years 1979 - 1999
9999999999999999^
1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999
■ ADM Employment -ConAgra Employment
Comparing ConAgra employment to the Food & Kindred Products
industry and the manufacturing sector, ConAgra employment levels have risen
consistently over the years. Industry and manufacturing employment levels, however, do
not demonstrate a similar trend. The former has remained relatively stagnant aside from a
19.2 percent dip in 1989, whereas the latter has declined steadily over the period. The
shares of ConAgra employment to industry and sector employment exhibited strong
growth throughout the 1980s, which also happened to be the time ADM was undergoing
their first indictment. The growth in employment shares tapered off slightly in the late
1990s following a decline that took place in 1996, which happens to correspond with
ADM’s second indictment (Figure 12).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 81
Figure 12. ConAgra employment as a share of manufacturing sector and industry employment for the years 1979 - 1999
6.0 % 0 . 5%
* 0.5% 5.0%
4.0% 0.3%
0% - 0.3%
2.0 % - 0.2
0% - 0 . 1%
0 % ~ 0 .0 % 19951979 1981 19831985 1987 1989 1991 1993 19951979 1997 1999 l^H S hare of Food & Kindred Products Industry Employment (SIC code 20) —•— Share ofM anufacturing SeclorEmployment
ConAgra sales display a similar pattern to employment levels. Throughout the period,
ConAgra sales continue to climb. Generally speaking, the same is true for sales at the
industry and sector level. Thus, ConAgra seems to be following the same overall trend
established by the industry and manufacturing sector. Between the beginning and ending
dates of ADM’s antitrust case, however, the average growth rate of ConAgra sales is 25.9
percent. Between 1979 and 1999, the rate is 17.3 percent. ADM, on the other hand, had
an average growth rate in sales of 6.2 percent between 1982 and 1991, and 6.1 percent
over the twenty-year period. The shares of ConAgra sales to the industry and sector also
increase over time (Figure 13). As seen with employment shares (Figure 12), much
growth takes place throughout ADM’s antitrust litigation period. However, beginning in
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 82
Figure 13. ConAgra sales as a share o f manufacturing sector and industry sales for the years 1979 - 1999
6.0% - 0.8%
- 0.7% 5.0% -
- 0.6%
4.0% - - 0.5%
3.0% | 0.4% I
~ 0.3% 2.0% - I 0.2%
1.0 % - 0 . 1%
1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999
■ ■ S h are of Food & Kindred Products Industry Sales (SIC code 20) —•— Share of Manufacturing Sector Sales
the early 1990s growth tapers off and this trend persists for the duration of the period.
Sales per employee generally trends upward throughout the period, aside from a sharp
decline in 1984. As seen from Figure 14, in terms of overall firm productivity, it does
appear that ConAgra was able to benefit from ADM’s antitrust indictment.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 83
Figure 14. ConAgra sales per employee for the years 1979 - 1999
Case Study 3: Merck and Schering-Plough
Merck & Co, Incorporated
Merck’s origins date back to a small apothecary in Darmstadt, Germany.
Frederick Jacob Merck acquired this apothecary in 1668, and this was where his vision
began. Merck & Co, Inc. was introduced to the U.S. in 1887 when it established its first
branch office in New York City that sold pharmaceuticals manufactured in Germany.
Since that time, Merck & Co. has grown to become one of the world’s largest full service
pharmaceutical companies. In addition to researching, developing, manufacturing, and
marketing human and animal health products, the firm also provides pharmaceutical
benefit services to more than 65 million Americans. Merck’s corporate headquarters are
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 84
located in Whitehouse Station, New Jersey, and its primary business segments are:
Human and Animal Health; and Environmental Health. In 2001, Merck’s sales exceeded
S47.7 billion.35
The Department of Justice filed a civil antitrust suit against Merck on
August 17, 1979 in a U.S. District Court in San Diego, California. The suit challenged the
proposed acquisition of United Kingdom firm, Alginate Industries, Ltd by Merck.
Alginate Industries is the leading producer of alginate, a seaweed extract, in the world.
The Kelco Division of Merck dominates the U.S. market in terms of alginate sales, and is
the world’s second leading producer of alginate. Merck and Alginate Industries sales of
alginate in 1978 were approximately S21 million and S2 million, respectively (Merck’s
alginate sales are approximately 0.05 percent of its total sales). Merck’s sales account for
80 percent of all U.S. alginate sales, and Alginate Industries sales constitute 8 percent of
U.S. sales.
According to the Trade Regulation Reporter, the Assistant Attorney
General in charge of the Antitrust Division, John H. Shenefield, asserted that the suit was
filed because the acquisition would allegedly suppress and eliminate competition in
several markets that use alginate as its principal chemical. These markets include
emulsifiers in buttered syrups, a foaming agent in a particular drug for the treatment of
reflux esophagitis, alginate thickeners used in textile printing and dyeing, a gelling agent
in certain dental impression compounds, and beer foam stabilizers. The government
requested the court to proclaim the acquisition illegal and declare a permanent injunction
55 A summary of Merck’s business activities was not available from FIS Online.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 85
against Merck. The injunction would preclude Merck from buying or retaining any
interest in the assets or stock of Alginate Industries. One year later, both sides reached a
settlement. On December 1, 1980, a consent decree was entered. According to Chemical
Week, the Court permitted Merck to retain ownership of Alginate Industries.36 However,
Merck was ordered to divest itself of Scotia Marine Products Limited, a wholly-owned
Canadian subsidiary that produces Algin in Nova Scotia. Furthermore, Merck is required
to provide information and assistance in creating a new company that could compete
effectively in the sale and production of alginate.37
Between 1979 and 1999, Merck employment has more than doubled,
starting at 30,800 in 1979 and growing to 62,300 in 1999. Merck’s antitrust litigation
occurred early and did not appear to have an effect on its employment levels since
employment continued to rise throughout this time. Employment in the Chemicals &
Allied Products industry also rose in 1979, but fell sharply for the next three years. This
sharp decline is likely to be caused by the two recessions between 1980 and 1982. A
slight recovery took place in 1984, but employment once again took a big hit for the next
two years to reach a low of slightly over one million workers. Merck employment also
fell by approximately 4,000 workers in 1985, but then stabilized and recovered slightly
beginning in the following year. Merck and industry employment do not follow similar
patterns, and in fact move in completely opposite directions after 1995. Merck
employment levels continue to rise, while industry employment falls. Employment levels
in the manufacturing sector exhibit a reverse trend to that displayed by Merck. Whereas
36 September 10, 1980.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 86
sector employment trends downward, Merck employment levels trend upward. Figure 15
shows Merck employment shares to industry and sector employment. During the antitrust
litigation period 1979-1980, employment shares do not display much growth. In fact,
throughout the twenty-year period employment shares do not show much growth until the
mid- to late-1990s. Beyond a brief stagnant period, which generally lasts until 1984,
Merck is able to rebound and recover through sharp growth in employment.
Merck’s sales and sales in the Chemicals & Allied Products industry
reveal a similar pattern. Industry sales experienced more instability than firm sales. In
fact, industry sales dropped 10.9 percent between 1981 and 1982— an outcome likely to
be caused from the 16-month recession. The year in which Merck was indicted for
antitrust, the Human and Animal Health business segment reported sales of 52,004.1
million, and the Environmental Health business segment reported sales of S380.5 million.
Alginate is likely to be included in the former business segment, which constituted
roughly 84.0 percent of total Merck sales.38 Sales in the Chemicals & Allied Products
industry showed an overall upward rise. Merck’s sales consistently climbed beginning in
1985 with an average growth rate of 14.6 percent. Up to that point, however, sales growth
was flat with an average rate of growth of barely 1.0 percent. It is worth noting that
Merck’s sales growth was approximately 5.0 percent from 1979-1980, which happens to
be the antitrust litigation period. Thus, Merck’s sales were not initially affected.
However, there may have been subsequent lagged impacts from the antitrust indictment
on sales growth.
37 According to the PR Newswire, September 2, 1980, Tuesday.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 87
Figure 15. Merck employment as a share o f manufacturing sector and industry employment for the years 1979 - 1999
0.35%
6.00% - 0.30%
5.00% - - 0.25%
4.00% - 0.20%
3.00% - 0.15%
2.00% - - 0 . 10%
1.0 0 % 0.05%
0 .00% 0. 00° / 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 ■ ■ S h a re of Chemicals & Allied Products Industry Employment (SIC code 28) —♦— Share of Manufacturing Sector Employment
Manufacturing sector sales are similar to industry sales in that there were
many more disturbances to sales growth that were impervious to firm sales. Firm sales
grew smoothly whereas industry and sector sales were more volatile. Figure 16 illustrates
Merck’s sales as a percentage o f both industry and sector sales. Merck’s share of industry
and sector sales remained relatively stable up until 1989. Beyond that time, their shares
increased sharply. Despite being indicted for antitrust, Merck’s share of manufacturing
sales grew approximately 17.9 percent from 1979 to 1980. During the same time, their
share of industry sales grew approximately 4.2 percent. For the next five years, the
38 Business segment data obtained from Merck’s 1990 Annual Report.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. average growth rates of manufacturing and industry sales shares are roughly 0.9 and 1.4
percent, respectively.
Figure 16. Merck sales as a share of manufacturing sector and industry sales for the years 1 9 7 9 - 1999
0.09 0.009
0.08 0.008
0.07 0.007
0.06 0.006
5 0.05 0.005
eu 0.04 0.004
0.03 0.003
0.02 0.002
0.01 0.001
1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 Share of Chemicals & A Ilied Products Industry Sales (SIC code 28) Share of Manufacturing Sector Sales
Sales per employee rises relatively consistently throughout the twenty-
year period except for a downward shift that occurs in 1993. As seen from Figure 17,
sales per employee remains stable up until 1984. After this time, sales per employee at
Merck exhibits an upward trend. Also observed from the previous graphs, there does not
appear to be an initial impact from an antitrust indictment. However for the next few
years, there is little growth in any of the economic indicators discussed. Beyond this
stagnant period, which generally lasts until 1984, Merck is able to rebound and recover
through sharp growth in both employment and sales.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 89
Figure 17. Merck sales per employee for the years 1979 - 1999
2347
480
420 -
360
300 -
240
180
x X--- 120 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999
Schering-Plough Corporation
The creation of Schering-Plough dates back to the late 1800s when
Schering AG, a German-based pharmaceutical and chemical company, created Schering
Corporation. This newly created corporation sold pharmaceutical products to U.S.
consumers, and became incorporated in 1928 in New York City, and in 1935 in New
Jersey. Throughout the 1940s, Schering Corporation established its reputation as more of
an American enterprise and less of a European-based company with operations in the
U.S. With continued success in the 1950s and 1960s, Schering Corporation became
incorporated on July 28, 1970 in New Jersey as Schering-Plough Corporation to
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 90
accelerate an anticipated merger with Plough, Inc., a worldwide manufacturer of
consumer products. The merger was finalized on January 16, 1971.
In 2001, Schering-Plough’s sales were nearly S10 billion. Although not as
large as Merck, the two companies still compete on many levels within the
pharmaceutical industry. Table 7 below summarizes Schering-Plough’s business
activities.
Table 7. Summary of Schering-Plough business activity Date Business Activity (a) (b) ______April 1970 Plough. Inc. acquired Master Bronze Powder Co., Inc., and Inland Coatings and Aerosol Corp. through an exchange of common shares September 1970 Shareholders of Schering Corporation and Plough Inc. approved a plan and agreement of merger whereby SCH Corp.. a wholly-owned sibsidiary of Co. was to merge into Schering and PLO, Inc., also a wholly-owned subsidiary of Co. was to merge into Plough, Inc., with the result that the two corporations became subsidiaries of Co. Each outstanding common share of Co. was converted into one common share of Co. and each common share of Plough, Inc. was converted into 1.3 Co. common shares. Each outstanding share of S2.20 convertible preferred stock of Plough, Inc. was converted into one S2.20 cumulative convertible preferred share series “A" of Co. January 1971 Merger between Schering Corporation and Plough Inc was consummated 1972 Organized P.T. Essex Indonesia, acquired AESCA chemisch-pharmazeutische G.m.b.H. (Austria), acquired majority control of Plough (Australia) Pty. Ltd., and liquidated Essex Pharma Pharmazeutika Ges m.b.H. (Austria), Webb Products Co. (Canada) Limited and Dicks-Armstrong-Pontius, Inc. (Tennessee) 1974 Acquired cosmetics business of Carl Hoeppner K.G., Dusseldorf, W. Germany, for cash; organized Essex Hellas (Greece), Essex Pharma (Greece) Productos Farmaceuticos Ecuatorianos (Ecuador), S.A. (Professa) Ecuador, Schereth Corporation (N.J.), Schering Antibiotic Corp. (N.J.), White Laboratories (Ireland) Limited (Bermuda); and liquidated Bayton Realty Corp. (N.J.), National Bio Serums Inc. (N.J.), Battiers Pharmacy (Tennessee), Schering Corp. Pan America S.A. (Panama), Schering Corp. (Panama) S.A. 1975 Organized Warrick Pharmaceuticals Limited (U.K.), Schering Brasil Industria, Comercio e Participacoes Ltda.; liquidated Sani-Swab Inc. (Ohio), Plough Philippines, Plopisa S.A.I.C. (Argentina), Plough Kardons, S.A. (Pern), Industria Quimica Plough (Chile) Ltda. 1976 Organized Schering Products Inc. (N.J.), Schering Laboratories (P.R.) N.J.; and liquidated Plough Pencil Company (Tennessee), Laboratories Plough Limitada (Brazil), N.V. Plough Belgium S.A. 1977 Organized Essex (Nigeria) Services Co. Ltd., Essex Laboratories (N.Z.) Limited (New Zealand), Schering Biochem Corp. (Del.), Scherpharm Corp. (Del.), Scheroid Inc. (Del.), Schering Industries Inc. (Del.); and liquidated White Laboratories (Ireland) Limited (Bermuda) and United Corporation (Tennessee)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 91
Table 7. Cont. Date Business Activity (a) (b) May 1979 Acquired through a subsidiary, Bums-Biotec April 1979 Merged with Scholl, Inc. Linder merger agreement, Scholl shareholders received one share of Co. S5.07 cumulative preferred stock. Series B for each Scholl common shares or, if they so elected, 0.968 of a share of common stock for each Scholl share June 1979 Completed purchase of Unifa, S.A., an Argentine pharmaceutical company, for cash 1979 Purchased Rimmel International Ltd. of London from International Telephone & Telegraph Corp. for cash July 1979 Purchased Kirby Pharmaceuticals Ltd. of Mildenhall. England, a manufacturer of injectable and ophthalmic drugs and proprietary medicines for cash September 1980 Acquired Wesley-Jessen Inc. for 526,000,000 cash 1981 Acquired Lofhis Bryan Co. and Irish Fine Co., Ireland February 1981 Schering Corp., a unit of Co., acquired Douglas Industries Inc., Lenexa Kansas tor 510,000,000 cash December 1981 Sold Adhesive Tape Co., subsidiary to Chemical Investors Inc. for 5 7,700.000 in cash and notes 1982 Acquired DNAX Limited for 529,800,000 consisting of 519,600.000 in cash and notes and 340,054 common shares March 198? Sold DAP, Inc. 1984 Sold Plough Broadcasting Co., Inc. July 1986 Merged Key Pharmaceuticals Inc., as a result of the merger, each share of Key common stock was exchanged for 0.265625 share of Co. common stock, with each whole share of Co. common stock issued with one attached common share purchase right August 1987 Completed the sale of its Dr. Scholl’s businesses in Europe and related assets in Latin America and the Far East to European Home Products PLC, the multinational British firm, for 5160,000,000. The sale included payment by Co. ofS7,000,000 to European Home Products for the performance of certain services following the August 14 closing October 1987 Sold the non-U.S. and non-Canadian operations of its subsidiary, Scholl Inc., to European Home Products PLC, the multinational British firm for 5160,000,000 November 1988 Acquired all rights in the U.S. and Japan to the Aquaflex line of soft contact lenses of The Cooper Companies, for approximately 535,000,000 in cash September 1989 Completed the sale of a majority interest in its Brazilian operations to Grupo Ache’ 1990 Sold its Maybelline cosmetics business October 1992 Co. and Sandoz Pharma Ltd. of Basel, Switzerland, announced a joint venture in the U.K. to market "Ieucomax”, a granulocyte macrophage colony stimulating factor. The product was jointly developed by Sandoz and Co. and will be co marketed in the EC by the two companies June 1995 Completed the sale of its Wesley-Jessen contact lens business, based in Chicago, III. To WJ Acquisition Corp. for 547,500,000 in cash February 1996 Acquired Canji, Inc. for 554,500,000 June 1997 Acquired the worldwide animal health business of Mallinckrodt Inc. for approximately 5490,000,000, which includes the assumption of debt and direct costs of the acquisition
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 92
Table 7. Cont. Date Business Activity (a) (b) June 2000 Acquired a majority interest in a joint venture with Takeda Chemical Industries, Ltd. to develop and market animal health products in Japan. The cost of the acquisition was approximately $48,000,000 December 2000 Co.’s subsidiary, Schering-Plough HealthCare Products sold its St. Josephr business to McNeil Consumer Healthcare Source: Company data report from FIS Online (http: www.fisonIine.com). Obtained on 10/9/01.
Schering-Plough shares the same industry as Merck, and acts as one of
their main competitors. Although similar in many ways, the companies differ in that
Schering-Plough was not indicted for violating antitrust laws. Figure 18 contrasts the two
companies’ employment levels. During the antitrust litigation period, both Schering-
Plough and Merck underwent employment growth. Merck increased its employment
levels by approximately 2.6 percent from 1979 to 1980 while Schering-Plough expanded
their employment levels by more than twice Merck’s growth rate—roughly 5.8 percent—
over the same period. For the rest of the decade, their employment patterns diverged.
Schering-Plough shed thousands of workers while Merck made modest employment
gains. There appears to be an initial effect on employment from the antitrust indictment.
Schering-Plough may have expanded its employment levels to gain market share or take
advantage of Merck’s weaker position. There may also be a lagged deterrent effect from
the indictment. Similar to ConAgra, Schering-Plough may have tempered its business
practices that led to the employment decline to avoid being indicted. For most of the
1990s, Schering-Plough’s employment rebounded, and the two pharmaceutical
companies followed generally equivalent employment paths.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 93
Figure 18. Comparison of employment levels between Merck and Schering-Plough for the years 1979 - 1999
- 27
- 25
o 45
1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 Merck Employment — S-P Employment
Comparing Schering-Plough’s employment levels to the Chemicals and
Allied Products industry and the manufacturing sector reveal that Schering-Plough’s
employment pattern was more a result of following industry and sector trends than of
Merck’s antitrust indictment. The firm followed the industry employment pattern up until
the late 1980s. Beyond that point, the patterns differ and move in entirely different
directions by the end of the period. Schering-Plough followed the sector’s employment
pattern more closely and for a longer period of time—up until the late 1990s—until again
the employment patterns diverge and move in reverse directions by 1999. Schering-
Plough’s shares of industry and sector employment remained relatively stable over the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 94
twenty-year period. Figure 19 shows that shares increased slightly during Merck’s
litigation period, but then declined rather steadily up until 1990. Over the entire period,
the average growth rates of Schering-Plough’s shares of industry and sector employment
are 0.5 and 1.1 percent, respectively. Schering-Plough exhibited strong employment
growth in the late 1990s. This growth is also reflected in Figure 19 as Schering-Plough’s
shares of industry and sector employment rise to levels higher than that obtained over the
entire period.
Figure 19. Schering-Plough employment as a share of manufacturing sector and industry employment for the years 1979 - 1999
3.0% ------" , ' - 0 /
2.5%
- 0. 12%
2.0 % 0. 10%
1.5% r 0.08%
* 0.06% 1.0%
0.04°/< 0.5% r 0.02%
0 .0% 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 Share of Chemicals & A Hied Products Industry Employment (SIC code 28) —♦— Share of Manufacturing Sector Employment
Schering-Plough’s sales tripled from approximately S2.6 billion to nearly
$9.0 billion over the period 1979 to 1999. During the time Merck underwent antitrust
litigation, Schering-Plough’s sales increased roughly 11.1 percent. Both industry and
sector sales exhibit more volatility than firm sales, but also increase steadily over time.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 95
Whereas Schering-Plough’s shares of industry and sector employment remained
relatively stagnant, the same is not true of their shares of sales. Figure 20 demonstrates
rising firm shares. In fact, the average growth rates of Schering-Plough’s shares of
industry and sector sales over the twenty-year period are 8.3 and 5.5 percent,
respectively. The remaining graph sheds light on the firm’s productivity by examining its
sales per employee. Up until 1984, growth is stagnant. As seen from Figure 21, sales per
employee does show a modest increase during Merck’s antitrust litigation period only to
decrease again the next year and stabilize. From 1985 forward, however, sales per
employee rises sharply and consistently.
Figure 20. Schering-Plough sales as a share of manufacturing sector and industry salesfor the years 1979 - 1999
15%
r 0 .20%
10 % -
r 0.15% 1.5%
0 .10% 1.0%
0.5% - r 0.05%
0.0% r 0 .00% 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 Share o f Chemicals & Allied Products Industry Sales (SIC code 28) — Share of Manufacturing Sector Sales
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 96
Figure 21. Schering-Plough sales per employee for the years 1979 - 1999
340
300 .
260 -
fl S i 3
180 •
140 •
100 1979 1981 1983 1985 1987 1989 1993
From the figures, it is evident that each measure—employment, sales, and
sales per employee—increased during Merck’s antitrust litigation period. Whether this
was due primarily to their competitor’s indictment or simply following industry and
sector trends is unclear. However, it does appear that Schering-Plough benefited and
improved its overall performance during this time.
Summary of Case Studies
In summary, six case studies were conducted in an effort to assess
individual firm responses to antitrust indictments. The first case study compared The
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 97
Stanley Works with Black & Decker. Both companies are involved in the tools and
hardware line of business, and are among each other’s top competitors. The Stanley
Works was indicted for allegedly fixing prices on architectural hinges sold in the U.S.
Litigation continued for three years until 1993 when The Stanley Works finally pleaded
no contest. Initially fined $8 million, their sentence was later revised to S5 million and
community service.
During the early stages of the indictment, employment fell. However, this
period of time also coincides with the recession of 1990-1991. Thus, it is unclear whether
the decline in employment was due to industry and sector downward trends due to the
recession or if it can also be attributed to the indictment. The Stanley Works’
employment patterns tend to generally follow industry and sector trends.
While The Stanley Works experienced employment losses, Black &
Decker’s employment levels increased in the early stages of their competitor’s antitrust
indictment. In fact, employment peaked in 1990— precisely the year the indictment was
issued. Beyond this time, however, employment suffered severe declines with an average
rate of growth of negative 6.3 percent. Sales at Black & Decker also peaked in 1990 only
to fall and remain stable for the remainder of the indictment period. Sales per employee
exhibited the same pattern as employment and sales in that productivity initially rose. For
the rest of the indictment period, though, this measure was relatively flat. Black &
Decker did not appear to experience deterrent effects from antitrust enforcement against
The Stanley Works. While the data analysis may suggest the firm took proactive action
when the indictment initially began, Black & Decker tempered their business practices
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 98
for the rest of the time. Unlike The Stanley Works, Black & Decker was not as sensitive
to industry and sector economic trends.
The second case study examined Archer-Daniels-Midland and ConAgra.
These two firms’ primary operations are in the Food & Kindred Products industry. ADM
was indicted at two different times between 1979 and 1999. The first indictment lasted
nine years, and the charges were for monopolization, restraint of trade, and acquisition.
The second lasted for one year, and accused ADM of fixing prices for lysine and citric
acid in international markets. The court’s decision regarding the first indictment was cast
in favor of ADM, whereas ADM pleaded guilty to the second within a year. ADM was
fined SI00 million.
ADM employment went counter to industry and sector trends. While
industry employment was stable and sector was declining, ADM employment rose
relatively consistently throughout the period. The indictments did not appear to have any
effect on employment. In fact, employment grew approximately 15.9 percent the year
following the second indictment where ADM was found guilty of antitrust violation.
ADM sales generally tended to follow industry and sector trends over the twenty-year
period. However, sales were growing at an average rate of roughly 6.1 percent compared
to 0.4 and 0.9 percent in the industry and sector, respectively. Sales per employee was
more revealing of a potential impact on firm performance from the indictment. Following
the $100 million fine, sales per employee fell by approximately 11.9 percent.
ConAgra and ADM’s employment trends are nearly the same. They differ
only in that beginning with the year the second indictment was issued, ConAgra’s
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. employment reverses its upward trend to a moderate descent. This action may have been
in response to ADM’s antitrust indictment. The same pattern developed for ConAgra’s
sales in that there was a moderate decline for the years following ADM’s second antitrust
indictment. ConAgra loosely followed industry and sector sales trends. Sales per
employee, on the other hand, exhibited an overall increase during both litigation periods.
Of all the cases, ADM and ConAgra seem to be the two companies most
sheltered from any adverse effects from antitrust enforcement. Both companies
experienced employment gains. The measure of firm productivity, sales per employee,
was more telling in that it did not exhibit such positive gains. In fact, the second
indictment led to a sharp decline in this indicator for ADM. The deterrent effect from
antitrust enforcement was also especially subtle in this case. The data analysis indicates
that the competing firm may be taking advantage of the precarious situation by trying to
surge ahead or gain more market share.
The final case study compares two pharmaceutical companies, Merck and
Schering-Plough. Antitrust action against Merck was taken in 1979 that challenged a
proposed acquisition of a U.K. firm. The two parties reached a settlement after two years.
In the end, Merck was able to retain ownership of the acquired firm in return for
assistance in establishing another company that would sell and produce alginate.
Merck employment levels consistently increased nearly throughout the
entire twenty-year period, which did not happen to be the pattern at either the industry or
sector levels. During the two years of the indictment, employment grew at a rate of 2.6
percent. Employment continued to rise until two years following the indictment when
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 100
employment levels fell by 1.2 percent. Merck sales followed the overall upward trend of
the industry and sector. Similar to employment, there is a modest decline in sales in the
years following the close of the antitrust case. Upon examining sales per employee, this
precise pattern emerges once again. Whether all these declines are solely the result of the
antitrust case is uncertain and unlikely. It is probable, however, that the antitrust
indictment had some impact on Merck’s economic performance in the years immediately
following the indictment.
Schering-Plough’s employment rose 5.8 percent during the time Merck
underwent antitrust litigation. Through the rest of the decade, however, employment
levels at Schering-Plough generally tended to follow the downward trends of the industry
and sector. Again, there is positive growth during and negative growth after the antitrust
case. By the 1990s, both Schering-Plough and Merck exhibited similar employment
patterns that included consistently positive growth from 1995 to the end of the period. As
did Merck, Schering-Plough generally followed the industry and sector sales pattern.
Moreover, the pattern of declining sales following antitrust litigation repeats itself.
Schering-Plough’s sales per employee also shows a modest decrease after the close of the
antitrust case. The declining trend eventually stabilizes and reverses itself by the mid-
1980s.
The case studies provided valuable information, and proved to be a
worthwhile undertaking. However it is questionable exactly how much of the fluctuations
in employment and sales can truly be attributed to the antitrust indictment. First, in most
of the cases, the sales from the product for which the firm underwent an indictment were
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 101
only a negligible percentage (less than 1 percent) of total sales. This being so, it is
unlikely that antitrust enforcement would have much of an impact on firm performance
since the product line contributes only a small amount to overall sales. Second, although
there were some lagged adverse effects on employment and sales for the indicted firms, it
is not clear what may have caused the downturn. Because the percentage of total sales is
less than 1 percent, it may be difficult to argue that the indictment played much of a role
in the employment and sales shifts. Finally, there are so many activities taking place
within the firm, not disclosed to the public, that it is nearly impossible to attribute firm
performance to a single primary cause.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 4
DATA AND ECONOMETRIC ISSUES
Trade Regulation Reporter39
The Trade Regulation Reporter provides data for the number and type of
antitrust indictments filed within each industry for the years 1979 to June 2001.40 This
document provides a concise synopsis of the indictment filed by the DOJ. It provides
pertinent information such as who was indicted, for what offense, for what product, and
the final outcome. Essentially, the Trade Regulation Reporter provides the facts and the
key players in the cases. Table 8 provides descriptive statistics on the antitrust cases in
the Manufacturing sector by major industry group. The industry is not readily identified
by SIC code, however, the company’s product line for which there was an indictment is
reported. The particular product line is then searched for in a detailed list of
manufacturing sector major industry groups at the 4-digit level to determine the
39 Also known as Commerce Clearing House Bluebooks.
40 Case data was collected through 2001, however the analysis was conducted for the period 1979 - 1999. The 2001 data is incomplete since the data was collected in and up to June 2001.
102
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 103
appropriate industry.41 Consequently, using the product lines the relevant cases are sorted
into the appropriate 2-digit SIC code.
Table 8. Descriptive statistics for antitrust lawsuit outcomes for the years 1979 through June 2001 (N = 532) Sum of: Victories Average Standard Govern Settle Dismis Un Duration Industrial Classification ment Defense ments sals resolved Total in Years (a) (b ) (c) (d) (e) (0 (g) (h) Food & Kindred Products (20) 86 12 21 2 13 134 0.955 Tobacco Products (21) I 0 2 0 0 3 1.000 Textile Mill Products (22) 2 0 I 0 0 3 1.000 Apparel and Other Textile Products (23) 13 0 2 0 1 16 0.687 Lumber and Wood 9 0 2 0 0 11 0.727 Products (24) Furniture and Fixtures (25) 2 0 1 0 0 ■> 1.333 Paper and Allied Products (26) 6 1 I 0 4 12 2.000 Printing and Publishing (27) 2 0 4 0 1 7 0.857 Chemicals and Allied Products (28) 27 0 13 0 2 42 0.238 Petroleum and Coal Products (29) 14 4 4 1 1 24 1.500 Rubber and Miscellaneous Plastics Products (30) 7 2 5 0 1 15 1.467 Leather and Leather Products (31) 0 0 0 0 0 0 0.000 Stone, Clay, Glass, and Concrete Products (32) 20 1 8 I 4 34 1.588 Primary Metal Industries (33) 22 8 1 2 2 35 1.943 Fabricated Metal Products (34) 22 2 5 2 1 32 0.875 Industrial Machinery and Equipment (35) 30 4 14 I 4 53 1.019 Electrical and Electronic Equipment (36) 19 4 9 2 3 37 1.351 Transportation Equipment (37) 25 0 2 I 6 34 2.000 Instruments and Related Products (38) 8 2 10 0 1 21 1.381 Miscellaneous Manufacturing Industries (39) 8 2 3 1 2 16 0.936 Grand Total 323 42 108 13 46 532 ■ , Note: Author’s analysis from data in the Trade Regulation Reporter.
41 The detailed list of manufacturing industries at the 4-digit level can be found on the U.S. Census website (www.census.uov).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 104
Three different aspects of each case were recorded. The first identifies the
type of legal suit, that is, civil or criminal. The second notes the kind of antitrust
violation.42 Finally, the outcome of the case was documented. There are four possible
outcomes: government victories, defense victories, settlements, or dismissals. In
instances where the case had not reached resolution by June 2001 (the time the data was
collected), the case was top-coded to end in 2001.43
In summary, the antitrust variable in the dataset is created from the Trade
Regulation Reporter. It is a dummy variable set equal to 1 when there was an indictment
initiated in a particular year in a specific industry. For instance, if the only antitrust case
initiated in 1990 occurred in the Food & Kindred Products industry, the value of the
antitrust variable would equal 1 for that industry and 0 for every other industry in 1990.
The antitrust variable would also have a value of 1 for every worker category in the Food
& Kindred Products industry since the indictment does not vary by worker. That is, each
of the 7 worker categories in the Food & Kindred Products industry in 1990 would have a
value of the antitrust variable equal to 1. Additionally, since the antitrust indictment
variable is a dummy, the value is 1 regardless of the number o f antitrust cases initiated in
that industry for that year.44
42 The guidelines specified previously dictate the types of violations included in the analysis.
43 This was done so that in case average duration was to be included in the model, the data point would not be considered missing.
44 Analyses on the number of antitrust indictments, and not just the presence of an indictment were also conducted. The results of these analyses, however, were not as fruitful as the current analyses.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 105
Annual Survey of Manufactures (ASM) and Economic Census
The ASM and Economic Census for the Manufacturing sector, also called
the Census of Manufactures, provide data on number of establishments, number of
employees, labor expenses, number of production workers, and capital expenditures. The
Economic Census is conducted every 5 years on years ending in 2 or 7, and the ASM is
conducted in between Census years. Capital expenditures are the sum of new and used.
There are two components of new and used capital expenditures. The first is Buildings
and Other Structures and Machinery. The second is Equipment. These variables represent
fixed costs to the firm. Although the Census Bureau switched from SIC to NAICS coding
in 1997, it did create a bridge for the 1992 and 1997 Census of Manufactures that
reported data by both SIC and NAICS codes. This document, however, was created only
for 1997 data and no subsequent reporting by both SIC and NAICS was undertaken by
the Census. The Bridge report contains data for number of establishments, number of
employees and labor expense for this single year. However, it does not include
information on capital expenditures, number of employees, labor expense, nor number of
production employees.45
The last SIC codes revision occurred in 1987 when approximately 20 new
service industries were added to the classification system and a few new industries were
45 Bridge report data was obtained on-line at http://www.census.gov/epcd/ec97brdg/INDXSIC2.htm. Obtained on 6/19/01. The Bridge report did contain information on Paid Employees and Annual Payroll. These data series, however, are for All Establishments and not exclusively Operating Manufacturing Establishments. Additionally, some values were suppressed or given in ranges to prevent breaches in confidentiality.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 106
added to manufacturing to reflect technological changes occurring in that sector. This
revision does not pose substantial problems to this analysis since the 2-digit major
industry groups remain largely intact. A problem does arise, however, for data after 1997
because of the crossover from SIC to NAICS. In 1992, the Office of Management and
Budget (OMB) formed the U.S. Economic Classification Policy Committee (ECPC). This
committee was chaired by the Bureau of Economic Analysis (BEA), and staffed by BEA,
the Bureau of Labor Statistics, and the Census Bureau. The Committee admitted that the
break in the data from pre-1998 to 1998-forward will “cause major disruptions in the
availability of ... time series information, not only for individual industries that are
redefined but also for the broad sectors, like manufacturing and retailing.”40 In fact, the
Census Bureau admitted that it would be nearly impossible to make the data consistent
across the years.47
Since data were not available for the years following 1996, data values
were generated using extrapolation techniques for the latter years 1997 to 1999.
Additionally, because the Census Bureau does report these variables by NAICS code, the
SIC share to total was calculated for 1996, and then applied to the total values for 1997
through 1999. This procedure was conducted for number of employees, labor expense,
number of production workers, and capital expenditures.
44 http: Avww.census.gov. epcdw.'ww naicsnsr.html. Obtained on 6/12/01.
47 This information was relayed by the Chief of the Forms, Publications, & Customer Services Branch in the Manufacturing and Construction Division.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 107
For example, to obtain employee counts for 1997 through 1999, the sum
of employees for the manufacturing sector in 1996 was calculated. Then, the shares of
employees for each major industry group to the total manufacturing sector for 1996 were
obtained. As mentioned previously. Census changes from SIC code to NAICS code in
1997. Total manufacturing employment for 1997 through 1999 are still reported, but now
in NAICS code which does not directly correspond to SIC code. In order to maintain the
time series consistent in 2-digit SIC code, employment counts for the latter years were
derived. This was done by applying the calculated industry shares for 1996 to Census’
reported total manufacturing employment (in NAICS code) for the years 1997 through
1999.
The Census Bureau provides establishment counts only for the Economic
Census years when the entire manufacturing community is canvassed. In the years in-
between the Economic Census, a sample survey is conducted and establishment counts
would be meaningless since only the number of establishments selected to participate in
the survey are represented. 48 Thus, a linear interpolation method was used to derive
values for non-Census years. Additionally, while establishment counts were available by
SIC code in 1997, they were only reported in NAICS thereafter. In order to complete the
time series, a log-linear regression model was specified. The model estimated the log of
the number of establishments (in thousands) as a function of real GDP and time trend.
The 5-year average growth rate of firms was based upon the interpolated values and the
4i! This survey is the Annual Survey of Manufactures.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 108
number of establishments forecast from 1998 to 2003 obtained from the regression
equation below.4'1
LN (Number of Establishments) =a + Pi*Real GDP P:*year (13)
Prior to the NAICS change, employees working in auxiliary
establishments within the Manufacturing sector were included in the manufacturing
reports. After the change, auxiliary' establishment employees were placed in a separate
NAICS industry based on the function of the auxiliary establishment. Thus, the number
of employees working in auxiliary establishments within the Manufacturing sector can no
longer be obtained. The last year in which this data is reported is 1997.'° For the
remaining two years, the share of auxiliary establishment employees to operating
manufacturing establishment employees was obtained from 1997 and applied to 1998 and
1999. This derived number was used for All Employees and then the shares by SIC code
in 1996 was applied to the years 1997 to 1999. The auxiliary sector in the NAICS is
defined as "establishments with payroll primarily engaged in providing services to one or
more establishments of the same enterprise. These establishments generally do not
produce any products nor provide services for customers outside the enterprise, but may
do so as a secondary activity. "'1
The Herfmdahl-Hirschmann Index (IT HI) was first published for the
Manufacturing sector in 1982 by the Census Bureau in their Census of Manufactures
w Data series begins in 1977 and ends in 1997.
This value is reported in Table 2 of the I99~ Bridge Between XAICS ami SIC publication produced by the U.S. Census Bureau.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 109
Subject Series report entitled "Concentration Ratios in Manufacturing." The HHI is
reported only during Census years. HHI data in the relevant time period of this
dissertation, then, exist only for the years 1982. 1987 and 1992. While HHI data from the
1997 Census of Manufactures are available, these data are not consistent with the prior
years due to the conversion from SIC to NAICS as described earlier.
Census' calculation of the HHI was conducted on the 4-digit SIC level,
except in 1987 when calculations were made at both the 2-digit and 4-digit levels. In
order to obtain a 2-digit SIC HHI measure, a concentration proxy was created. This proxy
is a weighted average of the 4-digit HHI. weighted by Value of Shipments also reported
at the 4-digit level.'* In order to obtain HHI data for the remaining years, a semi-log
regression model was conducted. The model estimated the natural log of HHI on time as
follows:'^
Ln (HHI) = aP*year (14)
For the years the Census Bureau does report the HHI. it is given in percentage terms. This
index is also calculated only for the 50 largest companies in that particular year. As a
result, the reported HHI is not the true value of the HHI. but an estimate. Nonetheless, the
disparity is unlikely to be large, and it still serves as a good indicator of concentration.
M 1997 Economic Census, p. 5.
Although HHI data exists in 1987 at the 2-digit level, these values were not used. Data were available at the 4-digit level for all years (1982. 1987. and 1992) to create the w eighted average, and in order to keep this analysis consistent these values were employed.
5- Gujarati (1988) suggests that the log-lin model is more useful than other models when one of the regressors is time for meaningful interpretation of the P coefficient. In this particular log-lin model. P measures the relative change in HHI due to an absolute change in time.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 110
The determination of the 50 largest companies is based upon value of shipments for that
year.
Table 9. Descriptive statistics for employment analysis ; Standard Variable N Mean i Deviation Minimum Maximum (a) ( b) (c) (d) (e) Table 10. Descriptive statistics for wage analysis Standard Variable N Mean Deviation Minimum Maximum (a) ( b) (c) (d) (e) (0 All 2.928 16.146 5.530 4.786 40.073 Technicians & Related Support 413 15.981 2.684 6.857 27.353 Professional Specialty 420 21.313 4.090 10.777 37.624 Executive. Administrative. & 420 22.741 3.155 16.391 33.114 Managerial Sales 420 19.020 3.791 9.905 40.073 Administrative Support & Clerical 420 11.747 1.757 7.959 19.297 Service 415 10.236 2.298 4.786 19.874 Production 420 11.913 2.566 6.890 19.722 Real GDP 2.940 6.508 1.201 4.901 8.876 Weighted Average HHI 2.940 667.530 440.556 128.124 2496.017 Capital-to-Labor Expense Ratio 2.940 2.109 1.419 0.460 9.280 Average 5-year Firm Growth Rate 2.940 0.119 1.624 -5.437 4.879 ATR Dummv 2.940 0.452 0.498 0.000 1.000 Source: Author's analysis Note: Real GDP is measured in hundreds of billions. Weighted average HHI and ATR dummy are measured in units. Capital-to-labor expense ratio and average 5-year firm growth rate are reported in percentages. The remaining variables are hourly wage rates. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Current Population Survey (CPS) The CPS is a monthly survey sponsored jointly by the Bureau of the Census and the Bureau of Labor Statistics. The survey uses a scientifically selected sample of approximately 50.000 households, and is administered by the Bureau of the Census. The sample is selected to represent the civilian non-institutionalized population aged 15 years and over. However, data analysis published by Census focus on those ages 16 and over. The survey has been conducted for more than 50 years, and is the primary source of information on the labor force characteristics of the I'.S. population. The sample provides estimates for the entire nation, in addition to individual states, cities, and regional geographic areas. The CPS provides estimates of numerous indicators such as employment, unemployment, earnings, and hours of work by a variety of demographic characteristics. Estimates are also available by occupation, industry, and class of w orker.'4 The CPS is a survey, thus the limitations inherent with surveys such as non-response, untruthful responses, misunderstanding of the question asked, and too little response categories in answering the questions are present. There are a few limitations to the CPS that are particularly noteworthy to this dissertation. First, the occupational title is self-reported. That is. the respondent may classify himself or herself to be in a certain occupational category while another would be more fitting. Another limitation is that the 54 For more information about this data, see www.bls.census.iiov. cps. cpsmain.htm. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. survey periodically undergoes revision. This characteristic poses problems for time series analysis because a question may be rephrased or the response categories changed. This problem was encountered in this dissertation for Industry and Major Occupation variables. The coding for Industry underwent a major change in 1983. The coding for Major Occupation underwent a significant change in 1984.'' Pre-1984 and 1984-forward data are not comparable for major occupations. Thus, pre-1984 detailed occupational codes were combined to coincide with the major occupation categories used in the years 1984-forward. A further complication is that the detailed occupation codes changed after 1983. Thus, for the time period 1979 to 1982 one detailed occupation code listing was used: for 1983. another code listing was used: and finally from 1984 to 1999. the major occupation code listing was utilized. Careful comparison of codebooks pre- and post-revisions enabled criteria for uniformity to be formulated and implemented.'6 Despite the aforementioned limitations, the CPS provides information on the most representative sample of workers in the U.S. The sub-sample used to conduct the analysis below includes all wage and salary workers with valid wage and hours data. Additional restrictions were placed on this sub-sample to isolate workers. These restrictions are: ages 18-64: employed in the public or private sector (excludes unincorporated self- employed): hours worked within the valid range in the survey (1 -99 per week); and wage rate between of $.50 and $100 in 1989 CPI-IJ-X1-adjusted dollars 55 Industry and Major Occupation variables also underwent minor changes in 1992 and 1994. respectively. The changes that occurred in these years did not atTect this analysis. 56 Appendix E includes a detailed account of the industry and occupation codes conversion, making it comparable across years. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 113 The calculation of average wages by SIC code and occupation was straightforward, and obtained from the CPS. The CPS does not include a separate major occupation classification for production workers. Thus, the two major occupation categories used in the CPS most closely related, craft workers and laborers, were combined to form production workers for wage calculation purposes. In order to be consistent with the real GDP values, all wages are reported in 19% constant dollars.'' The number of employees by occupation category was derived using a different procedure. The proportion of workers in the aforementioned occupation categories was calculated for the Manufacturing sector by SIC codes. This was obtained for each worker category except for production workers. The Census Bureau reports the number of production workers by SIC codes in its Annual Survey of Manufactures and Census of Manufactures publications. Since these data are available readily, they were used directly in the analysis instead of being estimated. In order to obtain the number of employees involved in other areas of manufacturing then, production workers were subtracted from the total number of employees reported by the Census Bureau. The share of workers by occupation category was then applied to the remainder by corresponding SIC code. 5' Conversion to 1996 dollars was done using the U.S. Bureau of Labor Statistics reported Not Seasonally Adjusted Consumer Price Index for All Urban Consumers (Series ID: CUUR.OOOOSAO). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 114 National Income and Product Accounts Real GDP was obtained from the Bureau of Economic Analysis, National Income and Product Accounts for the years 1977 to 1999. The values reported and used in this analysis were in chained 1996 dollars. Real GDP values through 2010 were obtained through published projections. The Congressional Budget Office reports economic projections, notably real GDP, for 2001 through 2011. These numbers were then used to calculate expected levels of real GDP and the firm 5-year growth rate of firms.58 Limitations Associated with empirical modeling and data analysis are econometric issues. This dissertation encountered several problems. The first is potential misspecification. It is possible, and even likely, that the quantity and price models may be excluding a variable(s) that contributes to the determination of a relationship between antitrust enforcement and employment. While every effort was made to include the relevant variables, there may be some that were inadvertently omitted. This sin of omission creates a bias in the empirical results by attributing higher statistical significance (through lower standard errors and subsequently higher t-statistics) to variables included in the model. Omitting a variable(s) can also bias the sign and magnitude of the coefficient estimates if the variables included in the model are S8 http:, vvww.bea.doc.uov bea da nipaweb PopularTables.asn: and http:■Vwww.cbo.nov showdoc.cfm?index=2727&sequenee= 11. Both obtained on 6/12/01. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 115 correlated with those excluded.5'’ While this problem is explicitly theoretical rather than statistical, it is still of concern. Another source of bias in the quantity and price models is through the presence of the dependent variable on both sides of the equation. The capital-to-labor expenditure ratio is included in both models as a control variable. While the two components of this variable were reported and obtained from the Annual Survey of Manufactures and Economic Census, ultimately labor expenditure is derived from the product of employment and wages. Labor employment is the dependent variable in the quantity analysis, and wage is the dependent variable in the price analysis. The presence of the dependent variable on both sides of the reduced-form equation creates a downward bias on the capital-to-labor expenditure ratio variable since the dependent variable is located in the denominator of the ratio. While this bias is worthy of note, it is not a cause of major concern to the analysis since the capital-to-labor expenditure variable is not the variable of interest. It is included in this analysis simply as one of several control variables. Second, focusing on employment only in the manufacturing sector could possibly generate selection bias. That is, firms operating in this sector may differ inherently from firms conducting business in other sectors. Also, the data indicate that manufacturing employment has been on a relatively steady decline throughout the 1980s and 1990s, which may exacerbate firm employment responses (although the TREND variable should control for this). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 116 Third, one of the five assumptions of the classical linear regression (CLR) model was violated.60 Assumptions are made concerning the way in which the data are generated, and each of these assumptions needs to be sustained for simple OLS estimation techniques to be appropriately applied. According to Kennedy (1998), the five assumptions of the CLR model are the following: 1. The dependent variable can be calculated as a linear function of a specific set of independent variables, plus a disturbance term. 2. The expected value of the disturbance term is zero. 3. The disturbance terms all have the same variance and are not correlated with one another. 4. The observations on the independent variable can be considered fixed in repeated samples. 5. The number of observations is greater than the number of independent variables and that there are no exact linear relationships between the independent variables. Given these assumptions, the OLS estimator is the most desirable for several reasons. For one, there is low computation cost involved. All statistical packages, including some hand-held calculators, have this estimation technique available in its list of built in procedures. As a result, OLS estimation is both easy to perform and timely. Also, the OLS estimator is designed to decrease the sum of squared residuals, which is a desirable feature in and of itself, but provides an additional advantage in terms of a corresponding high R-squared. In the CLR model, the OLS estimator is the best linear unbiased estimator (BLUE). This suggests that the OLS estimator has the lowest variance as a single number, and the smallest variance-covariance matrix as a vector of numbers. 59 All the major variables affecting employment and wages were included in the model, therefore this is not likely to be a problem in this case. See Greene (1997) for more details. 60 For more detail on the CLR model, see Kennedy (1998). The following discussion is from Kennedy (1998) and Greene (1997). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 117 Given that the OLS estimator is unbiased, it is also unbiased in very large samples. Consequently, the OLS estimator is consistent. For these reasons and others, OLS is the preferred estimation technique. Unfortunately, the third assumption of the CLR model was violated in this analysis. That is, the disturbance terms do not have uniform variance and are correlated with one another. The variance-covariance matrix of the disturbance vector demonstrates the variances of the individual disturbances on the diagonal, and the covariances between the disturbances on the off-diagonal. If the diagonals of this matrix are non-uniform, the errors are heteroskedastic. Intuitively, this suggests that the error term is believed to be coming from a different distribution for each observation in the dataset. If the off- diagonals of the variance-covariance matrix are non-zero, this implies that the errors are autocorrelated. That is, in repeated samples the disturbance term for one observation is believed to be correlated with the disturbance term for another observation. In the presence of heteroskedasticity and autocorrelation, the disturbances are said to be nonspherical. The CLR model is no longer valid and the generalized linear regression model (GLR) becomes the appropriate model. While the OLS estimator was the estimator of choice in the CLR model, the GLS estimator is preferred in the GLR model. This is because the OLS estimator no longer has the lowest variance in the presence of nonspherical disturbances, and subsequently is not the BLUE. The GLS estimator can be shown to be the BLUE by taking into account the nonsphericalness of the disturbances. The GLS estimator recognizes that some disturbances may be larger due to their large variance Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 118 (heteroskedasticity), and that some disturbances may be influenced or affected by other disturbances (autocorrelation). Thus, the GLS procedure gives a lower weight to the observations that have larger disturbance variances, and ultimately computes a weighted sum of the squared residuals. While both OLS and GLS estimation seek to minimize the sum of squared residuals, only the GLS estimator makes use of all information and adjusts for nonspherical disturbances. Failing to correct for serial correlation in the data results in several undesirable features. First, autocorrelation produces inaccurate estimates. Depending upon the correlation, positive or negative, the coefficient estimates obtained from OLS regression would either over-estimate or under-estimate the true relationship. Second, serial correlation in the data results in a higher R-squared than would be obtained otherwise. This is due to the fact that OLS estimation provides a better fit to the data than the true relationship because of the influence of large residuals on other residuals. Finally, serial correlation leads to an under-estimated variance, and consequently, a higher level of statistical significance. Thus, certain variables may be found to be statistically significant when, in the true relationship, they are not. Even in the presence of heteroskedasticity, the OLS estimator is still unbiased, consistent, and asymptotically normally distributed. However, as mentioned earlier, it is no longer the best linear unbiased estimator. For this reason, the GLS estimator is preferred over the OLS estimator because it is more efficient. That is, the sampling distribution of the GLS estimator has a smaller variance. As mentioned above, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 119 GLS attributes less weight to high-variance observations, therefore the variation of the GLS regression line from the true regression line will be lower when compared to OLS. In this analysis, the disturbances were both correlated with one another, and did not have uniform variance in all cases except one. In that one single case, autocorrelation was present, but not heteroskedasticity. The Likelihood Ratio (LR) test was used to test for heteroskedasticity by estimating the model twice—once imposing the restriction of homoskedasticity and once without the restriction. The LR test tests whether the log likelihoods of the restricted model less the unrestricted model (LnLR - LnLuR) are significantly different from zero. Autocorrelation was tested by comparing the empirical results of the same model estimated once allowing for serial correlation and once without.61 If no serial correlation was present, the empirical results should be identical. In each instance, the results differed. The results of both tests are contained in Appendix F. Due to nonspherical disturbances in the data, GLS estimation techniques on panel data were used in this analysis. Fourth, conducting the analysis on observations in the dataset at the 2-digit level of SIC codes may be too broad and overly aggregated to obtain any employment or wage effects. For instance, an antitrust case in one small segment of the industry is likely to have little impact on the entire industry at the 2-digit level. Performing the analysis at the 3- or 4-digit level of SIC codes, however, does not provide for a better alternative. 61 It is assumed that the correlation parameter is common for all panels in the data. The empirical results of serial correlation with the same correlation parameter were compared to the empirical results with a unique correlation parameter for each panel. The two estimations produced results similar enough that a common correlation parameter seemed more intuitively plausible. That is, the serial correlation present in the data is likely to be the prolonged influence of random shocks to the overall manufacturing sector rather than a Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 120 This is because at this disaggregate level, many of the detailed industries would always have an antitrust indictment dummy variable equal to 0. Again, the outcome would be to realize very little effects from antitrust enforcement. Finally, in the case studies, the product involved in the antitrust violation may be a trivial share of overall sales to the company. In this circumstance, the indictment would likely have little impact on the firm’s practices and overall performance. Also, the competing firm, who did not undergo an indictment but is examined in the case studies to determine deterrent effects, may not be the top competitor in the particular product for which the indicted firm is being investigated. As a result, there may be no deterrent effects to be captured. Again, the data analysis may be conducted at a level too aggregated to draw any conclusions other than broad generalizations. shock occurring in one major industry group that would affect another industry group. Thus, the correlation is expected to be a product o f the time series aspect o f the data. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 5 EMPIRICAL RESULTS FROM INDUSTRY ANALYSES Analysis of effects on employment levels The previous analyses examined firm-specific responses to antitrust enforcement. These analyses look at government intervention on a broader scale by examining the effect of antitrust enforcement on employment and wages within the manufacturing sector as a whole. As explained previously, there are two primary components of the empirical analysis. Of primary note in each analysis are the antitrust variables. The first component examines the quantity effect of antitrust enforcement on the labor market through the estimation of five model specifications (Models 1 through 5). The second component considers the wage effect of antitrust enforcement through the use of nearly identical model specifications (Models 6 through 10). The models are similar in that they all are specified in terms of levels rather than differences.62 62 Difference analyses were also conducted on employment and the natural log of average wage levels. The difference analyses are included in Appendix C and D, respectively. Additionally, quantity and price analyses were performed on the number of antitrust indictments issued, not just the presence of an indictment. The results of these analyses were not as strong, and thus not included in this dissertation. 121 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 122 As mentioned in Chapter 3, tests were conducted for heteroskedasticity and autocorrelation. In each model, serial correlation was present, but in one case, the errors were homoskedastic. The results of the test can be found in Appendix F. Due to the nonspherical disturbances, OLS estimation techniques would not be appropriate since they do not provide the best estimate of the true relationship of the independent variables to the dependent variable. Thus, GLS regression techniques on panel data were used for these analyses. Employment counts in the analyses are measured in the thousands. Real GDP is measured in hundreds of billions. Weighted average HHI is measured in units from 0 to 10,000. The remaining non-dummy variables—capital-to-labor expenditure ratio, and the industry’s 5-year average growth rate of firms—are all recorded in percentage points, and not decimals. Table 11 below shows the empirical results from each model. Appendix A contains the STATA programming code producing the estimates, and the results of the Likelihood Ratio tests. Table 11. Estimates of the effect of antitrust enforcement on number of employees ______(in thousands, N = 2,660) ______Dependent 1 nrioble: Sumber o f Employees Model Independent 1 driobtes: ■y i 4 5 (a) (b) (c) (d) (e) (0 Real GDP 0027 0.053 0.233 1 547 -0 162 (0306) (0.316) (0.680) (0866) (0787) Weighted Average HHI -0.006 -0 008 0006 -0.019 0.009 (0.003) • (0 003) •• (0.005) (0.009) •• (0005) Capital-to-Labor Expense Ratio -0.341 -0.347 -0.366 -0 802 -0 305 (0.066) • •• (0.068) ••• (0.115) •• (0.182) • •• (0.123) Average 5-vear Firm Growth Rate 3 618 3.654 8.557 11.902 9.901 (0.641) ••• (0 662) ••• (1 097) •• (1.779) (1186) ATR Dummy 1.672 1.734 7 178 5 291 7.894 (0.725) •• (0.749) •• (1.650) •• (2.083) •• (1.990) • •• 1-Ycar Lagged ATR Dummy 1.883 -0.481 145.965 21.171 254.987 (0.784) •• (1.238) (20.496) •• (15.034) (26498) ••• 2-Year Lagged ATR Dummy 0.265 0.289 4.325 2.691 4.462 (0.720) (0.743) (I 634) •• (2.073) (1.982) Trend -0.205 -0.279 -I 032 -4.135 -0 269 (0.661) (0.683) (1455) (1.877) •• (1.682) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 123 Table 11. Cont. Dependent Variable: Sumber o f Employees Model Independent 4 triable* I | i 4 5 (a) (b) | (c) (d) (c) (0 Interaction between I- Year Lag,ted A TR Dumm y and: Weighted Average HHI 0.004 0 007 0.043 0 003 (0002) •* (0 004) •• (0012) ••• (0012) Worker Category: Technicians &. Related Support — -151.973 ~ -229 499 (20.683) (24 138) Professional Specialty — -140.602 -212.647 (20670) (24.076) ••• Executive. Administrative. & — -135 260 — -204 606 Managerial (20 579) *•• (23 944) ••• Sales — -152 341 *~ -230.661 (20683) (24 146) Administrative Support & Clerical — -135 309 -204 294 (20566) (23 926) • •• Service — -155 300 — -235.086 (20.735) (24216) Major Industry Group Tobacco Products — ” -121 317 -48.704 (SIC code 21 1 (25 589) ••• (25 256) Textile Mill Products _ — — -42.210 -37 87? (SIC code 22) (14.253) ••• (11.651) • •• Apparel & Other Textile Products —“ — -48.384 -49 924 (SIC code 23) (15 354) (11957) ••• Lumber & Wood Products — —— -35.201 -47 847 (SIC code 24) (15 451) •• (12 543) • •• f urniture & Fixtures — — -33 498 -42 609 (SIC code 25) (14918) •• (12417) ••• Paper & Allied Products — “ ~ •45.700 -42 657 (SIC code 26) (15 193) ••• (12.191) ••• Printing <& Publishing — __ -25 171 -18 417 (SIC code 27) (14.580) • (II 905) Chemicals & Allied Products — —— -42.050 -20061 (SIC code 28) (13684) ••• (9 994) • Petroleum & Coal Products — — — -36 195 -41.575 (SIC code 29) (14132) ••• (11096) ••• Rubber & Miscellaneous Plastics — ~ -30.987 -41.022 Products (SIC code 30) (13 986) •• (11006) Stone. Clay. & Glass Products —— — -52.964 -46 060 (SIC code 32) (12 740) ••• (9 462) •** Primary Metal Industries —— — -60 501 -47 431 (SIC code 33) (14.139) ••• (10217) ••• Fabricated Metal Products —— “ -37.314 -31.732 (SIC code 34) (14 120) ••• (10635) ••• Industrial Machinery & Equipment — — — -16 705 9.270 (SIC code 35) (14.379) (10 467) Electronic & Other Electric — — — -47.002 -11 249 Equipment (SIC code 36) (13.010) ••• (9 635) Transportation Equipment — —— -65 073 -25.754 (SIC code 37) (16.715) ••• (14818) • Instruments & Related Products — — — -49 202 -38.386 (SIC code 38) (12.885) ••• (9 789) Miscellaneous Manufacturing —— — -38.167 -50 298 Industries (SIC code 39) (14.646) ••• (11 534) • •• Rho 0.953 0949 0.658 0.886 0567 R-square 0.017 0 021 0239 0 158 0326 Source: Author’s analysis Note: Standard errors are in parentheses. The Leather and Leather Products industry, SIC code 31, is omitted because no antitrust cases were brought forth in this industry during the relevant time period. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 124 Model 1: Analysis of employment levels - Core model The first model examines how antitrust enforcement affects employment levels for all w'orkers across all industries, with the cross section being worker category. The unit of observation is employment, of a particular type of occupation in a certain industry in a representative year, measured in thousands. The influence from an antitrust indictment on a particular category of workers is not isolated. Rather, this analysis examines the overall impact on all workers in the manufacturing sector. This model contains both heteroskedastic and autocorrelated errors. The null hypothesis of homoskedasticity was rejected at a high degree of statistical significance. As seen from table 11, antitrust enforcement does influence employment. In fact, there is a positive impact on employment from an antitrust indictment in the first year. Specifically, in a representative year the presence of an antitrust indictment increases employment for the typical worker in an average manufacturing industry by approximately 1,700 workers. Moreover, this effect is statistically significant at the 5 percent level. While initially this may be good news for workers, it is worthwhile to explore what happens beyond the first year of enforcement. In fact, the data suggest that for the year following the indictment, employment levels rise by an additional 1,900 workers. There is still growth in employment levels in the second year following the indictment, but this effect is not statistically significant. Overall, the increase in a representative industry’s employment growth due to antitrust indictments is approximately 3,800 workers (including the second year employment increase of roughly Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 125 200 workers). Statistically significant employment gains from antitrust enforcement appear in the same year as the indictment and persist the following year. The data analysis reveals that real GDP and manufacturing employment are positively correlated, but not statistically significant. This relationship is consistent with the anticipated results because a strong and/or growing economy would presumably lead to increased production. Increased production, in turn, typically requires additional inputs into the production process. Consequently, new jobs would be created in a strong economy. In a given year, as the market environment becomes more concentrated within a typical industry, there is an adverse effect on employment levels. A 100-point increase in the average concentration index reduces overall employment by nearly 600 workers. This effect is statistically significant at the 10 percent level. The analysis suggests that concentrated industries, marked by a few leading firms that control a large share of the market may be output restricting, thereby not requiring as much labor—an input into the production process. An increase in the capital-to-labor expense ratio corresponds to a decrease in employment in a given year. This effect is statistically significant at the 1 percent level. As firms allocate more monies to capital investment rather than labor by 10 percent, there is an overall decrease in manufacturing employment for the average industry by nearly 3,500 workers. This relationship was anticipated since there is substitutability between these two factor inputs. An increase in capital investment implies that the industry is becoming more capital-intensive, and thus less labor may be necessary Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 126 in the manufacturing production process. As a result, labor is being substituted away in favor of capital, and this leads to a decline in employment levels.63 As the industry’s average 5-year growth rate in the number of firms increases, so does the level of employment. As an industry grows, more firms are entering, more output is being produced, and more factor inputs are required. On average, a 10 percent increase in an industry’s 5-year average growth rate of firms increases the number of jobs available for a representative worker in a typical industry by over 36,000. This positive employment effect was anticipated, and is significant at the 1 percent level. Finally, the trend variable reflects the general decline of employment in the manufacturing sector over this period. Model 2: Analysis of employment levels, including industry concentration interaction effect Model 2 adds an interaction term between the 1-year lagged antitrust dummy and industry concentration into the model. This variable examines the effect on the antitrust/employment relationship of an increase in the concentration ratio. Due to the inclusion of the interaction variables, the estimate of the 1-year lagged antitrust variable is no longer meaningful on its own. This is because the estimate is now influenced by the estimates from interacting this variable with others.64 63 Recall this variable may reflect a bias due to the presence of the dependent variable in the creation of this variable. 64 A simplified version of the employment equation in Model 2 is E = a + P LI ATR + 6LIATRHHI + yZ where Z represents the remaining variables. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 127 Including the interaction variable results in the once lagged antitrust variable losing its statistical significance. However, the interaction variable between industry concentration and the 1-year lagged antitrust indictment variable is positively correlated with employment levels, and statistically significant at the 5 percent level. This effect is consistent with the notion that antitrust enforcement is more likely to be warranted in industries with high levels of concentration. In these instances, more output and employment expansion would be the expected outcome. The data analysis suggest that given a 100-point rise in the HHI, the previous year’s antitrust indictment against firms in this industry would increase the employment impact for the average industry by roughly an additional 400 workers. Recall, this employment effect is to be considered in conjunction with the effect from the once lagged antitrust variable on its own. Consistent with Model 1, the relationship between antitrust enforcement and employment is initially positive and similar in magnitude. Employment increases by over 1,700 workers in the same year as the indictment. At the mean value of concentration (from tables 9 and 10), the one-year lagged employment effect is a increase of approximately 2,200 workers.65 This result is consistent with Model 1 where the effect of the once lagged antitrust variable was also positive. This suggests that incorporating the market environment into the analysis amplifies the effect of an antitrust indictment on Including occupation interaction variables, the employment equation in Model 3 becomes E = a + [a + b*occ 1 + c*occ2 + d*occ3 + e*occ4 + Pocc5 + g*occ6 + 6HHI]L1 ATR + yZ In Model 1, the p estimate for LI ATR was equal to a. In Model 2, P = a + 6HHI. In Model 3, P = a + b*occl+ c*occ2 + d*occ3 + e*occ4 + f*occ5 + g*occ6 + oHHI. From this example, it is clear that the coefficient estimate for LI ATR is not meaningful on its own in Models 2 through 5 and 7 through 10. 65 This value was obtained by the following calculation: 4 (667.53) - 481 = 2189.12. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 128 employment by roughly 500 workers at the average level of industry concentration. The twice-lagged antitrust indictment variable is again not statistically significant. The remaining variables, except for real GDP and trend, are statistically significant and exhibit the same relationship with employment as in the core model. Model 3: Analysis of employment levels, including worker category interaction effects Model 3 expands upon Model 2 by including interaction dummy variables between 1-year lagged antitrust indictment and occupation. This model evaluates the differential effect of antitrust enforcement on employment levels for a particular group of workers (relative to the omitted category, production workers). It is interesting to examine if certain types of workers are more affected by an antitrust indictment in their industry. If all workers are not equally affected then what group of workers is impacted most? This model contained both heteroskedastic and autocorrelated disturbances. The GLS estimation procedure takes this into account. All three antitrust variables are statistically significant in this model. The data analysis suggests that an antitrust indictment initially raises employment for the average worker in a typical industry by over 7,000. Two years following the indictment, employment growth is positive, but not as large. Specifically, employment increases by nearly 4,500. This is a reduction in growth of approximately 2,500 workers. Both of these effects are statistically significant at the 1 percent level. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 129 Evaluating the once lagged antitrust indictment variable requires taking into consideration the additional impacts on employment from the interaction variables between this variable and industry concentration and worker categories. Both the once lagged antitrust variable and its interaction with the market environment are statistically significant and positively correlated with employment levels. This implies that the average worker in a typical industry generally benefits from antitrust enforcement. The interaction dummies between 1-year lagged antitrust indictment and worker category revealed that production workers are the group of workers most benefited by antitrust enforcement in the previous year. Relative to this group, all six remaining workers categories for the average industry in the manufacturing sector experience employment losses. The magnitude of the additional employment impact due to antitrust enforcement in the previous year ranges from a decrease of roughly 135,000 Executive, Administrative, & Managerial workers to a decrease of approximately 155,000 Service employees, relative to Production workers. All of these coefficients are statistically significant at the 1 percent level. This outcome suggests that there is, in fact, a temporary component to the employment effect. First and foremost, this implies that when a firm undergoes investigation for violating antitrust laws, there is an employment impact. Then, the effect on employment varies by worker category. There is a disproportionate effect on employment opportunities for each type of worker, and this disproportionate impact is not completely random. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 130 The category affected most negatively is Service. Recall, Service workers comprise the low wage worker category. It is interesting to discover that Production workers fare better than non-Production workers from antitrust enforcement. Although interesting, this result is a reasonable outcome. If an antitrust indictment leads to an increase in competition, this would generally amount to an increase in the number of firms and consequently an increase in total production. Increased production levels require more Production workers. It is also worth noting that the two categories most benefited by antitrust enforcement, next to Production workers, are Executive, Administrative, & Managerial and Administrative Support & Clerical workers. These two groups, although at the two extremes in terms of both skills and wages, are necessary for a new firm to function. Model 4: Analysis of employment levels, including major industry group interaction effects Model 4 incrementally builds upon Model 2 by including additional interaction variables for I-year lagged antitrust indictment and type of industry. Instead of evaluating the effect of antitrust enforcement on a particular group of workers, the effect on a particular industry within the manufacturing sector is examined. Once again, tests for autocorrelation and heteroskedasticity were conducted. The null hypothesis of homoskedasticity was rejected at a high degree of statistical significance. Thus, serial correlation in the disturbances was present, but not heteroskedasticity. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 131 Antitrust enforcement increases employment in the same year of the indictment by roughly 5,000 workers. This effect decreased in statistical significance compared to Model 3 from the 1 percent to the 5 percent level. Excluding the interaction variables, the remaining antitrust dummy variables no longer have a statistically significant effect on employment. Real GDP is positive and, for the first time, statistically significant. The remaining non-interaction variables are statistically significant, and exhibit the same relationship as was established in earlier models. Such similar outcomes provide support for the robustness of the analyses. The additional effect on employment from the joint effect of industry concentration and antitrust enforcement is positive and statistically significant at the 1 percent level. Thus, when an indictment is appropriately applied, there are positive employment gains to the average worker in a typical manufacturing industry. Whereas Model 3 examined the effect by worker category without considering the industry in which the worker is employed, this model examines the effect by industry group without considering the type of worker. Model 5, to be discussed next, brings both of these components together. This analysis examines how antitrust enforcement affects workers in a particular manufacturing industry. The omitted category is the Food & Kindred Products industry. Of the statistically significant effects, the results indicate that workers in all other industries fare worse from antitrust enforcement in the previous year. On average for a given year, the additional employment impact from an antitrust indictment in the previous year ranges from losses of roughly 25,000 in the Printing & Publishing industry Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 132 to roughly 121,000 in the Tobacco Products industry, relative to the Food & Kindred Products industry. These coefficients are statistically significant at the 10- and 1 percent level, respectively. The data analysis suggests that not only does antitrust enforcement have different effects by type of worker, as was found in Model 3, but also by type of industry in which the worker is employed. Let us now consider the effects on both industry and worker simultaneously. Model 5: Analysis of employment levels, including both worker category and major industry group interaction effects The final model of the quantity analyses is a combination of all the previous models. In addition to evaluating the effect of antitrust enforcement on a particular group of workers, the effect on a particular industry within the manufacturing sector is also examined. This model was tested for heteroskedasticity and autocorrelated disturbances. The presence of both were found in this model. Similar to Model 3, all three antitrust variables are statistically significant. Consistent with all the previous models, there is a positive effect on average industry employment levels in the same year in which the antitrust indictment was issued. In a given year, employment increases for the average worker in a typical industry by nearly 8,000. Employment levels due to an antitrust indictment in the previous year also rise, and this effect is statistically significant at the 1 percent level. Again, this estimate alone does not provide very meaningful information due to the interaction variables. On average, two years after the antitrust indictment employment still grows but falls by Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 133 nearly half of the original increase. The growth in employment tapers off to an increase of approximately 4,500 workers. These three effects are statistically significant at either the I - or 5 percent level, as seen from table 12. The effect on worker category is consistent with that obtained from Model 3. The additional impact on employment of the once lagged antitrust variable is most positive for the omitted category of Production workers. As seen previously, the workers who experience the greatest employment losses are Service workers. The workers who experience the next best employment gains (after Production workers) are those in the Executive, Administrative, & Managerial and Administrative Support & Clerical worker categories. At the industry level, somewhat different results are obtained. Relative to workers in the Food & Kindred Products industry, the additional impact from an antitrust indictment in the previous year on employment ranges from a loss of approximately 50,000 to a gain of roughly 9,000. The industries creating this range are the Miscellaneous Manufacturing and the Industrial Machinery & Equipment industries, respectively. Although statistically insignificant, the only positive employment effect from the industry interaction variables was found in the Industrial Machinery & Equipment industry. The industry that experienced the least additional adverse employment effect is the Chemicals & Allied Products industry. This effect is statistically significant at the 5 percent level. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 134 Analysis of effects on average wages A comprehensive analysis takes into consideration not only the effects of antitrust enforcement on the quantity of labor, but also the price of labor. The second level of analyses examines the possibility of a wage effect from antitrust enforcement. The price of labor is measured by hourly wages, and the regression analysis is conducted on the natural log of average hourly wages in 1996 constant dollars.66 Other than the dependent variable, the econometric specification for the price analyses is identical to the quantity analyses. Similar to the quantity analyses, tests were conducted for heteroskedasticity and autocorrelation. Just as before, the Likelihood Ratio test was employed for heteroskedasticity. Then, in order to determine whether autocorrelated disturbances were also present, the empirical results from a restricted model of no autocorrelation was compared to an unrestricted model of autocorrelation. If the coefficients in the two models are considerably different, then that would suggest the presence of autocorrelation. The results of these tests can be found in Appendix F. Each of the remaining models contained both heteroskedastic and autocorrelated errors. Model 6: Analysis of average wages - core model (natural log specifications) The first price model examines whether there is a wage effect from antitrust enforcement across the manufacturing sector. The LR test rejected the null 66 In order to be consistent with the quantity analysis, a first differences model for price was also conducted. This model examines changes in the natural log of wages as a function of changes in the other variables. The results from this analysis are contained in Appendix D. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 135 hypothesis of homoskedastic errors at a high level of statistical significance. Autocorrelation was also found to be present in the model. As a result, GLS estimation techniques were employed in the analysis. None of the antitrust variables are statistically significant in this analysis. The data analysis suggests that antitrust enforcement does not have an impact on the price of labor. The industry concentration variable, on the other hand, is positive and statistically significant at the 1 percent level. As the degree of concentration within an industry rises, so does the wage rate. This implies that a less competitive market generates a higher wage to the average worker. The data analysis suggests that workers in industries with escalating concentration have sufficient bargaining power, and are able to extract rents for themselves through higher wages. Recall from the employment analyses, increasing average industry concentration reduces overall employment. Combining these findings leads to an interesting result because it implies that whereas overall employment is lower in a less competitive market, those employed receive higher wages, on average. Specifically, a 100-point increase in the weighted average of the concentration index contributes to approximately a 1.5 percent increase in average wages in a typical manufacturing industry. On average, a rise in the capital-to-labor expenditure ratio also drives up wages. In a given year, a percentage point increase in capital investment over labor increases wages for the average worker by roughly 3.6 percent. This effect is statistically significant at the 1 percent level. As capital investment increases, capital is substituted for labor, however increased capital use may increase the marginal product of labor. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 136 Neoclassical labor economic theory purports that workers are paid the value of their marginal product. Embracing this theory, as the value of the marginal product of labor increases, so will the wage rate. Finally, the industry’s 5-year average growth rate is positively correlated with wages. Again, this outcome is consistent with what was anticipated. A growing industry is associated with growth in the number of firms and entrants into the market. This surge would shift out the labor demand curve, thereby increasing the wage rate offered in the labor market. In fact, a percentage point increase in the 5-year industry growth rate raises average wages in a typical industry by roughly 1.5 percent. This equates to a rise in average wages for a common worker of close to 25 cents in 1996 dollars.67 This effect is also statistically significant at the 1 percent level. Table 12. Estimates of the effect of antitrust enforcement on the natural log of average wages (in 1996 constant dollars, N = 2,650) Dependent I'arisbie: S etu n l Lor of Avenge H'mges Model Independent V arimbles: 6 ? 1 9 10 (a) (b> (c) 67 This calculation is based upon the mean of average wages across all industries being S16.146 as seen in Table 10. For further information regarding the interpretation of the coefficient, see Lewis (1986). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 137 Table 12. Cont. Dependent Variable: Satura! Log o f Average Wages Model Independent Variables: 6 T 1 9 10 U> (b) 1C) (d) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 138 Model 7: Analysis of average wages, including industry concentration interaction effect GLS estimation techniques were again used due to heteroskedastic and autocorrelated disturbances. Model 7 expands the core model by incorporating the effect that lagged antitrust enforcement in the current market environment may have on average wages. As was seen in Model 6, none of the key variables of interest are statistically significant. This is also proved to be the case for this model. The three statistically significant variables in Model 6 are also statistically significant in this model. Moreover, the same relationships to wages are demonstrated in this model as in the previous. Model 8: Analysis of average wages, including worker category interaction effects Including interaction variables for 1-year lagged antitrust indictment and occupation led to statistically significant results for some of the antitrust variables. Table 12 reveals that the coefficient on the once lagged antitrust variable is negative. This effect, however, must be taken in conjunction with its interaction with the occupation groups. Relative to Production workers, the additional effect from the once lagged antitrust variable is positive for most occupation groups. The three groups that realized negative wage effects from the previous year’s antitrust indictment are Production, Administrative Support & Clerical, and Service workers.68 In addition to being in the low-wage category, these workers tend to be less skilled in comparison to the other 68 The effects on these latter two worker categories, are not significantly different from those of Production workers. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 139 occupation categories examined. This result is consistent with the quantity analysis in Model 3 where it was determined that Service workers experienced the greatest employment losses. The analysis indicates that, on average for a given year, low-wage workers—particularly Service workers—receive the least benefit from antitrust enforcement in a typical industry of the manufacturing sector. The data analysis indicates there is a disproportionate influence on wages for different categories of workers, which gives credence to a temporary employment effect. The positive wage effects are all statistically significant at the 1 percent level. The category of workers who experience the greatest additional wage gain, 17.3 percent, from an antitrust indictment last year is Executive, Administrative & Managerial—the high-wage workers. Combining this percent increase with the decline in wages from the once lagged antitrust variable results in an average increase of 12.4 percent in wages for Executive, Administrative & Managerial workers in a representative industry.69 For Professional Specialty workers, the rise in wages from antitrust enforcement a year prior is the second highest at 10.2 percent. On average, the effect on wages from an antitrust indictment in the previous year for Technicians & Related Support workers is to increase their wages by 4.2 percent. Sales workers also benefit from an antitrust indictment in the previous year. 69 The 12.4 value is obtained from the percentage gain of 17.3 obtained for Executive, Managerial, & Administrative workers less the overall decline in wages of 4.9 from the 1 -year lagged antitrust variable. The value of the coefficient from the interaction variable between the once lagged antitrust indictment dummy and industry concentration is not statistically significant. Consequently, its value is essentially zero. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 140 On average, the effect of the 1-year lagged antitrust variable on these workers is to increase wages by 5.4 percent. Model 9: Analysis of average wages, including industry interaction effects Instead of looking at differential effects by worker category, this analysis examines the potential variation in wages from an antitrust indictment on a particular industry. Using GLS estimation, which allows for heteroskedastic and autocorrelated errors, the empirical results suggest that there is little to no effect of last year’s antitrust indictments on average wages across any of the major industry groups. This analysis also suggests that there is no wage effect from antitrust enforcement at any time period— contemporaneous or lagged. Model 10: Analysis of average wages, including both worker category and major industry group interaction effects The final model of the wage analyses examines both worker category and major industry group effects simultaneously. The GLS regression estimates reveal that Real GDP is positive and statistically significant for the first time in the wage analyses. The data analysis suggests that a one unit increase in Real GDP increases the natural log of average wages for a typical worker in a representative industry by approximately 5.5 percent. The same pattern that existed in Model 8 for the once lagged antitrust dummy and worker category interactions presents itself in this model. The high-wage workers— Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 141 specifically, the Executive, Administrative & Managerial occupation category— experience the greatest wage gain from antitrust enforcement the year prior. On average, their wages rise by approximately 11.2 percent.70 The next highest wage gainers are Professional Specialty workers, also of the high-wage worker category, who see their wages increase by roughly 8.9 percent. These effects are statistically significant at the 1 percent level. The medium wage workers. Technicians & Related Support and Sales, also experience additional wage gain relative to Production workers. These workers realize gains of roughly 1.8 and 3.6 percent, respectively. The wage losers are again Production, Administrative Support & Clerical, and Service workers. The empirical estimation suggests that industry-specific wage effects from an antitrust indictment in the previous year exist. The analysis is conducted relative to the Food & Kindred Products industry. O f the 7 statistically significant wage effects, workers in all of these industries experience wage gains from antitrust enforcement in the previous year. Although statistically insignificant, workers in the Apparel & Other Textile Products industry are the only ones who experience wage losses. On average, the wage effects augment the impact of the 1-year lagged antitrust variable, and range from a 6.3 percentage point increase in log wages in the Electronic & Other Electric Equipment industry to a 19.2 percentage point increase in the Tobacco Products industry. These influences are statistically significant at the 10- and 5 percent level, respectively. Recall from Model 4 of the employment analysis that workers in the Tobacco Products industry suffered employment losses, relative to workers in the Food & Kindred Products Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 142 industry, from antitrust enforcement last year. This is consistent with the wage gain seen from this model since a reduction in the supply of labor would correspond with a rise in wages, ceteris paribus. In summary, the 10 models capture both aspects of labor market analyses—price and quantity. The data analyses suggest that under certain model specifications, employment and wage effects from antitrust enforcement exist. Furthermore, under certain model specifications worker- and industry-specific effects emerge from the data. On average, Iow-wage workers (primarily Service) tend to be the ones most adversely impacted in terms of both employment opportunities and wage gain, whereas high-wage workers (primarily Executive. Managerial & Administrative) benefit the most from government intervention. These results suggest the existence of a temporary employment effect in that there is a disproportionate influence on employment and wages depending on the specific type of worker. The analyses also reveal employment and wage effects from antitrust enforcement that vary by industry. Workers in the Tobacco Products industry experience the highest wage gain at the expense of declining employment from an antitrust indictment in the previous year (relative to workers in the Food & Kindred Products industry). 70 This value is obtained by subtracting 8.3 from 19.5. Again, the interaction of industry concentration and once lagged antitrust is essentially zero. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 6 CONCLUSION The purpose of this dissertation is to re-examine the role of government intervention in the marketplace using employment and wage data from the U.S. manufacturing sector over the period 1979 through 1999. The existence of monopoly power in the market is undesirable for the consumer in most cases. However, the effect on the labor market is less clear. According to economic theory, unfettered competition in a free market benefits consumers through lower prices, improved quality, and better choices. Antitrust enforcement enables firms to compete on a level playing field without anti-competitive restraints or influences. This analysis extends beyond this theory by empirically estimating quantity and price effects on the labor market as a result of antitrust enforcement. Specifically, the number of workers and their average wages are studied. Due to restructuring and litigation, are certain workers possessing particular skill sets more likely to be affected than others? How large are the employment and wage impacts from antitrust enforcement? The answers to these questions and others were sought after in this research project, and answered. This dissertation was conducted on two levels: firm and industry. As stated previously, the firm analysis uses a case study approach to examining employment effects while the industry analysis examines employment and wage effects from antitrust enforcement across the manufacturing sector. The individual firm study employed 143 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 144 descriptive data analysis, while the industry analysis was conducted using formal econometric analysis techniques. For this reason, at the level of the individual firm, one must be cautious in attributing impacts. There were three recessions that took place over the 1979 to 1999 time period. As the data depicted (Figures 1 through 21), these business cycles inevitably had an effect on the overall industry and manufacturing sector, and they more than likely had an effect on the individual firm. Additionally, there is the general tendency for the firm to follow industry and sector trends. Furthermore, since the individual firm is a component of both the industry and sector, this makes it increasingly difficult to identify what caused the oscillations in the firm-level economic data. Finally, within each firm there is a multitude of specific events taking place that may or may not be known to the public, which would have an impact on their economic performance. It is impossible to know of all these occurrences, and it is likely only a few are reported in their summary of business activities. Despite these shortcomings, there was still valuable information to be gleaned from the firm-level analyses. The firm analysis initiates and introduces the investigation into employment effects from antitrust indictments. In the instance o f the case studies, the true indicator of an impact was firm productivity, or sales per employee. The other firm performance measures, employment and sales, often were highly sensitive to industry and sector trends. As a result, it is difficult to isolate an effect from antitrust enforcement. The measure for firm productivity remained relatively stagnant or declined slightly in all cases except for ADM and ConAgra. ConAgra’s productivity increased while ADM experienced much volatility, but no discemable trend. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 145 A consistent pattern that emerged from the case studies was an immediate rise and subsequent decline in competitor employment as soon as the antitrust indictment was issued. This is consistent with what was obtained in the industry analysis, and could be an outcome of the competing firm maneuvering to improve its market position and increase its competitiveness. In addition to lagged adverse employment effects occurring when the antitrust case was issued, the case studies also provided evidence for a lagged adverse employment effect at the conclusion of the antitrust case. In fact, this was true for half of the firms in the case studies. Specifically, Merck, Schering-Plough, ADM and The Stanley Works all showed employment declines a few years after the antitrust case was closed. The two firms showing declining employment that began during the antitrust litigation period were ConAgra and Black & Decker. The industry analysis completes and provides a comprehensive assessment of the examination of employment effects from antitrust indictments. These results reveal that antitrust enforcement does, in fact, affect the labor market. Regardless of how the models are specified, an employment effect still prevails. The empirical estimation indicates that antitrust intervention generally benefits manufacturing sector employment. Specifically, there is an average increase in employment of over 1,500 workers. The increase in employment levels persists for the two following years beyond the initial indictment. However, the rate of increase does not. The analyses reveal that, in most cases, the growth in employment surges one year after the indictment, but then tapers off considerably in the following years. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 146 The occupation group that gained the least from an antitrust indictment in the previous year is Service workers. Comparing average wages among the seven occupation categories studied, these workers also tend to earn relatively lower wages. Relative to the Food & Kindred Products industry, there were no industries that obtained statistically significant employment gains from antitrust enforcement. Those in the Tobacco Products and Miscellaneous Manufacturing industry experienced the greatest additional adverse effect from antitrust enforcement in the previous year. Conversely, workers in the Printing & Publishing and Chemicals & Allied Products industries underwent the least number of job losses. In addition to quantity effects, the empirical estimations reveal there are price effects in the labor market from antitrust enforcement. These price effects emerge from the data analysis when controls for occupation are introduced. Unlike the results obtained in the quantity analysis, Production workers experienced little wage benefit from government intervention. This result is consistent with economic theory in that it suggests that as the supply of labor increases (as was the case for Production workers in the quantity analysis), the price of labor falls, ceteris paribus. While Production workers experience the greatest additional benefit— in terms of employment opportunities— from the previous year’s antitrust indictment, they forgo wage gains. Relative to Production workers, Executive, Administrative & Managerial and Professional Specialty workers encountered the highest additional positive wage effect from antitrust enforcement in the previous year. Interestingly, these workers also tend to be high skilled, and constitute the high-wage worker category. Compared to workers in the Food & Kindred Products Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 147 industry, workers in the Electronic & Other Electrical Equipment industry experienced the lowest positive additional wage effect. Conversely, workers in the Tobacco Products industry saw the greatest additional increase in wages the year following an antitrust indictment. Recall this industry also experienced the highest additional employment losses from a once lagged antitrust indictment. Consistent with economic theory, this implies that as the supply of labor decreases, the price of labor increases, ceteris paribus. The results of this dissertation differ markedly from the aforementioned study conducted by Shughart and Tollison that directly examined the employment consequences of antitrust enforcement. While the data analysis from this study shows strong initial and weaker subsequent employment gains, their study results indicate negative employment consequences from antitrust indictments. A major explanation for these differences is that two very separate approaches were taken. First, Shughart and Tollison examined the effect of antitrust enforcement on the rate of unemployment for the entire U.S. economy. This dissertation examined the same effect for employment levels in the U.S., but only for the manufacturing sector. Second, the model specifications were dissimilar. While both model specifications sought to include factors that might have an impact on employment, Shughart and Tollison’s model captured only economy-wide influences on overall employment. In contrast, this dissertation included variables specific to manufacturing sector employment. For example, industry concentration proved to be an important factor in manufacturing employment levels. However, it is unclear whether this variable would have any effect, or even be appropriate, for general employment analysis. Additionally, Shughart and Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 148 Tollison used OLS estimation techniques whereas this dissertation used generalized least squares model estimation. Finally, two different time periods were examined in the separate studies. Shughart and Tollison examined the period from 1890 to 1981. This dissertation focused on the twenty-year period beginning in 1979 and ending in 1999.71 As seen from the figures in Chapter 3, employment levels fluctuate widely over time. Thus, one time span may capture certain employment trends not exhibited in other periods. For these reasons, obtaining differing results from the two studies is not surprising. The resounding theme from these analyses is that antitrust enforcement does not occur without having an impact on the labor market. As seen from the industry analysis, high-wage workers who tend to possess specific human capital benefit the most from antitrust enforcement in terms of both quantity and price, whereas low-wage workers who embody general human capital benefit the least. These outcomes provide support for a temporary component of the employment effect, and indicate that firms respond to government intervention in ways that ultimately affect labor in terms of both quantity and price. As seen from the firm analysis, lagged adverse employment effects occur during and after the antitrust indictment period. This dissertation provides strong evidence for employment and wage consequences derived from government intervention in the marketplace. 71 Anecdotally, the latter period seems to correspond to an era of greater influence by economists. Their influence increases the likelihood that true anti-competitive behavior would be targeted by antitrust with the expectation of output- and employment-enhancing results. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 149 The results of this investigation are important to antitrust enforcement policies because this information can be used to restructure antitrust policy to minimize adverse employment and wage effects or optimize benefits to workers. One area that was not fully investigated was the industry-specific findings. What is it about those particular industries that resulted in employment and wage gains as opposed to declines? The answer to this question, although beyond the scope of this dissertation, is a worthwhile research endeavor. Much more research needs to take place concerning the ramifications of government intervention in capitalistic markets, and their subsequent effects on the labor market. While this dissertation only began to scratch the surface, it is toward that end that this study is dedicated. Continued research in this area could be conducted on many levels. While this dissertation only examined whether an antitrust indictment occurred, employment effects by the type of antitrust violation committed is also worth considering. Alternatively, examination by type of outcome should also be considered. That is, if the antitrust indictment filed ended in a dismissal, settlement, victory for the government, or defense. Even further investigation could include not only the incidence of an antitrust action, as was done in this dissertation, but the number of actions initiated in each industry. Other aspects worth exploring are the effect on employment in other sectors of the economy aside from manufacturing. Currently, the data are unavailable to perform analyses on sectors other than manufacturing. Finally, different empirical models and econometric methods could be applied. For instance, seemingly unrelated regression techniques would be a viable task to invoke. Future research surrounding labor market Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 150 effects from government intervention hold infinite possibilities, and is indeed a worthwhile undertaking. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 151 APPENDIX A: ANALYSIS OF EFFECTS ON EMPLOYMENT LEVELS Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 152 log: u:\dissertation\atr_levels_final_gls.log log type: text opened on: 12 Jul 2002, 16:59:45 . / ‘ THIS ANALYSIS EVALUATES THE EFFECT OF ATR INDICTMENTS ON EMPLOYMENT BY OCCUPATION USING CROSS-SECTION TIME SERIES TECHNIQUES. THE VAR SICYEAR WAS CREATED TO ACCOMMODATE FOR THE INABILITY OF STATA TO COMPUTE THE SIC CODE AND OCCUPATION AND YEAR VARIATION THE VAR EMPLOYMT REPRESENTS NUMBER OF EMPLOYEES BY OCCUPATION AND SIC CODE * / . use *\\\sparkynt\userfiles\ypho\diss\sas_programs\datasets\bigatr_final.dta', clear . egen occsic=concat(occ sic) /'concatenates sic and occ variables*/ . egen sicyear=concat(sic year) /'concatenates sic and year variables*/ . egen occyear=concat(occ year) /'concatenates occ and year variables*/ . egen sicyearocc=concat(sic year occ) /'concatenates sic, year, and occ variables*/ . encode occsic, gen(OCCSIC) /'transforms character value to numeric*/ . encode sicyear, gen(SICYEAR) /'transforms character value to numeric*/ . encode occyear, gen(OCCYEAR) /'transforms character value to numeric*/ . encode sicye a ro cc, gen(SICYEAROCC) /'tra n s fo rm s c h a ra c te r value to n u m e ric '/ . iis occ /'allows one to take the difference without subtracting the 1999'/ . tsset OCCSIC year /'value from 1979 across 2 difference SIC codes*/ panel v a ria b le : OCCSIC, 1 to 140 time variable: year, 1979 to 1999 . replace avghhi=avghhi*100 /'converting hhi to 1 to 10000 range'/ (2940 re a l changes made) . gen trend=year-1978 . gen demploymt=d.employmt (140 missing values generated) . gen davghhi=d.avghhi (140 missing values generated) . gen dk_lexp=d.k_lexp (140 missing values generated) . replace realgdp=realgdp/100 (2940 re a l changes made) . gen drealgdp=d.realgdp (140 missing values generated) . replace davghhi=davghhi*iO (2800 re a l changes made) . replace atrhhi=atr*avghhi (1855 re a l changes made) . gen llatrhhi=liatr*avghhi (140 missing values generated) . gen 13atr=l.12atr (420 missing values generated) . gen 14atr=1.13atr (560 missing values generated) . gen 15atr=1.14atr (700 missing values generated) . gen 16atr=1.15atr (840 missing values generated) . gen 17atr=1.16atr (980 missing values generated) . replace atrtypel=l if type1n2==i /'the way atrtypei was defined previously was ncorrect because atr_case.sas takes the firs t.s ic when sorted by year and sic code to prevent repeated sic and year values, this code uses the typeln2 var which was created in Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 153 excel to identify sic codes and years with both types of atr case types to change the value to 1 if the first.sic code kept a 0 instead of 1*/ (168 real changes made) . replace atrtype2=1 if typeln2==1 (168 real changes made) . gen H atrtl=l.atrtypel /•lags antitrust type 1*/ (140 missing values generated) . gen Hatrt2=l.atrtype2 /•lags antitrust type 2*/ (140 missing values generated) . gen I2atrt1=l.l1atrtl /•twice lags antitrust type 1*/ (280 missing values generated) . gen 12atrt2=l.Hatrt2 /*twice lags antitrust type 1*/ (280 missing values generated) . gen Hatrt1hhi=Hatrt1'avghhi /‘ interaction between lagged atr type 1 and contemporaneous concentration*/ (140 missing values generated) . gen I1atrt2hhi=l1atrt2*avghhi /•interaction between atr type 2 and contemporaneous concentration*/ (140 missing values generated) . gen workcat=l if occ==5 | occ==6 | occ==7 /*low wage worker category*/ (1680 missing values generated) . re p la ce w orkcat=2 i f occ==1 | occ==4 / ‘ medium wage w orker c a te g o ry */ (840 real changes made) . replace workcat=3 if occ==2 | occ==3 ,'*high wage worker category*/ (840 re a l changes made) . / ‘ ORIGINALLY, ATRDOCC# VAR WAS EQUAL TO THE PRODUCT OF L1ATR AND DOCCM. NOW THAT WE HAVE TWICE LAGGED AND CONTEMPORANEOUS VARIABLES IN THE ANALYSIS, THE VAR NAME IS CHANGED REFLECT AS SUCH*/ . rename a trd o c c l lla t r d o c d . rename atrdocc2 l1atrdocc2 . rename a trdocc3 H a trd o cc3 . rename atrdocc4 Hatrdocc4 . rename atrdoccS llatrdoccS . rename atrdocc6 11atrdoccS . rename atrdocc7 Hatrdocc7 . /'ORIGINALLY, ATRDSIC# VAR WAS EQUAL TO THE PRODUCT OF L1ATR AND DSIC#. NOW THAT WE HAVE TWICE LAGGED AND CONTEMPORANEOUS VARIABLES IN THE ANALYSIS, THE VAR NAME IS CHANGED REFLECT AS SUCH''/ . rename atrd 2 0 H a trd 2 0 . rename atrd21 H a trd 2 i . rename atrd22 I1atrd22 . rename atrd 2 3 H a trd 2 3 . rename atrd24 I1atrd24 . rename atrd 2 5 H a trd 2 5 . rename atrd26 Hatrd26 . rename atrd 2 7 H a trd 2 7 . rename atrd28 Hatrd28 . rename atrd 2 9 I1 a trd 2 9 . rename atrd30 Hatrd30 . rename a trd 3 2 I1 a trd 3 2 . rename atrd 3 3 H a trd 3 3 . rename atrd 3 4 I1 a trd 3 4 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 154 . rename atrd35 I1atrd35 . rename atrd36 I1atrd36 . rename atrd37 Hatrd37 . rename atrd38 Hatrd38 . rename atrd39 Hatrd39 . gen byte low= workcat==1 . gen byte med= workcat==2 . gen byte high= workcat==3 . gen Hatrlow=llatr'low (140 missing values generated) . gen I1atrmed=l1atr*med (140 missing values generated) . gen 11 atrhi=Hatr*high (140 missing values generated) . label define workfmt 1 "Low wage worker" 2 "Medium Wage Worker* 3 "High Wage Worker" . label define occfmt 1 "Technicians" 2 "Prof Speclty Occ" 3 "Mgrs & Admin" 4 "Sales" 5 "Admin Supp, Cler" 6 "Service" 7 "Prodn" . label define sicfmt 20 "Food & Kindred" 21 "Tobacco Mfrs" 22 "Textil M ill Products" 23 •Apparel & Other Textile Products" 24 'Lumber & Wood Products" 25 “Furniture & Fixtures" 26 "Paper & A llied Products" 27 "Printing & Publishing" 28 'Chemicals & Allied Products" 29 "Petroleum & Coal Products" 30 "Rubber & Misc* 31 "Leather & leather Products" 32 "Stone,Clay,Glass & Concrete" 33 "Primary Metal" 34 "Fabricated Metal" 35 "Industrial Machinery & Equip" 36 "Electrical & Electronic" 37 "Transportation Equip" 38 "Instruments & Related" 39 "Misc Mfrg Inds" . label define atrtypes 1 "Monopoly, Premerger N otification Failure, Acquisitions, Joint Ventures" 2 "Price fixing, Restraint of Trade, Bid Rigging, Territorial Allocation, Restricting Output" . label values occ occfmt . label values workcat workfmt . label values sic sicfmt . i i s occ . ts s e t occ SICYEAR panel variable: occ, 1 to 7 tim e v a ria b le : SICYEAR, 1 to 420 . set matsize 700 . /'ANALYSIS ON ATR*/ . xtgls employmt realgdp avghhi k_lexp frm5yrte atr H atr 12atr trend, force corr(arl) panel(hetero) Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: heteroskedastic Correlation: common AR(1) coefficient for a ll panels (0.9526) E stim ated covariances = 7 Number o f obs 2660 E stim ated a u to c o rre la tio n s = 1 Number o f groups 7 Estimated coefficients = 9 No. of time periods 380 Wald c h i2 (8 ) 119.08 Log likelihood = -11777.45 Prob > chi2 0.0000 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 155 employmt j Coef. Std. E rr. z P»|Z | [95% Conf. In te r v a l] re a lg d p | .0268728 .3058843 0.09 0.930 -.5726493 .626395 a vg h h i | - .0058898 .0030188 -1 .95 0.051 -.0118064 .0000269 k_lexp | -.3411054 .0663354 -5.14 0.000 -.4711204 -.2110904 frm S y rte | 3.618375 .6407607 5.65 0.000 2.362508 4.874243 a t r | 1.671556 7254583 2.30 0.021 .2496838 3.093428 H a t r | 1.888032 .7842129 2.41 0.016 .3510029 3.425061 1 2 a tr | .2650273 .720144 0.37 0.713 -1.146429 1.676484 trend | -.2045087 .6610827 -0.31 0.757 -1.500207 1.09119 _cons | 52.17242 14.68412 3.55 0.000 23.39208 80.95276 . /‘ creating R-square see. Greene 1997 p. 509*/ . predict xml, xb (280 missing values generated) . gen double residml = employmt - xml /‘this command obtains the residuals*/ (280 missing values generated) . matrix accum Aml=residm1, noconstant /’ this command creates the scalar of (y-Xb)'(y- Xb) * / (obs=2660) • 9en y_ybar=employmt-134.766 /‘this command creates y-ybar*/ . gen sqy_ybar=y_ybar"2 . sum sqy_ybar Variable | Obs Mean Std. Oev. Min Max +■...... sqy_ybar I 2940 62162.33 203194.1 .0351979 2267235 . return lis t s c a la rs : r(N ) 2940 r(sum_w) 2940 r(mean) 62162.32836442971 r(V a r) 41287822525.8298 r(s d ) 203194.0514036516 r(m in ) .0351979322731495 r(m ax) 2267235 r(sum ) 182757245.3914233 . scalar sumy=r(sum) . matrix rsq l=1 -(Am1/sumy) . matrix lis t rsql symmetric rs q l[1,11 c1 r1 .01749245 . xtgls employmt realgdp avghhi Hatrhhi k_lexp frmSyrte atr lia tr 12atr trend,force corr(arl) panel(hetero) Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: heteroskedastic Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 156 Correlation: common AR(1) coefficient for a ll panels (0.9490) Estimated covariances = 7 Number o f obs = 2660 Estimated autocorrelations = 1 Number o f groups = 7 Estimated coefficients = 10 No. of tim e p e rio d s= 380 Wald ch i2 (9 ) = 123.81 Log likelihood = -11808.68 Prob > chi2 = 0.0000 employmt | Coef. Std. Err. z P>|zl [95% C onf. In te r v a l) realgdp | .0525643 .3158957 0.17 0.868 -.56658 .6717085 avghhi | - .007679 .0032079 -2.39 0.017 -.0139663 -.0013916 lla t r h h i | .003715 .0015252 2.44 0.015 .0007256 .0067043 k_lexp | - .3470077 .0684199 -5.07 0.000 -.4811082 -.2129072 frm S yrte | 3.653903 .6617586 5.52 0.000 2.35688 4.950926 a tr | 1.733836 .7489772 2.31 0.021 .2658682 3.201805 H a t r | -.4813831 1.287682 -0.37 0.709 -3.005192 2.042426 1 2 a tr | .2885223 .7434385 0.39 0.698 -1.16859 1.745635 trend | -.2786572 .6827885 -0.41 0.683 -1.616898 1.059584 _cons | 53.70756 14.92505 3.60 0.000 24.45499 82.96012 . p re d ic t xm2, xb (280 missing values generated) . gen double residm2 = employmt - xm2 (280 missing values generated) . matrix accum Am2=residm2, noconstant (obs=2660) . matrix rsq2=l-(Am2/sumy) . matrix lis t rsq2 symmetric rsq2[1,1] c l rl .02098138 . xtgls employmt realgdp avghhi k_lexp frmSyrte atr H atr 12atr 13atr 14atr ISatr 16atr 17atr, force corr(arl) panel(hetero) Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: heteroskedastic Correlation: common Afl(1) coefficient for a ll panels (0.9601) E stim ated covariances = 7 Number o f obs 1960 Estimated autocorrelations = 1 Number o f groups 7 Estimated coefficients = 13 No. of time periods 280 Wald c h i2 (12) 175.55 Log likelihood = -8820.466 Prob > chi2 0.0000 employmt | Coef. Std. Err. z P>|Z| [95% C onf. In te r v a l) realgdp | -.1952502 .0739732 -2 .6 4 0.008 -.3402351 -.0502653 avghhi | -. 0025042 .0034743 -0 .7 2 0.471 -.0093136 .0043053 k_lexp | -.2942349 .1026796 -2 .8 7 0.004 -.4954833 -.0929865 frm S yrte | 5.799372 .9597184 6.04 0.000 3.918358 7.680385 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 157 a t r | 4.987936 .9709127 5.14 0.000 3.084983 6.89089 H a t r | 4.283625 1.085787 3.95 0.000 2.155521 6.411729 1 2 a tr | 4.514065 1.141552 3.95 0.000 2.276665 6.751465 1 3 a tr | 4.362274 1.158567 3.77 0.000 2.091525 6.633023 1 4 a tr | 1.713142 1.204811 1.42 0.155 -.6482438 4.074528 1 5 a tr | 1.819691 1.237142 1.47 0.141 -.6050635 4.244445 16atr | 1.994541 1.169879 1.70 0.088 -.2983799 4.287462 1 7 a tr | -.5080421 1.013213 -0.50 0.616 -2.493904 1.477819 _cons | 58.25176 11.96412 4.87 0.000 34.80251 81.701 . xtgls employmt realgdp avghhi Hatrhhi k_lexp frmSyrte atr Hatr 1 2 a tr H a t r Hatrdocc2 Hatrdocc3 Hatrdocc4 !1atrdocc5 Hatrdocc6 trend, force corr(ar1) panel(hetero) Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: heteroskedastic Correlation: common AR(1) coefficient for all panels (0.6581) Estimated covariances 7 Number o f obs 2660 Estimated autocorrelations 1 Number o f groups 7 Estimated coefficients 16 No. o f tim e periods 380 Wald chi2(15) 228.77 Log lik e lih o o d = -13476.75 Prob >chi2 0.0000 employmt | Coef. Std. Err. z P >|z| [95% Conf . Interval) realgdp | .2326345 .6796088 0.34 0.732 -1.099374 1.564643 avghhi | .006165 .004606 1 .34 0.181 -.0028626 .0151926 H a tr h h i | .0071254 .0035601 2.00 0.045 .0001478 .0141031 k_lexp | -.3658009 .114842 3.19 0.001 -.5908871 - .1407147 frm 5 y rte | 8.55656 1.097363 7.80 0.000 6.405768 10.70735 atr | 7.1784 1.649939 4.35 0.000 3.944578 10.41222 Hatr | 145.9653 20.49587 7.12 0.000 105.7941 186.1365 12atr | 4.325355 1.633933 2.65 0.008 1.122905 7.527806 Hatrdoccl | -151.9726 20.68256 7.35 0.000 -192.5097 -111.4355 Hatrdocc2 | -140.6018 20.66977 6.80 0.000 -181.1138 -100.0898 Hatrdocc3 | -135.26 20.57934 6.57 0.000 -175.5947 -94.92522 l1atrdocc4 | -152.3408 20.68328 7.37 0.000 -192.8792 -111.8023 11atrdoccS | -135.3087 20.56551 6.58 0.000 -175.6164 -95.00107 Hatrdocc6 | -155.3004 20.7354 7.49 0.000 -195.941 -114.6598 tre n d | -1.031661 1.454911 0.71 0.478 -3.883235 1.819913 _cons | 53.01947 28.56642 1.86 0.063 -2.969675 109.0086 . p re d ic t xm3, xb (280 missing values generated) . gen double residm3 = employmt - xm3 (280 missing values generated) . m a trix accum Am3=residm3, noconstant (obs=2660) . matrix rsq3=1- (Am3/sumy) . matrix lis t rsq3 symmetric rsq3[1,1] Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 158 C1 r1 .23922962 . xtgls employmt realgdp avghhi Hatrhhi k_lexp frmSyrte atr H atr 12atr H atrdocd l1atrdocc2 l1atrdocc3 l1atrdocc4 HatrdoccS Hatrdocc6 I1atrd21 Hatrd22 I1atrd23 1atrd24 Hatrd25 Hatrd26 I1atrd27 I1atrd28 I1atrd29 I1atrd30 Hatrd32 I1atrd33 Hatrd34 Hatrd35 Hatrd36 Hatrd37 Hatrd38 Hatrd39 trend, force corr(arl) panel(hetero) Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: heteroskedastic Correlation: common A8(1) coefficient for a ll panels (0.5672) E stim ated cova ria n ce s = 7 Number o f obs = 2660 Estimated autocorrelations = 1 Number of groups = 7 Estimated coefficients = 34 No. of time periods= 380 Wald chi2(33) = 424.97 Log likelihood = -13870.83 Prob > chi2 = 0.0000 employmt | Coef. S td. E rr. z P»|Z| [95% Conf. In te r v a l] realgdp | -.162089 .787457 -0.21 0.837 -1.705476 1.381298 avghhi | .0086786 .0047491 1.83 0.068 -.0006295 .0179867 H a tr h h i | .0027591 .0119635 0.23 0.818 -.020689 .0262071 k _ le xp | -.3054616 .1225508 -2.49 0.013 -.5456567 .0652665 frm S y rte | 9.900712 1.186153 8.35 0.000 7.575896 12.22553 atr | 7.894047 1.989617 3.97 0.000 3.99447 11.79362 H a t r | 254.9866 26.49771 9.62 0.000 203.052 306.9211 1 2 a tr | 4.46172 1.982013 2.25 0.024 .5770449 8.346395 H a tr d o c d | -229.4985 24.1377 -9.51 0.000 -276.8075 -182.1894 l1atrdocc2 | -212.6468 24.07648 -8.83 0.000 -259.8359 -165.4578 l1atrdocc3 | -204.6058 23.94445 -8.55 0.000 -251.5361 -157.6756 Hatrdocc4 | -230.6611 24.14552 -9.55 0.000 -277.9854 -183.3367 HatrdoccS | -204.2935 23.92614 -8.54 0.000 -251.1879 -157.3991 Hatrdocc6 | -235.0859 24.21609 -9.71 0.000 -282.5486 -187.6233 Hatrd2l | -48.70422 25.25603 -1.93 0.054 -98.20513 .796698 I1 a trd 2 2 | -37.87658 11.65141 -3.25 0.001 -60.71292 -15.04024 Hatrd23 | -49.9238 11.95743 -4.18 0.000 -73.35993 -26.48768 Hatrd24 | -47.84687 12.54309 -3.81 0.000 -72.43088 -23.26286 H a trd 2 5 | -42.60873 12.41688 -3.43 0.001 -66.94536 -18.2721 Hatrd26 | -42.65714 12.19092 -3.50 0.000 -66.55089 -18.76338 I1 a trd 2 7 | -18.41661 11.9053 -1.55 0.122 -41 .75058 4.917349 Hatrd28 | -20.06055 9.993721 -2.01 0.045 -39.64788 - .4732131 I1atrd29 | -41.57477 11.09576 -3.75 0.000 -63.32205 -19.82748 Hatrd30 | -41.02168 11.00571 -3.73 0.000 -62.59249 -19.45088 Hatrd32 | -46.06015 9.461767 -4.87 0.000 -64.60488 -27.51543 Hatrd33 | -47.4311 10.21727 -4.64 0.000 -67.45658 -27.40563 H a trd 3 4 | -31.73247 10.63496 -2.98 0.003 -52.57661 -10.88834 Hatrd35 | 9.269542 10.46731 0.89 0.376 -11.24601 29.7851 H a trd 3 6 | -11.24935 9.635002 -1.17 0.243 -30.1336 7.634911 I1 a trd 3 7 | -25.75422 14.81795 -1.74 0.082 -54.79687 3.288432 Hatrd38 | -38.38593 9.788642 -3.92 0.000 -57.57132 -19.20054 I1 a trd 3 9 | -50.29806 11.53351 -4.36 0.000 -72.90333 -27.6928 trend | -.2685311 1.681898 -0.16 0.873 -3.564991 3.027929 _cons | 67.40163 33.10311 2.04 0.042 2.52073 132.2825 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 159 . p re d ic t xm4, xb (280 missing values generated) . gen double residm4 = employmt - xm4 (280 missing values generated) . m a trix accum Am4=residm4, noconstant (obs=2660) . matrix rsq4=1- (Am4/sumy) . matrix lis t rsq4 symmetric rsq4(1,l) C1 n .32600606 . xtgls employmt realgdp avghhi Hatrhhi k_lexp frmSyrte atr H atr 12atr I1atrd2l Hatrd22 I1atrd23 11atrd24 llatrd25 I1atrd26 Hatrd27 llatrd28 llatrd29 I1atrd30 Hatrd32 l1atrd33 I1atrd34 I1atrd35 I1atrd36 I1atrd37 Hatrd38 I1atrd39 trend, force corr(arl) Cross-sectional tim e-series FGLS regression Coefficients: generalized least squares Panels: homoscedastic Correlation: common AR(1) coefficient for a ll panels (0.8862) Estimated covariances 1 Number o f obs = 2660 Estimated autocorrelations 1 Number o f groups = 7 Estimated coefficients 28 No. o f tim e periods= 380 Wald chi2(27) 227.83 Log lik e lih o o d = -14128.83 Prob > chi2 = 0.0000 employmt | Coef. Std. Err. z P>|Z| [95% Conf. In te rv a l) realgdp | 1 .547352 .8659796 1.79 0.074 -.1499371 3.24464 avghhi | -.0193719 .0087139 -2.22 0.026 -.0364508 -.002293 Hatrhhi | .0426116 .0116765 3.65 0.000 .019726 .0654972 k_lexp | -.8017243 .1818986 -4.41 0.000 -1.158239 -.4452097 frm5yrte | 11.90183 1.778947 6.69 0.000 8.415156 15.3885 a tr | 5.290996 2.083349 2.54 0.011 1.207706 9.374286 Hatr | 21.17094 15.03412 1 .41 0.159 -8.295404 50.63728 12atr | 2.691087 2.072528 1.30 0.194 -1.370994 6.753168 H a trd 2 1 | -121.3173 25.58942 -4.74 0.000 -171.4717 -71.16299 H a trd 2 2 | -42.21047 14.25346 -2.96 0.003 -70.14674 -14.2742 I1 a trd 2 3 | -48.38441 15.35377 -3.15 0.002 -78.47724 -18.29158 l1 a trd 2 4 | -35.20073 15.45138 -2.28 0.023 -65.48487 -4.916581 l1 a trd 2 5 | -33.49804 14.91796 -2.25 0.025 -62.7367 -4.259377 I1atrd26 | -45.70038 15.19296 -3.01 0.003 -75.47803 -15.92273 H a trd 2 7 | -25.17105 14.58037 -1.73 0.084 -53.74805 3.405937 I1 a trd 2 8 | -42.0504 13.68412 -3.07 0.002 -68.87077 -15.23002 H a trd 2 9 | -36.19467 14.13236 -2.56 0.010 -63.8936 -8.495751 l1 a trd 3 0 | -30.98653 13.98552 -2.22 0.027 -58.39764 -3.575413 H a trd 3 2 | -52.96362 12.73966 -4.16 0.000 -77.93289 -27.99435 I1 a trd 3 3 | -60.50117 14.1387 -4.28 0.000 -88.21252 -32.78983 H a trd 3 4 | -37.31376 14.11993 -2.64 0.008 -64.98832 -9.639206 H a trd 3 5 | -16.70527 14.37918 -1.16 0.245 -44.88794 11.4774 I1 a trd 3 6 | -47.00167 13.01043 -3.61 0.000 -72.50165 -21.5017 I1atrd37 | -65.0726 16.71493 -3.89 0.000 -97.83327 -32.31194 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 160 H a trd 3 8 | •49.20207 12.88542 -3.82 0.000 -74.45702 -23.94711 I1 a trd 3 9 | -38.1672 14.64648 -2.61 0.009 -66.87376 -9.460632 tre n d | -4.134986 1.877058 -2.20 0.028 -7.813951 -.4560202 _cons | 98.51253 37.19881 2.65 0.008 25.60421 171.4209 . p re d ic t xm5, xb (280 missing values generated) . gen double residmS = employmt - xm5 (280 missing values generated) . m a trix accum Am5=residm5, noconstant (obs=2660) . matrix rsq5=1-(Am5/sumy) . matrix lis t rsq5 symmetric rsq5[1,1) c1 r1 .15848928 . summarize employmt realgdp avghhi H atrhhi k_lexp frmSyrte atr H atr 12atr Hatrdocd l1atrdocc2 HatrdoccS l1atrdocc4 HatrdoccS l1atrdocc6 Hatrd2l I1atrd22 I1atrd23 Hatrd24 Hatrd25 Hatrd26 Hatrd27 Hatrd28 I1atrd29 Hatrd30 I1atrd32 Hatrd33 l1atrd34 I1atrd35 I1atrd36 l1atrd37 Hatrd38 I1atrd39 V a ria b le | Obs Mean S td . Dev. Min Max employmt | 2940 134.7656 249.3662 0 1640.5 realgdp | 2940 65.084 12.01119 49.009 88.758 avghhi | 2940 667.5299 440.556 128.1236 2496.017 H a tr h h i | 2800 311.7569 439.2919 0 2443.248 k_lexp | 2940 21.08745 14.1851 4.598701 92.79851 frm S yrte | 2940 .118852 1.624088 -5.437162 4.879242 a t r | 2940 .452381 .4978119 0 1 H a t r | 2800 .455 .4980598 0 1 1 2 a tr | 2660 .4552632 .4980882 0 1 H a trd o c d | 2800 .065 .2465699 0 1 l1atrdocc2 | 2800 .065 .2465699 0 1 Hatrdocc3 | 2800 .065 .2465699 0 1 Hatrdocc4 | 2800 .065 .2465699 0 1 l1atrdocc5 | 2800 .065 .2465699 0 1 l1atrdocc6 | 2800 .065 .2465699 0 1 H a tr d 2 l | 2800 .005 .0705463 0 1 I1 a trd 2 2 | 2800 .0075 .0862926 0 1 I1 a trd 2 3 | 2800 .0175 .1311484 0 1 H a trd 2 4 | 2800 .015 .1215742 0 1 lla tr d 2 5 | 2800 .0075 .0862926 0 1 H a trd 2 6 | 2800 .0125 .1111223 0 1 I1 a trd 2 7 | 2800 .0125 .1111223 0 1 I1 a trd 2 8 | 2800 .0325 .1773555 0 1 I1 a trd 2 9 | 2800 .025 .1561528 0 1 lla tr d 3 0 | 2800 .0225 .1483294 0 1 H a trd 3 2 | 2800 .03 .1706177 0 1 H a trd 3 3 | 2800 .04 .1959942 0 1 H a trd 3 4 | 2800 .04 .1959942 0 1 I1 a trd 3 5 | 2800 .035 .1838126 0 1 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 161 l1 a trd 3 6 | 2800 .0325 .1773555 0 1 l1atrd37 | 2800 .0325 .1773555 0 1 11atrd38 | 2800 .025 .1561528 0 1 I1atrd39 I 2800 .0175 .1311484 0 1 . /'TH IS ANALYSIS EVALUATES THE EFFECT OF ATR INDICTMENTS ON EMPLOYMENT BY OCCUPATION USING SEEMINGLY UNRELATED REGRESSION TECHNIQUES. THE VAR OCCYEAR WAS CREATED TO ACCOMMODATE FOR THE INABILITY OF STATA TO COMPUTE THE SIC CODE AND OCCUPATION AND YEAR VARIATION THE VAR EMPLOYMT REPRESENTS NUMBER OF EMPLOYEES BY OCCUPATION AND SIC CODE*/ . i i s s ic . ts s e t s ic OCCYEAR panel variable: sic, 20 to 39 tim e v a ria b le : OCCYEAR, 1 to 147 . / ‘ ANALYSIS ON TWICE LAGGED ATR VARIABLE AND LAGGED ATRHHI*/ . sureg (employmt = realgdp avghhi Hatrhhi k_lexp frmSyrte atr H atr 12atr Hatrdocd l1atrdocc2 Hatrdocc3 Hatrdocc4 HatrdoccS l1atrdocc6 H atrd2l Hatrd22 Hatrd23 I1atrd24 I1atrd25 Hatrd26 Hatrd27 I1atrd28 I1atrd29 I1atrd30 I1atrd32 I1atrd33 I1atrd34 I1atrd35 Hatrd36 I1atrd37 I1atrd38 I1atrd39 trend) Seemingly unrelated regression Equation Obs Parms RMSE ‘ R -sq’ c h i2 P employmt 2660 33 173.3592 0.4981 2640.325 0.0000 | Coef. Std. Err. z P>|z| [95% Conf. Interval) employmt realgdp | 1.589283 2.244758 0.71 0.479 -2.810363 5.988929 avghhi | -.0011237 .0107288 -0.10 0.917 -.0221518 .0199045 H a tr h h i | -.0537163 .0484399 -1.11 0.267 -.1486568 .0412242 k_lexp | -.6746324 .3004057 -2.25 0.025 -1.263417 -.085848 frm S yrte | 16.8509 2.779757 6.06 0.000 11.40268 22.29913 a t r | 16.20015 7.920954 2.05 0.041 .6753659 31.72494 H a t r | 724.565 43.23616 16.76 0.000 639.8237 809.3063 1 2 a tr | 14.54236 7.935858 1 .83 0.067 -1.01164 30.09635 H a trd o c d | -698.7145 18.6938 -37.38 0.000 -735.3537 -662.0753 Hatrdocc2 | -648.5642 18.6938 -34.69 0.000 -685.2033 -611.925 l1atrdocc3 | -624.315 18.6938 -33.40 0.000 -660.9542 -587.6759 l1atrdocc4 | -704.8069 18.6938 -37.70 0.000 -741.4461 -668.1677 HatrdoccS | -622.0505 18.6938 -33.28 0.000 -658.6897 -585.4113 Hatrdocc6 | -718.1906 18.6938 -38.42 0.000 -754.8298 -681.5514 Hatrd2i | -46.8385 102.5519 -0.46 0.648 -247.8365 154.1595 l1atrd22 | -111.4732 43.04823 -2.59 0.010 -195.8461 -27.10018 Hatrd23 | -109.3738 32.57332 -3.36 0.001 -173.2163 -45.53126 I1 a trd 2 4 | -162.5342 41.31633 -3.93 0.000 -243.5127 -81.55566 I1atrd25 | -194.451 47.08508 -4.13 0.000 -286.736 -102.1659 I1atrd26 | -123.6101 34.72874 -3.56 0.000 -191.6772 -55.54303 I1 a trd 2 7 | -52.98312 41.57757 -1 .27 0.203 -134.4737 28.50741 I1atrd28 | -79.26493 25.53326 -3.10 0.002 -129.3092 -29.22066 Hatrd29 | -178.9202 34.82853 -5.14 0.000 -247.1829 -110.6576 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Ilatrd30 | -142.1474 35.64257 -3.99 0.000 -212.0056 -72.28926 Hatrd32 | -146.4404 25.53988 -5.73 0.000 -196.4976 -96.38315 I1atrd33 | -115.3591 23.7983 -4.85 0.000 -162.003 -68.71531 H a trd 3 4 | -52.50455 29.67054 -1 .77 0.077 -110.6577 5.648638 I1atrd35 | 33.94024 25.84104 1.31 0.189 -16.70726 84.58775 H a trd 3 6 | 20.73608 24.77743 0.84 0.403 -27.82679 69.29895 H a trd 3 7 | 45.38744 55.65675 0.82 0.415 -63.69779 154.4727 H a trd 3 8 | -149.4323 30.05005 -4.97 0.000 -208.3293 -90.53525 H a trd 3 9 | -206.4675 34.13901 -6.05 0.000 -273.3787 -139.5563 trend | -4.186384 4.698856 -0.89 0.373 -13.39597 5.023204 _cons | 58.29757 95.05921 0.61 0.540 -128.0151 244.6102 . log close log: u:\dissertation\atr_levels_final_gls.log log type: text closed on: 12 Jul 2002, 17:00:05 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 163 APPENDIX B: ANALYSIS OF EFFECTS ON THE NATURAL LOG OF AVERAGE WAGES Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 164 log: u:\dissertation\atr_logwage_levels_final_gls.log log type: text opened on: 12 Jul 2002, 16:53:43 . / ‘ THIS ANALYSIS EVALUATES THE EFFECT OF ATR INDICTMENTS ON EMPLOYMENT BY OCCUPATION USING CROSS-SECTION TIME SERIES TECHNIQUES. THE VAR SICYEAR WAS CREATED TO ACCOMMODATE FOR THE INABILITY OF STATA TO COMPUTE THE SIC CODE AND OCCUPATION AND YEAR VARIATION THE VAR EMPLOYMT REPRESENTS NUMBER OF EMPLOYEES BY OCCUPATION AND SIC CODE * / . use *\\\sparkynt\userfiles\ypho\diss\sas_programs\datasets\bigatr_final.dta", clear . egen occsic=concat(occ sic) /’concatenates sic and occ variables*/ . egen sicyear=concat(sic year) / ’concatenates sic and year variables*/ . egen occyear=concat(occ year) / ’concatenates occ and year variables*/ . egen sicyearocc=concat(sic year occ) /’ concatenates sic, year, and occ variables*/ . encode occsic, gen(OCCSIC) /’transforms character value to numeric*/ . encode sicyear, gen(SICYEAR) / ’transforms character value to numeric*/ . encode occyear, gen(OCCYEAR) /’ transforms character value to numeric*/ . encode s icye a ro cc, gen(SICYEAROCC) / ’ transform s c h a ra c te r value to num eric*/ . iis occ /’allows one to take the difference without subtracting the 1999*/ . tsset OCCSIC year /’value from 1979 across 2 difference SIC codes*/ panel variable: OCCSIC, 1 to 140 time variable: year, 1979 to 1999 . /’converting current wage to wages in 1996 constant dollars*/ . gen wage=(wavgwg/(74/l56.9)) if year==l979 (2801 m issing values generated) . replace wage=(wavgwg/(82.4/156.9)) if year==1980 (139 re a l changes made) . replace wage=(wavgwg/(90.9/156.9)) if year==1981 (139 re a l changes made) . replace wage=(wavgwg/(96.5/156.9)) if year==1982 (139 re a l changes made) . replace wage=(wavgwg/(99.6/156.9)) if year==l983 (140 re a l changes made) . replace wage=(wavgwg/(103.9/156.9)) if year==1984 (140 re a l changes made) . replace wage=(wavgwg/(107.6/156.9)) if year==l985 (140 r e a l changes made) . replace wage=(wavgwg/(109.6/156.9)) if year==l986 (139 r e a l changes made) . replace wage=(wavgwg/(113.6/156.9)) if year==l987 (140 re a l changes made) . replace wage=(wavgwg/(118.3/156.9)) if year==l988 (139 r e a l changes made) . replace wage=(wavgwg/(l24/156.9)) if year==1989 (140 r e a l changes made) . replace wage=(wavgwg/(l30,.7/1 56,•9)) i f year= =1990 (140 r e a l changes made) . replace wage=(wavgwg/(l36. .2/1 56,■9)) i f year= =1991 (139 re a l changes made) . replace wage=(wavgwg/(l40. .3/1 56. 9)) i f year= =1992 (140 re a l changes made) . replace wage=(wavgwg/(l44. 5/1 56. 9)) i f year= =1993 (140 re a l changes made) . replace wage=(wavgwg/(148. 2/1 56. 9)) i f year= =1994 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 165 (140 real changes made) . replace wage=(wavgwg/(152.4/l56.9)) if year==1995 (140 real changes made) . replace wage=(wavgwg/(156.9/156.9)) if year==1996 (138 real changes made) . replace wage=(wavgwg/(160.5/156.9)) if year==1997 (139 real changes made) . replace wage=(wavgwg/(l63/l56.9)) if year==l998 (139 real changes made) . replace wage=(wavgwg/(166.6/156.9)) if year==1999 (139 real changes made) . replace avghhi=avghhi'l00 (2940 real changes made) . gen trend=year-i978 . gen lnwage=log(wage) (12 missing values generated) . gen dlnwage=d.Inwage (159 missing values generated) . gen davghhi=d.avghhi (140 missing values generated) . gen dk_lexp=d.k_lexp (140 missing values generated) . replace k_lexp=k_lexp/i0 (2940 real changes made) . replace realgdp=realgdp/l000 (2940 real changes made) . gen drealgdp=d.realgdp (140 missing values generated) . replace drealgdp=drealgdp*lO (2800 real changes made) . replace davghhi=davghhi*iO (2800 real changes made) . replace atrhhi=atr*avghhi (1855 real changes made) . gen Hatrhhi=Hatr*avghhi (140 missing values generated) . replace atrtypel=l if typeln2==1 /'the way atrtypel was defined previously was incorrect because atr_case.sas takes the firs t.s ic when sorted by year and sic code to prevent repeated sic and year values, this code uses the type1n2 var which was created in excel to identify sic codes and years with both types of atr case types to change the value to 1 if the first.sic code kept a 0 instead of 1'/ (168 real changes made) . replace atrtype2=1 if typeln2==1 (168 real changes made) . gen llatrtl=l.atrtypel /'lags antitrust type 1'/ (140 missing values generated) . gen Hatrt2=l.atrtype2 /'lags antitrust type 2*/ (140 missing values generated) . gen 12atrtl=l.liatrtl /'tw ice lags antitrust type 1*/ (280 missing values generated) . gen 12atrt2=l.Hatrt2 /'tw ice lags antitrust type 1*/ (280 missing values generated) . gen l1atrtlhhi=Uatrt1 'avghhi /'interaction between lagged atr type 1 and contemporaneous concentration*/ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 166 (140 missing values generated) . gen Hatrt2hhi=Hatrt2*avghhi /‘ interaction between atr type 2 and contemporaneous concentration*/ (140 missing values generated) . gen workcat=1 if occ==5 | occ==6 | occ==7 /‘low wage worker category*/ (1680 missing values generated) . re p la c e w orkcat=2 i f occ==1 | occ==4 / ‘ medium wage w orker c a te g o ry */ (840 real changes made) . replace workcat=3 if occ==2 | occ==3 /‘high wage worker category*/ (840 real changes made) . / ‘ ORIGINALLY, ATRDOCC# VAR WAS EQUAL TO THE PRODUCT OF L1ATR AND DOCC#. NOW THAT WE HAVE TWICE LAGGED AND CONTEMPORANEOUS VARIABLES IN THE ANALYSIS, THE VAR NAME IS CHANGED REFLECT AS SUCH*/ . rename atrdoccl Hatrdoccl . rename atrdocc2 Hatrdocc2 . rename atrdocc3 11atrdocc3 . rename atrdocc4 Hatrdocc4 . rename atrdocc5 Hatrdocc5 . rename atrdocc6 Hatrdocc6 . rename atrdocc7 11atrdocc7 . / ‘ ORIGINALLY, ATRDSIC# VAR WAS EQUAL TO THE PRODUCT OF L1ATR AND OSIC#. NOW THAT WE HAVE TWICE LAGGED AND CONTEMPORANEOUS VARIABLES IN THE ANALYSIS, THE VAR NAME IS CHANGED REFLECT AS SUCH*/ . rename a trd 2 0 I1 a trd 2 0 . rename atrd21 H a trd 2 1 . rename a trd 2 2 H a trd 2 2 . rename atrd23 H a trd 2 3 . rename a trd 2 4 H a trd 2 4 . rename a trd 2 5 H a trd 2 5 . rename atrd26 I1 a trd 2 6 . rename a trd 2 7 H a trd 2 7 . rename atrd28 I1 a trd 2 8 . rename atrd29 I1 a trd 2 9 . rename a trd 3 0 H a trd 3 0 . rename atrd32 I1 a trd 3 2 . rename atrd33 H a trd 3 3 . rename atrd34 H a trd 3 4 . rename atrd35 H a trd 3 5 . rename atrd36 lla trd 3 6 . rename atrd37 H a trd 3 7 . rename a trd 3 8 H a trd 3 8 . rename atrd39 I1 a trd 3 9 . gen byte low= workcat==l . gen byte med= workcat==2 . gen byte high= workcat==3 . gen Hatrlow=l1atr*low (140 missing values generated) . gen llatrmed=llatr*med (140 missing values generated) . gen H atrhi=liatr*high (140 missing values generated) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 167 . label define workfmt 1 “Low wage worker" 2 ‘Medium Wage Worker" 3 "High Wage Worker* . label define occfmt 1 "Technicians* 2 "Prof Specify Occ" 3 "Mgrs & Admin" 4 "Sales* 5 "Admin Supp, C le r" 6 "S ervice" 7 "Prodn" . label define sicfmt 20 "Food & Kindred" 21 "Tobacco Mfrs* 22 "Textil M ill Products" 23 ■Apparel & Other Textile Products" 24 "Lumber & Wood Products" 25 "Furniture & Fixtures" 26 "Paper & A llied Products" 27 "Printing & Publishing" 28 "Chemicals & A llied Products* 29 "Petroleum & Coal Products" 30 "Rubber & Misc" 31 "Leather & leather Products' 32 ‘Stone,Clay,Glass & Concrete" 33 "Primary Metal" 34 "Fabricated Metal" 35 "Industrial Machinery & Equip" 36 "Electrical & Electronic" 37 "Transportation Equip* 38 "Instruments & R elated" 39 "M isc M frg Inds" . label define atrtypes 1 "Monopoly, Premerger Notification Failure, Acquisitions, Joint Ventures" 2 ‘Price fixing, Restraint of Trade, Bid Rigging, Territorial Allocation, Restricting Output" . label values occ occfmt . label values workcat workfmt . label values sic sicfmt . i i s occ . ts s e t occ SICYEAR panel variable: occ, 1 to 7 time variable: SICYEAR, 1 to 420 . set matsize 700 . /"ANALYSIS ON LAGGED ATRHHI AND ATR CASE TYPES*/ . xtgls lnwage realgdp avghhi k_lexp frmSyrte atr lla tr 12atr trend, force corr(arl) panel(hetero) Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: heteroskedastic Correlation: common AH(1) coefficient for a ll panels (0.8193) Estim ated co variances = 7 Number o f obs = 2650 Estim ated a u to c o rre la tio n s = 1 Number o f groups = 7 Estimated coefficients = 9 Obs per group: min = 375 avg = 378.5849 max = 380 Wald c h i2 (8 ) = 139.21 Log lik e lih o o d = 1445.796 Prob > c h i2 = 0.0000 lnwage | Coef. Std. E rr. z P »|z| (95% Conf. IntervalI realgdp | .0225884 .022248 1.02 0.310 -.0210169 .0661936 avghhi | .0001453 .0000188 7.73 0.000 .0001084 .0001821 k_lexp | .0359429 .0044608 8.06 0.000 .0271998 .0446859 frmSyrte | .0151206 .0042825 3.53 0.000 .006727 .0235142 a tr | -.0030295 .0053024 -0.57 0.568 -.0134219 . 0073629 H a t r | -.0033635 .0056669 -0.59 0.553 -.0144704 .0077434 12atr | .0031869 .0052639 0.61 0.545 -.0071302 .0135039 trend | -.0052476 .004796 -1.09 0.274 -.0146477 .0041524 _cons | 2.384679 .0941204 25.34 0.000 2.200206 2.569151 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 168 . p re d ic t xm6, xb (280 missing values generated) . gen double residm6 = lnwage - xm6 /'this command obtains the residuals*/ (290 missing values generated) . matrix accum Am6=residm6, noconstant /‘ this command creates the scalar of (y-Xb)'(y- X b )*/ (obs=2650) • 9en y_ybar=lnwage-2.722 /‘this command creates y-ybar*/ (12 missing values generated) . gen sqy_ybar=v_ybar*2 (12 missing values generated) . sum sqy_ybar V a ria b le | Obs Mean S td . Oev. Min Max sqy_ybar | 2928 .1233233 .1354898 8.53e-10 1.337 . return lis t s c a la rs : r(N ) = 2928 r(sum_w) = 2928 r(mean) = .1233233178510188 r(Var) = .0183574920790862 r(sd) = .1354898227878618 r(min) = 8.53173698356e-10 r(max) = 1.336999773979187 r(sum) = 361.0906746677831 . scalar sumy=r(sum) . m a trix rs q 6 = l• (Am6/sumy) . matrix lis t rsq6 symmetric rsq6[l,1] C1 n .10796648 . xtgls lnwage realgdp avghhi H atrhhi k_lexp frmSyrte atr H atr 12atr trend, force corr(arl) panel(hetero) Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: neteroskedastic Correlation: common AR(1) coefficient for a ll panels (0.8169) Estim ated covariances = 7 Number o f obs = 2650 Estim ated a u to c o rre la tio n s = 1 Number o f groups = 7 Estimated coefficients = 10 Obs per group: min = 375 avg = 378.5849 max = 380 Wald chi2(9) = 139.44 Log likelihood = 1440.8 Prob > chi2 = 0.0000 lnwage | Coef. Std. Err. z P>|z| (95% Conf. Interval) .. — — + ...... realgdp I .0227887 .0223407 1.02 0.308 -.0209982 .0665756 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 169 avghhi | .0001452 .0000195 7.43 0.000 .0001069 .0001835 Hatrhhi | 3.90e-07 .000011 0.04 0.972 -.0000212 .000022 k_lexp | .0360441 .0044684 8.07 0.000 .0272862 .044802 frm S yrte | .015152 .0042984 3.53 0.000 .0067272 .0235767 a tr | -.0029793 .0053237 -0.56 0.576 -.0134135 .007455 l l a t r | -.0035621 .0092056 -0.39 0.699 - .0216048 .0144806 1 2 a tr | .0032412 .0052846 0.61 0.540 - .0071164 .0135988 tre n d | -.0052919 .0048162 -1.10 0.272 -.0147315 .0041476 cons | 2.384176 .0944583 25.24 0.000 2.199041 2.569311 . p re d ic t xm7, xb (280 missing values generated) . gen double residm7 = lnwage - xm7 (290 missing values generated) . matrix accum Am7=residm7, noconstant (obs=2650) . matrix rsq7=1-(Am7/sumy) . matrix lis t rsq7 sym m etric r s q 7 [ t,11 c1 n .1086609 . xtgls lnwage realgdp avghhi H atrhhi k_lexp frmSyrte atr lla tr 12atr Hatrdoccl Hatrdocc2 Hatrdocc3 Hatrdocc4 HatrdoccS Hatrdocc6 trend, force corr(arl) panel(hetero) Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: heteroskedastic Correlation: common AR(1) coefficient for a ll panels (0.5973) Estimated covariances = 7 Number o f obs = 2650 Estimated autocorrelations = 1 Number o f groups = 7 Estimated coefficients = 16 Obs per group: min = 375 avg = 378.5849 max = 380 Wald c h i2 (15) 342.78 og likelihood = 928.7902 Prob > chi2 = 0.0000 lnwage | Coef. Std. Err. z P >iz| [95% Conf. In te r v a l] realgdp | .0466611 .0285887 1.63 0.103 - .0093717 .102694 avghhi | .0001546 .0000179 8.65 0.000 .0001196 .0001897 Hatrhhi | -7.33e-06 .0000155 -0.47 0.636 -.0000376 .000023 k_lexp | .044937 .0045302 9.92 0.000 .036058 .053816 frm S yrte | .0180201 .0043436 4.15 0.000 .0095069 .0265333 atr | .0033901 .0070494 0.48 0.631 -.0104265 .0172067 llatr | -.049321 .0168223 -2.93 0.003 -.0822921 -.0163498 12atr | .0075812 .0069726 1.09 0.277 -.0060849 .0212472 Hatrdoccl | .0911884 .0196774 4.63 0.000 .0526215 .1297554 Hatrdocc2 | .1511501 .0227968 6.63 0.000 .1064692 .195831 Hatrdocc3 | .1734414 .021437 8.09 0.000 .1314256 .2154572 Hatrdocc4 | .1029319 .0239691 4.29 0.000 .0559534 .1499105 HatrdoccS | -.01088 .0182184 -0.60 0.550 -.0465874 .0248274 !1atrdocc6 | -.0342119 .0259856 -1 .32 0.188 -.0851428 .0167191 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 170 trend | -.0109047 .0061039 -1.79 0.074 -.0228682 .0010588 _COns I 2.290183 .12018 19.06 0.000 2.054634 2.525731 . p re d ic t xm8, xb (280 missing values generated) . gen double residm8 = lnwage - xm8 (290 missing values generated) . m a trix accum Am8=residm8, noconstant (obs=2650) . m a trix rsq8=1- (Am8/sumy) . matrix lis t rsq8 symmetric rsq8[1,1] C1 n .27440945 . xtgls lnwage realgdp avghhi H atrhhi k_lexp frmSyrte atr lla tr 12atr Hatrdoccl l1atrdocc2 l1atrdocc3 llatrdocc4 HatrdoccS HatrdoccS 11atrd2l Hatrd22 I1atrd23 Hatrd24 Hatrd25 I1atrd26 I1atrd27 Hatrd28 I1atrd29 Hatrd30 Hatrd32 Hatrd33 Hatrd34 Hatrd35 I1atrd36 I1atrd37 Hatrd38 Hatrd39 trend, force corr(arl) panel(hetero) Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: heteroskedastic Correlation: common AR(1) coefficient for a ll panels (0.5716) E stim ated covariances = 7 Number o f obs = 2650 E stim ated a u to c o rre la tio n s = 1 Number o f groups = 7 Estimated coefficients = 34 Obs per group: min = 375 avg = 378.5849 max = 380 Wald c h i2 (3 3 ) = 404.72 Log likelihood = 878.9047 Prob > chi2 = 0.0000 lnwage | Coef. S td. E rr. Z P>|Z| [95% Conf. In te rv a l] realgdp | .0548368 .029596 1.85 0.064 -.0031703 .1128439 avghhi | .0001523 .000018 8.46 0.000 .0001171 .0001876 H a tr h h i | -.0000523 .0000449 -1.16 0.244 -.0001402 .0000357 k_lexp | .0440081 .004633 9.50 0.000 .0349277 .0530886 frm S yrte | .0163396 .004504 3.63 0.000 .0075119 .0251674 a tr | .0069951 .0074712 0.94 0.349 -.0076482 .0216385 llatr | -.082584 .0465467 -1 .77 0.076 -.1738139 .0086459 1 2 a tr | .010335 .0074448 1.39 0.165 -.0042566 .0249266 Hatrdoccl | .1012935 .0199395 5.08 0.000 .0622127 .1403742 Hatrdocc2 | .1720204 .0232429 7.40 0.000 .1264652 .2175756 Hatrdocc3 | .1952805 .0222143 8.79 0.000 .1517412 .2388198 Hatrdocc4 | .1186374 .0241828 4.91 0.000 .07124 .1660348 HatrdoccS | -.0105144 .0187818 -0.56 0.576 -.0473261 .0262972 Hatrdocc6 | -.0396214 .0266301 -1.49 0.137 -.0918154 .0125726 I1atrd21 | .1920854 .0947652 2.03 0.043 .006349 .3778218 I1 a trd 2 2 | .0513494 .0438029 1.17 0.241 -.0345027 .1372014 I1 a trd 2 3 | -.0167102 .0450244 -0.37 0.711 -.1049565 .071536 H a trd 2 4 | .0406009 .0471736 0.86 0.389 -.0518575 .1330594 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 171 H a trd 2 5 | .0187687 .0466696 0.40 0.688 -.072702 .1102394 I1atrd26 | .0433813 .0458795 0.95 0.344 -.0465409 .1333035 I1 a trd 2 7 | .0313086 .0447679 0.70 0.484 -.0564348 .1190521 lia trd 2 B | .08887 .0376608 2.36 0.018 .015056 .1626839 I1atrd29 | .0595921 .0417578 1.43 0.154 - .0222517 .1414358 Hatrd30 | .0387865 .0414148 0.94 0.349 - .0423849 .119958 Hatrd32 | .0411941 .0356443 1.16 0.248 -.0286674 .1110556 H a trd 3 3 | .0702757 .0385298 1.82 0.068 -.0052413 .1457927 H a trd 3 4 | .0545893 .0400585 1 .36 0.173 -.0239238 .1331025 Hatrd35 | .1010909 .0394563 2.56 0.010 .023758 .1784239 H a trd 3 6 | .0634606 .0363038 1.75 0.080 -.0076935 .1346146 I1 a trd 3 7 | .0954063 .0556634 1 .71 0.087 -.013692 .2045045 H a trd 3 8 | .0946628 .0368594 2.57 0.010 .0224197 .1669059 H a trd 3 9 | .0310986 .0434069 0.72 0.474 -.0539774 .1161746 trend | -.012449 .0063257 -1.97 0.049 -.0248472 -.0000509 cons I 2.258128 .1243592 16.16 0.000 2.014388 2.501868 . p re d ic t xm9, xb (280 missing values generated) . gen double residm9 = lnwage - xm9 (290 missing values generated) . matrix accum Am9=residm9, noconstant (obs=2650) . matrix rsq9=l-(Am9/sumy) . matrix lis t rsq9 symmetric rsq9[1,1] C1 rl .30168457 . xtgls lnwage realgdp avghhi H atrhhi k_lexp frmSyrte atr lla tr 12atr I1atrd2l I1atrd22 Hatrd23 Hatrd24 I1atrd25 I1atrd26 Hatrd27 llatrd28 I1atrd29 I1atrd30 I1atrd32 l1atrd33 Hatrd34 Hatrd35 I1atrd36 Hatrd37 Hatrd38 llatrd39 trend, force corr(an) panel(hetero) Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: heteroskedastic Correlation: common AR(1) coefficient for a ll panels (0.8022) E stim ated covariances = 7 Number o f obs = 2650 E stim ated a u to c o rre la tio n s = 1 Number o f groups = 7 Estimated coefficients = 28 Obs per group: min = 375 avg = 378.5849 max = 380 Wald chi2(27) = 158.14 Log likelihood = 1417.795 Prob > chi2 = 0.0000 lnwage | C oef. S td. E rr. z P>|Z| [95% Conf. In te r v a l] realgdp | .0261237 .0233432 1.12 0.263 -.0196282 .0718756 avghhi | .0001455 .0000204 7.12 0.000 .0001055 .0001856 H a tr h h i | -.0000336 .000032 -1.05 0.294 -.0000964 .0000292 k_lexp | .0358559 .004575 7.84 0.000 .026889 .0448229 frm S yrte | .0139885 .004474 3.13 0.002 .0052197 .0227574 a t r | -.0026292 .0056341 -0.47 0.641 -.0136718 .0084134 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 172 l l a t r | .0208719 .0384647 0.5 4 0.587 -.0545175 .0962613 1 2 a tr | .0039592 .0056138 0.71 0.481 -.0070436 .014962 I1 a trd 2 l | .1066551 .0695092 1.53 0.125 -.0295805 .2428906 H a trd 2 2 | .0085433 .0367006 0.2 3 0.816 -.0633885 .0804751 I1 a trd 2 3 | -.0426935 .0394542 -1.08 0.279 -.1200223 .0346353 I1 a trd 2 4 | .0261922 .0398102 0 .66 0.511 -.0518344 .1042187 Hatrd25 | -.0260554 .0385327 -0 .6 8 0.499 -.1015782 .0494674 I1 a trd 2 6 | -.0167123 .0392458 -0.43 0.670 -.0936327 .0602082 H a trd 2 7 | -.0250392 .0375682 -0 .6 7 0.505 -.0986715 .048593 H a trd 2 8 | .016608 .0345739 0 .4 8 0.631 -.0511557 .0843716 H a trd 2 9 | -.0156473 .0361739 -0 .4 3 0.665 -.0865469 .0552523 H a trd 3 0 | -.0214523 .0357377 -0 .6 0 0.548 -.0914969 .0485923 I1 a trd 3 2 | -.013798 .0323207 -0.43 0.669 -.0771454 .0495495 Hatrd33 | -.0123521 .0359138 -0 .3 4 0.731 -.0827418 .0580376 H a trd 3 4 | -.017558 .0359479 -0 .4 9 0.625 -.0880145 .0528985 I1 a trd 3 5 | .0160804 .0364209 0.44 0.659 -.0553033 .087464 I1 a trd 3 6 ( -.0048368 .0330446 -0 .1 5 0.884 -.069603 .0599294 I1 a trd 3 7 | .0239149 .0440444 0.5 4 0.587 -.0624105 .1102402 I1 a trd 3 8 | .0255571 .0327282 0.7 8 0.435 -.038589 .0897031 I1 a trd 3 9 | -.002409 .0375551 -0 .06 0.949 -.0760156 .0711976 trend | -.0057124 .0050473 -1.13 0.258 -.0156049 .0041801 _cons | 2.369515 .0984292 24.07 0.000 2.176597 2.562432 . predict xmlO, xb (280 missing values generated) . gen double residmlO = lnwage - xm10 (290 missing values generated) . m atrix accum Am10=residml0, noconstant (obs=2650) . matrix rsqlO=1- (Am10/sumy) . matrix lis t rsqlO symmetric rsql0[1,1J c1 n .1128554 . summarize wage realgdp avghhi H atrhhi k_lexp frmSyrte atr lla tr 12atr Hatrdoccl !1atrdocc2 l1atrdocc3 l1atrdocc4 HatrdoccS Hatrdocc6 Hatrd2l Hatrd22 I1atrd23 Hatrd24 I1atrd25 I1atrd26 I1atrd27 I1atrd28 Hatrd29 Hatrd30 I1atrd32 Hatrd33 Hatrd34 Hatrd35 I1atrd36 I1atrd37 Hatrd38 Hatrd39 trend V a ria b le | Obs Mean S td. Oev. Min Max wage | 2928 16.14622 5.530118 4.786086 40.07283 re a lg d p | 2940 6.5084 1.201119 4.9009 8.8758 a vghhi | 2940 667.5299 440.556 128.1236 2496.017 H a tr h h i | 2800 311.7569 439.2919 0 2443.248 k_ le xp | 2940 2.108745 1.41851 .4598701 9.279851 frm S y rte | 2940 .118852 1.624088 -5.437162 4.879242 a t r | 2940 .452381 .4978119 0 1 l l a t r | 2800 .455 .4980598 0 1 1 2 a tr | 2660 .4552632 .4980882 0 1 Hatrdoccl | 2800 .065 .2465699 0 1 Hatrdocc2 | 2800 .065 .2465699 0 1 Hatrdocc3 | 2800 .065 .2465699 0 1 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 173 Hatrdocc4 | 2800 .065 .2465699 0 1 HatrdoccS | 2800 .065 .2465699 0 1 l1atrdocc6 | 2800 .065 .2465699 0 1 11atrd2l | 2800 .005 .0705463 0 1 I1 a trd 2 2 | 2800 .0075 .0862926 0 1 I1 a trd 2 3 | 2800 .0175 .1311484 0 1 I1 a trd 2 4 | 2800 .015 .1215742 0 1 I1 a trd 2 5 | 2800 .0075 .0862926 0 1 I1 a trd 2 6 | 2800 .0125 .1111223 0 1 I1 a trd 2 7 | 2800 .0125 .1111223 0 1 I1 a trd 2 8 | 2800 .0325 .1773555 0 1 I1 a trd 2 9 | 2800 .025 .1561528 0 1 I1atrd30 | 2800 .0225 .1483294 0 1 I1 a trd 3 2 | 2800 .03 .1706177 0 1 I1 a trd 3 3 | 2800 .04 .1959942 0 1 H a trd 3 4 | 2800 .04 .1959942 0 1 Hatrd35 | 2800 .035 .1838126 0 1 11atrd36 | 2800 .0325 .1773555 0 1 H a trd 3 7 | 2800 .0325 .1773555 0 1 Hatrd38 | 2800 .025 .1561528 0 1 I1atrd39 | 2800 .0175 .1311484 0 1 tre n d | 2940 11 6.056331 1 21 . by occ: summarize wage -> occ = Technicians Variable | Obs Mean Std. Dev. Min Max wage | 413 15.98102 2.684062 6.856915 27.35315 -> occ = Prof Speclty Occ V a ria b le | Obs Mean S td . Dev. Min Max wage | 420 21.31252 4.089957 10.77676 37.62443 -> occ = Mgrs & Admin V a ria b le | Obs Mean Std. Dev. Min Max wage | 420 22.74068 3.15546 16.39114 33.11391 -> occ = Sales Variable | Obs Mean Std. Dev. Min Max wage | 420 19.02018 3.79095 9.904864 40.07283 -> occ = Admin Supp, C le r Variable | Obs Mean Std. Dev. Min Max wage | 420 11.74675 1.757084 7.958673 19.29726 -> occ = Service Variable | Obs Mean Std. Dev. Min Max wage I 415 10.23639 2.297712 4.786086 19.87438 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 174 > occ = Prodn Variable | Obs Mean Std. Dev. Min Max wage | 420 11.9129 2.566068 6.890295 19.72178 . / ‘ ANALYSIS ON TWICE LAGGED ATR VARIABLE WITH AVG WAGE CATEGORIES INSTEAD OF OCC CATEGS AND LAGGED ATRHHI*/ . iis workcat . ts s e t w orkcat SICYEAROCC panel variable: workcat, 1 to 3 time variable: SICYEAROCC, 1 to 2940, but with gaps . xtgls lnwage realgdp avghhi k_lexp frmSyrte atr lla tr 12atr,force corr(arl) panel(hetero) Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: heteroskedastic Correlation: common AR(1) coefficient for a ll panels (0.6841) Estim ated cova ria n ce s = 3 Number of obs = 2650 Estimated autocorrelations = 1 Number of groups = 3 Estimated coefficients = 8 Obs per group: min = 755 avg = 919.1887 max = 1135 Wald c h i2 (7 ) = 73.39 Log likelihood = 370.6168 Prob > chi2 = 0.0000 lnwage | Coef. Std. Err. z P>|z| (95% C onf. In te r v a l) realgdp | - .0006145 .0068012 -0.09 0.928 - .0139447 .0127156 avghhi | .0001566 .0000272 5.75 0.000 .0001032 .00021 k_lexp | .0452582 .0075747 5.97 0.000 .030412 .0601043 frm S yrte | .0155975 .0072055 2.16 0.030 .0014751 .02972 a tr | .0006038 .0123503 0.05 0.961 -.0236024 .02481 l l a t r | -.0027996 .0125228 -0.22 0.823 -.0273439 .0217446 1 2 a tr | .0071033 .0121793 0.58 0.560 -.0167677 .0309743 _cons | 2.519835 .051161 49.25 0.000 2.419561 2.620109 . xtgls lnwage realgdp avghhi H atrhhi k_lexp frmSyrte atr lla tr 12atr, force corr(arl) panel(hetero) Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: heteroskedastic Correlation: common AR(1) coefficient for a ll panels (0.6839) Estimated covariances = Number o f obs = 2650 Estimated autocorrelations = Number o f groups = 3 Estimated coefficients = Obs per group: min = 755 avg = 919.1887 max = 1135 Wald c h i2 (8 ) = 74.10 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 175 Log lik e lih o o d = 370.7134 Prob >chi2 = 0.0000 lnwage | Coef. Std. Err. z P>|z| (95% C onf. In te r v a l] realgdp | - .0005909 .0068001 -0.09 0.931 - .0139189 .012737 avghhi | .000166 .0000297 5.59 0.000 .0001078 .0002242 Hatrhhi | - .0000219 .0000274 -0.80 0.425 -.0000756 .0000318 k_lexp | .045205 .0075718 5.97 0.000 .0303645 .0600455 frmSyrte | .0163029 .0072566 2.25 0.025 .0020802 .0305256 atr | .0007707 .0123507 0.06 0.950 -.0234361 .0249776 llatr | .011497 .0218566 0.53 0.599 - .0313412 .0543352 12atr | .0070345 .0121783 0.58 0.564 -.0168345 .0309035 cons | 2.51359 .0517656 48.56 0.000 2.412131 2.615049 . xtgls lnwage realgdp avghhi Hatrhhi k_lexp frmSyrte atr lla tr 12atr llatrmed H atrhi, force corr(arl) panel(hetero) Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: heteroskedastic Correlation: common AR(1) coefficient for a ll panels (0.5358) Estimated covariances = 3 Number of obs = 2650 Estimated autocorrelations = 1 Number of groups = 3 Estimated coefficients = 11 Obs per group: min = 755 avg = 919.1887 max = 1135 Wald c h i2 ( 10) 421.96 Log lik e lih o o d = 302.0225 Prob > ch i2 = 0.0000 lnwage | Coef. S td. E rr. z P>|z| [95% Conf. In te r v a l] realgdp | -.0075689 .0062183 -1 .22 0.224 -.0197566 .0046188 avghhi | .0001655 .0000236 7.02 0.000 .0001193 .0002116 H a tr h h i | -.0000308 .0000277 -1.11 0.267 -.0000852 .0000236 k_lexp | .0505973 .0059717 8.47 0.000 . 038893 .0623016 frmSyrte | .0171315 .0057514 2.98 0.003 .005859 .0284041 a tr | .0116774 .0121097 0.96 0.335 -.0120573 .035412 l l a t r | -.1758887 .0248631 -7.07 0.000 -.2246195 -.127158 12atr | .0132189 .0119103 1.11 0.267 - .0101249 .0365626 lla trm e d | .2566045 .0256053 10.02 0.000 .2064191 .30679 H a t r h i | .3800962 .0229496 16.56 0.000 .3351159 .4250766 _cons | 2.546174 .045191 56.34 0.000 2.457601 2.634747 . xtgls lnwage realgdp avghhi Hatrhhi k_lexp frmSyrte atr lla tr 12atr Hatrd2l Hatrd22 I1atrd23 I1atrd24 I1atrd25 Hatrd26 I1atrd27 I1atrd28 I1atrd29 I1atrd30 I1atrd32 I1atrd33 Hatrd34 l1atrd35 Hatrd36 I1atrd37 Hatrd38 I1atrd39,force corr(arl) panel(hetero) Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: heteroskedastic Correlation: common AR(1) coefficient for a ll panels (0.6695) Estimated covariances = 3 Number of obs = 2650 Estimated autocorrelations = 1 Number of groups = 3 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 176 Estimated coefficients = 27 Obs per group: min = 755 avg = 919.1887 max = 1135 Wald c h i2 (2 6 ) 93.97 Log lik e lih o o d = 362.3707 Prob > chi2 = 0.0000 lnwage | C oef. S td . E rr. z P> |z | (95% C onf. In te rv a l) realgdp | .0004016 .0069195 0.06 0.954 -.0131603 .0139635 avghhi | .0001646 .0000296 5.56 0.000 .0001065 .0002226 H a tr h h i | -.0000665 .0000774 -0.86 0.391 -.0002182 .0000852 k_lexp | .0434445 .0077068 5.64 0.000 .0283393 .0585496 frm S y rte | .015319 .0074668 2.05 0.040 .0006843 .0299537 a t r | .0036963 .0128013 0.29 0.773 -.0213937 .0287864 l l a t r | -.0108332 .0765626 -0.14 0.887 -.1608933 .1392268 1 2 a tr | .0090444 .0127062 0.71 0.477 -.0158592 .033948 I1atrd2l | .1733104 .1635292 1 .06 0.289 -.1472009 .4938217 H a trd 2 2 | .0690956 .0746546 0.93 0.355 -.0772247 .2154158 I1 a trd 2 3 | -.0864497 .0760374 -1 .14 0.256 -.2354803 .0625808 I1 a trd 2 4 | .0497259 .0803846 0.62 0.536 -.1078251 .2072769 I1 a trd 2 5 | -.0118154 .0796847 -0.15 0.882 -.1679946 .1443637 I1 a trd 2 6 | .0334762 .0774876 0.43 0.666 -.1183967 .1853491 H a trd 2 7 | .0141752 .0762326 0.19 0.852 -.1352379 .1635883 I1 a trd 2 8 | .1000364 .0633881 1.58 0.115 -.024202 .2242748 H a trd 2 9 | .0554682 .070813 0.78 0.433 -.0833227 .1942592 I1 a trd 3 0 | .0435711 .0702892 0.62 0.535 -.0941932 .1813354 I1 a trd 3 2 | .0301634 .0601403 0.50 0.616 -.0877095 .1480363 I1 a trd 3 3 j .0913286 .0644115 1.42 0.156 -.0349156 .2175728 I1 a trd 3 4 | .0620483 .0675198 0.92 0.358 -.0702881 .1943847 I1 a trd 3 5 | .1127211 .0662613 1.70 0.089 -.0171486 .2425907 H a trd 3 6 | .0776218 .0611619 1.27 0.204 -.0422533 .197497 H a trd 3 7 i .0925825 .0952628 0.97 0.331 -.0941292 .2792941 I1atrd38 | .081744 .0623055 1.31 0.190 -.0403725 .2038606 H a trd 3 9 | .0110581 .0734436 0.15 0.880 -.1328888 .1550049 _cons | 2.508715 .0524454 47.83 0.000 2.405923 2.611506 . xtgls lnwage realgdp avghhi H atrhhi k_lexp frmSyrte atr lla tr 12atr llatrmed H atrhi Hatrd21 Hatrd22 Hatrd23 I1atrd24 I1atrd25 Hatrd26 Hatrd27 Hatrd28 Hatrd29 I1atrd30 Hatrd32 Hatrd33 I1atrd34 I1atrd35 I1atrd36 I1atrd37 Hatrd38 11 atrd39,force corr(arl) panel(hetero) Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: heteroskedastic Correlation: common AR(1) coefficient for a ll panels (0.5136) Estim ated covariances = 3 Number o f obs = 2650 Estim ated a u to c o rre la tio n s = 1 Number o f groups = 3 Estimated coefficients = 29 Obs per group: min = 755 avg = 919.1887 max = 1135 Wald c h i2 (2 8 ) 502.31 Log lik e lih o o d = 296.1801 Prob > c h i2 = 0.0000 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 177 lnwage | Coef. std. Err. z P*!2! [95% Conf. IntervalJ re a lg d p | -.006833 .0062323 -1.10 0.273 -.0190481 .005382 avghhi | .0001642 .0000234 7.02 0.000 .0001184 .00021 H a tr h h i | -.0000971 .0000781 -1.24 0.214 -.0002502 .0000559 k_lexp | .0475013 .0063314 7.50 0.000 .0350919 .0599107 frm S yrte | .0154626 .0060495 2.56 0.011 .0036059 .0273194 a t r | .0149064 .0125581 1.19 0.235 -.0097071 .0395198 l l a t r | -.2226594 .0723544 -3.08 0.002 -.3644714 -.0808473 1 2 a tr | .0140696 .0124492 1.13 0.258 -.0103304 .0384696 lla trm e d | .2703592 .025039 10.80 0.000 .2212837 .3194347 H a t r h i | .4025619 .022718 17.72 0.000 .3580354 .4470884 H a tr d 2l | .208451 .1636317 1.27 0.203 -.1122612 .5291632 H a tr d 22 | .0665238 .0707985 0.94 0.347 -.0722388 .2052864 H a trd 2 3 | -.0587616 .0662847 -0.89 0.375 -.1886773 .0711541 H a trd 2 4 | .0351724 .0743713 0.47 0.636 -.1105927 .1809375 l1 a trd 2 5 | -.0069764 .0763967 -0.09 0.927 -.1567112 .1427583 I 1 a tr d 26 | .0616834 .0688951 0.90 0.371 -.0733486 .1967154 11atrd27 | .0302531 .0715631 0.42 0.672 -.1100079 .1705141 H a tr d 28 | .1382686 .0539436 2.56 0.010 .032541 .2439961 I1 a trd 2 9 | .0911067 .0641635 1.42 0.156 -.0346514 .2168648 I1 a trd 3 0 | .0577386 .0643081 0.90 0.369 -.0683029 .1837801 I1 a trd 3 2 | .0612462 .0520412 1 .18 0.239 -.0407527 .163245 11atrd33 | .1304481 .0530171 2.46 0.014 .0265364 .2343598 I1 a trd 3 4 | .0915224 .0587131 1 .56 0.119 -.0235531 .2065979 H a trd 3 5 | .1606347 .055561 2.89 0.004 .0517371 .2695324 I1 a trd 3 6 | .1203938 .0522215 2.31 0.021 .0180416 .222746 lla tr d 3 7 | .1541455 .0922778 1.67 0.095 -.0267157 .3350067 H a trd 3 8 | .1220008 .0558075 2.19 0.029 .0126202 .2313814 l1 a trd 3 9 | .0177172 .0657915 0.27 0.788 -.1112318 .1466661 _cons | 2.54644 .0453905 56.10 0.000 2.457476 2.635403 . / ‘ THIS ANALYSIS EVALUATES THE EFFECT OF ATR INDICTMENTS ON EMPLOYMENT by OCCUPATION > USING SEEMINGLY UNRELATED REGRESSION TECHNIQUES. THE VAR OCCYEAR WA? CREATED TO ACCOMMODATE FOR THE INABILITY OF STATA TO COMPUTE THE SIC CODE AND OCPUPATION ^ YEAn VARIATION. THE VAR EMPLOYMT REPRESENTS NUMBER OF EMPLOYEES BY OCCUPATION AND SIC CODE*/ . i l s s ic . ts s e t s ic OCCYEAR pir.el .ariable: sic, 20 to 39 time variable: OCCYEAR, 1 to 147 . / ‘ ANALYSIS ON TWICE LAGGED ATR VARIABLE AND LAGGED ATRHHI*/ . sureg (lnwage = realgdp avghhi Hatrhhi k_lexp frmSyrte atr lla tr l2atr Hatrdoccl Hatrdocc2 Hatrdocc3 Hatrdocc4 HatrdoccS HatrdoccS llatrd2l H atrd22 Hatrd23 11atrd24 Hatrd25 I1atrd26 I1atrd27 I1atrd28 11atrd29 Hatrd30 llatrdC?2 Hatrd33 Hatrd34 Hatrd35 Hatrd36 Hatrd37 Hatrd38 Hatrd39 trend) Seemingly unrelated regression Equation Obs Parms RMSE *R-sq" c h i2 P Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 178 lnwage 2650 33 .2694759 0.4196 1915.538 0.0000 I Coef. Std. Err. z P>|z| [95% Conf. I n t e r v a l] lnwage | realgdp | .0884659 .0349594 2.53 0.011 .0199467 .1569851 avghhi | .0001618 .0000168 9.66 0.000 .000129 .0001947 Hatrhhi | -.0001634 .0000753 -2.17 0.030 -.0003111 - .0000158 k_lexp | .048097 .0046954 10.24 0.000 .0388942 .0572998 frm S yrte | .0123698 .004359 2.84 0.005 .0038263 .0209133 a t r | .0228338 .0123151 1 .85 0.064 -.0013033 .046971 lla tr | -.2638514 .0672352 -3.92 0.000 -.3956299 - .1320729 1 2 a tr | .0105529 .0123411 0.86 0.392 -.0136353 .034741 Hatrdoccl | .3102476 .0290583 10.68 0.000 .2532943 .3672009 l1atrdocc2 | .6030962 .0290583 20.75 0.000 .5461429 .6600494 l1atrdocc3 | .646395 .0290583 22.24 0.000 .5894417 .7033482 l1atrdocc4 | .4434839 .0290583 15.26 0.000 .3865306 .5004372 11atrdocc5 | -.002741 .0290583 -0.09 0.925 -.0596942 .0542123 l1atrdocc6 | - .1513592 .0290583 -5.21 0.000 -.2083125 - .0944059 I1atrd21 | .1759262 .1594319 1 .10 0.270 -.1365546 .488407 I1atrd22 | .0426025 .0669201 0.64 0.524 -.0885586 .1737635 I1atrd23 | -.0242596 .0506377 -0.48 0.632 -.1235077 .0749885 H a trd 2 4 | .0154133 .0642314 0.24 0.810 -.1104779 .1413045 I1atrd25 | -.0068489 .0732034 -0.09 0.925 -.150325 .1366272 H a trd 2 6 | .1217181 .0539962 2.25 0.024 .0158875 .2275488 I1 a trd 2 7 | .042745 .064645 0.66 0.508 -.0839568 .1694469 I1 a trd 2 8 | .1946008 .0397021 4.90 0.000 .1167861 .2724154 I1 a trd 2 9 | .1406006 .0541995 2.59 0.009 .0343716 .2468296 I1 a trd 3 0 | .0707836 .0554268 1 .28 0.202 -.0378508 .1794181 I1 a trd 3 2 | .1030628 .0397013 2.60 0.009 .0252497 .1808759 I1atrd33 | .1687482 .0369941 4.56 0.000 .0962411 .2412554 I1 a trd 3 4 | .1139184 .0461264 2.47 0.014 .0235125 .2043244 H a trd 3 5 | .2163684 .0401736 5.39 0.000 .1376297 .2951071 I1atrd36 | .1778123 .0385176 4.62 0.000 .1023192 .2533054 I1 a trd 3 7 | .2814405 .0865238 3.25 0.001 .111857 .4510241 Hatrd38 | .2023723 .0467506 4.33 0.000 .1107429 .2940017 H a trd 3 9 | .0390438 .0530763 0.74 0.462 -.0649838 .1430714 trend | -.022624 .0073178 -3.09 0.002 -.0369666 - .0082815 _cons | 2.169406 .1480707 14.65 0.000 1.879192 2.459619 . log close lo g : u: \dissertation\atr_logwage_levels_final_gls.log lo g typ e : te x t closed on: 12 Jul 2002, 16:54:08 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 179 APPENDIX C: DIFFERENCE ANALYSIS OF EFFECTS ON EMPLOYMENT LEVELS Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 180 log: u:\dissertation\atr_empchanges_gls.log log type: text opened on: 23 Jul 2002, 11:06:02 . /'T H IS ANALYSIS EVALUATES THE EFFECT OF ATR INDICTMENTS ON EMPLOYMENT BY OCCUPATION USING CROSS-SECTION TIME SERIES TECHNIQUES. THE VAR SICYEAR WAS CREATED TO ACCOMMODATE FOR THE INABILITY OF STATA TO COMPUTE THE SIC CODE AND OCCUPATION AND YEAR VARIATION. THE VAR EMPLOYMT REPRESENTS NUMBER OF EMPLOYEES BY OCCUPATION AND SIC CODE*/ . use 'y:\ypho\diss\sas_programs\datasets\bigatr_final.dta", clear . egen occsic=concat(occ sic) /'concatenates sic and occ variables*/ . egen sicyear=concat(sic year) /'concatenates sic and year variables*/ . egen occyear=concat(occ year) /'concatenates occ and year variables*/ . egen sicyearocc=concat(sic year occ) /'concatenates sic, year, and occ variables*/ . encode occsic, gen(OCCSIC) /'transforms character value to numeric*/ . encode sicyear, gen(SICYEAR) /‘ transforms character value to numeric*/ . encode occyear, gen(OCCYEAR) /'transforms character value to numeric*/ . encode sicye a ro cc, gen(SICYEAROCC) /'tra n s fo rm s c h a ra c te r va lue to n u m e ric*/ . iis occ /'allows one to take the difference without subtracting the 1999*/ . tsset OCCSIC year /'value from 1979 across 2 difference SIC codes*/ panel v a ria b le : OCCSIC, 1 to 140 time variable: year, 1979 to 1999 . replace avghhi=avghhi*lOO (2940 re a l changes made) . gen demploymt=d.employmt (140 missing values generated) . gen davghhi=d.avghhi (140 missing values generated) . gen dk_lexp=d.k_lexp (140 missing values generated) . replace realgdp=realgdp/l00 (2940 real changes made) . gen drealgdp=d.realgdp (140 missing values generated) . replace davghhi=davghhi*10 (2800 re a l changes made) . replace atrhhi=atr*avghhi (1855 real changes made) . gen llatrhhi=llatr*avghhi (140 missing values generated) . gen 1 3 a tr= 1 .1 2 a tr (420 missing values generated) . replace atrtypel =1 if type1n2==l /'the way atrtypel was defined previously was incorrect because atr_case.sas takes the firs t.s ic when sorted by year and sic code to prevent repeated sic and year values, this code uses the typeln2 var which was created in excel to identify sic codes and years with both types of atr case types to change the value to 1 if the first.sic code kept a 0 instead of 1*/ (168 re a l changes made) . replace atrtype2=i if typeln2==i (168 real changes made) . gen I1atrtl=l.atrtypel /'lags antitrust type 1*/ (140 missing values generated) . gen Hatrt2=l.atrtype2 /‘lags antitrust type 2*/ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 181 (140 missing values generated) . gen I2atrt1=l.llatrt1 /‘twice lags antitrust type 1*/ (280 missing values generated) . gen 12atrt2=l.llatrt2 /‘twice lags antitrust type 1*/ (280 missing values generated) . gen lla trt1 h h i= lla trtl‘avghhi /‘interaction between lagged atr type 1 and contemporaneous concentration*/ (140 missing values generated) . gen H atrt2hhi=llatrt2*avghhi /‘ interaction between atr type 2 and contemporaneous concentration*/ (140 missing values generated) . gen workcat=l if occ==5 | occ==6 | occ==7 /‘low wage worker category*/ (1680 missing values generated) . replace workcat=2 if o c c ==1 | occ==4 / ‘ medium wage w orker ca te g o ry*/ (840 re a l changes made) . replace workcat=3 if occ==2 | occ==3 /‘high wage worker category*/ (840 re a l changes made) . / ‘ ORIGINALLY, ATROOCC# VAR WAS EQUAL TO THE PRODUCT OF L1ATR AND DOCC#. NOW THAT WE HAVE TWICE LAGGED AND CONTEMPORANEOUS VARIABLES IN THE ANALYSIS, THE VAR NAME IS CHANGED REFLECT AS SUCH*/ . rename a trd o c c i H a trd o c c l . rename a trd o cc2 H a trd o c c 2 . rename a trd o cc3 H a trd o c c 3 . rename a trd o cc4 l1 a trd o c c 4 . rename a trd o cc5 H a trd o c c S . rename a trd o cc6 lla trd o c c 6 . rename a trdoccT H a trd o c c 7 . / ‘ ORIGINALLY, ATRDSIC# VAR WAS EQUAL TO THE PRODUCT OF L1ATR AND DSIC#. NOW THAT WE HAVE TWICE LAGGED AND CONTEMPORANEOUS VARIABLES IN THE ANALYSIS, THE VAR NAME IS CHANGED REFLECT AS SUCH*/ rename a trd 2 0 I1 a trd 2 0 rename atrd2l Hatrd21 rename a trd 2 2 H a trd 2 2 rename a trd 2 3 H a trd 2 3 r ename atrd 2 4 I1 a trd 2 4 rename a trd 2 5 I1 a trd 2 5 rename a trd 2 6 H a trd 2 6 rename a trd 2 7 H a trd 2 7 rename a trd 2 8 I1 a trd 2 8 rename atrd29 Hatrd29 rename atrd 3 0 H a trd 3 0 rename atrd 3 2 I1 a trd 3 2 rename a trd 3 3 H a trd 3 3 rename atrd 3 4 I1 a trd 3 4 rename a trd 3 5 H a trd 3 5 rename a trd 3 6 H a trd 3 6 rename a trd 3 7 H a trd 3 7 rename a trd 3 8 I1 a trd 3 8 rename atrd39 I1atrd39 gen byte low= w orkcat== gen byte med= w orkcat==l gen byte high== w o rk c a t— Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 182 . gen liatrlow=llatr*low (140 missing values generated) . gen I1atrmed=l1atr*med (140 missing values generated) . gen 11atrhi=llatr*high (140 missing values generated) . label define workfmt 1 “Low wageworker" 2 "Medium Wage Worker" 3 "High Wage Worker" . label define occfmt 1 "Technicians" 2 "Prof Speclty Occ" 3 "Mgrs & Admin" 4 "Sales" 5 "Admin Supp, Cler" 6 "Service" 7 "Prodn" . label define sicfmt 20 "Food & Kindred" 21 "Tobacco Mfrs" 22 "Textil M ill Products" 23 •Apparel & Other Textile Products" 24 ‘ Lumber & Wood Products* 25 "Furniture & Fixtures" 26 "Paper & Allied Products" 27 "Printing & Publishing" 28 'Chemicals & A llied Products" 29 "Petroleum & Coal Products" 30 "Rubber & Misc* 31 "Leather & leather Products" 32 "Stone,Clay.Glass & Concrete" 33 "Primary Metal* 34 "Fabricated Metal* 35 "Industrial Machinery & Equip" 36 "Electrical & Electronic’ 37 "Transportation Equip" 38 "Instruments & R elated" 39 "Misc M frg Inds" . label define atrtypes 1 "Monopoly, Premerger Notification Failure, Acquisitions, Joint Ventures" 2 "Price fixing, Restraint of Trade, Bid Rigging, Territorial Allocation, Restricting Output" . label values occ occfmt . label values workcat workfmt . label values sic sicfmt . i i s occ . ts s e t occ SICYEAR panel variable: occ, 1 to 7 time variable: SICYEAR, 1 to 420 . set m atsize 700 . / ‘ ANALYSIS ON ATR*/ . xtgls demploymt drealgdp davghhi dk_lexp frmSyrte atr H atr 12atr Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: homoscedastic Correlation: no autocorrelation Estimated covariances = 1 Number of obs = 2660 Estim ated a u to c o rre la tio n s = 0 Number o f groups = 7 Estimated coefficients = 8 No. of time periods= 380 Wald c h i2 (7 ) = 94.09 Log lik e lih o o d = -11087.27 Prob > Chi2 = 0.0000 demploymt | Coef. Std. Err. z P>|z| [95% Conf. In te r v a l) drealgdp | 1.389333 .2533221 5.48 0.000 .8928305 1.885835 davghhi | - .0051465 .0008346 -6.17 0.000 -.0067822 -.0035108 dk_lexp | .2379058 .0861591 2.76 0.006 .0690371 . 4067745 frmSyrte | .5068357 .1976681 2.56 0.010 .1194134 .894258 a t r | 1.235061 .673837 1 .83 0.067 -.0856348 2.555757 H a t r | -1.527466 .6751012 -2.26 0.024 -2.85064 -.204292 1 2 a tr | -1.757893 .6664898 -2.64 0.008 -3.064189 - .4515971 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 183 cons | -2.804285 .7265172 -3.86 0.000 -4.228232 -1.380337 . / ‘ creating fl-square see. Greene 1997 p. 509*/ . p re d ic t xm l, xb (280 missing values generated) . gen double residml = demploymt - xml /•th is command obtains the residuals*/ (280 missing values generated) . matrix accum Am1=residm1, noconstant /*this command creates the scalar of (y-Xb)'(y- xb) * / (obs=2660) • 9en y_ybar=demploymt-134.766 /‘this command creates y-ybar*/ (140 missing values generated) . gen sqy_ybar=y_ybar*2 (140 missing values generated) . sum sqy_ybar Variable | Obs Mean Std. Oev. Min Max sqy_ybar | 2800 18722.48 5264.435 1051.964 114085.9 . return list s c a la rs : r(N ) = 2800 r(sum_w) = 2600 r(mean) = 18722.47851497105 r(V a r) = 27714280.49691663 r(s d ) = 5264.435439524036 r(m in ) = 1051.964233398438 r(max) = 114085.8671875 r(sum ) = 52422939.84191895 . scalar sumy=r(sum) . matrix rsql=l-(Am1/sumy) . matrix lis t rsql symmetric rsql[ 1 , 1 ] cl r1 .98760372 . xtgls demploymt drealgdp davghhi H atrhhi dk_lexp frmSyrte atr H atr 12atr Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: homoscedastic Correlation: no autocorrelation Estimated covariances = 1 Number o f obs = 2660 Estimated autocorrelations = 0 Number o f groups = 7 Estimated coefficients = 9 No. o f tim e p e riods= 380 Wald chi2(8) = 94.09 Log lik e lih o o d = -11087.27 Prob > chi2 = 0.0000 demploymt | Coef. Std. Err. P>|z| [95% Conf. Interval) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 184 drealgdp | 1.390047 .253682 5.48 0.000 .8928398 1.887255 davghhi | -.0051598 .0008716 -5 .92 0.000 -.0068681 - .0034515 llatrhhi | .0000632 .0011942 0.05 0.958 -.0022774 .0024038 dk_lexp | .2379941 .0861752 2.76 0.006 .0690938 .4068943 frm5yrte | .5065113 .197763 2.56 0.010 .1189029 .8941198 atr | 1.234202 .6740322 1.83 0.067 - .0868766 2.555281 Hatr | -1.568762 1.032048 -1 .52 0.128 -3.591538 .4540142 I2 a tr | -1.759144 .666909 -2.64 0.008 -3.066262 -.4520267 _cons | -2.80466 .7265513 -3 .8 6 0.000 -4.228674 -1.380645 . p re d ic t xra2, xb (280 missing values generated) . gen double residm2 = demploymt - xm2 (280 missing values generated) . matrix accum Am2=residm2, noconstant (obs=2660) . matrix rsq2=1- (Am2/sumy) . matrix lis t rsq2 symmetric rsq2[1,1) c1 rl .98760373 . xtgls demploymt drealgdp davghhi H atrhhi dk_lexp frmSyrte atr H atr 12atr Hatrdoccl l1atrdocc2 l1atrdocc3 Hatrdocc4 HatrdoccS l1atrdocc6 Cross-sectional tim e-series FGLS regression Coefficients: generalized least squares Panels: homoscedastic Correlation: no autocorrelation Estim ated cova ria n ce s = 1 Number o f obs = 2660 Estimated autocorrelations = 0 Number of groups = 7 Estimated coefficients = 15 No. of time periods= 380 Wald c h i2 ( l4 ) = 124.08 Log lik e lih o o d = -11072.86 Prob > chi2 = 0.0000 demploymt | Coef. Std. Err. z P>|z| [95% Conf. In te r v a l] drealgdp | 1.390047 .2523121 5.51 0.000 .8955247 1.88457 davghhi | -.0051598 .0008669 -5.95 0.000 -.0068589 -.0034607 H a tr h h i | .0000632 .0011878 0.05 0.958 -.0022648 .0023911 dk_lexp | .2379941 .0857099 2.78 0.005 .0700058 .4059823 frm S yrte | .5065113 .1966951 2.58 0.010 .1209959 .8920267 atr | 1.234202 .6703926 1 .84 0.066 -.079743 2.548148 H a tr | -6.38728 1.502664 -4.25 0.000 -9.332448 -3.442112 12atr | -1.759144 .6633078 -2.65 0.008 -3.059204 -.459085 Hatrdoccl | 5.8237 1.67635 3.47 0.001 2.538115 9.109285 Hatrdocc2 | 5.239578 1.67635 3.13 0.002 1.953992 8.525163 Hatrdocc3 | 6.97638 1.67635 4.16 0.000 3.690795 10.26197 Hatrdocc4 | 6.92042 1.67635 4.13 0.000 3.634835 10.20601 H atrdoccS | 2.535913 1.67635 1.51 0.130 -.749672 5.821498 Hatrdocc6 | 6.233636 1.67635 3.72 0.000 2.948051 9.519221 _cons | -2.80466 .7226281 -3.88 0.000 -4.220985 -1.388335 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 185 . p re d ic t xm3, xb (280 missing values generated) . gen double residm 3 = demploymt - xm3 (280 missing values generated) . m a trix accum Am3=residm3, noconstant (obs=2660) . matrix rsq3=l-(Am3/sumy) . matrix lis t rsq3 symmetric rsq3[1,1] c1 r1 .98773725 . xtgls demploymt drealgdp davghhi H atrhhi dk_lexp frmSyrte atr H atr 12atr Hatrdoccl Hatrdocc2 Hatrdocc3 Hatrdocc4 HatrdoccS Hatrdocc6 Hatrd21 Hatrd22 I1atrd23 I1atrd24 Hatrd25 Hatrd26 I1atrd27 I1atrd28 I1atrd29 I1atrd30 I1atrd32 Hatrd33 l1atrd34 Hatrd35 Hatrd36 Hatrd37 Hatrd38 I1atrd39 Cross-sectional tim e-series FGLS regression Coefficients: generalized least squares Panels: homoscedastic Correlation: no autocorrelation Estimated covariances = 1 Number o f obs = 2660 Estimated autocorrelations = 0 Number o f groups = 7 Estimated coefficients = 33 No. o f tim e p e riods= 380 Wald chi2(32) = 152.55 Log likelihood = -11059.33 Prob > chi2 = 0.0000 demploymt | Coef. S td . E rr. z P>|Z| [95% Conf. Interval) drealgdp | 1.432797 .2547369 5.62 0.000 .9335223 1.932072 davghhi | -.0054128 .0008892 -6.09 0.000 -.0071556 -.00367 H a tr h h i | -.009402 .0041982 -2.24 0.025 -.0176304 -.0011737 dk_lexp | .2464127 .0869115 2.84 0.005 .0760692 .4167562 frm S yrte | .6290663 .2267143 2.77 0.006 .1847145 1.073418 a t r | 1.390481 .6992701 1.99 0.047 .0199365 2.761025 H a t r | 2.820763 3.754064 0.75 0.452 -4.537068 10.17859 1 2 a tr | -1.791148 .6949433 -2.58 0.010 -3.153212 -.4290841 Hatrdoccl | 5.8237 1.667843 3.49 0.000 2.554787 9.092613 Hatrdocc2 | 5.239578 1.667843 3.14 0.002 1.970665 8.50849 Hatrdocc3 [ 6.97638 1.667843 4.18 0.000 3.707468 10.24529 Hatrdocc4 | 6.92042 1.667843 4.15 0.000 3.651508 10.18933 HatrdoccS | 2.535913 1.667843 1.52 0.128 -.7329995 5.804826 l1atrdocc6 | 6.233636 1.667843 3.74 0.000 2.964723 9.502549 I1 a trd21 | 24.15337 9.129701 2.65 0.008 6.259489 42.04726 I1atrd22 | -4.446989 3.811322 -1 .17 0.243 -11.91704 3.023064 I1atrd23 | -5.202879 2.866336 -1.82 0.069 -10.82079 .4150364 I1 a trd 2 4 | -7.949569 3.653035 -2.18 0.030 -15.10939 -.7897512 I1atrd25 | -3.223379 4.155425 -0.78 0.438 -11.36786 4.921105 I1 a trd 2 6 | -1.178116 3.029526 -0.39 0.697 -7.115879 4.759647 Hatrd27 | -8.560911 3.648752 -2.35 0.019 -15.71233 -1.409489 I1atrd28 | -1.942163 2.245093 -0.87 0.387 -6.342463 2.458138 I1atrd29 | -4.049608 2.819355 -1 .44 0.151 -9.575441 1.476226 I1atrd30 | -3.285239 3.127511 -1 .05 0.294 -9.415047 2.844569 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 186 I1atrd32 | -1.184061 2.279343 -0.52 0.603 -5.651492 3.28337 I1 a trd 3 3 | -5.102434 2.122387 -2.40 0.016 -9.262236 -.9426324 lla trd 3 4 | -5.51996 2.614576 -2.11 0.035 -10.64443 -.3954858 Hatrd35 | -4.45111 2.288622 -1.94 0.052 -8.936727 .0345072 Hatrd36 | -6.044815 2.211082 -2.73 0.006 -10.37846 -1.711174 Hatrd37 | 8.750894 4.943991 1 .77 0.077 -.939151 18.44094 H a trd 3 8 | -6.486941 2.64699 -2.45 0.014 -11.67495 -1.298935 H a trd 3 9 | -4.090536 2.993001 -1 .37 0.172 -9.95671 1.775638 _cons | -2.90821 .7403394 -3.93 0.000 -4.359249 -1.457172 . p re d ic t xra4, xb (280 missing values generated) . gen double residm4 = demploymt - xm4 (280 missing values generated) . matrix accum Am4=residm4, noconstant (obs=2660) . matrix rsq4=i-(Am4/sumy) . matrix lis t rsq4 symmetric rsq4(l,1] C1 r1 .98786138 . xtgls demploymt drealgdp davghhi Hatrhhi dk_lexp frm5yrte atr H atr 12atr Hatrd2l I1atrd22 llatrd23 11atrd24 I1atrd25 I1atrd26 I1atrd27 Hatrd28 Hatrd29 Hatrd30 11atrd32 I1atrd33 Hatrd34 llatrd35 I1atrd36 Hatrd37 llatrd38 Hatrd39 Cross-sectional tim e-series FGLS regression Coefficients: generalized least squares Panels: homoscedastic Correlation: no autocorrelation Estimated covariances 1 Number o f obs = 2660 Estimated autocorrelations 0 Number o f groups = 7 Estimated coefficients 27 No. o f tim e periods= 380 Wald c h i2 (2 6 ) 121.95 Log lik e lih o o d = -11073.88 Prob > chi2 = 0.0000 demploymt | Coef. Std. Err. Z P>|z| [95% Conf. In te r v a l] drealgdp | 1.432797 .256134 5.59 0.000 .930784 1.934811 davghhi j -.0054128 .0008941 -6.05 0.000 -.0071651 .0036604 H a trh h i | - .009402 .0042213 -2.23 0.026 -.0176756 -.0011285 dk_lexp | .2464127 .0873882 2.82 0.005 .075135 .4176904 frm 5 yrte | .6290663 .2279577 2.76 0.006 .1822774 1.075855 a tr | 1.390481 .7031053 1.98 0.048 .0124197 2.768542 H a tr | 7.639281 3.611473 2.12 0.034 .5609238 14.71764 12atr | -1.791148 .6987548 -2.56 0.010 -3.160682 -.4216139 H a trd 2 1 | 24.15337 9.179773 2.63 0.009 6.16135 42.1454 H a trd 2 2 | -4.446989 3.832225 -1.16 0.246 -11.95801 3.064034 Hatrd23 | -5.202879 2.882056 -1.81 0.071 -10.8516 .4458479 H a trd 2 4 | -7.949569 3.673071 -2.16 0.030 -15.14866 -.7504831 Hatrd25 | -3.223379 4.178216 -0.77 0.440 -11.41253 4.965773 I1atrd26 | -1.178116 3.046142 -0.39 0.699 -7.148444 4.792213 H a trd 2 7 | -8.560911 3.668764 -2.33 0.020 -15.75156 -1 .370267 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 187 H a trd 2 8 ( -1.942163 2.257406 -0.86 0.390 -6.366597 2.482272 H a trd 2 9 | -4.049608 2.834818 -1.43 0.153 -9.605748 1.506533 11atrd30 | -3.285239 3.144664 -1 .04 0.296 -9.448666 2.878188 Hatrd32 | -1.184061 2.291845 -0.52 0.605 -5.675994 3.307872 I1 a trd33 | -5.102434 2.134027 -2.39 0.017 -9.285051 -.9198179 lla trd 3 4 | -5.51996 2.628915 -2.10 0.036 -10.67254 -.3673805 H a trd 3 5 | -4.45111 2.301174 -1.93 0.053 -8.961329 .0591086 llatrd36 l -6.044815 2.223209 -2.72 0.007 -10.40222 -1.687406 H a trd 3 7 | 8.750894 4.971107 1 .76 0.078 -.9922962 18.49408 lla trd 3 8 | -6.486941 2.661508 -2.44 0.015 -11.7034 -1.270482 I1atrd39 | -4.090536 3.009416 -1 .36 0.174 -9.988883 1.807811 _cons | -2.90821 .7443999 -3.91 0.000 -4.367207 -1.449213 . p re d ic t xm5, xb (280 missing values generated) . gen double residm5 = demploymt - xm5 (280 missing values generated) . matrix accum Am5=residm5, noconstant (obs=2660) . m a trix rs q 5 = i- (Am5/sumy) . matrix lis t rsq5 symmetric rsq5[1,11 C1 r 1 .98772787 . summarize demploymt drealgdp davghhi H atrhhi dk_lexp frmSyrte atr H atr 12atr Hatrdoccl Hatrdocc2 l1atrdocc3 Hatrdocc4 Hatrdocc5 Hatrdocc6 Hatrd21 Hatrd22 I1atrd23 I1atrd24 Hatrd25 I1atrd26 I1atrd27 I1atrd28 l1atrd29 Hatrd30 Hatrd32 I1atrd33 I1atrd34 Hatrd35 Hatrd36 11atrd37 Hatrd38 l1atrd39 V a ria b le | Obs Mean S td . Dev. Min Max demploymt | 2800 -1.102224 16.1987 -203 167.2 drealgdp | 2800 1.98185 1.265558 -1.017 3.729 davghhi | 2800 48.40384 361.2067 -1891.933 2224.073 H a tr h h i | 2800 311.7569 439.2919 0 2443.248 dk_lexp | 2800 .0667728 3.502241 -21.1852 20.28664 frm S yrte | 2940 .118852 1.624088 -5.437162 4.879242 a tr | 2940 .452381 .4978119 0 1 H a tr | 2800 .455 .4980598 0 1 12atr | 2660 .4552632 .4980882 0 1 H a trd o c c l | 2800 .065 .2465699 0 1 lla trd o c c 2 | 2800 .065 .2465699 0 1 H a trd o cc3 | 2800 .065 .2465699 0 1 H a trd o c c 4 | 2800 .065 .2465699 0 1 H atrdoccS | 2800 .065 .2465699 0 1 H a trd o cc6 | 2800 .065 .2465699 0 1 I1atrd21 | 2800 .005 .0705463 0 1 Hatrd22 | 2800 .0075 .0862926 0 1 I1 a trd 2 3 | 2800 .0175 .1311484 0 1 I1 a trd 2 4 | 2800 .015 .1215742 0 1 I1 a trd25 | 2800 .0075 .0862926 0 1 H a trd 2 6 | 2800 .0125 .1111223 0 1 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 188 H a trd 2 7 | 2800 .0125 .1111223 0 I1 a trd 2 8 | 2800 .0325 .1773555 0 H a trd 2 9 | 2800 .025 .1561528 0 I1 a trd 3 0 | 2800 .0225 .1483294 0 H a trd 3 2 | 2800 .03 .1706177 0 I1 a trd 3 3 | 2800 .04 .1959942 0 I1 a trd 3 4 | 2800 .04 .1959942 0 I1 a trd 3 5 | 2800 .035 .1838126 0 I1 a trd 3 6 | 2800 .0325 .1773555 0 H a trd 3 7 | 2800 .0325 .1773555 0 H a trd 3 8 | 2800 .025 .1561528 0 H a trd 3 9 | 2800 .0175 .1311484 0 /•ANALYSIS ON TWICE LAGGED ATR VARIABLE WITH AVG WAGE AND LAGGED ATRHHI*/ . iis workcat . ts s e t w orkcat SICYEAROCC panel variable: workcat, 1 to 3 time variable: SICYEAROCC, 1 to 2940, but with gaps . xtgls demploymt drealgdp davghhi dk_lexp frmSyrte atr Jlatr 12atr Cross-sectional tine-series FGLS regression Coefficients: generalized least squares Panels: homoscedastic Correlation: no autocorrelation Estimated covariances = 1 Number of obs = 2660 Estimated autocorrelations = 0 Number o f groups = 3 Estimated coefficients = 8 Obs per group: min = 760 avg = 922.8571 max = 1140 Wald C hi2(7) 94.09 Log lik e lih o o d = -11087.27 Prob > chi2 0.0000 demploymt | C o e f. Std. E rr. z P>|z| (95% Conf. In te r v a l] drealgdp | 1.389333 .2533221 5.48 0.000 .8928305 1.885835 davghhi [ - .0051465 .0008346 -6.17 0.000 -.0067822 -.0035108 dk_lexp | .2379058 .0861591 2.76 0.006 .0690371 .4067745 frm 5 y rte | . 5068357 .1976681 2.56 0.010 .1194134 .894258 a t r | 1.235061 .673837 1.83 0.067 -.0856348 2.555757 H a t r | -1.527466 .6751012 -2.26 0.024 -2.85064 -.204292 1 2 a tr | -1.757893 .6664898 -2.64 0.008 -3.064189 -.4515971 _cons | -2.804285 .7265172 -3.86 0.000 -4.228232 -1.380337 . xtgls demploymt drealgdp davghhi H atrhhi dk_lexp frm5yrte atr H atr 12atr Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: homoscedastic Correlation: no autocorrelation Estim ated covariances = 1 Number o f obs = 2660 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 189 Estim ated a u to c o rre la tio n s = 0 Number o f groups = 3 Estimated coefficients = 9 Obs per group: min = 760 avg = 922.8571 max = 1140 Wald chi2(8) = 94.09 Log lik e lih o o d = -11087.27 Prob > chi2 = 0.0000 demploymt | Coef. Std. E rr. z P *|z | [95% Conf. In te r v a l] drealgdp | 1.390047 .253682 5.48 0.000 .8928398 1.887255 davghhi | -.0051598 .0008716 -5.92 0.000 -.0068681 -.0034515 H a tr h h i | .0000632 .0011942 0.05 0.958 -.0022774 .0024038 dk_lexp | .2379941 .0861752 2.76 0.006 .0690938 .4068943 frm 5 y rte | .5065113 .197763 2.56 0.010 .1189029 .8941198 a t r | 1.234202 .6740322 1.83 0.067 -.0868766 2.555281 H a t r | -1.568762 1.032048 -1.52 0.128 -3.591538 .4540142 1 2 a tr | -1.759144 .666909 -2.64 0.008 -3.066262 - .4520267 _cons | -2.80466 .7265513 -3.86 0.000 -4.228674 -1.380645 . xtgls demploymt drealgdp davghhi H atrhhi dk_lexp frmSyrte atr H atr 12atr Hatrmed H a t r h i Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: homoscedastic Correlation: no autocorrelation Estimated covariances = 1 Number o f obs = 2660 Estimated autocorrelations = 0 Number o f groups = 3 Estimated coefficients = 11 Obs per group: min = 760 avg = 922.8571 max = 1140 Wald chi2(l0) = 107.96 Log lik e lih o o d = -11080.59 Prob > chi2 = 0.0000 demploymt | Coef. Std. E rr. z P>|z| [95% Conf. In te r v a l] drealgdp | 1.390047 .2530457 5.49 0.000 .8940869 1.886008 davghhi | -.0051598 .0008694 -5.93 0.000 -.0068638 -.0034558 H a tr h h i | .0000632 .0011912 0.05 0.958 -.0022715 .0023979 dk_lexp | .2379941 .0859591 2.77 0.006 .0695174 .4064707 frm S yrte | .5065113 .197267 2.57 0.010 .1198751 .8931476 a t r | 1.234202 .6723416 1 .84 0.066 -.0835631 2.551968 H a t r | -3.464097 1.152813 -3.00 0.003 -5.723569 •1.204625 1 2 a tr | -1.759144 .6652362 -2.64 0.008 -3.062983 -.4553053 H atrm ed | 3.448877 1.085225 3.18 0.001 1.321875 5.575879 H a t r h i | 3.184796 1.085225 2.93 0.003 1.057794 5.311798 _cons | -2.80466 .724729 -3.87 0.000 -4.225102 -1.384217 . xtgls demploymt drealgdp davghhi H atrhhi dk_lexp frm5yrte atr H atr 12atr Hatrmed H atrhi I1atrd21 Hatrd22 l1atrd23 I1atrd24 I1atrd25 Hatrd26 llatrd27 Hatrd28 I1atrd29 Hatrd30 I1atrd32 l1atrd33 l1atrd34 I1atrd35 Hatrd36 I1atrd37 I1atrd38 I1atrd39 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 190 Cross-sectional tirae-series FGLS regression Coefficients: generalized least squares Panels: homoscedastic Correlation: no autocorrelation Estimated covariances = 1 Number o f obs = 2660 Estimated autocorrelations = 0 Number o f groups = 3 Estimated coefficients = 29 Obs per group: min = 760 avg = 922.8571 max = 1140 Wald chi2(28) = 136.10 Log lik e lih o o d = -11067.13 Prob > chi2 = 0.0000 demploymt | C oef. Std. E rr. z P >|z| [95% Conf. In te r v a l] drealgdp | 1.432797 .255485 5.61 0.000 .9320559 1.933539 davghhi | -.0054128 .0008918 -6.07 0.000 -.0071607 -.0036648 Hatrhhi | -.009402 .0042106 -2.23 0.026 -.0176546 -.0011495 dk_lexp | .2464127 .0871668 2.83 0.005 .0755689 .4172565 frm 5 yrte | .6290663 .2273801 2.77 0.006 .1834094 1.074723 a t r | 1.390481 .7013239 1.98 0.047 .0159112 2.76505 H a t r | 5.743946 3.639123 1.58 0.114 1.388603 12.8765 1 2 a tr | -1.791148 .6969844 -2.57 0.010 -3.157212 -.4250837 H atrm ed | 3.448877 1.07975 3.19 0.001 1.332606 5.565148 H a tr h i | 3.184796 1.07975 2.95 0.003 1.068524 5.301067 H a trd 2 l | 24.15337 9.156515 2.64 0.008 6.206935 42.09981 Hatrd22 | -4.446989 3.822515 -1.16 0.245 -11.93898 3.045004 I1 a trd 2 3 | -5.202879 2.874754 -1.81 0.070 -10.83729 .4315362 l1 a trd 2 4 | -7.949569 3.663764 -2.17 0.030 -15.13042 -.7687228 I1atrd25 | -3.223379 4.16763 -0.77 0.439 -11.39178 4.945025 I1atrd26 | -1.178116 3.038424 -0.39 0.698 -7.133318 4.777086 I1 a trd 2 7 | -8.560911 3.659468 -2.34 0.019 -15.73334 -1 .388485 H a trd 2 8 | -1.942163 2.251687 -0.86 0.388 -6.355387 2.471062 I1atrd29 | -4.049608 2.827635 -1.43 0.152 -9.591671 1.492456 H a trd 3 0 | -3.285239 3.136696 -1.05 0.295 -9.433051 2.862572 I1 a trd 3 2 | -1.184061 2.286038 -0.52 0.604 -5.664613 3.296491 H a trd 3 3 | -5.102434 2.12862 -2.40 0.017 -9.274454 -.930415 I1 a trd 3 4 | -5.51996 2.622255 -2.11 0.035 -10.65948 -.3804352 I1 a trd 3 5 | -4.45111 2.295344 -1.94 0.052 -8.949902 .0476814 Hatrd36 | -6.044815 2.217576 -2.73 0.006 -10.39118 -1.698446 Hatrd37 | 8.750894 4.958512 1.76 0.078 -.9676106 18.4694 I1atrd38 | -6.486941 2.654764 -2.44 0.015 -11.69018 -1.283698 H a trd 3 9 | -4.090536 3.001791 -1.36 0.173 -9.973939 1.792866 _cons | -2.90821 .7425138 -3.92 0.000 -4.363511 -1.45291 . /'T H IS ANALYSIS EVALUATES THE EFFECT OF ATR INDICTMENTS ON EMPLOYMENT BY OCCUPATION > USING SEEMINGLY UNRELATED REGRESSION TECHNIQUES. THE VAR OCCYEAR WAS CREATED TO ACCOMMODATE FOR THE INABILITY OF STATA TO COMPUTE THE SIC CODE AND OCCUPATION AND YEAR VARIATION. THE VAR EMPLOYMT REPRESENTS NUMBER OF EMPLOYEES BY OCCUPATION AND SIC CODE*/ . i i s s ic . ts s e t s ic OCCYEAR panel v a ria b le : s ic , 20 to 39 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 191 tim e v a ria b le : OCCYEAR, 1 to 147 . / ‘ ANALYSIS ON TWICE LAGGED ATR VARIABLE AND LAGGED ATRHHI*/ . sureg (demploymt = drealgdp davghhi H atrhhi dk_lexp frmSyrte atr H atr 12atr llatrdocd l1atrdocc2 l1atrdocc3 l1atrdocc4 HatrdoccS Hatrdocc6 Hatrd21 I1atrd22 Hatrd23 Hatrd24 llatrd25 l1atrd26 I1atrd27 llatrd28 I1atrd29 I1atrd30 I1atrd32 Hatrd33 l1atrd34 I1atrd35 I1atrd36 I1atrd37 Hatrd38 I1atrd39) Seemingly unrelated regression Equation Obs Parms RMSE -R -s q ‘ chi2 P demploymt 2660 32 15.46694 0. 0542 152 .5506 0.0000 I Coef. Std. Err z P>|Z| [95% Conf. Interval] demploymt | drealgdp | 1.432797 .2547369 5.62 0.000 .9335223 1.932072 davghhi | -.0054128 .0008892 -6.09 0.000 -.0071556 -.00367 Hatrhhi | -.009402 .0041982 -2.24 0.025 -.0176304 -.0011737 dk_lexp | .2464127 .0869115 2.84 0.005 .0760692 .4167562 frmSyrte | .6290663 .2267143 2.77 0.006 .1847145 1.073418 a t r | 1.390481 .6992701 1.99 0.047 .0199365 2.761025 Hatr | 2.820763 3.754064 0.75 0.452 -4.537068 10.17859 12atr | -1.791148 .6949433 -2.58 0.010 -3.153212 -.4290841 llatrdocd | 5.8237 1.667843 3.49 0.000 2.554787 9.092613 Hatrdocc2 | 5.239578 1.667843 3.14 0.002 1.970665 8.50849 llatrdocc3 | 6.97638 1.667843 4.18 0.000 3.707468 10.24529 l1atrdocc4 | 6.92042 1.667843 4.15 0.000 3.651508 10.18933 HatrdoccS | 2.535913 1.667843 1 .52 0.128 -.7329995 5.804826 l1atrdocc6 | 6.233636 1.667843 3.74 0.000 2.964723 9.502549 H a tr d 21 | 24.15337 9.129701 2.65 0.008 6.259489 42.04726 H a trd 2 2 | -4.446989 3.811322 -1.17 0.243 -11.91704 3.023064 I1 a trd 2 3 | -5.202879 2.866336 -1.82 0.069 -10.82079 .4150364 Hatrd24 | -7.949569 3.653035 -2.18 0.030 -15.10939 -.7897512 I1 a trd 2 5 | -3.223379 4.155425 -0.78 0.438 -11 .36786 4.921105 I1atrd26 | -1.178116 3.029526 -0.39 0.697 -7.115879 4.759647 I1 a trd 2 7 | -8.560911 3.648752 -2.35 0.019 -15.71233 -1.409489 H a trd 2 8 | -1.942163 2.245093 -0.87 0.387 -6.342463 2.458138 H a trd 2 9 | -4.049608 2.819355 -1.44 0.151 -9.575441 1.476226 I1 a trd 3 0 | -3.285239 3.127511 -1.05 0.294 -9.415047 2.844569 l1 a trd 3 2 | -1.184061 2.279343 -0.52 0.603 -5.651492 3.28337 l1 a trd 3 3 | -5.102434 2.122387 -2.40 0.016 -9.262236 -.9426324 I1 a trd 3 4 | -5.51996 2.614576 -2.11 0.035 -10.64443 -.3954858 I1 a trd 3 5 | -4.45111 2.288622 -1.94 0.052 -8.936727 .0345072 Hatrd36 | -6.044815 2.211082 -2.73 0.006 -10.37846 -1.711174 l1atrd37 | 8.750894 4.943991 1.77 0.077 -.939151 18.44094 I1 a trd 3 8 | -6.486941 2.64699 -2.45 0.014 -11 .67495 -1.298935 H a trd 3 9 | -4.090536 2.993001 -1.37 0.172 -9.95671 1.775638 _cons | -2.90821 .7403394 -3.93 0.000 -4.359249 -1.457172 . log close log: u:\dissertation\atr_empchanges_gls.log Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 192 log type: text closed on: 23 Jul 2002, 11:06:10 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 193 APPENDIX D: DIFFERENCE ANALYSIS OF EFFECTS ON THE NATURAL LOG OF AVERAGE WAGES Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 194 log: u:\dissertation\atr_logwage_changes_gls.log log type: text opened on: 23 J u l 2002, 11:53:48 . /'T H IS ANALYSIS EVALUATES THE EFFECT OF ATR INDICTMENTS ON EMPLOYMENT BY OCCUPATION > USING CROSS-SECTION TIME SERIES TECHNIQUES. THE VAR SICYEAR WAS CREATED TO ACCOMMODATE FOR THE INABILITY OF STATA TO COMPUTE THE SIC CODE AND OCCUPATION AND YEAR VARIATION. THE VAR EMPLOYMT REPRESENTS NUMBER OF EMPLOYEES BY OCCUPATION AND SIC COOE * / . use ■y:\ypho\diss\sas_programs\datasets\bigatr_final.dta', clear . egen occsic=concat(occ sic) /‘concatenates sic and occ variables*/ . egen sicyear=concat(sic year) /‘concatenates sic and year variables*/ . egen occyear=concat(occ year) /‘concatenates occ and year variables*/ . egen sicyearocc=concat(sic year occ) /‘concatenates sic, year, and occ variables*/ . encode occsic, gen(OCCSIC) /‘transforms character value to numeric*/ . encode sicyear, gen(SICYEAR) / ‘ transforms character value to numeric*/ . encode occyear, gen(OCCYEAR) /‘ transforms character value to numeric*/ . encode sicyearocc, gen(SICYEAROCC) / ‘ transforms character value to numeric*/ . i i s occ /•allows one to take the difference without subtracting the 1999*/ . ts s e t OCCSIC year /•v a lu e from 1979 across 2 d iffe re n c e SIC codes*/ panel variable: OCCSIC, 1 to 140 time variable: year, 1979 to 1999 . / ‘ c o n v e rtin g c u rre n t wage to wages in 1996 co n sta n t d o lla r s * / . gen wage=(wavgwg/(74/l56.9)) if year==l979 (2801 m issing va lu e s generated) . replace wage=(wavgwg/(82.4/156.9)) if year==l980 (139 re a l changes made) . replace wage=(wavgwg/(90.9/156.9)) if year==l98l (139 re a l changes made) . replace wage=(wavgwg/(96.5/156.9)) if year==1982 (139 re a l changes made) . replace wage=(wavgwg/(99.6/156.9)) if year==1983 (140 re a l changes made) . replace wage=(wavgwg/(103.9/156.9)) if year==1984 (140 re a l changes made) . replace wage=(wavgwg/(107.6/156.9)) if year==l985 (140 re a l changes made) . replace wage=(wavgwg/(109.6/156.9)) if year==1986 (139 re a l changes made) . replace wage=(wavgwg/(113.6/156.9)) if year==l987 (140 re a l changes made) . replace wage=(wavgwg/(H8.3/156.9)) if year==1988 (139 re a l changes made) . replace wage=(wavgwg/(124/156.9)) if year==l989 (140 re a l changes made) . replace wage=(wavgwg/(i30.7/l56.9)) if year==l990 (140 re a l changes made) . replace wage=(wavgwg/(i36.2/l56.9)) if year==l99l (139 re a l changes made) . replace wage=(wavgwg/(140.3/156.9)) if year==1992 (140 re a l changes made) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 195 . replace wage=(wavgwg/(l44.5/l56.9)) if year==l993 (140 re a l changes made) . replace wage=(wavgwg/(148.2/156.9)) if year==1994 (140 re a l changes made) . replace wage=(wavgwg/(l52.4/i56.9)) if year==l995 (140 re a l changes made) . replace wage=(wavgwg/(l56.9/i56.9)) if year==l996 (138 real changes made) . replace wage=(wavgwg/(160.5/156.9)) if year==1997 (139 re a l changes made) . replace wage=(wavgwg/(163/156.9)) if year==1998 (139 re a l changes made) . replace wage=(wavgwg/(166.6/156.9)) if year==l999 (139 re a l changes made) . replace avghhi=avghhi*l00 (2940 re a l changes made) . gen lnwage=log(wage) (12 missing values generated) . gen dlnwage=d.Inwage (159 missing values generated) . replace dlnwage=dlnwage*l00 (2781 re a l changes made) . gen davghhi=d.avghhi (140 missing values generated) . replace k_lexp=k_lexp/lO (2940 re a l changes made) . gen dk_lexp=d.k_lexp (140 missing values generated) . replace realgdp=realgdp/1000 (2940 re a l changes made) . gen drealgdp=d.realgdp (140 missing values generated) . replace drealgdp=drealgdp*10 (2800 re a l changes made) . replace davghhi=davghhi*10 (2800 re a l changes made) . replace atrhhi=atr*avghhi (1855 re a l changes made) . gen Hatrhhi=Hatr*avghhi (140 missing values generated) . replace atrtypel=i if typeln2==1 /‘the way atrtypel was defined previously was incorrect because atr_case.sas takes the firs t.s ic when sorted by year and sic code to prevent repeated sic and year values, this code uses the typeln2 var which wascreated in excel to identify sic codes and years with both types of atr case types to change the value to 1 if the first.sic code kept a 0 instead of 1*/ (168 re a l changes made) . replace atrtype2=l if typeln2==l (168 re a l changes made) . gen Hatrtl=l.atrtypel /‘lags antitrust type 1*/ (140 missing values generated) . gen I1atrt2=l.atrtype2 /'lags antitrust type 2*/ (140 missing values generated) . gen 12atrtl=l.Hatrtl /‘twice lags antitrust type 1*/ (280 missing values generated) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 196 . gen 12atrt2=l.llatrt2 /•twice lags antitrust type 1 */ (280 missing values generated) . gen I1atrtlhhi=llatrtl*avghhi /•interaction between lagged atr type 1 and contemporaneous concentration*/ (140 missing values generated) . gen Hatrt2hhi=Hatrt2*avghhi /•interaction between atr type 2 and contemporaneous concentration*/ (140 missing values generated) . gen workcat=1 if occ==5 | occ==6 | occ==7 /‘low wage worker category*/ (1680 missing values generated) . replace w orkcat=2 i f occ==1 | occ==4 / ‘ medium wage w orker c a te g o ry */ (840 re a l changes made) . replace workcat=3 if occ==2 | occ==3 /‘high wage worker category*/ (840 re a l changes made) . / ‘ ORIGINALLY, ATRDOCC# VAR WAS EQUAL TO THE PRODUCT OF L1ATR AND DOCC#. NOW THAT WE HAVE TWICE LAGGED AND CONTEMPORANEOUS VARIABLES IN THE ANALYSIS, THE VAR NAME IS CHANGED REFLECT AS SUCH*/ . rename a trd o c c l H a trd o c c l . rename a trd o cc2 H a trd o c c 2 . rename a trd o cc3 H a trd o c c 3 . rename a trd o cc4 H a trd o c c 4 . rename a trd o c c 5 H a trd o c c S . rename a trd o cc6 H a trd o c c 6 . rename a trd o cc7 H a trd o c c 7 /•ORIGINALLY, ATRDSIC# VAR WAS EQUAL TO THE PRODUCT OF LIATR AND DSIC#. NOW THAT WE HAVE TWICE LAGGED AND CONTEMPORANEOUS VARIABLES IN THE ANALYSIS, THE VAR NAME IS CHANGED REFLECT AS SUCH*/ . rename atrd20 Hatrd20 . rename atrd21 H a trd 2 1 . rename a trd 2 2 H a trd 2 2 . rename a trd 2 3 H a trd 2 3 . rename atrd24 I1atrd24 . rename a trd 2 5 I1 a trd 2 5 . rename a trd 2 6 I1 a trd 2 6 . rename atrd27 I1atrd27 . rename a trd 2 8 H a trd 2 8 . rename atrd29 I1atrd29 . rename atrd30 I1atrd30 . rename a trd 3 2 I1 a trd 3 2 . rename atrd33 I1atrd33 . rename atrd34 I1atrd34 . rename atrd35 I1atrd35 . rename atrd36 I1atrd36 . rename atrd37 Hatrd37 . rename a trd 3 8 I1 a trd 3 8 . rename a trd 3 9 H a trd 3 9 . gen byte low= w orkcat==l . gen byte med= workcat==2 . gen byte high= workcat==3 . gen Hatrlow=llatr*low (140 missing values generated) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 197 . gen I1atrmed=l1atr*med (140 missing values generated) . gen I1atrhi=liatr*high (140 missing values generated) . label define workfmt 1 “Low wage worker" 2 ‘Medium Wage Worker" 3 "High Wage Worker" . label define occfmt 1 "Technicians" 2 "Prof Speclty Occ" 3 “Mgrs & Admin" 4 "Sales* 5 "Admin Supp, C le r" 6 "S e rv ic e " 7 "Prodn" . label define sicfmt 20 "Food & Kindred" 21 "Tobacco Mfrs" 22 "Textil M ill Products* 23 "Apparel & Other Textile Products" 24 "Lumber & Wood Products* 25 "Furniture & Fixtures" 26 "Paper & Allied Products’ 27 "Printing & Publishing" 28 "Chemicals & A llied Products* 29 "Petroleum & Coal Products" 30 "Rubber & Misc" 31 "Leather & leather Products" 32 "Stone,Clay,Glass & Concrete* 33 "Primary Metal" 34 "Fabricated Metal" 35 "Industrial Machinery & Equip" 36 "E lectrical & Electronic" 37 "Transportation Equip* 38 "Instruments & Related" 39 "Misc Mfrg Inds* . label define atrtypes 1 "Monopoly, Premerger Notification Failure, Acquisitions, Joint Ventures" 2 "Price fixing, Restraint of Trade, Bid Rigging, T erritorial Allocation, Restricting Output" . label values occ occfmt . label values workcat workfmt . label values sic sicfmt . save "y:\ypho\diss\sas_programs\datasets\atrocc_wg.dta", replace file y:\ypho\diss\sas_programs\datasets\atrocc_wg.dta saved use 'y :\ypho\diss\sas_programs\datasets\atrocc_wg.dta" i i s occ ts s e t occ SICYEAR panel variable: occ, 1 to 7 tim e v a ria b le : SICYEAR, 1 to 420 set m atsize 700 /•ANALYSIS ON LAGGED ATRHHI AND ATR CASE TYPES*/ xtgls dlnwage drealgdp davghhi dk_lexp frm5yrte atr H atr 12atr Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: homoscedastic Correlation: no autocorrelation Estimated covariances = 1 Number o f obs = 2643 Estimated autocorrelations = 0 Number o f groups = 7 Estimated coefficients = 8 Obs per group: min = 370 avg = 377.6122 max = 380 Wald c h i2 (7 ) = 15.06 Log likelihood = -10903.37 Prob > chi2 = 0.0352 dlnwage | Coef. S td . E rr. z P>|z| [95% C onf. In te r v a l] drealgdp | .399531 .2436502 1 .64 0.101 -.0780147 .8770767 davghhi | .000977 .0008034 1 .22 0.224 -.0005977 .0025517 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 198 dk_lexp | -2.296497 .8321691 -2.76 0.006 -3.927518 -.6654756 frm 5 yrte | .0728794 .1919606 0.38 0.704 -.3033565 .4491154 atr | -.8693626 .6461013 -1.35 0.178 -2.135698 .3969728 Hatr | -.5457432 .6472736 -0.84 0.399 -1.814376 .7228898 12atr | .498571 .6389365 0.78 0.435 -.7537216 1.750864 _cons | -.420862 .6989572 -0.60 0.547 -1.790793 .9490689 . /'creating R-square see. Greene 1997 p. 509*/ . p re d ic t xm l, xb (280 missing values generated) . gen double residml = dlnwage - xml /•th is command obtains the residuals*/ (297 missing values generated) . matrix accum Am1=residml, noconstant /’this command creates the scalar of (y-Xb)’(y- Xb ) • / (ObS=2643) • 9en y_ybar=dlnwage -134.766 /'this command creates y-ybar*/ (159 missing values generated) . gen sqy_ybar=y_ybar*2 (159 missing values generated) . sum sqy_ybar Variable | Obs Mean S td. Dev. Min Max sqy_ybar | 2781 18407.24 4060.423 733.4113 59044.25 return lis t sc a la rs : r(N ) 2781 r(sum_w) 2781 r(mean) 18407.24168989119 r(V a r) 16487034.28820921 r(s d ) 4060.422919870443 r(m in ) 733.4113159179688 r(m ax) 59044.25390625 r(sum ) 51190539.1395874 . scalar sumy=r(sum) . matrix rsq1=l - (Am1/sumy) . matrix lis t rsql symmetric rsq l[1,1] c1 r1 .98842049 . xtgls dlnwage drealgdp davghhi Hatrhhi dklexp frmSyrte atr H atr 12atr Cross-sectional tim e-series FGLS regression Coefficients: generalized least squares Panels: homoscedastic Correlation: no autocorrelation Estimated covariances = 1 Number o f obs = 2643 Estim ated a u to c o rre la tio n s = 0 Number o f groups = 7 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 199 Estimated coefficients = 9 Obs per group: min = 370 avg = 377.6122 max = 380 Wald c h i2 (8 ) = 15.07 Log lik e lih o o d = -10903.37 Prob > c h i2 = 0.0578 dlnwage | Coef. Std. Err. z P >|z| (95% Conf. In te r v a l] drealgdp | .4007645 .2440072 1.64 0.101 -.0774808 .8790098 davghhi j .0009543 .0008394 1.14 0.256 - .0006909 .0025996 H a tr h h i | .000107 .0011448 0.09 0.926 -.0021368 .0023507 dk_lexp | -2.294848 .8323549 -2.76 0.006 -3.926233 -.6634621 frm S yrte | .0723072 .192058 0.38 0.707 -.3041194 .4487339 a tr | -.8707915 .6462812 -1.35 0.178 -2.137479 .3958964 H a t r | - .6156929 .9896226 -0.62 0.534 -2.555318 1.323932 1 2 a tr | .4964524 .6393376 0.78 0.437 -.7566262 1.749531 cons | -.4215146 .6989909 -0.60 0.546 -1.791512 .9484824 . p re d ic t xm2, xb (280 missing values generated) . gen double residm2 = dlnwage - xm2 (297 missing values generated) . matrix accum Am2=residm2, noconstant (obs=2643) . matrix rsq2=1-(Am2/sumy) . matrix lis t rsq2 symmetric rsq2(1,1J c1 r1 .98842053 . xtgls dlnwage drealgdp davghhi Hatrhhi dk_lexp frmSyrte atr H atr 12atr llatrdocd Hatrdocc2 Hatrdocc3 Hatrdocc4 HatrdoccS Hatrdocc6 Cross-sectional tim e-series FGLS regression Coefficients: generalized least squares Panels: homoscedastic Correlation: no autocorrelation Estimated covariances = 1 Number o f obs = 2643 Estimated autocorrelations = 0 Number o f groups = 7 Estimated coefficients = 15 Obs per group: min = 370 avg = 377.6122 max = 380 Wald chi2(14) = 16.40 Log lik e lih o o d = -10902.71 Prob > chi2 = 0.2898 dlnwage | Coef. S td. E rr. z P>|Z| (95% Conf. In t e r v a l] drealgdp | .4007645 .2439463 1.64 0.100 -.0773615 .8788905 davghhi | .0009543 .0008392 1 .14 0.255 -.0006905 .0025991 H a tr h h i | .000107 .0011445 0.09 0.926 -.0021362 .0023502 dk_lexp | -2.294848 .8321473 -2.76 0.006 -3.925826 -.6638688 frm S yrte | .0723072 .1920101 0.38 0.706 -.3040256 .4486401 atr | -.8707915 .6461201 -1.35 0.178 -2.137164 .3955806 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 200 H a t r | -.9974167 1.447742 -0.69 0.491 -3.834939 1.840106 1 2 a tr | .4964524 .6391782 0.78 0.437 -.7563138 1.749219 llatrdocd | 1.01741 1.614481 0.63 0.529 -2.146915 4.181736 Hatrdocc2 | 1.063287 1.614481 0.66 0.510 -2.101039 4.227612 Hatrdocc3 | .4682526 1.614481 0.29 0.772 -2.696073 3.632578 Hatrdocc4 | -.4380183 1.614481 -0.27 0.786 -3.602344 2.726307 HatrdoccS | .2353341 1.614481 0.15 0.884 -2.928991 3.39966 Hatrdocc6 | .3258018 1.614481 0.20 0.840 -2.839524 3.490127 cons | -.4215146 .6988166 -0.60 0.546 -1.79117 .9481408 . p re d ic t xoi3, xb (280 missing values generated) . gen double residm3 = dlnwage - xm3 (297 missing values generated) . m a trix accum Am3=residm3, noconstant (obs=2643) . matrix rsq3=1-(Am3/sumy) . matrix lis t rsq3 symmetric rsq3(1,1] c1 r1 .9884263 . xtgls dlnwage drealgdp davghhi Hatrhhi dk_lexp frmSyrte atr H atr 12atr llatrdocd l1atrdocc2 Hatrdocc3 l1atrdocc4 HatrdoccS Hatrdocc6 I1atrd2l Hatrd22 Hatrd23 Hatrd24 I1atrd25 Hatrd26 I1atrd27 l1atrd28 llatrd29 I1atrd30 I1atrd32 Hatrd33 Hatrd34 I1atrd35 Hatrd36 Hatrd37 l1atrd38 I1atrd39 Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: homoscedastic Correlation: no autocorrelation Estimated covariances = 1 Number o f obs = 2643 Estimated autocorrelations = 0 Number o f groups = 7 Estimated coefficients = 33 Obs per group: min = 370 avg = 377.6122 max = 380 Wald Chi2(32) = 24.95 Log lik e lih o o d = -10898.46 Prob > chi2 = 0.8081 dlnwage | Coef. S td. E rr. z P>|z| [95% Conf. In te r v a l] d re a lg d p | .4458835 .2471611 1.80 0.071 -.0385434 .9303105 davghhi | .0012441 .000864 1.44 0.150 -.0004493 .0029375 l la t r h h i | -.0008856 .0040581 -0.22 0.827 -.0088393 .007068 dk_lexp | -2.096365 .8465899 -2.48 0.013 -3.755651 -.4370797 frm S y rte | .0574468 .2227961 0.26 0.797 -.3792256 .4941191 a t r | -.883736 .676382 -1.31 0.191 -2.20942 .4419484 H a t r | .1304534 3.629649 0.04 0.971 -6.983529 7.244436 1 2 a tr | .5285678 .6719907 0.79 0.432 -.7885098 1.845646 llatrdocd | 1.01741 1.611892 0.63 0.528 -2.141839 4.176659 Hatrdocc2 | 1.063287 1.611892 0.66 0.509 -2.095963 4.222536 Hatrdocc3 | .4682526 1.611892 0.29 0.771 -2.690997 3.627502 !1atrdocc4 | -.4380183 1.611892 -0.27 0.786 -3.597268 2.721231 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. HatrdoccS | .2353341 1.611892 0.15 0.884 -2.923915 3.394583 Hatrdocc6 | .3258018 1.611892 0.20 0.840 -2.833448 3.485051 H a trd 2 1 | 8.060634 8.825146 0.91 0.361 -9.236334 25.3576 I1 a trd 2 2 | .8172924 3.683801 0.22 0.824 -6.402825 8.03741 H a trd 2 3 | -.4114367 2.770504 -0.15 0.882 -5.841525 5.018651 I1 a trd 2 4 | .6958498 3.531275 0.20 0.844 -6.225322 7.617022 H a trd 2 5 | -5.728247 4.017793 -1.43 0.154 -13.60298 2.146483 H a trd 2 6 | -2.284511 2.928247 -0.78 0.435 -8.023769 3.454748 lla trd 2 7 | -1.031324 3.528125 -0.29 0.770 -7.946322 5.883674 11atrd28 | -.1469463 2.170077 -0.07 0.946 -4.400219 4.106327 H a trd 2 9 | -2.743606 2.724821 -1.01 0.314 -8.084157 2.596945 llatrd30 | -.0316161 3.02471 -0.01 0.992 -5.959939 5.896707 H a trd 3 2 | -1.084338 2.203068 -0.49 0.623 -5.402272 3.233595 I1 a trd 3 3 | -.7344953 2.051266 -0.36 0.720 -4.754902 3.285911 I1atrd34 | -.5303141 2.527526 -0.21 0.834 -5.484175 4.423546 H a trd 3 5 | -.9594432 2.212467 -0.43 0.665 -5.295798 3.376912 H a trd 3 6 | .6154846 2.137239 0.29 0.773 -3.573427 4.804396 I1 a trd 3 7 ] -.4679754 4.778521 -0.10 0.922 -9.833705 8.897755 11atrd38 | 1.547217 2.562832 0.60 0.546 -3.475841 6.570274 I1atrd39 | -.7663889 2.893573 -0.26 0.791 -6.437688 4.904911 cons | -.5293822 .718051 -0.74 0.461 -1.936736 .877972 . predict xm4, xb (280 missing values generated) . gen double residm4 = dlnwage - xm4 (297 missing values generated) . matrix accum Am4=residm4, noconstant (obs=2643) . matrix rsq4=l -(Am4/sumy) . matrix lis t rsq4 symmetric rsq4(1,1) cl r1 .98846341 . xtgls dlnwage drealgdp davghhi Hatrhhi dk_lexp frmSyrte atr H atr 12atr llatrd21 l1atrd22 I1atrd23 I1atrd24 I1atrd25 I1atrd26 Hatrd27 Hatrd28 Hatrd29 I1atrd30 I1atrd32 Hatrd33 Hatrd34 Hatrd35 11atrd36 I1atrd37 Hatrd38 I1atrd39 Cross-sectional tim e-series FGLS regression Coefficients: generalized least squares Panels: homoscedastic Correlation: no autocorrelation Estimated covariances = 1 Number o f obs = 2643 Estimated autocorrelations = 0 Number o f groups = 7 Estimated coefficients = 27 Obs per group: min = 370 avg = 377.6122 max = 380 Wald ch i2 (2 6 ) = 23.62 Log lik e lih o o d = -10899.13 Prob > chi2 = 0.5980 dlnwage | Coef. Std. Err. z P>|z| (95% Conf. Interval] drealgdp I .4458835 .247223 1.80 0.071 -.0386646 .9304317 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 202 davghhi | .0012441 .0008642 1 .44 0.150 -.0004497 .0029379 Hatrhhi | -.0008856 .0040591 -0.22 0.827 -.0088413 .00707 dk_lexp | -2.096365 .8468017 -2.48 0.013 -3.756066 - .4366646 frmSyrte | .0574468 .2228519 0.26 0.797 -.3793348 .4942284 atr | -.883736 .6765512 -1.31 0.191 -2.209752 .4422801 Hatr | .5121772 3.473741 0.15 0.883 -6.296231 7.320585 12atr | .5285678 .6721589 0.79 0.432 -.7888394 1.845975 H a tr d 21 | 8.060634 8.827354 0.91 0.361 -9.240662 25.36193 11atrd22 | .8172924 3.684723 0.22 0.824 -6.404632 8.039217 I1 a trd 2 3 | -.4114367 2.771197 -0.15 0.882 -5.842883 5.02001 Hatrd24 | .6958498 3.532159 0.20 0.844 -6.227054 7.618754 I1 a trd 2 5 | -5.728247 4.018798 -1.43 0.154 -13.60495 2.148453 H a trd 2 6 | -2.284511 2.928979 -0.78 0.435 -8.025205 3.456184 I1 a trd 2 7 | -1.031324 3.529008 -0.29 0.770 -7.948053 5.885404 I1 a trd 2 8 | -.1469463 2.17062 -0.07 0.946 -4.401283 4.107391 I1 a trd 2 9 | -2.743606 2.725503 -1.01 0.314 -8.085493 2.598281 I1 a trd 3 0 | -.0316161 3.025467 -0.01 0.992 -5.961423 5.89819 I1 a trd 3 2 | -1.084338 2.203619 -0.49 0.623 -5.403352 3.234676 H a trd 3 3 | -.7344953 2.051779 -0.36 0.720 -4.755908 3.286917 H a trd 3 4 | -.5303141 2.528159 -0.21 0.834 -5.485414 4.424786 I1 a trd 3 5 | -.9594432 2.21302 -0.43 0.665 -5.296883 3.377997 I1 a trd 3 6 | .6154846 2.137774 0.29 0.773 -3.574475 4.805444 I1 a trd 3 7 | -.4679754 4.779717 -0.10 0.922 -9.836049 8.900098 I1 a trd 3 8 | 1.547217 2.563473 0.60 0.546 -3.477097 6.571531 I1atrd39 | -.7663889 2.894297 -0.26 0.791 -6.439107 4.906329 _cons | -.5293822 .7182307 -0.74 0.461 -1.937088 .8783241 . p re d ic t xm5, xb (280 missing values generated) . gen double residm5 = dlnwage - xm5 (297 missing values generated) . m a trix accum Am5=residm5, noconstant (obs=2643) . matrix rsq5=1- (Am5/sumy) . matrix lis t rsq5 symmetric rsq5[1,1) C1 n .98845763 . summarize dlnwage drealgdp davghhi Hatrhhi dk_lexp frmSyrte atr H atr 12atr llatrdocd Hatrdocc2 Hatrdocc3 l1atrdocc4 HatrdoccS Hatrdocc6 l1atrd21 I1atrd22 Hatrd23 Hatrd24 Hatrd25 Hatrd26 I1atrd27 11atrd28 Hatrd29 I1atrd30 Hatrd32 Hatrd33 I1atrd34 Hatrd35 I1atrd36 Hatrd37 Hatrd38 I1atrd39 Variable | Obs Mean Std. Dev. Min Max dlnwage | 2781 -.0874443 14.8953 -108.2242 107.6844 drealgdp | 2800 1.98185 1.265558 -1.017 3.729 davghhi | 2800 48.40384 361.2067 -1891.933 2224.073 Hatrhhi | 2800 311.7569 439.2919 0 2443.248 dk_lexp | 2800 .0066773 .3502241 -2.11852 2.028664 frm5yrte | 2940 .118852 1.624088 -5.437162 4.879242 a tr | 2940 .452381 .4978119 0 1 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. H a t r 2800 .455 .4980598 0 12atr 2660 .4552632 .4980882 0 lla t r d o c d 2800 .065 .2465699 0 llatrdocc2 2800 .065 .2465699 0 l1atrdocc3 2800 .065 .2465699 0 H a trd o c c 4 2800 .065 .2465699 0 H a trd o c c S 2800 .065 .2465699 0 l1atrdocc6 2800 .065 .2465699 0 I1 a trd 2 l 2800 .005 .0705463 0 11atrd22 2800 .0075 .0862926 0 lla tr d 2 3 2800 .0175 .1311484 0 H a trd 2 4 2800 .015 .1215742 0 H a trd 2 5 2800 .0075 .0862926 0 11atrd26 2800 .0125 .1111223 0 I1 a trd 2 7 2800 .0125 .1111223 0 lla tr d 2 8 2800 .0325 .1773555 0 H a trd 2 9 2800 .025 .1561528 0 H a trd 3 0 2800 .0225 .1483294 0 H a trd 3 2 2800 .03 .1706177 0 I1 a trd 3 3 2800 .04 .1959942 0 H a trd 3 4 2800 .04 .1959942 0 11atrd35 2800 .035 .1838126 0 I1 a trd 3 6 2800 .0325 .1773555 0 11atrd37 2800 .0325 .1773555 0 11atrd38 2800 .025 .1561528 0 H a trd 3 9 2800 .0175 .1311484 0 . /* > THIS ANALYSIS EVALUATES THE EFFECT OF ATR INDICTMENTS ON EMPLOYMENT BY OCCUPATION USING SEEMINGLY UNRELATED REGRESSION TECHNIQUES. THE VAR OCCYEAR WAS CREATED TO ACCOMMODATE FOR THE INABILITY OF STATA TO COMPUTE THE SIC CODE AND OCCUPATION AND YEAR VARIATION.THE VAR EMPLOYMT REPRESENTS NUMBER OF EMPLOYEES BY OCCUPATION AND SIC CODE * / . i i s s ic . ts s e t s ic OCCYEAR panel v a ria b le : s ic , 20 to 39 time variable: OCCYEAR, 1 to 147 . / ‘ ANALYSIS ON TWICE LAGGED ATR VARIABLE AND LAGGED ATRHHI*/ . sureg (dlnwage = drealgdp davghhi H atrhhi dk_lexp frmSyrte atr H atr 12atr llatrdocd Hatrdocc2 Hatrdocc3 Hatrdocc4 HatrdoccS Hatrdocc6 Hatrd2l Hatrd22 Hatrd23 Hatrd24 I1atrd25 I1atrd26 Hatrd27 Hatrd28 Hatrd29 Hatrd30 I1atrd32 I1atrd33 I1atrd34 Hatrd35 I1atrd36 I1atrd37 I1atrd38 Hatrd39) Seemingly unrelated regression E quation Obs Parms RMSE "R-sq" ch i2 P dlnwage 2643 32 14.94807 0.0094 24.94965 0.8081 | Coef. Std. Err. z P>iz| [95% Conf. Interval] dlnwage [ drealgdp | .4458835 .2471611 1.80 0.071 -.0385434 .9303105 davghhi I .0012441 .000864 1.44 0.150 -.0004493 .0029375 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. H a tr h h i | -.0008856 .0040581 -0 .2 2 0.827 -.0088393 .007068 dk_lexp | -2.096365 .8465899 -2 .4 8 0.013 -3.755651 -.4370797 frm S y rte | .0574468 .2227961 0.2 6 0.797 -.3792256 .4941191 a t r | -.883736 .676382 -1.31 0.191 -2.20942 .4419484 H a t r | .1304534 3.629649 0 .04 0.971 •6.983529 7.244436 1 2 a tr | .5285678 .6719907 0 .79 0.432 -.7885098 1.845646 llatrdocd | 1 .01741 1.611892 0 .6 3 0.528 -2.141839 4.176659 Hatrdocc2 | 1.063287 1.611892 0 .66 0.509 -2.095963 4.222536 Hatrdocc3 | .4682526 1.611892 0.29 0.771 -2.690997 3.627502 Hatrdocc4 | -.4380183 1.611892 -0.27 0.786 -3.597268 2.721231 H a trd o c c S | .2353341 1.611892 0.15 0.884 -2.923915 3.394583 Hatrdocc6 | .3258018 1.611892 0.20 0.840 -2.833448 3.485051 I1atrd21 | 8.060634 8.825146 0.91 0.361 -9.236334 25.3576 I1 a trd 2 2 | .8172924 3.683801 0.22 0.824 -6.402825 8.03741 I1 a trd 2 3 | -.4114367 2.770504 -0 .1 5 0.882 -5.841525 5.018651 H a trd 2 4 | .6958498 3.531275 0.2 0 0.844 -6.225322 7.617022 Hatrd25 | -5.728247 4.017793 -1 .43 0.154 -13.60298 2.146483 I1 a trd 2 6 | -2.284511 2.928247 -0 .7 8 0.435 -8.023769 3.454748 I1 a trd 2 7 | -1.031324 3.528125 -0.29 0.770 -7.946322 5.883674 I1 a trd 2 8 | -.1469463 2.170077 -0 .0 7 0.946 -4.400219 4.106327 l1 a trd 2 9 | -2.743606 2.724821 -1 .01 0.314 -8.084157 2.596945 H a trd 3 0 | -.0316161 3.02471 -0.01 0.992 -5.959939 5.896707 I1 a trd 3 2 | -1.084338 2.203068 -0 .4 9 0.623 -5.402272 3.233595 I1 a trd 3 3 | -.7344953 2.051266 -0.36 0.720 -4.754902 3.285911 I1 a trd 3 4 | -.5303141 2.527526 -0.21 0.834 -5.484175 4.423546 H a trd 3 5 | -.9594432 2.212467 -0 .4 3 0.665 -5.295798 3.376912 H a trd 3 6 | .6154846 2.137239 0 .2 9 0.773 -3.573427 4.804396 lla tr d 3 7 | -.4679754 4.778521 -0.10 0.922 -9.833705 8.897755 H a trd 3 8 | 1.547217 2.562832 0 .6 0 0.546 -3.475841 6.570274 H a trd 3 9 | -.7663889 2.893573 -0.26 0.791 -6.437688 4.904911 _cons | -.5293822 .718051 -0 .7 4 0.461 -1.936736 .877972 . log close log: u: \dissertation\atr_logwage_changes_gls. log log type: text closed on: 23 Jul 2002, 11:53:57 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 205 APPENDIX E: CONVERSION OF DETAILED INDUSTRY AND OCCUPATION CODES Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 206 libname in •\\sparkynt\fdata0\org2\swawage'; libname out ' \\sparkynt\userfiles\ypho\diss\sas progranis\datasets'; options obs=max symbolgen mprint; %globai year start stop; %let start=79; %let stop=99; proc format; value OCc79fmt 80-85,150-173=1 1-79,86-149,174-195=2 201-245=3 260-296=4 301-395=5 901-984=6 /*401-575=7 601-785=8 801-846=9*/; value occ83fmt 203-235=1 43-199=2 3-42=3 243-302=4 303-402=5 403-472=6 /*503-702=7 703-889=8 473-499=9*/; value occ84fmt 3=1 2=2 1=3 4=4 5=5 6-8=6 /*9=7 10-12=8 13=9*/; value occfmt 1='Technicians’ 2='Prof Speclty Occ' 3='Mgrs & Admin' 4='Sales' 5='Admin Supp, Cler' 6='Service' /*7='Craft' 8='Laborer' 9='Farm'*/; %macro names; %do j=&start %to &stop; wocc&j %end; %mend; %macro names2; %do j=&start %to &stop; uocc&j %end; %mend; %macro yr7 9 ; mj occ=put(occode,occ79fmt.); %mend; %macro yr8 3 ; m j occ=put(occode,occ83fmt.); %mend; %macro yr8 4 ; mjocc=put(majocc,occ84fmt.); %mend; %macro occ; %do year=&start %to Sstop; data majocc; set in.wage&year.c(keep=majocc occode orgwt indcode); orgwt=orgwt/12; format majocc 3.; %if &year<=82 %then %do; %yr79 %end; %else %if &year=83 %then %do; Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 207 %yr83 %end; %else %do; %yr84 %end; /*due to a change in the survey, indcodes and occodes were changed in the cps in 1983*/ % if &year<=82 %then %do; “ f o r years 1982 and p r io r ; if 107<=indcode<=398; if 268<=indcode<=298 then sic=20; e lse if indcode=299 then sic=2l; e lse i f 307<=indcode<=318 then s ic = 2 2 ; e lse i f 319<=indcode<=327 then sic= 2 3 ; e lse if 107<=indcode<=109 then sic= 2 4 ; e lse i f indcode=H8 then sic=25; else i f 328<=indcode<=337 then sic= 2 6 ; e lse if 338<=indcode<=339 then sic=27; e lse if 347<=indcode<=369 then sic= 2 8 ; else if 377<=indcode<=378 then sic= 2 9 ; e lse i f 379<=indcode<=387 then sic= 3 0 ; e lse if 388<=indcode<=398 then sic=31; e lse if il9<=indcode<=l38 then s ic = 3 2 ; e lse i f 139<=indcode<=149 then sic=3 3 ; else if l57<=indcode<=l69 then sic=34; else if 177<=indcode<=l98 then sic= 3 5 ; else if 199<=indcode<=209 then sic=3 6 ; e ls e if 219<=indcode<=238 then sic=37; e lse if 239<=indcode<=258 then sic=3 8 ; e ls e if indcode=259 then sic=39; e lse s ic =1000; %end; %else %do; if 100<=indcode<=392; **for years 1983 to present; if I00<=indcode<=i22 then sic=20; e lse i f indcode=l30 then s ic = 2 l; e lse i f 132<=indcode<=i50 then sic=2 2 ; e lse if l5K=indcode<=l52 then sic=2 3 ; e lse i f 230<=indcode<=24l then sic=24; e lse i f indcode=242 then :sic=25; e ls e if I60<=indcode<=i62 then sic=2 6 ; e lse i f 17K=indcode<=l72 then sic=27; e ls e if I80<=indcode<=l92 then s ic =28; else if 200<=indcode<=20l then sic=29; e lse if 2l0<=indcode<=2l2 then sic=30; e lse if 220<=indcode<=222 then s ic = 3 l; e lse if 250<=indcode<=262 then sic=32; e lse if 270<=indcode<=280 then sic=3 3 ; e lse if 281<=indcode<=301 then sic= 3 4 ; e lse if 3l0<=indcode<=332 then sic=3 5 ; Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 208 else if 340<=indcode<=350 then sic=36; else if 351<=indcode<=370 then sic=37; else if 37K=indcode<=382 then sic=38; else if 390<=indcode<=392 then sic=39; else sic=l000; %end; if mjocc=''r or mjocc='2' or mjocc='3' or mjocc=‘4’ or mjocc^S' or mjocc='6'; /‘keeps o n ly the relevant worker categories*/ if mjocc='1‘ then tech=l; else tech=0; if mjocc='2' then prof=1; else prof=0; if mjocc=,3' then mgr=l; else mgr=0; if mjocc='4' then sales=l; else sales=0; if n1]0cc=,5• then clencal=l; else clerical=0; if mjocc='6' then serviced; else service=0; /* if rajocc='7’ or mjocc='8' or mjocc=’9‘ then oth=1; else oth=0; */ proc sort data=majocc; by s ic ; proc means data=majocc noprint; var tech prof mgr sales clerical service /‘oth*/; c la s s s ic ; weight orgwt; output out=wocc&year mean=wtech wprof wmgr wsales wcler wsvc / ‘woth*/ sum-wntech wnprof wnmgr wnsale wncler wnsvc / ‘wnoth*/ sumwgt=sumwgt; data wocc&year; set wocc&year; year=l9&year; proc print data=wocc&year; title 2 "Weighted Percentage of Workers in Manufacturing Sector by Majocc for the I9 4 y e a r*; proc means data=majocc noprint; var tech prof mgr sales clerical service /‘oth*/; c la s s s ic ; Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 209 output out=uocc4year mean=utech uprof uragr usales ucler usvc / ‘ uoth*/ sura=untech unprof unmgr unsale under unsvc / ‘ unoth*/; data uocc&year; set uocc&year; year=l9&year; proc print data=uocc&year; title 2 "Unweighted Percentage of Workers in Manufacturing Sector by Majocc for the 19&year*; %end; Smend occ; %occ; data wmajocc; set %names; data umajocc; set %names2; proc sort data=wmajocc; by year s ic ; proc sort data=umajocc; by year s ic ; data raajocd (drop=_type freq_); merge wmajocc umajocc; by year s ic ; if sic=. then sic=l9; / ‘total # of employees by occupation category for a particular y e a r*/ data out.majocc; set majoccl; Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 14«?ersar.al Services, Exc. Private Hhlds 15-Entertainment And Recreation. Services 16-Hospitals 17-Medical Services, Exc. Hospitals 13=Educational Services 19-Social Services 20-Other Professional Services 21-Forestry And Fisheries 22=?ublic Administration 23-Armed Forces MAJOCC Type: numerical, categorical Range: 1979-1983:C-13 1984-1993:1-15 1994-present:1-14 Years Available: 1979-present Definition: Major Occupation 1979-1983 I-prof, tech and kindred 2=managers except farm 3=sales 4=clerical 5-craftsmen 6-operavtives except transport 7»transport equip operators 6-nor. farm laborers 9-private hh services lC-all othr svce workers II-farm and farm managers 12-farm laborers and foremen 13-no previous ft work experience 1984-1993 1-Executive, Admin, i Managerial 2=Professional Specialty Occs 3-Technicians And Related Support 4-Sales Occs 5-Admin. Support Occs, Incl. Clerical 6-Private Household Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7«?rotective Service 8=0ther services 9»Precision Prod., Craft & Repair 10»Machine Opers, Assemblers & Inspectors ll=Transportation And Material Moving 12=Kandlers,equip Cleaners, helpers, laborers 13“Farming, Forestry And Fishing l4*Armed Forces 15»no previous experience 1994-present l«Executive, Admin, & Managerial 2*?rcfessional Specialty Cccs 3“Technicians And Related Support 4=Sales 5“Admin. Support Occs, Incl. Clerical 6=Privace Household 7»Protective Service 8»Service Occs, Exc. Protective St Hhld 9«Precisicn Prod., Craft & Repair 10*Machine Opers, Assemblers & Inspectors ll=Transportation And Material Moving 12“Handlers,equip Cleaners, helpers, labcrrs 13=Farmir.g, Forestry And Fishing 14=Armed Forces INDCODS Type: Range: 1979-1982:17-937 1983-present: 10-991 Years Available: Definition.-. For 1979-1982, contains the 3 digit Industry Classification from the 1970 Census, and for 1983 on, contains 3 digit Industry Classification from the 1980 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 212 % m i d w - I APPENDIX B I INDUSTRY CLiSSXPIClTIOH (Numbers in p aren th eses a re th e SIC code equivalents) Census Cede AGRICULTURE, FORESTRY, AND FISHERIES 017 Agricultural production (01) 018 Agricultural services, exc. h o r tic u ltu r a l (07 except 0713 and 073) 019 Horticultural services (073) 027 F orestry (08) 023 F ish eries (09) AUXBG 047 fle ta l m ining -(10) 048 Coal mining (11, '12) 045 Crude petroleum and natural gas extractions (13) 057 Ncnnetallic mining and quarrying, exc. fu e l (14) CONSTRUCTION General building contractors (15) General contractors, exc. building (16) Special trace contractors (17) Not specified construction BiHUFlCTOBXBG Durable soods Lumber and wood p ro d u c ts, ex c. f u r n itu r e Logging (241) Sawmills,. planing mills, and mill work (242, 243) M iscellaneous wood p ro d u cts (244, 249) F u rn itu re and fix tu re s (25) Stone, clay, and glass products Glass and glass products (321-323) Cement, concrete* gypsum, and p la s te r products (324, 327) Structural clay products (325) Pottery and related products (326) Biscellantoijs nonmetallic mineral-and ston e Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. I 213 products (328, 325) Metal industries ■139 Blast furnaces, steel works, rolling and finishing mills (3312, 3313) r Other primary ircn and steel industries (3315-3317, 332, 3391, part 3395) I Primary aluminum industries (3334, part 334, 3352, 3361, part 3392, part 3355) > Other primary nonferrous industries (3331-333, 3339, Dart 334, 3351, 3356, 3357, 3362, 3369, part 3392, part 3*359) r Cutlery, hand tools, and other baadvare (342) 3 Fabricated structural metal products (344) J Screw machine products (345) 7 Metal stamping* (346) i Siscellaneoas fabricated metal Drocucts (341, 343, 347, 348, 349) .. 169 Not specified metal industries Bachiaery, exceot electrical r Engines ancl turbines (351) I Farm machinery and equipment (352) 9 Construction.andmaterial handling machines (353) 1 Betalworking machinery (3 54) 8 Office and accounting machines (357 e x c . 3573) 35 8 Electronic computing equipment (3573) 7 flacbinery, excl electrlca'l, n - e . c . (355, 356, 353, 359) 8 Not specified aachinery Electrical machinery, equipment, and supplies 9 Household appliances (363! 7 Radio, T.V.,,‘ and communication equipment (365, 366) 8 Electrical machine, equicnent, and supplies, n.e.c. (361, 362, 364, 367, 36*9) _ 209 Not specified electrical machinery, equipment, and supplies Transportation ••guipaent 9 B otor vehicles and motor vehicle equipment (371) 7 Aircraft and parts (372) 8 Ship and boat* building and repairing (373) 9 Railroad locomotives anil equipment (374) 7 Mobile d w e llin g s and campers (3791) 6 Cycles and miscellaneous transportation equipment (375, 3799) Professional and photographic equipment, and watches 9 Scientific and controlling instruments (381, 382) 7 Optical and health services supplies (383, 384, 385) 8 Photographic equipment and supplies (386) 9 la t c h e s , clocks, and clock-work-operated devices (387) 7 Not specified professional equipment 8 Ordnance (19) 9 Miscellaneous manufacturing industries (39) . Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. r o o d and kindred products I# 2 66 Keat products (201) 2 6 9 D airy products (202 276 Canning and preserving fruits, vegetables, and sea foods (293) " 279 Grain-mill products (204, 0713) 1s> 287 Bakery products (205) 288 C onfectionery and r e la te d nroducts (207) 28 9 Beverage industries (205) 297 Miscellaneous food preoaratior. and kindred products (206, 209) 2 9 8 Not specified food industries ^ 299 Tobacco m anufactures (21) Textile mill products .307 Knitting‘mills (225) 308 Dyeing and finishing textiles, exc. wool and k n it goods (226) 309 F lo o r co v erin g s, exc. h ard s u rfa c e (227) l l 317 Yam, thread, and fabric mills (221-224, 228) '-Jl 8 Miscellaneous textile mill products (229) Apparel and other fabricated textile products 319 Apparel and accessories (231-238) 327 Miscellaneous fabricated textile products (239) n [ Paper and allied products r ~ 22B Pulp, paper,'and paperboard m ills .(261-263, 265) Miscellaneous paper and pulp products (264) & L is Paperboard co n tain ers ana boxes (265) Printing, publishing, and allied industries r338 Newspaper publishing and p r in tin g (271) f t L 339 Printing, publishing, and allied industries, except newspapers (272-279) Chemicals and aliiea products ’ 347 Industrial chemicals (281) 348 Plastics, synthetics and resins, exc. fibers (282, exc. 2823 and 2824) 349 Synthetic fibers (2823, 2824) 357 Drugs and medicines (283) 358 Soaps and cosm etics (284) 359 Paints, varnishes, and related products (285) 367 Agricultural chemicals (287) 368 Blscellaaeous chemicals (286, 289) '-369 Not specified chemicals and allied products Petroleum and coal products Petroleum re fin in g (291) V 378 Miscellaneous petroleum and coal products (295, 299) Rubber and miscellaneous plastic products 3 7 9 Rubber products (301-303, 306) 3 8 7 M iscellaneous p la s tic p ro d u cts (307) r Leather and leather products r 388 Tanned, curried, and finished leather (311) Footwear, except rubber (313, 314) 1 15 Leather products, wxc. footwear (312, 315- 317, 319) ' 1-398B ■ot specified manufacturing industries Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 215 w63-^aMjvdr Industrial Classification System £qw vdtnt numeric codes follow the- alphabetic codes. Either a d e mey be used. depending on the processing method. Num&eri in ptrem heen following tfto industry cat •gents its the SIC definitions. The abQrivtation ~pt" means "part" and ~na.fi.” mean* ~not where daaiRtd.* In d u e Indue* try Industry attgery try industry eattvony . codr cod* AGRICULTURE, FORESTRY, ANO FISHERIES- MANU FACTU R! NG-Coit. A (QIC) Agricultural production. c ro » (61) Nondurable Goods—&rt. S 011 Agricultural production. livestock (C2) P ioer and allied produca / C2Q Agricultural wrvice*. s*aat horticultural (07, except !60 j # Pulp, paper, and peperbcerd mills (261*263.256) 076) 1ST V? M lnllaneoui paper and pulp produca (264) 021 Ho rtfcuirural uirrices (078) 162 Jit Paperboard containers and boxee (265) 030 Forestry (08) t Printing, publishing, and allied (ribuecries Q31 Fishing, hunting, and trapping (OS) C (171) Newspaper pubtitfiing and printing (271) 172 17.4 Printing, publishing. and titled industries, exceot ftowipaoeri (272*279) MINING Chemicals and allied products Plastics, fyrtthetia. and resins (282) 040 M«tal mining (10) 181 Orngt (283). 041 Coal mining (11.12) 182 3*8 Soaps and eosmetia ( 2341 042 Crude petroleum and naturil gas extraction (131 190 3“ Paints, vamianes, and refaud produca (265) 350 Nonmataliic mining and quarrying, cxapt fuel (14)' Agricultural chemicals (287) Industrial ana miscellaneous chemicals (281,286. 289) „ Petroleum and coal produca 8 10601 CONSTRUCTION (1 5 .1 6 ,1?1 j *-200 ^ Petroleum refining 1291) fc.2G1 Miscellaneous petroleum and eoei produca (295, 299) MANUFACTURING Rubber and miscellaneous plastics products r*210 5*3 Tires and inner tubes (301} Nondunbto Good. I 211 5^4 Other rubber products, and Qiastics footwear and I belting (302*304. 306) Focd and kindred producti V> V212 Miscellareout plastics products 13071 P*100 ZiS Meet prcduea (301) Leather and leather produca 101 w Dairy produca (202) 3 6 $ Leather tanning end f inisning (311) 102 W Canned and prtarved fruits and vtgrtables (2C3) Footwear, exact rubOer and plastic (313.314) n o rrt G a in mill produce 1204) Leather produca, except footwear (315*317.319) 8 akery products (205) Sugar and confectionery products (206) S e w a g e industries (206) Durable Goods Miscellaneous food preparations and kindred prod* Lumber and wood preducts. sxapt furniture ucts (207,209) Logging (241) Net specified food industr ies Sawmills, planing mills, and miilwork (242.243) Tcbaeeo manufactures (21) Wood buildings end mobile homes (2451 T e x tla mill products MisoKaneoua wood produca (244.249) Knitting miila (225) ^ **Vi42 II Furniture and fixtures (25) Oyting and flniihinf witiiM. tu tc t wool and Stane, dev, giaa. and concrete products knit joada (2361 250 \A Glass and giaa produca (321 *323) 141 3*1 Floor raw ing, .joast hard lurfaa (237) 251 a t Cement, concrete, gypsum, and piaster products 142 3! Yam . » fM d . and fabric milla (338.231-234) (324.327) 150 life M ln llan w u a t u t i l t mill produca (239) 283 i 'Jb Structural clay p roduca (325) Apevti and otftar fimiiwd ttitila produca * 1 281 iVr Posery and related produca (3261 Aoecraf and aooai riai. a x o s t knit (231-23) ^262 i'hk: Miscellaneous nonm eallic mineral and r s n e prod* 52 Jit MioUanooua faoricmd taittl. produca (239) uca (328,329) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 216 livdussM CM ficatian System Indus Indus* try Industry category try Industry category cod* MANUFACTURING—Con. TRANSPORTATION, COMMUNICATIONS. AND Durable Goada-Con. OTHER PUBLIC UTILITIES Metal industries Transportation 2 73 I ? ' Blaat furnaces. steelworks. rolling and finishing 400 Railroads (40) m illi 1331) 401 Bus service and urban transit (41. except 412) 271 l*f? Iron and i t x l foundries 1332) 402 Taxicab sarvia (412) 272 W* Primary aluminum industrial 1333d, pt 33d, 3353- 1 410 Trucking ten ia (421,423) 3355.3331) I 411 Warehousing and storage (422) 230 Otnar primary matal Industrlas 13331-3333.3339, 412 U.S. Postal S ervia (43) pt 33d. 3351. 3356. 3357. 3362. 3369. 3391 420 Water transoortabcn (44) -.2 3 1 Cutlary. hand tools, and other hardwire ( 342) 421 Air transportation (45) I 282 |J 3 Fabricated structural matal products 1344) 422 Pipe lines, e x a p t natural gas (461 290 <9 Screw mactiina products (3dS) 432 Services inddental to transportation (47) 291 Matal forgings and stampings (3461 Communications 2 92 Hr? Ordnanca 13481 r ddo Radio and television broadcasting (483) * 300 /if? Miscallanaous fabricated matal products (341,343. I 441 Telephone (wire and radial (481) 347.349) I 442 Telegraph and mitalleneoui communication « V 301 I if1] Not ipscifiad matal Industries (482.489) Utilities and sanitary servias Macftin«ry. occfit tl«ctnat r 460 Electric light and power (491 1 >310 i f ? Engine* »nd turtainM (3511 I 461 Gas and steam supply tyftems (492.496) 311 Firm machinery mdtquiomtnt (352) ' 462 Electric and gas. and other combinations (492! 312 irf Construction md mattriii handling macrnnts I-S3: ! 470 Vltter supply in i irrigation [4.94,497) r32C i f t M ttilw orking m»cftiflery (35*1 i lr' Sanitary services I4S5I 321 it6 Cffict md Kcounting maoitnw (357. except 35731 472 Not specified utilities 322 Electronic camouting equipment (3573) 331 13? Mecftinery. e*«ot electrical. n.».e. (355.356. 3 56.359) WHOLESALE TRAOE 332 v Noe specified machinery Electrical machinery, equipment. and tupptiet Durable Goods Household aooliances (363) 341 2d? Radio. TV. and communication eouiomant (365. 366)/' 50^ Motor vehicles and equipment (SOI) Furniture and home furnishings (502) 342 2 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Code Industry 123-129 not used 130 Tobacco m anufactures (21) 131 not used 132-150 Textile mill products 132 Knitting m ills (225) 133-139 not used 140 Dyeing and finishing textiles, except wool and knit goods (226) 141 C arpets an d rugs (227) 142 Y am , thread, and fabric mills (221-224,228) 143-149 not used 150 Miscellaneous textile mill products (229) 151-152 A pparel and o th er finished textile products 151 Apparel and accessories, except knit (231-238) 152 Miscellaneous fabricated textile products (239) 153-159 not used 160-162 P aper and allied products 160 Pulp, paper, and paperboard mills (261-263) 161 Miscellaneous paper and palp products (267) 162 P aperboard containers and boxes (265) 163-170 not used 171-172 Printing, publishing, and allied industries 171 Newspaper publishing and printing (271) 172 Printing, publishing, and allied industries, except newspapers (272-2791 173-179 not used 180-192 Chemicals and allied products 180 Plastics, synthetics, and resins (282) 181 Drugs (283) 182 Soaps and cosmetics (284) 183-189 not used 190 Paints, varnishes, and related products (285) 191 A gricultural chemicals (287) 192 Industrial and miscellaneous chemicals (281,286,289) 193-199 not used 200-201 Petroleum and coal products 200 Petroleum refining (291) 201 Miscellaneous petroleum and coal products (295,2991 202-209 not used 210-212 Rubber and miscellaneous plastics products 210 Tires and inner tubes (301) 211 Other rubber products, and plastics footwear and belting (302-306) 212 Miscellaneous plastics products (308) 213-219 not used 220-222 L eather and leather products 220 L eather tanning and finishing (311) 221 Footwear, except rubber and plastic (313,314) 222 Leather products, except footwear (315-317,319) 223-229 not used 230-392 D U R A B LE G O O D S 230-241 Lum ber and w ood products, except furniture 230 Logging (241) 231 Sawmills, planing mills, and millwork (242,243) 232 Wood buildings and mobile homes (245) 233-240 not used 241 Miscellaneous wood products (244,249) 242 Furniture and fixtures (25) 243-249 not used Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. B P P B M U H I H INDUSTRY CLASSIFICATION Industry Classification Codes for Detailed Industry (3-digit)' There are 236 categories for the employed, with 1 additional category for the experienced unemployed. These categories are aggregated into 51 detailed groups and 23 major groups . (Numbers in parentheses are the 1987 SIC code equivalent; see Executive Office of the President, Office of Management and Budget, Standard Industrial Classification Manual, 1887. “Pt” means part, “n.e.c." means not elsewhere classified) Code Industry 000-009 not used 010-030 AGRICULTURE 010 Agricultural production, crops (01) 011 Agricultural production, livestock (02) 012 Veterinary services (074) 013-019 not used 020 Landscape and horticultural services (078) 021-029 not used 030 Agricultural services, n.e.c. (071,072,075, 076) 031-032 FORESTRY AND FISHERIES 031 Forestry (08) 032 Fishing, hunting, and trapping (09) 033-039 not used 040-050 MINING 040 Metal mining (10) 041 Coalm ining (12) 042 Oil and gas extraction (13) 043-049 not used 050 Nonmetallic mining and quarrying except fuel (14) 051-059 ' not used 060 • CONSTRUCTION (15,16,17) 061-099 not used 100-392 MANUFACTURING 100-222 NONDURABLE GOODS 100-122 Food and kindred products 100 Meat products (201) 101 Dairy products (202) 102 Canned, frozen and preserved fruits and vegetables (203) 103-109 not used 110 Grain mill products (204) HI Bakery products (205) 112 Sugar and confectionery products (206) 113-119 not used 120 Beverage industries (208) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Code Industry 250-262 Stone, day, glass, and concrete products 250 Glass and glass products (321-323) 251 Cement, concrete, gypsum, and plaster products (324,327) 252 Structural day products (325) 253-260 not used 261 Pottery and related products (326) 262 Miscellaneous nonmetallic mineral and stone products (328,329) 263-269 not used 270-301 Metal industries 270 Blast furnaces, steelworks, rolling and finishing mills (331) 271 Iron and steel foundries (332) 272 Primary aluminum industries (3334, part 334,3353-3355,3363,3365) 273-279 not used 280 Other primary metal industries (3331, 3339, part 334, 3351, 3356,3357, 3364,3366, 3369,339) 281 Cutlery, handtools, and general hardware (342) 282 Fabricated structural metal products (344) 283-289 not used 290 Screw machine products (345) 291 Metal forgings and stampings (346) 292 Ordnance (348) , 293-299 not used 300 Miscellaneous fabricated metal products (341, 343,347, 349) 301 Not specified metal industries 302-309 not used 310-332 Machinery and computing equipment 310 Engines and turbines (351) 311 Farm machinery and equipment (352) 312 Construction and material handling machines (353) 313-319 not used 320 Metalworking machinery (354) 321 Office and accounting machines (3578,3579) 322 Computers and related equipment (3571-3577) 323-330 not used 331 Machinery, except electrical, rnex. (355,356,358,359) 332 Not specified machinery 333-339 not used 340-350 Electrical machinery, equipment, and supplies 340 Household appliances (363) 341 Radio, TV, and communication equipment (365,366) 342 Electrical machinery, equipment, and supplies, n.e.c. (361,362,364,367,369) 343-349 not used 350 Not specified electrical machinery, equipment, and supplies 351-370 Transportation equipment 351 . -Motor vehicles and motor vehicle equipment (371) 352 Aircraft and parts (372) 353-359 not used 360 Ship and boat building and repairing (373) 361 Railroad locomotives and equipment (374) 362 Guided missiles, space vehicles, and parts (376) 363-369 not used 370 Cycles and miscellaneous transportation equipment (375,379) 371-381 Professional and photographic equipment, and watches 371 Scientific and controlling instruments (381,382 except 3827) 372 Medical, dental, and optical instruments and supplies (3827,384,335) 373-379 not used 330 Photographic equipment and supplies (386) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 220 Code Industry 381 Watches, docks, and clockwork operated devices (387) 382-389 not wed 390 Toys, amusement, and sporting goods (394) 391 Miscellaneous manufacturing industries (39 except 394) 392 Not specified manufacturing industries 393-399 _ not used 400-472 TRANSPORTATION, COMMUNICATIONS, AND OTHER PUBLIC UnLTTIES 400-432 TRANSPORTATION 400 Railroads (40) 401 Bus service and urban transit (41, except 412) 402 Taxicab service (412) 403-409 not used 410 Trucking service (421, 423) 411 Warehousing and storage (422) 412 U.S. Postal Service (43) 413-419 not used 420 Water transportation (44) 421 Air transportation (45) 422 Pipe lines, except natural gas (46) 423-431 not used ' , 432 Services incidental to transportation (47) 433-439 not used 440-442 COMMUNICATIONS 440 Radio and television broadcasting and cable (483, 484) 441 Telephone communications (481) 442 Telegraph and miscellaneous communications services (482,489) 443-449 not used 450-472 UTUJnES AND SANITARY SERVICES 450 Electric light and power (491) 451 Gas and steam supply systems (492,496) 452 Electric and gas, and other combinations (493) 453469 not used 470 Water supply and irrigation (494,497) 471 Sanitary services (495) 472 N ot specified utilities 473499 not used 500-571 WHOLESALE TRADE 500-532 Durable Goods 200 Motor vehicles and equipment (501) 201 Furniture and home furnishings (202) 202 Lumber and construction materials (503) 503-509 not used 210 Professional and commercial equipment and supplies (504) 211 Metals and minerals, except petroleum (505) 512 Electrical goods (506) 513-520 not used 221 Hardware, plumbing and heating supplies (507) 522-529 not used 230 Machinery, equipment, and supplies (508) 231 Scrap and waste materials (5093) 232 M iscellaneous wholesale, durable goods (509 except 50931 533-539 not used 540-571 Nondurable Goods INDUSTRY CLASSIFICATION CODES FOR DETAILED INDUSTRY Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Code Industry 540 Paper and paper product* (511) 541 Drugs, chemicals and allied products (512,516) 542 Apparel, fabrics, and notions (513) 543*549 not used 550 Groceries and related products (514) 551 Farm-product raw materials (515) 552 Petroleum products (517) 553*559 not used 560 Alcoholic beverages (518) 561 Farm supplies (5191) 562 Miscellaneous wholesale, nondurable goods (5192-5199) 563-570 not used 571 Not specified wholesale trade 572-579 not used 580-691 RETAIL TRADE 580 Lumber and building material retailing (521,523) 581 Hardware stores (525) 582 Retail nurseries and garden stores (526) 583-589 not used 590 Mobile home dealers (527) 591 D epartm ent stores (531) 592 Variety stores (533) 593-599 not used 600 Miscellaneous general merchandise stares (539) 601 Grocery stores (541) 602 Dairy products stores (545) 603-609 not used 610 Retail bakeries (546) 611 Food stores, n.ex. (542, 543,544, 549) 612 Motor vehicle dealers (551,552) 613-619 not used 620 Auto and home supply stores (553) 621 Gasoline service stations (554) 622 Miscellaneous vehicle dealers (555, 556,557,559) 623 Apparel and accessory stores, except shoe (56, except 566) 624-629 not used 630 Shoe stores (566) 631 Furniture and home furnishings stores (571) 632 Household appliance stores (572) 633 Radio, TV, and computer stares (5731,5734) 634-639 not used 640 Music stores (5735,5736) 641 Eating and drinking places (58) 642 Drug stores (591) 643-649 not used 650 Liquor stores (592) 651 Sporting goods, bicycles, and hobby stores (5941,5945,5946) 652 Book and stationery stores (5942,5943) 653-659 not used 660 Jewelry stores (5944) 661 Gift, novelty, and souvenir shops (5947) 662 Sewing, needlework and piece goods stores (5949) 663 Catalog and mail order houses (5961) 664-669 not used 670 Vending machine operators (5962) 671 Direct selling establishments (5963) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 222 Code Industry 672 Fuel dealers (598) 673-680 not used 681 Retail florists (5992) 682 Miscellaneous retail stores (593,5948,5993-5995,5999) 683-690 not used 691 Not specified retail trade 692-699 not used 700-712 FINANCE, INSURANCE, AND REAL ESTATE 700 Banking (60 except 603 and 606) 701 Savings institutions, including credit unions (603, 606) 702 Credit agencies, n.e.c. (61) 703-709 not used 710 Security, commodity brokerage, and investment companies (62,67) 711 Insurance (63,64) 712 Real estate, including real estate-insurance offices (65) 713-720 not used 721-760 BUSINESS AND REPAIR SERVICES 721 Advertising (731} 722 Services to dwellings and other buildings (734) 723-730 not used 731 Personnel supply services (736) 732 Computer and data processing services (737) 733-739 not used 740 ' Detective and protective services (7381,7382) • 741 Business services, n.e.c. (732,733,735,7383-7389) 742 Automotive rental and leasing, without drivers (751) 743-749 not used 750 Automotive parking and carwashes (752, 7542) 751 Automotive repair and related services (753,7549) 752 Electrical repair shops (762,7694) 753-759 not used 760 Miscellaneous repair services (763,764,7692,7699) 761-791 PERSONAL SERVICES 761 PRIVATE HOUSEHOLDS (88) 762-791 PERSONAL SERVICES, EXCEPT PRIVATE HOUSEHOLD 762 Hotels and motels (701) 763-769 not used 770 Lodging places, except hotels and motels (702,703,704) 771 Laundry, cleaning, and garment services (721 except part 7219) 772 Beauty shops (723) 773-779 not used 780 Barber shops (724) 781 Funeral service and crematories (726) 782 Shoe repair shops (725) 783-789 not used 790 Dressmaking shops (part 7219) 791 Miscellaneous personal services (722,729) 792-799 not used 800-810 ENTERTAINMENT AND RECREATION SERVICES 800 Theaters and motion pictures (781-783,792) 801 Video tape rental (784) INDUSTRY CLASSIFICATION CODES FOR DETAILED INDUSTRY Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Code Industry 802 Bowling centers (793) 803-809 not used 810 Miscellaneous entertainment and recreation services (791,794,799) 811 not used 812-893 PROFESSIONAL AND RELATED SERVICES 812-830 MEDICAL SERVICES, EXCEPT HOSPITALS 812 Offices and clinics of physicians (801, 803) 813-819 not used 820 Offices and clinics of dentists (802) 821 Offices and clinics of chiropractors (8041) 822 Offices and clinics of optometrists (8042) 823-829 not used 830 Offices and clinics of health practitioners, n.ex. (8043, 8049) 831 HOSPITALS (806) 832-840 MEDICAL SERVICES, EXCEPT HOSPITALS (Continued) 832 Nursing and personal care facilities (805) 833-839 not used 840 H ealth services, n.e.c. (807, 808,809) 841 OTHER PROFESSIONAL SERVICES (also includes codes 872-893) 841 _ Legal services (81) 842-860 EDUCATIONAL SERVICES 842 Elementary and secondary schools (821) 843-849 not used 850 Colleges and universities (822) 851 Vocational schools (824) 852 Libraries (823) 853-859 not used 860 Educational services, rne.c. (829) 861-871 SOCIAL SERVICES 861 Job training and vocational rehabilitation services (833) 862 Child day care services (part 835) 863 Family child care homes (part 835) 864-869 not used 870 Residential care facilities, without nursing (836) 871 Social services, n.e.c. (832,839) 872-893 OTHER PROFESSIONAL SERVICES (Also indudes code 840) 872 Museums, art galleries, and zoos (84) 873 Labonmions (863) 874-879 not used 880 Religious organizations (866) 881 Membership organizations, n.e.c. (861,862,864,865,869) 882 Engineering, architectural, and surveying services (871) 883-889 not used 890 Accounting, auditing, and bookkeeping services (872) 891 Research, development, and testing services (873) 892 Management and public relations services (874) 893 Miscellaneous professional and related services (899) 894-899 not used Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 224 Code Industry 900-932 PUBLIC ADMINISTRATION 900 Executive and legislative offices (911-913) 901 General government, n.e.c. (919) 902-909 not used 910 Justice, public order, and safety (92) 911-920 not used 921 Public finance, taxation, and monetary policy (93) 922 Administration of human resources programs (94) 923-929 not used 930 Administration of environmental quality and housing programs (95) 931 Administration of economic programs (96) 932 National security and international affairs (97) 933-990 not used 991 Assigned to persons whose labor force status is unemployed and whose last job was Arm ed Forces INDUSTRY CLASSIFICATION CODES FOR DETAILED INDUSTRY Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 225 APPENDIX A Occucasj.oa Classification C an su s -Code PROFESSIONAL, TECHNICAL, AN) KINDRED WORKERS 001 Accountants 0C2 Architects Conputer specialists 003 Computer programmers 004 Computer systems analysts 005 Computer specialists, n.e-c. Engineers 006 Aeronautical and astronautical eigineers 010 Chemical engineers O H Civil engineers 012 Electrical ar.d electronic engine its 013 Industrial engineers 014 Hechanical'engineers 015 tietallurgical and materials ergireers 020 Mining engineers 021 Petroleum engineers 022 Sales engineers 023 Engineers, n ,e .c . 024 Farm management advisors 025 Foresters and conservationists 026 Home management advisors Lawyers and judges 030 Judges 031 Lawyers Librarians, archivists, and curators 032 Librarians 033 Archivists and curators Batieaatical specialists 034 Actuaries 035 Bath* a&ti clans 036 Statisticians Life and physical scientists 042 Agricultural scientists 043 Atmospheric and space scientists 044 Biological scientists 045 Chemists 051 Geologists 052 aarine scientists 053 Physicists and astronomers 054 Life and physical scientists, a. i.e. 055 Operations and systems researchers an I analysts 056 Personnel and labor relations workers Physicians, dentists, and related pra:titioners 064 chiropractors 062 Daatiats Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 0S3 Optometrists 064 Pharmacists 065 Physicians, medical and osteopatiic 071 Podiatrists 072 Veterinarians 073 Health practitioners, n.e.c. _ Nurses, dietitians, and therapists . 074 Dietitians 075 Registered nurses I 076 Therapists ' Health technologists and technicians ^*080 Clinical laboratory technologist; and technicians 081 Dental hygienists i 082 Health record technologists and technicians t i p !!? Radiologic technologists and technicians V* 084 Theracy assistants 085 Health technologists and technicians, n.e.c. — Religious workers f~ 086 Clergymen ! 090 Religious workers, o-a.c. Social scientists 091 Economists 092 Political scientists 093 Psychologists 094 Sociologists 095 Urban and regional planners I 096 Social scientists, n.e.c. Social and recreation workers ' 100 Social workers I *01 Recreation workers ! Teachers, college and university Agriculture teachers Atmospheric, earth, marine, and ipace teachers Biology teachers Chemistry teachers Physics teachers Engineering teachers Batheaatics teachers Health specialties teachers Psychology teachers Business and commerce teachers Economic teachers History teachers Sociology teachers Social science teachers, a.e.c. Art, drama, and music teachers Coaches and physical education t lackers Education teachers English teachers Foreign language teachers Home economics teachers Law teachers Theology teachers Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 134 Trade, industrial, and technical teachers Miscellaneous teachers, college tod univers ity 140 Teachers, college and university, subject not specified Teachers, except college and u n iv e r s ity Adult education teachers Elementary school teachers Prekindergarten and kindergarten teachers Secondary school teachers Teachers, except college and university, s. e. c. Engineering and science technicians ISO Agriculture and biological technicians, exc eoo health Chemical technicians Draftsmen Electrical and electronic engine iring techn ctans Industrial engineering t e c h n i c ia n Bechanical engineering technicia is Batheaatical technicians Surveyors Engineering and science technicims, n.e.c. Technicians, except health, and er.ginrering .and.science Airplane pilots Air traffic controllers Embalmers Flight engineers Radio operators Tool programmers, numerical control Technicians, n . e . c . Vocational and educational counselors Writers, artists, and entertainers Actors Athletes and kindred workers Authors Dancers Designers Editors and reporters B u s ic ia iu and composers Painters and sculptors Photographers Public relations men and publicity writers Radio and television announcers Writers, artists, and entertainers, n . e . c . Research workers, not specified MANAGERS AND ADMINISTRATORS, JXC tPT FARM 1 201 Assessors, controllers, and treasurer,-; 1 local public administration 1 202 Bank officers and financial managers Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Buyers and shippers, fa i t . products Buyers, wholesale and retail trade Credit men Funeral directors Health administrators Construction inspectors, public administration Inspectors, except construction, public administration Saoagers and superintendents, fcuildinj Office managers, n„e.c. Officers, pilots, and pursers; ship Officials and administrators; public administration, n.e. c. Officials of lodges, societies, and aiioas Postmasters and mail superintendents Purchasing agents and buyers, n.e.c,. Railroad conductors Restaurant, cafeteria, and bar managers Sales managers and department heads”, retail trade Sales managers, except retail trade School administrators, college School administrators, elementary and secondary Managers and administrators, n-a-c. SALES YORKERS Advertising agents and salesmen Auctioneers Demonstrators Hucksters and peddlers Insurance agents, brokers, and underwriters Newsboys Real estate agents and brokers Stock and bond salesmen Salesmen and sales clerks, a.e.c. Sales representatives, manufacturing industries Sales representatives, wholesale trade Sales clerks, retail trade Salesmen, retail trade Salesmen of services and constraitios Sales >orkers - allocated CLERICAL AND KINDRED YORKERS Bank tellers Billing clerks Bookkeepers Cashiers Clerical assistants, social welfare Clerical supervisors, jue.c. Collectors, ’bill and account Counter clerks, except food Dispatchers and starters, vehicle Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 229 320 Enumerators and interviewers 321 Estimators and investigators, n.e^c. 323 Exoediters and production controllers 325 File clerics 326 Insurance adjusters, examiners, and investigators 330 Library attendants and assistants 331 flail carriers, post office 332 Hail handlers, except post office 333 lessengers and office boys 334 Satar readers, utilities Office machine operators 341 Bookkeeping and billing machine iperators 342 Calculating machine operators 343 Computer and peripheral equipment operators 344 Duplicating machine operatcri 345 Key punch o p erato rs 350 Tabulating machine operators 355 Office machine operators, tue.o. 360 Payroll and timekeeping clerks / 361 Postal clerks 362 Proofreaders 363 Real estate appraisers 364 Receptionists" Secretaries 370 Secretaries, lecal 379 Secretaries, meSical 372’ Secretaries',- n—ere. 374 Shipping and receiving clerks 375 Statistical clerks 376 Stenographers 381 Stock clerks and storekeepers 382 -Teacher aides, exc. school monitors 383 Telegraph messengers 384 Telegraph operators 385 Telephone operators 390 Ticket, station, and express agents 391 Typists 392 Ueighers 394 Biscellaneons clerical workers 395 Not specified clerical workers CE1FISBES AND KINDRED UORXERS 401 Automobile accessories installers 402 Bakers 403 Blacksmiths 404 Boilermakers 405 Bookbinders 410 Brickmasons and stonemasons <*11 Brickmasons and stonemasons, apprentices 412 Bulldozer operators 413 Cabinetmakers Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ■»15 Carpenters 416 Carpenter apprentices Carpet installers 421 Cement and concrete finishers 422 Compositors and typesetters 423 Printing trades apprentices, eic. pressmen 424 Cranemen, derrickmen, and hoistien 425 Decorators and window dressers 426 Dental laboratory technicians 430 Electricians 431 Electrician apprentices 433 Electric power' linemen and cableses 434 Electrotypers and stereorypers 435 Engravers, exc. photoencravers 436 Excavating, grading, and road machine operators,- exc. "bulldozer 440 Floor layers, exc. tile setters 441 Foremen, n.e.c. 442 Pcrgemer. and hammermen 443 Furniture and wood finishers 444 Furriers 445 Glaziers 446 Heat treaters, annealers, and temperers 450 Inspectors, scalers, and graders:"log and lumber 452 Inspectors, o.e.c. 453 Jewelers and watchmakers 454 Job and die setters,'metal 455 Locomotive engineers 456 Locomotive firemen 461 Hachir.ists 462 Hachinists aoprentices Bechaaics and repairmen 470 Air conditioning, heating, and refrigeration 471 Aircraft 472 Automobile body repairmen 473 Automobile mechanics 474 Automobile mechanic apprentices 475 Data processing machine repairmen 480 Farm implement 481 Heavy equipment mechanics, lacl. diesel 482 Household appliance and accessory installers and mechanics 483 Loom fixers 484 Office machine 485 Radio and television 486 Railroad and car shop 491 Sechanic, exc. auto, apprentices 492 Miscellaneous mechanics and repairmen 495 Not specified mechanics and repairmen 501 Billets; grain, flour, and feed 50L Millwrignts 503 Holders, metal 504 Holder apprentices Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Metier, picture projectionists Opticians and lens grinders and polishers Painters, construction and maintenance Painter apprentices Paperhangers Pattern and model makers, e x c . paper Photoengravers and lithographers Piano and organ tuners and repairmen Plasters Plasterer aoprer.tices Plumbers ancTpipe fitters Plumber and pipe fitter apprentices Power station operators Pressmen and plate printers, printing Pressman apprentices Rollers ana finishers, metal Roofers and slaters Sheetmetal workers and tinsmiths Sheetmetal apprentices Shipfitters Shoe repairmen Sign painters and letterers Stationary engineers Stone cutters and stone carvers Structural zetal craftsmen Tailors Telephone installers and ' repairmen - Telephone linemen and splicers Tile* setters Tool and die makers Tool and die maker apprentices Upholsterers Specified craft apprentices, a . e . c . Not specified apprentices Craftsmen and kindred workers, a.c. F o n e r members of the Armed Forces OPERATIVE, EXCEPT TRANSPORT 601 Asbestos and insulation workers 602 Assemblers 603 Slasters and powdermen 604 Bottling and canning operatives 605 Chainmen, roda«n, and axieo, surveying 610 Checkers, examiners, and inspectors, manufacturing y 611 Clothing ironers and pressers 612 Cutting operatives, me.e. 6-13 Dressmakers and seamstresses, except factory 614 Drillers, earth T 6 15 Dry tail installers and lathers 6i0 Dyers Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 232 621 Filers, polishers, senders, and buffers 622 Fumaceraen, sneltermen, and pourers 623 Garage workers and gas station attendants 624 Graders and sorters, manufacturing 625 Produce graders and packers, exc. factory and farm 626 Heaters, aatal 630 Laundry and dry cleaning operatives, a.e.c. 639 Beat cutters and butchers, 'exc. manufacturing 633 Heat cutters and butchers, manufacturing 634 Heat wrappers, retail trade 635 Betal platers 636 Milliners 640 dine operatives, n.e.c. 6 TRANSPORT BQ0IPB28T OPERATIVES 701 Boatmen and canalaen 703 Busdrivers 704 Conductors and motormen, urban rail transit 705 Deliverymen and routenen 706 Fork lirt and tow motor ooeratives 7,10 Botorxea; mine, factory, logging camp, etc.. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 716 Parking attendants 712 Railroad brakeraen 7'13 Railroad switchmen 71 4 Taxicab drivers and chauffeurs 715 Truck drivers LASORERS, EXCEPT FARM 740 Animal caretakers, exc, farm 750 Carpenters1 helpers 751 Construction laborers, etc. carpenters' helpers 752 Fishermen and oystemen 753 Freight and material handlers 754 Garbage collectors 755 Gardeners and grour.dskeepers, exc. farm 760 Longshoremen and stevedores 761 Lumbermen, raftsmen, and woodchoppers 762 Stcckhandlers 763 Teamsters 764 Vehicle washers and equipment cleaners 770 Warehousemen, n.e.c. 780 tiiscellaneous laborers 785 Not soecified laborers F133EBS AND FlHfi MANAGERS 801 Farmers (owners and tenants) 802 Farm manacers FARM LA30RERS AND FARM FOREMEN 821 Farm foremen 822 Far* laborers, wage workers 823 Farm laborers, unpaid family workers 824 Farm service laborers, self-employed SERVICE 80BK2SS, EXC. PRIVATE HOUSEHOLD Cleaning service workers 901 Chambermaids and maids, exc. private households 902 Cleaners and charwomen 903 Janitors and sextons Food service workers 910 Bartenders 911 3usboys 91 2 Cooks, exc. private household 9'13 Dishwashers * T 414 Food counter and fountain workers 915 Waiters Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 916 rood service workers, n-e.c., exc. private household Health service workers 921 Dental assistants 922 Health aides, exc. nursing 923 Health trainees 92* Day nidwives 925 Nursing aides, orderlies, and attendants 926 Practical nurses Personal service workers 931 Airline stewardesses 932 Attendants, recreation and amusement 933 Attendants, personal service, n.e.c. 934 Baggage porters and bellhops 935 Barbers 940 Boarding and lcdginghcuse keepers 541 Bootblacks ¥ 942 Child care Yorkers, exc. private household 943 Elevator operators 944 Hairdressers and cosmetologists 94 5 Personal service apprentices 950 Housekeepers, exc. private household .952 School ■ onitors 953 Ushers, recreation and amusement 554 Welfare service aides Protective service workers 950 Crossing guards and bridge tenders 951 Firemen,” fire protection” 952 Guards and watchmen 953 Harshals and constables 964 Folicemen and detectives 965 Sheriffs and bailiffs PRIVATE HOUSEHOLD WORKERS 930 Child care workers, private household 991 Cooks, private household 982 Housekeepers, private household 963 Laundresses, private household 554 Balds and servants, private household Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 235 1^83 o Ccodl {Aeizui-zJ ^ N.7 Coding for: January 19B3-1991 occ and ocdyr 1984-1990 occ20 February 1984,1990 occ March 19S3-1991 occ and occiyr June 1983-1991 occ October 1983-1991 occ November 19B3-1991 occ compose 1994-1989 occ Outgoing 1983-1931 occ and occx 1980 Census of Fcpulaticn - Occupation Classification (Numbers in parentheses are the 1930 SCC code equivalent; “Ft" means part; "n.e.c" means not elsewhere classified. The numbers were verified by Technical Paper 59, Bureau of Census, Feb 39! . Census Code not used c-2 MANAGERIAL AND PROFESSIONAL SPECIALTY OCCUPATIONS 3-199 Executive, administrative, and managerial occupations 3 -3 7 Legislators (1111 3 Chief executives 6 general administrators, public administration 4 ( 112) Administrators and officials, public administration (1132-1139) 5 Administrators, protective services (1131) g Financial managers (122) 7 Personnel and labor relations managers (123) 8 Purchasing managers (124) 9 not used 1 0 -1 2 Managers, marketing, advertising; and public relations (125) 13 "Administrators, education and related fields (128) 14 Managers, medicine and health (131) 15 Managers, properties and real estate (1353) is Postmasters and mail superintendents (1344) n Funeral directors (pt 1359) 13 Managers and administrators, n.e.c. (121, 128, 127, 132-139, 19 exc. 1344, 1353, pt 1359) not used 2 0 -2 2 Management related occupations 23-37 Accountants and auditors (1412) 23 Underwriters (1414) 24 Other financial officers (1415, 1419) 25 Management analysts (142) 28 Personnel, training, and labor relations specialists (143) 27 Purchasing agents and buyers, farm products (1443) 28 Buyers, wholesale and retail trade except farm products (1442) 29 not used 30-32 Purchasing agents and buyers, n.e.c. (14491 33 Business and promotion agents (145) 34 Construction inspectors (1472) 35 Inspectors and compliance officers, exc. construction (1473! 36 Management related occupations, n.e.c. (149) 37 not used 38-42 Professional specialty occupations 4 3 -1 9 9 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 236 Engineers, architects, and surveyors 43-63 Architects (161! 43 Engineers 44-59 Aerospace (1622) 44 Metallurgical and materials (1623) 45 Mining (1624) 46 Petroleum (1625! 47 Chemical (1626) 48 Nuclear (1627) 49 not used 50-52 Civil (1628) 53 Agricultural (1632) 54 Electrical and electronic (1633, 163S) 55 Industrial (1634) 56 Mechanical (1635) 57 Karine and naval architects (1637) 58 Engineers, n.e.c. (163 9) 59 not used 60-62 Surveyors and napping scientists (164) 63 Mathematical and computer scientists 64-68 Computer systems analysts and scientists (171) 64 Operations and systems researchers and analysts (172) 65 Actuaries (1732) 66 I Statisticians (1733) 67 Mathematical scientists, n.e.c. (1739) 63 Natural scientists 69-83 Physicists and astronomers (1842, 1843) 69 I not used 70-72 Chemists, except biochemists (1845) 73 Atmospheric and space scientists (1846! 74 I Geologists and geodesists (1847) 75 Physical scientists, n.e.c. (1849) 76 Agricultural and food scientists (1853) 77 I Biological and life scientists (1854) 78 Forestry and conservation scientists (1852) 79 not used 80-82 Medical scientists (1855) 83 I Health diagnosing occupations 84-89 Physicians (261) 84 Dentists (262) 85 1 Veterinarians (27) 86 Optometrists (2 8 1 ! 87 Podiatrists (283! 68 I Health diagnosing practitioners, n.e.c. (289) 89 not used 90-94 Health assessment and treating occupations 95-106 Registered nurses (29) 95 I Pharmacists (301) 96 Dietitians (302) 97 therapists 98-105 I Inhalation therapists (3C31) 98 Occupational therapists (3032) 99 not used 100-102 1 Physical therapists (3033) 103 Speech therapists (3034) 104 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 237 Therapists, n.e.c. (3039) 105 Physicians' assistants (304) ICS not used 107-112 Teachers, postsecondary 113-154 Earth, environmental, and marine science teachers (2212) 113 Biological science teachers (2213) 114 Chemistry teachers (2214) 115 Physics teachers (2215) IIS S’atural science teachers, n.e.c. (22IS) 117 Psychology teachers (2217) 118 Economics teachers (2218! 119 not used 12 0-122 History teachers (2222) 123 Political science teachers (2223) 124 Sociology teachers (2224) 125 Social science teachers, n.e.c. (2225) 12S Engineering teachers (222S) 127 Mathematical science teachers (2227) 1 2 s Computer science teachers (2228) 129 not used 130-132 Medical science teachers (2231) 133 Health specialties teachers (2232) 134 Business, commerce, and marketing teachers (2233! 135 Agriculture and forestry teachers (2234) 136 Art, drama, and music teachers (223 5) 13 7 Physical education teachers (2236) 133 Education teachers (2237! 139 not used 140-142 English teachers (223S) 143 Foreign language teachers (2242) 144 Law teachers (2243) 145 Social work teachers (2244) 146 Theology teachers (2245) 147 Trade and industrial teachers (2246) 148 Home economics teachers (2247) 149 not used 150-152 Teachers, postsecondary, h.e.c. (2249) 153 Postsecondary teachers, subject not specified 154 Teachers, except postsecondary 155-159 Teachers, prekir.dergarten and kindergarten (231) 155 Teachers, elementary school (232) 156 Teachers,, secondary school (233) 157 Teachers, special education (235) 158 Teachers, n.e.c. (236, 239) 159 not used 160-162 Counselors,educational and vocational (24) 163 Librarians, archivists, and curators 164-165 Librarians (251) 164 Archivists and curators (252) 165 Social scientists and urban planners 166-173 Economists (1912) 166 Psychologists (1915) , 157 Sociologists (1916! i 168 Social scientists, n.e.c. (1913, 1914, 1S19) 169 not used 170-172 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 238 Urban planners (192) 173 Social, recrearion, and religious workers 174-177 Social workers (2032) \ 174 Recreation workers (2033) 175 Clergy (2042) 176 Religious workers, n.e.c. (2049) 177 lawyers and judges 1 7 8 -1 7 9 Lawyers (211! 178 Judges (212) 179 not used 1 8 0 - 1 3 2 Writers, artists, entertainers, and athletes 1S3-199 Authors (321) 1 8 3 Technical writers (398! 134 Designers (322) I85 Musicians and composers (323) 166 Actors and directors 1324) 1 8 7 Painters,sculptors,craft-artists, i artist prir.t-makers (325) 188 Photographers (328) 189 not used 190-192 Dancers (327) 193 Artists, performers, and related wcrkers, n.e.c. (328, 329) 194 Editors and reporters (331) 195 not used 196 Public relations specialists (332) 197 Announcers (333! 198 Athletes (3 4 ) 199 not used 2 0 0 -2 0 2 TECHNICAL, SALES, AND ADMINISTRATIVE SUPPORT OCCUPATIONS 203 -389 Technicians and related support occupations 203-235 Health technologists and technicians 203-208 Clinical laboratory technologists and technicians (382) 203 Dental hygienists (383! 204 Health record technologists and technicians (384) 205 Radiology technicians (385) 208 Licensed practical nurses (368) 207 Health Technologists and technicians, n.e.c. (389) 208 not used 209-212 Technologists and technicians, except health 213-235 Engineering and related technologists and technicians 213-218 Electrical and electronic technicians (3711) 213 Industrial engineering technicians (37121 214 Mechanical engineering technicians (3713) 215 Engineering technicians, n.e.c. (3719) 218 Drafting occupations (372) 217 Surveying and mapping technicians (373) 218 not used 219-222 Science technicians 223-225 Biological technicians (382) 223 Chemical technicians (3 331) 224 Science technicians, n.e.c. (3832, 3833, 384, 389) 225 - Technicians; except health, engineering, and science 228-235 Airplane pilots and navigators (825) ■ 226 Air traffic controllers (392) 227 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 239 Broadcast equipment operators (393) 223 Computer programmers (3971, 3972) 229 not used 230-232 Tool programmers, numerical control (3 9 7 4 ) 233 Legal assistants (39S) 234 Technicians, n.e.c. (399) 239 not used 236-242 Sales occupations 243-235 Supervisors and proprietors, sales occupations (40) 243 not used 244-252 Sales representatives, finance and business services 253-257 Insurance sales occupations (4122) 253 Real estate sales occupations (4123) 254 Securities 6 financial services sales occupations (4124) 255 Advertising and related sales occupations (4153) 256 Sales occupations, ocher business services (4152) 257 Sales representatives, commodities except retail 258-259 Sales engineers (421) 258 Sales representatives, mining, manufacturing, 6 wholesale 259 (423, 424) not used 260-262 Sales workers, retail and personal services 263-278 Sales workers, motor vehicles and boats (4342, 4 3 4 4 ) 263 Sales workers, apparel (4346) 264 Sales workers, shoes (4351) 265 Sales workers, furniture and home furnishings (4348) 266 Sales workers; radio, TV, hi-fi, S appliances (4343,4352! 267 Sales workers, hardware and building supplies (4353) 268 Sales workers, parts (4367) 269 not used 270-273 Sales workers, other commodities (4345, 4 3 4 7 , 4 3 5 4 , 4356, 274 4359, 4362, 4369) Sales counter clerks (4363) 275 Cashiers (4364) 276 Street and door-co-docr sales workers (4366) 277 News vendors (4365) 278 not used 279-282 Sales related occupations 283-28 5 Demonstrators, promoters and models, sales (445) 283 Auctioneers (4 4 7 ) 284 Sales support occupations, n.e.c. (4 4 4 , 446, 449) 285 not used 286-302 Administrative support occupations, including clerical 303-389 Supervisors, administrative support occupations 303-307 Supervisors, general office (4511,4513,4514,4516,4519,4529) 303 Supervisors, computer equipment operators (4512) 304 Supervisors, financial records processing (4521) 305 Chief communications operators (4523) 306 Supervisors; distribution, scheduling, and adjusting clerks 307 (4522, 4524-4528) Computer equipment operators 308-3 09 Computer operators (4612) 308 Peripheral equipment operators (4613) 309 not used 310-312 Secretaries, stenographers and typists 313-315 I Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Secretaries (4622) 313 Stenographers (4623) 314 Typists (4S24) 315 Information clerks 316-•323 Interviewers (4642) 316 Hotel clerks (4S43) 317 Transportation tickec and reservation agents (4644) 318 Receptionists (464S) 319 not used 320- 322 Information clerks, n.e.c. (4S49) 323 not used 324 Records processing occupations, except financial 325-■336 Classified-ad clerks (4662! 325 Correspondence clerks (4663) 326 Order clerks (4664) 327 Fersonr.el clerks, except payroll and timekeeping (4692) 323 Library clerks (4694) 329 not used 330- 334 File clerks (4696) 335 Records clerks (4599) 336 Financial records processing occupations 337- 344 Bookkeepers, accounting, and auditing clerks (4712) 337 Payroll and timekeeping clerks (4713) 333 Billing clerks (4715) 339 not used 340- 342 Cost and rate clerks (4716) 343 3illing, posting, ar.d calculating machine operators (4718) 344 duplicating, mail and other office machine operators 345- 347 Duplicating machine operators (4722) 345 Mail preparing and paper handling machine operators (4723) 346 office machine operators, n.e.c. (4729) 347 Communications equipment operators 348- 353 Telephone operators (4732) 348 Telegraphers (4733) 349 r.ct used 350- 352 Communications equipment operators, n.e.c. (4739) 353 Mail and message distributing occupations 354- 357 Postal clerks, exc. mail carriers (4742) 354 Kail carriers, postal service (4743) 355 Kail clerks, exc. postal service (4744) 356 Messengers (4745! 357 not used 358 Material recording, scheduling, and distributing clerks, n.e.c. 359- 374 Dispatchers (4751) 359 not used 360- 362 Production coordinators (4752) 363 Traffic, shipping, and receiving clerks (4753) 364 Stock and inventory clerks (4754) 365 Meter readers (4755) 366 not used 367 Weighers, measurers, and checkers (4756) 368 Samplers (4757) 369 not used 370-•372 Expediters (4758) 373 Material recording, scheduling, & distributing clerks, n.e.c. 374 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 241 (4759) Adjusters and investigators 37 5 -375 Insurance adjusters, examiners, and investigators (4782) 375 Investigators and adjusters, except insurance (4783) 37s Eligibility clerks, social welfare (4784) 377 Bill and account collectors (478S) 378 Miscellaneous administrative support occupations 379-385 General office clerks (453) 3 79 not used 380-382 Bank tellers (4791) 333 Proofreaders (4792) 384 Data-er.try keyers (47931 395 Statistical clerks (4794) 395 Teachers aides (4795) 387 not used 388 Administrative support occupations, n.e.c. (4787, 4799! 339 net used 390-402 SERVICE OCCUPATIONS 403-459 Private household occupations 403-407 Launderers and iror.ers (503) 403 Cooks, private household (504) 404 Housekeepers and butlers (505! 405 Child care workers, private household (505) 406 Private household cleaners and servants (502, 507, 509) 407 not used 408-412 Protective service occupations 413-427 Supervisors, protective service occupations 413-415 Supervisors, firefighting t fireprevention occupations (5111) 413 Supervisors, police and detectives (5112) 414 Supervisors, guards (5113) 415 Firefighting and fire prevention occupations 415-417 Fire inspection and fire prevention occupations (5122) 416 Firefighting occupations (5123) 417 Police and detectives 418-424 Police and detectives, public service (5132) 418 not used 419-422 Sheriffs, bailiffs, and other law enforcement officers (5134) 423 Correctional institution officers (5133) 424 Guards 425-427 Crossing guards (5142) 425 Guards and police, exc. public service (5144) 425 Protective service occupations, n.e.c. (5149) 427 not used 428-432 Service occupations, except protective and household 433-469 Food preparation and service occupations 433-444 Supervisors, food preparation and service occupations (5211) 433 Bartenders (5212) 434 Waiters and waitresses (5213) 435 Cooks, except short order (5214) 436 Short-order cocks (5215) 437 / Food counter, fountain and related occupations (5216) 438 Kitchen workers, food preparation (5217) 439 not used 44Q-442 I Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Waiters'/waitresses' assistants (5213) 443 Miscellaneous food preparation occupations (5229) 444 Health service occupations 4 4 5 -4 4 7 Dental assistants (5232! 445 Health aides, except nursing (5233) 44 s Nursing aides, orderlies, and attendants (5235) 447 Cleaning and building service occupations, except household 448-4 5s Supervisors, cleaning and building service workers (5241) 448 Maids and housemen (5242,5249! 443 not used 450-452 Janitors and cleaners (5244) 453 Elevator operators (5245) 454 Pest control occupations (5246) 455 Personal service occupations 456-459 Supervisors, personal service occupations (5251) 456 Sarbers (5252! 457 Hairdressers and cosmetologists (5253) 4 5a Attendants, amusement and recreation facilities (5254) 459 not used 460-462 Guides (5255) 463 Ushers (52 56) 464 Public transportation attendants (5257) 465 Baggage porters and bellhops (5262) 466 Welfare service aides (5263) 467 Child care workers, except private household (5264) 466 Personal service occupations, n.e.c. (S25B, 5269) 469 not used 470-472 FARMING, FORESTRY, AND FISHING OCCUPATIONS 473 -499 Farm operators and managers 473-476 Farmers, except horticultural (5512-5514) 473 Horticultural specialty farmers (5515) 474 Managers, farms, except horticultural (S522-SS24) 475 Managers, horticultural specialty farms (5525) 476 Other agricultural and related occupations 4 77-48 9 Farm occupations, except managerial 477-484 Supervisors, farm workers (5611) 477 not used 478 Farm workers (5612-5617) 479 not used 480-432 Marine life cultivation workers (5618) 483 Nursery workers (S619) 484 Related agricultural occupations 485-489 Supervisors, related agricultural occupations (5621) 485 Groundskeepers and gardeners, except farm (5622) 486 Animal caretakers, except farm (5624) 487 Graders and sorters, agricultural products (5625) 486 Inspectors, agricultural products (5627) 489 not used 490-493 Forestry and logging occupations 494-496 Supervisors, forestry and logging workers (571) 494 Forestry workers, except logging (572) 495 Timber cutting and logging occupations (573, 579) 496 Fishers, hunters, and trappers 497-499 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Captains and other officers, fishing vessels (pt 8241) 497 Fishers (583) 498 Hunters and trappers (534) 499 not used SCC-5C2 PRECISION PRODUCTION, CRAFT. AND REPAIR OCCUPATIONS 5C3-S59 Mechanics and repairers 5 C3 -5 4 9 Supervisors, mechanics and repairers (S3) 503 i nct used 5C4 Mechanics and repairers, except supervisors 505-549 Vehicle and mobile equipment mechanics and repairers 505-517 , Automobile mechanics (pt 6111) 5C5 ! Automobile mechanic apprentices (pt Sill) 5 0 s Bus, trucic, and stationary engine mechanics (6112) SC7 Aircraft engine mechanics (6113) 508 1 Small engine repairers (6114) 5 C9 1 not used S10-513 Automobile body and related repairers (6115) 514 1 Aircraft mechanics, exc. engine (6116) =19 j Heavy equipment mechanics (6117) 919 Farm equipment mechanics (6118) 917 Industrial machinery repairers (613) 9 1 a Machinery maintenance occupations (614) 519 not used 520-522 Electrical and electronic equipment repairers 523-533 Electronic repairers, communications i industrial equipment 523 I (6151, 6153, 6155) not used 924 1 Data processing equipment repairers (6154) 925 1 Household appliance and power tool repairers (6156) 526 Telephone line installers and repairers (6157) 527 . not used 528 ^ Telephone installers and repairers (6158) 529 not used 530-532 Miscellaneous electrical and electronic equipment repairers 533 1 (6152, 6159) " Heating, air conditioning, and refrigeration mechanics (616) 534 Miscellaneous mechanics and repairers 535-549 Camera, watch, 6 musical instrument repairers (6171, 6172) 535 ] Locksmiths and safe repairers (6173) 539 not used 537 Office machine repairers (6174) S3B I Mechanical controls and valve repairers (6175! 539 not used 540-542 Elevator installers and repairers (6176) 543 I Millwrights (6178) 544 . not used 545-546 Specified mechanics and repairers, n.e.c. (6177, 6179) 547 not used 5 4 9 S Not specified mechanics and repairers 549 not used , S50-S52 Construction trades * 553-559 I Supervisors, construction occupations 553-558 * Supervisors; brickmasons, stonemasons, and tile setters (6312) 553 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. I Supervisors. carpenters and related workers (6313) 554 P Supervisors, electricians 6 power transmission installers 555 (613*1 Supervisors; painters, paperkangers, and plasterers (6315) 556 Supervisors plumbers, pipefitters, and steamfitters (6316) 557 P Supervisors n.e.c. (6311, 63131 558 not used 559-562 Construction trades, except supervisors 563-599 ¥ Srickmasons and stonemasons (pt 6412, pt 6413) 563 Brickmason and stonemason apprentices (pt 6412, pt 6413) 584 Tile setters, hard and scft (6414, pt 6462) 565 W Carpet installers (pt 6462) 565 Carpenters (pt 6422) 567 not used 568 P Carpenter apprentices (pt 6422) 569 not used 570-572 Drywall installers (6424) 573 not used 574 P Electricians (pt 6432) 575 Electrician apprentices (pt 6432) 576 Electrical power installers and repairers (6433) 577 P not used 573 Painters, construction ar.d maintenance (6442) 579 not used 580-982 I Paperkangers (644 3) 583 Plasterers (6444! 584 Plumbers, pipefitters, and steax.fitters (pt 645) 585 not used 586 P Plumber, pipefitter, and steamfitter apprentices (pt 645) 537 Concrete and terrazzo finishers (6453) 588 Glaziers (6464) 589 P r.ct used 590-592 Insulation workers (5465) 593 Paving, surfacing, and tamping equipment operators (6456) 594 P Roofers (6468) 595 Skeetmetal duct installers (6472) 596 Structural metal workers (6473) 597 Drillers, earth (6474) 598 P Construction trades, n.e.c. (6467, 6475, 6476, 6479) 599 not used 600-612 Extractive occupations 613-617 I Supervisors, extractive occupations (532) 613 Drillers, oil well (652) 614 Explosives workers (653) 615 I Mining machine operators (654) 616 Mining occupations, n.e.c. (656) 617 not used 618-632 Precision production occupations 633-699 Supervisors, production occupations (67, 71) 633 Precision metal working occupations 634-655 Tool and die makers (pt 6811) 634 Tool and die maker apprentices (pt 6811) 635 Precision assemblers, metal (6812) 636 Machinists (pt 6313) 637 not used 638 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Machinist apprentices (pt 6B13) 639 net used 640-642 Boilermakers (6814) 643 Precision grinders, filers, and tool sharpeners (6816) 544 Patternmakers and model makers, metal (6817) 645 Lay-out workers (6821) 646 Precious stones and metals workers (jewelers) (6822, 6 8 6 6 ) 647 1 not used 648 Engravers, metal (6 823) 64 9 not used 650-652 ( Sheet mecai workers (pt 6624) 653 Sheet metal worker apprentices (pt 68241 654 . Miscellaneous precision metal workers (6829) 655 Precision woodworking occupations 656-659 Patternmakers and model makers, wood (6331) 656 Cabinet makers and bench carpenters (6832) 657 j Furniture and wood finishers (6835) 653 ' Miscellaneous precision woodworkers (6839) 659 not used 660-665 I Frecisicn textile, apparel and furnishings machine wcrkers 666-674 Dressmakers (pc 68 52, pt 7752) 666 Tailors (pt 6852) 667 I Cpholsterers (6853) 663 Shoe repairers (6854) 669 not used 670-672 Apparel and fabric patternmakers (6856) 673 I Miscellaneous precision apparel 1 fabric wcrkers 674 (6859, pt 7752 ) Precision workers, assorted materials 675-684 I Hand molders and shapers, except jewelers (6861) 675 Patternmakers, lay-out workers, and cutters (6862! 676 Optical goods workers (6e64, pt 7 4 7 7 , pt 7677) 677 ( Dental laboratory and medical appliance technicians C6S65) 678 Bookbinders (6844) 679 not used 680-682 Electrical and electronic equipment assemblers (6857) 683 ( Miscellaneous precision workers, n.e.c. (6869) 684 not used 685 Precision food production occupations 636-688 I Butchers and meat cutters (6371) 686 J Bakers (6872) 687 Food batchmakers (6873, 6879) 683 I Precisian inspectors, testers, and related workers 689-693 Inspectors, testers, and graders (£831, 828) 6S5 not used 690-692 Adjusters and calibrators (66B2) 693 I Plant and system operators 694-699 * Water and sewage treatment plant operators (691) 694 Power plant operators (pt 693) 695 ( Stationary engineers (pt 693, 7663) 696 Miscellaneousnot used plant and system cperators (692, 694, 695, 696) 697-698 699 [ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 246 hoc u s e d 700-702 OPERATORS, FABRICATORS, AND LABORERS 703-889 Machine operators, assemblers, and inspectors 703-799 Machine operators and tenders, except precision 703-779 Metalworking and plastic working machine operators 703-715 Lathe and turning machine set-up operators (7312! 703 Lathe and turning machine operators (7512) 701 m Milling and planing machine operators (7313, 7513) 705 Punching and stamping press machine operators 70’g (7314, 7317, 7514, 7517) m Rolling machine operators (731S, 751S) 707 Drilling and boring machine operators (7318, 7518) 70S Grinding, abrading, buffing, 4 polishing machine operators 709 m <7322, 7324, 7522) not used 710-712 Forging machine operators (7319, 7519) 713 Numerical control machine operators (732S) 714 m Miscellaneous metal, plastic, stone, i glass working 715 machine operators (7329, 7529 ) not used 716 m Fabricating machine operanors, n.e.c. (7339, 7539) 717 not used 718 Metal and plastic processing machine operators 719-725 E Molding and casting machine operators (7315,7342,7515,7542) 719 not used 720-722 Metal plating machine operators (7343, 7543) 723 Heat treating equipment operators (7344, 7544) 724 V Miscellaneous metal 6 plastic processing machine operators 725 (7349, 7549) Woodworking machine operators 726-73 3 II Wood lathe, routing, 6 planing machine operators 726 (7431, 7432, 7631, 7632) Sawing machine operators (7433, 7633) 727 f Shaping and joining machine operators (7435, 7635) 728 Nail and tacking machine.operators (7636) 729 not used 730-732 Miscellaneous woodworking machine operators 733 R (7434, 7439, 7634, 7639) Printing machine operators 734-737 Printing machine operators (7443, 76431 734 I Photoengravers and lithographers (6842, 7444, 7644) 735 Typesetters and compositors (6841,7642) 736 Miscellaneous printing machine operators (664S, 7449, 7649) 737 I Textile, apparel, and furnishings machine operators 738-749 Winding and twisting machine operators (7451, 7S51) 738 Knitting, looping, taping, t weaving machine operators 739 (7452,7652) I not used 740-742 Textile cutting machine operators (7654) 743 Textile sewing machine operators (7655) 744 I Shoe machine operators* (7656) 745 not used “ 746 Pressing machine operators (7657) 747 I Laundering and dry cleaning machine operators (6855, 7658) 748 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 247 Miscellaneous textile machine operators (7459, 7659) 743 not used 750-752 Machine operators 753-779 Cementing and gluing machine operators (7661) 753 Packaging and filling operators (7462, 7652) 754 Extruding and forming machine operators (7463, 7663) 755 Mixing and blending machine operators (7664) 756 Separating, filtering, and clarifying machine operators 757 (7476. 7656, 7676 ) Compressing and compacting machine operators (7467, 7667) 758 Painting and paint spraying machine operators (7669! 759 not used 760-752 Roasting and baking machine operators, food (7472. 7672) 763 Washing, cleaning, and pickling machine operators (7673; 764 Folding machine operators (7474, 7674) 755 Furnace, kiln, and oven operators, exc. food (7675! 766 not used 757 Crushing and grinding machine operators (pt 7477. pt 7677) 753 Slicing and cutting machine operators (7478, 7678) 769 not used 770-772 Motion picture projectionists (pt 7479) 773 Photographic process machine operators (68S3, 6868, 7671) 774 not used 775-776 Miscellaneous machine operators, n.e.c. (pt 7479,7665,7579) 777 not used 7 7 3 Machine operators, not specified 779 not used 7eo-782 Fabricators, assemblers, and hand working occupations 783-795 Welders and cutters (7332, 7532, 7714) 783 Solderers and brazers (7333, 7533, 7717) 794 Assemblers £772, 774) 785 Eand cutting and trimming occupations (7753) 786 Hand molding, casting, and forming occupations (7754, 7755! 787 not used 788 Hand painters, coating, and decorating occupations (7756) 7B3 not used 790-792 Hand engraving and printing occupations (7757) 793 Hand engraving and polishing occupations (7758) 794 Miscellaneous hand working occupations (7759) 795 Production inspectors, testers, samplers, and weighers 796-753 Production inspectors, checkers, and examiners (782, 787) 796 Production testers (783) 797 Production samplers and weighers (784) 79 a Graders, and sorters, exc. agricultural (785) 79 s not used 800-6C2 transportation and material moving occupations 803-859 Kotor vehicle operators 803-814 Supervisors, motor vehicle operators (8111) 803 Truck drivers, heavy (8212, e213) 804 Truck drivers, light (8214) 80S Driver-sales workers (8219) 806 not used 807 Bus drivers (8215) 808 Taxi cab drivers and chauffeurs (8216) 809 not used 810-812 I Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 248 P Parking lot attendants (874) 813 P Motcr transportation occupations, n.e.c. (8219) 614 not used 815-822 Transportation occupations, except motor vehicles 823-834 P Hail transportation occupations 823-826 Railroad conductors and yardnasters (8113) 823 locomotive operating occupations (8232) 824 Railroad brake, signal, and switch operators (8233) 825 F-1 Rail vehicle operators, n.e.c. (8238) 826 not used 827 Water transportation occupations 628-834 P Ship captains & mates, except £ i s h i r . g boats (pt 8241, 8242) 828 Sailors and deckhands (8243) 829 not used 832-832 P Marine engineers (8244) 833 Bridge, lock, and lightbcuse tenders (8245) 834 not used 835-342 ■ Material moving equipment operators 843-859 Supervisors, material moving equipment operators (812) 843 Operating engineers (8312) 844 ■ longshore equipment operators (8313) 845 P not used 846-847 Hoist and winch operators (8314) 848 * Crane and tower operators (8315) 849 I- not used 853-652 Excavating and loading machine operators (831S) 853 not used 854 I Grader, dozer, and scraper operators (8317) 855 P Industrial truck and cractor equipment operators (8318) 856 not used 857-85a | Miscellaneous material moving equipment operators (8319) 859 P not used 860-862 Handlers, equipment cleaners, helpers, and laborers 863-869 ■ Supervisors, handlers, equipment cleaners, and laborers, n.e.c. 863 P (85) Helpers, mechanics and repairers (883) 864 ■ Helpers, Construction and Extractive Occupations 865-867 1 Helpers, construction trades (8641-8S4S, 8648) 865 r Helpers, surveyor (aS46) 866 Helpers, extractive occupations (865) 867 1 not used 868 v Construction laborers (871) 869 not used 870-872 ■ Production helpers (861, 562) 873 1 net used 874 Freight, stock, and material handlers 875-883 ■ Garbage collectors (8721) 875 Stevedores (8723) 876 i Stock handlers and baggers (3724) 677 Machine feeders and offbearers (3725) 878 1 net used 879-882 i Freight, stocks, and material handlers, n.e.c. (8726) 633 not used 334 i Garage and service station related occupation (873) 685 not used 886 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission 249 m i ' 2 o o o Vehicle washers and equipment cleaners.(875! w 887 Hand packers and packagers (8 7 S 1 ) ntrrAfi 880 Laborers, except construction•(8769) 0 VQC 0 3 9 not used 890-994 Assigned to persons whose labcr force status is unemployed 90S and whose last job was Armed Forces not used 9CS-9CS Assigned to persons whose labor force status is ur.eraplcyed 909 and who last worked in 1974 or earlier. not used 913♦ last job was in Armed Forces (1992) 951 January 1992-1953 occ 1992 oec20 February 1994-1999 occ and oce2jb 1994-1599 oec20 1995 occbse 1996,1999 occten March 1992-2033 occ and ccclyr June 1992-1998 occ and occ2jb October 1992-1998 occ 1994-1998 occ2jb November 1992-1998 occ 1994-1998 occ2jb Basic 1998-2030 occ and occ2jb Comp Use 1993-1998 occ 1994-1998 occ2jb Outgoing 1992-2000 occ and occx 1994-2000 occ2jb 1990 Census of Population - Occupation Classification (Numbers in parenthese are the 1990 SOC code equivalent; "Pt" means part; "n.e.c" means not elsewhere classified.) Census Code not used 0-2 MANAGERIAL AND PROFESSIONAL SPECIALTY OCCUPATIONS 3-199 Executive, administrative, and managerial occupations 3-37 Legislators (1 1 1 ) 3 Chief executives 6 general administrators, public administration 4 ( 1 1 2 ) Administrators and officials, public administration (1132-1139) 5 Administrators, protective services (1131) s Financial managers (122) ; 7 Personnel and labor relations managers (123) 8 Purchasing managers (124) 9 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 250 APPENDIX F: TESTS FOR HETEROSKEDASTICITY AND AUTOCORRELATION IN THE DATA Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. log: u:\dissertation\emp_levels_tests.log log type: text opened on: 20 Jun 2002, 18:51:50 . / ‘ THIS ANALYSIS EVALUATES THE EFFECT OF ATR INDICTMENTS ON EMPLOYMENT BY OCCUPATION USING CROSS-SECTION TIME SERIES TECHNIQUES. THE VAR SICYEAR WAS CREATED TO ACCOMMODATE FOR THE INABILITY OF STATA TO COMPUTE THE SIC CODE AND OCCUPATION AND YEAR VARIATION. VAR EMPLOYMT REPRESENTS NUMBER OF EMPLOYEES BY OCCUPATION AND SIC CODE * / . use ■\\\sparkynt\userfiles\ypho\diss\sas_programs\datasets\bigatr_final.dta‘ , clear . egen occsic=concat(occ sic) /‘concatenates sic and occ variables*/ . egen sicyear=concat(sic year) / ‘concatenates sic and year variables*/ . egen occyear=concat(occ year) / ‘concatenates occ and year variables*/ . egen sicyearocc=concat(sic year occ) /‘concatenates sic, year, and occ variables*/ . encode occsic, gen(OCCSIC) /‘transforms character value to numeric*/ . encode sicyear, gen(SICYEAR) / ‘ transforms character value to numeric*/ . encode occyear, gen(OCCYEAR) /‘ transforms character value to numeric*/ . encode s ic y e a ro c c , gen(SICYEAROCC) / ‘ transform s c h a ra c te r va lu e to n u m e ric*/ . iis occ /‘allows one to take the difference without subtracting the 1999*/ . tsset OCCSIC year /‘value from 1979 across 2 difference SIC codes*/ panel variable: OCCSIC, 1 to 140 time variable: year, 1979 to 1999 . replace avghhi=avghhi‘ lOO /‘converting hhi to 1 to 10000 range*/ (2940 re a l changes made) . gen tre n d = y e a r-1978 . gen demploymt=d.employmt (140 missing values generated) . gen davghhi=d.avghhi (140 missing values generated) . gen dk_lexp=d.k_lexp (140 missing values generated) . replace realgdp=realgdp/100 (2940 re a l changes made) . gen drealgdp=d.realgdp (140 missing values generated) . replace davghhi=davghhi*lO (2800 re a l changes made) . replace atrhhi=atr*avghhi (1855 re a l changes made) . gen llatrhhi=Hatr*avghhi (140 missing values generated) . gen 13atr=1.12atr (420 missing values generated) . gen 14atr=1.13atr (560 missing values generated) . gen 15atr=1.14atr (700 missing values generated) . gen 16atr=1.15atr (840 missing values generated) . gen 17atr=1.16atr (980 missing values generated) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 252 . replace atrtypel=1 if typem2==i /’the way atrtypel was defined previously was in c o rre c t because a tr_ ca se .sa s takes th e f i r s t . s i c when s o rte d by year and s ic code to prevent repeated sic and year values, this code uses the type1n2 var which was created in excel to identify sic codes and years with both types of atr case types to change the value to 1 if the first.sic code kept a 0 instead of 1*/ (168 re a l changes made) . replace atrtype2=l if type1n2==1 (168 re a l changes made) . gen Hatrt1=l.atrtypel /•lags antitrust type 1*/ (140 missing values generated) . gen Hatrt2=l.atrtype2 /•lags antitrust type 2*/ (140 missing values generated) . gen 12atrtl=l.H atrtl /•twice lags antitrust type 1*/ (280 missing values generated) . gen 12atrt2=l.Hatrt2 /•twice lags antitrust type 1*/ (280 missing values generated) . gen Hatrt1hhi=l1atrt1*avghhi /•interaction between lagged atr type 1 and contemporaneous concentration*/ (140 missing values generated) . gen Hatrt2hhi=Hatrt2*avghhi /•interaction between atr type 2 and contemporaneous concentration*/ (140 missing values generated) . gen workcat=1 if occ==5 | occ==6 | occ==7 /‘low wage worker category*/ (1680 missing values generated) . replace workcat=2 if o c c ==1 | occ==4 / ‘ medium wage worker ca te g o ry */ (840 re a l changes made) . replace workcat=3 if occ==2 | occ==3 /'high wage worker category*/ (840 re a l changes made) . / ‘ ORIGINALLY, ATRDOCC# VAR WAS EQUAL TO THE PRODUCT OF LIATR AND DOCC#. NOW THAT WE HAVE TWICE LAGGED AND CONTEMPORANEOUS VARIABLES IN THE ANALYSIS, THE VAR NAME IS CHANGED REFLECT AS SUCH*/ . rename a trd o c c l H a trd o c c l . rename atrdocc2 lla trd o c c 2 . rename atrdocc3 11atrdocc3 . rename atrdocc4 H a trd o c c 4 . rename atrdoccS H a trd o c c 5 . rename atrdoccS H a trd o c c 6 . rename atrdocc7 H a trd o c c 7 . / ‘ ORIGINALLY, ATRDSIC# VAR WAS EQUAL TO THE PRODUCT OF L1ATR AND OSIC#. NOW THAT WE HAVE TWICE LAGGED AND CONTEMPORANEOUS VARIABLES IN THE ANALYSIS, THE VAR NAME IS CHANGED REFLECT AS SUCH'*/ . rename atrd20 H a trd 2 0 . rename atrd21 H a tr d 2 l . rename atrd22 H a trd 2 2 . rename atrd23 H a trd 2 3 . rename atrd24 H a trd 2 4 . rename atrd25 H a trd 2 5 . rename atrd26 H a trd 2 6 . rename atrd27 H a trd 2 7 . rename atrd28 I1 a trd 2 8 . rename atrd29 I1atrd29 . rename atrd30 H a trd 3 0 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 253 rename atrd32 Hatrd32 rename atrd33 Hatrd33 rename atrd34 H a trd 3 4 rename atrd35 I1atrd35 rename atrd36 Hatrd36 rename atrd37 I1 a trd 3 7 rename atrd38 H a trd 3 8 rename atrd39 I1 a trd 3 9 gen b yte low= w orkcat==l gen byte med= workcat==2 . gen byte high= workcat==3 . gen H a trlo w = l1 a tr* lo w (140 missing values generated) . gen I1atrm ed=l1atr*m ed (140 missing values generated) . gen lla t r h i= H a t r * h ig h (140 missing values generated) . label define workfmt 1 “Low wage worker" 2 "Medium Wage Worker* 3 "High Wage Worker" . label define occfmt 1 "Technicians* 2 "Prof Speclty Occ" 3 "Mgrs & Admin" 4 "Sales" 5 "Admin Supp, Cler" 6 "Service" 7 "Prodn" . label define sicfmt 20 "Food & Kindred’ 21 "Tobacco Mfrs" 22 "Textil M ill Products' 23 "Apparel & Other Textile Products" 24 "Lumber & Wood Products" 25 "Furniture & Fixtures" 26 "Paper & Allied Products" 27 "Printing & Publishing" 28 "Chemicals & A llied Products" 29 "Petroleum & Coal Products’ 30 "Rubber & Misc" 31 "Leather & leather Products" 32 'Stone,Clay.Glass & Concrete" 33 "Primary Metal" 34 "Fabricated Metal" 35 "Industrial Machinery & Equip" 36 'E lectrical & Electronic" 37 "Transportation Equip" 38 “Instruments & R elated" 39 "Misc M frg Inds" . label define atrtypes 1 "Monopoly, Premerger Notification Failure, Acquisitions, Joint Ventures" 2 "Price fixing, Restraint of Trade, Bid Rigging, Territorial Allocation, Restricting Output" . label values occ occfmt . label values workcat workfmt . label values sic sicfmt . i i s occ . ts s e t occ SICYEAR panel variable: occ, 1 to 7 time variable: SICYEAR, 1 to 420 . set matsize 700 . /'testing for autocorrelation and heteroskedasticity using lrtest heteroskedasticity f i r s t * / . xtgls employmt realgdp avghhi k_lexp frm5yrte atr H atr 12atr trend, force panel(hetero) Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: heteroskedastic Correlation: no autocorrelation Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 254 Estimated covariances = 7 Number o f obs = 2660 Estimated autocorrelations = 0 Number o f groups = 7 Estimated coefficients = 9 No. o f tim e periods= 380 Wald c h i2 (8 ) 223.15 Log lik e lih o o d = -16070.41 Prob > chi2 = 0.0000 employmt | C oef. Std. E rr. z P>|Z| [95% Conf. Interval] re a lg d p | 1.403058 1.27948 1.10 0.273 -1.104678 3.910793 avghhi | .0159728 .0046575 3.43 0.001 .0068442 .0251014 k_lexp | -.4427554 .1410244 -3.14 0.002 -.7191581 -.1663526 frmSyrte | 11.32637 1.334178 8.49 0.000 8.711432 13.94131 a t r | 18.85556 4.382795 4.30 0.000 10.26544 27.44568 l i a t r | 16.88832 4.397182 3.84 0.000 8.27 25.50664 1 2 a tr | 16.10029 4.3491 3.70 0.000 7 576216 24.62437 trend | -4.144907 2.669777 -1.55 0.121 -9.377573 1.087759 _cons | -1.070727 54.29247 -0.02 0.984 -107.482 105.3406 . lrtest, saving(O) . xtgls employmt realgdp avghhi k_lexp frmSyrte atr lia tr 12atr trend, force Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: homoscedastic Correlation: no autocorrelation Estimated covariances = 1 Number o f obs = 2660 Estimated autocorrelations = 0 Number o f groups = 7 Estimated coefficients = 9 No. o f tim e periods= 380 Wald c h i2 (8 ) 114.84 Log lik e lih o o d -18348.39 Prob > chi2 = 0 .0 0 0 0 employmt | Coef. Std. E rr. z P>|z| [95% Conf. In t e r v a l] re a lg d p | 4.214686 3.028808 1.39 0.164 -1.721669 10.15104 avghhi | .0164357 .0110254 1 .49 0.136 -.0051737 .038045 k_ le xp | -1.343262 .3338354 -4.02 0.000 -1.997568 -.6889571 frm5yrte | 15.28844 3.158289 4.84 0.000 9.098303 21.47857 a t r | 31.46115 10.37503 3.03 0.002 11.12647 51.79583 H a t r | 31.70023 10.40908 3.05 0.002 11.2988 52.10167 1 2 a tr | 32.79819 10.29526 3.19 0.001 12.61984 52.97654 tre n d | -10.63902 6.319941 -1 .68 0.092 -23.02588 1.747837 _cons | -49.20382 128.5221 -0.38 0.702 -301.1025 202.6948 . local df=e(N_g) - 1 . lrtest, d f('d f) Xtgls: likelihood-ratio test c h i2 (6 ) 4555.97 Prob > chi2 0.0000 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 255 . xtgls employmt realgdp avghhi H atrhhi k_lexp frmSyrte atr lia tr 12atr trend.force panel(hetero) Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: heteroskedastic Correlation: no autocorrelation Estimated covariances 7 Number o f obs = 2660 Estimated autocorrelations 0 Number o f groups = 7 Estimated coefficients 10 No. of time periods= 380 Wald chi2(9) 227.69 Log lik e lih o o d = -16070.09 Prob > chi2 = 0.0000 employmt | Coef. Std. Err. Z P >|z| [95% Conf. In te r v a l] realgdp | 1.439754 1.280299 1.12 0.261 -1.069586 3.949093 avghhi | .007878 .006119 1.29 0.198 -.004115 .0198709 H a tr h h i | .0199873 .0097146 2.06 0.040 .0009471 .0390274 k_lexp | -.4385533 .1411195 -3.11 0.002 -.7151424 -.1619643 frm S yrte | 10.60254 1.382974 7.67 0.000 7.891956 13.31312 a t r | 18.62873 4.386749 4.25 0.000 10.03086 27.2266 liatr | 3.878734 7.707875 0.50 0.615 -11.22842 18.98589 1 2 a tr | 16.10643 4.351514 3.70 0.000 7.577623 24.63525 tre n d | -4.165838 2.671267 1 .56 0.119 -9.401425 1.06975 _cons | 2.144089 54.34624 0.04 0.969 -104.3726 108.6608 . lrtest, saving(O) . xtgls employmt realgdp avghhi H atrhhi k_lexp frmSyrte atr lia tr 12atr trend.force Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: homoscedastic Correlation: no autocorrelation Estimated covariances 1 Number o f obs = 2660 Estimated autocorrelations 0 Number o f groups = 7 Estimated coefficients 10 No. of time periods= 380 Wald c h i2 (9 ) = 119.08 Log lik e lih o o d = -18346.36 Prob > chi2 = 0.0000 employmt | Coef. Std. Err. z P>|z| [95% Conf. In te r v a l] realgdp | 4.294134 3.026757 1.42 0.156 -1.6382 10.22647 avghhi | - .0024523 .0144659 -0.17 0.865 -.0308049 .0259003 H a tr h h i | .0462711 .0229662 2.01 0.044 .0012582 .0912841 k_lexp | -1.332889 .3336207 -4.00 0.000 -1.986774 -.6790043 frm S yrte | 13.56705 3.269492 4.15 0.000 7.158962 19.97514 a t r | 30.91067 10.37072 2.98 0.003 10.58443 51.23691 l i a t r | 1.555348 18.2222 0.09 0.932 -34.15951 37.27021 1 2 a tr | 32.77727 10.28742 3.19 0.001 12.61429 52.94024 tre n d | -10.67331 6.315148 -1.69 0.091 -23.05077 1.704157 _cons | -41.5686 128.48 -0.32 0.746 -293.3848 210.2476 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 256 . local df=e(N_g) - 1 . lrtest, df( df') Xtgls: likelihood-ratio test Chi2(6) = 4552.55 Prob > chi2 = 0.0000 . xtgls employmt realgdp avghhi H atrhhi k_lexp frmSyrte atr lia tr 12atr liatrdoccl Hatrdocc2 Hatrdocc3 Hatrdocc4 HatrdoccS llatrdocc6 trend, force panel(hetero) Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: heteroskedastic Correlation: no autocorrelation Estimated covariances = 7 Number o f obs 2660 Estimated autocorrelations = 0 Number of groups = 7 Estimated coefficients = 16 No. o f time periods= 380 Wald c h i2 (15) 968.71 Log likelihood = -15439 .4 Prob > c h i2 = 0.0000 employmt | C oef. S td. E rr z P>|z| [95% C onf. In te r v a l) realgdp | 1.248892 1.018827 1.23 0.220 -.7479727 3.245756 avghhi | .0069844 .0048693 1.43 0.151 -.0025593 .0165281 llatrhhi | .0176327 .0077306 2.28 0.023 .002481 .0327843 k_iexp | -.3788917 .112299 -3.37 0.001 -.5989937 -.1587896 frm S yrte | 9.296875 1.100533 8.45 0.000 7.139869 11.45388 a tr | 16.5213 3.490856 4.73 0.000 9.679354 23.36326 liatr | 663.6344 33.58213 19.76 0.000 597.8146 729.4541 1 2 a tr | 14.25135 3.462817 4.12 0.000 7.46435 21.03834 liatrdoccl | -698.7145 33.63058 -20.78 0.000 -764.6292 -632.7998 Hatrdocc2 | -648.5642 33.68875 -19.25 0.000 -714.5929 -582.5354 l1atrdocc3 | -624.315 33.4845 -18.64 0.000 -689.9435 -558.6866 Hatrdocc4 | -704.8069 33.63552 -20.95 0.000 -770.7313 -638.8825 HatrdoccS | -622.0505 33.482 -18.58 0.000 -687.674 -556.427 Hatrdocc6 | -718.1906 33.70097 -21.31 0.000 -784.2433 -652.1379 trend | -3.627753 2.125722 -1.71 0.088 -7.794092 .5385851 _cons | 2.480467 43.24726 0.06 0.954 -82.2826 87.24353 . lrtest, saving(O) . xtgls employmt realgdp avghhi H atrhhi k_lexp frmSyrte atr lia tr 12atr liatrdoccl Hatrdocc2 Hatrdocc3 Hatrdocc4 HatrdoccS Hatrdocc6 trend, force Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: homoscedastic Correlation: no autocorrelation Estimated covariances = 1 Number o f obs = 2660 Estimated autocorrelations = 0 Number o f groups = 7 Estimated coefficients = 16 No. o f tim e p e riods= 380 Wald chi2(15) = 2335.83 Log likelihood = -17566.34 Prob > chi2 = 0.0000 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 257 employmt | Coef. Std. Err. z P»|Z| [95% Conf. In te r v a l) realgdp | 4.294134 2.257481 1.90 0.057 -.130447 8.718715 avghhi | -.0024523 .0107893 -0.23 0.820 -.0235989 .0186942 H a tr h h i | .0462711 .0171291 2.70 0.007 .0126986 .0798436 k_lexp | -1.332889 .2488282 -5.36 0.000 -1.820583 -.8451946 frm S yrte | 13.56705 2.438523 5.56 0.000 8.787631 18.34647 a t r | 30.91067 7.734914 4.00 0.000 15.75051 46.07082 H a t r | 575.3613 18.53667 31.04 0.000 539.0301 611.6925 1 2 a tr | 32.77727 7.672787 4.27 0.000 17.73888 47.81565 liatrdoccl | -698.7145 19.25507 -36.29 0.000 -736.4538 -660.9753 Hatrdocc2 | -648.5642 19.25507 -33.68 0.000 -686.3034 -610.8249 l1atrdocc3 | -624.315 19.25507 -32.42 0.000 -662.0543 -586.5758 Hatrdocc4 | -704.8069 19.25507 -36.60 0.000 -742.5461 -667.0676 HatrdoccS | -622.0505 19.25507 -32.31 0.000 -659.7898 -584.3113 llatrdocc6 | -718.1906 19.25507 -37.30 0.000 -755.9299 -680.4514 tre n d | -10.67331 4.710099 -2.27 0.023 •19.90493 -1.441681 cons | -41.5686 95.82574 -0.43 0.664 -229.3836 146.2464 . local df=e(N_g) - 1 . lrtest, df( df ’) Xtgls: likelihood-ratio test chi2(6) = 4253.88 Prob > chi2 = 0.0000 . xtgls employmt realgdp avghhi H atrhhi k_lexp frm5yrte atr lia tr 12atr liatrdoccl l1atrdocc2 Hatrdocc3 Hatrdocc4 HatrdoccS l1atrdocc6 I1atrd21 I1atrd22 I1atrd23 Hatrd24 I1atrd25 I1atrd26 l1atrd27 I1atrd28 I1atrd29 I1atrd30 Hatrd32 Hatrd33 I1atrd34 I1atrd35 Hatrd36 Hatrd37 Hatrd38 Hatrd39 trend, force panel(hetero) Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: heteroskedastic Correlation: no autocorrelation Estim ated covariances = 7 Number o f obs = 2660 E stim ated a u to c o rre la tio n s = 0 Number o f groups = 7 Estimated coefficients = 34 No. of time periods= 380 Wald chi2(33) = 1447.27 Log lik e lih o o d = -15320.79 Prob >chi2 = 0.0000 employmt | Coef. Std. E rr. z P>|z| [95% Conf. Interval) re algdp | -.2889316 .9954367 -0.29 0.772 -2.239952 1.662089 avghhi | .0062681 .0047577 1 .32 0.188 -.0030568 .0155931 H a tr h h i | -.0425313 .0214806 -1.98 0.048 -.0846326 -.00043 k_lexp | -.1536963 .1332147 •1 .15 0.249 -.4147924 .1073997 frm S yrte | 10.6299 1.232682 8.62 0.000 8.213888 13.04591 a t r | 10.94051 3.512542 3.11 0.002 4.056054 17.82497 H a t r | 734.6659 36.80247 19.96 0.000 662.5344 806.7974 1 2 a tr | 7.149915 3.519151 2.03 0.042 .2525051 14.04732 liatrdoccl | -698.7145 32.56928 -21.45 0.000 -762.5491 -634.8799 Hatrdocc2 | -648.5642 32.48266 -19.97 0.000 -712.229 -584.8993 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Hatrdocc3 | -624.315 32.26302 -19.35 0.000 -687.5494 -561.0807 Hatrdocc4 | -704.8069 32.59212 -21.63 0.000 -768.6863 -640.9275 HatrdoccS | -622.0505 32.25327 -19.29 0.000 -685.2658 -558.8353 l1atrdocc6 | -718.1906 32.70601 -21.96 0.000 -782.2932 -654.088 H a tr d 2 l | -1.627227 45.47657 -0.04 0.971 -90.75967 87.50521 Hatrd22 | -68.4514 19.08971 -3.59 0.000 -105.8665 -31.03626 H a trd 2 3 | -78.67817 14.44462 -5.45 0.000 -106.9891 -50.36724 H a trd 2 4 | -101.7468 18.3217 -5.55 0.000 -137.6566 -65.83689 I1 a trd 2 5 | -107.5982 20.87985 -5 .1 5 0.000 -148.5219 -66.67441 I1 a trd 2 6 | -67.75926 15.40044 -4.40 0.000 -97.94356 -37.57496 I1 a trd 2 7 | 14.86221 18.43755 0.81 0.420 -21.27471 50.99914 I1 a trd 2 8 | -.4753084 11.32271 -0.04 0.967 -22.66741 21.71679 H a trd 2 9 | -89.97852 15.44469 -5.83 0.000 -120.2496 -59.70748 I1 a trd 3 0 | -89.58723 15.80567 -5.67 0.000 -120.5658 -58.60868 I1 a trd 3 2 | -75.08647 11.32564 -6.63 0.000 -97.28432 -52.88861 l1 a trd 3 3 | -66.77515 10.55334 -6.33 0.000 -87.45932 -46.09098 H a trd 3 4 | -40.61581 13.15738 -3 .09 0.002 -66.4038 -14.82781 I1 a trd 3 5 | 42.49845 11.45919 3.71 0.000 20.03885 64.95806 H a trd 3 6 | 36.21739 10.98754 3.30 0.001 14.68221 57.75256 I1 a trd 3 7 | 36.26305 24.68095 1.47 0.142 -12.11073 84.63683 I1 a trd 3 8 | •48.28489 13.32568 -3.62 0.000 -74.40274 -22.16705 I1 a trd 3 9 | -103.1481 15.13892 -6.81 0.000 -132.8199 -73.47641 tre n d | -.0873629 2.083705 -0.04 0.967 -4.171349 3.996623 _cons | 67.95101 42.15395 1.61 0.107 -14.66921 150.5712 . lrtest, saving(O) . xtgls employmt realgdp avghhi Hatrhhi k_lexp frmSyrte atr H atr 12atr liatrdoccl l1atrdocc2 Hatrdocc3 Hatrdocc4 HatrdoccS Hatrdocc6 I1atrd21 I1atrd22 Hatrd23 Hatrd24 I1atrd25 I1atrd26 Hatrd27 I1atrd28 I1atrd29 I1atrd30 I1atrd32 Hatrd33 I1atrd34 Hatrd35 Hatrd36 I1atrd37 I1atrd38 Hatrd39 trend, force Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: homoscedastic Correlation: no autocorrelation Estimated covariances = 1 Number o f obs = 2660 Estimated autocorrelations = 0 Number o f groups = 7 Estimated coefficients = 34 No. o f tim e p e rio d s= 380 Wald chi2(33) = 2640.33 Log lik e lih o o d = -17487.65 Prob > chi2 = 0.0000 employmt | C o e f. S td. E rr. z P>|Z| [95% C onf. In te r v a l] re a lg d p | 1.589283 2.244758 0.71 0.479 -2.810363 5.988929 avghhi | -.0011237 .0107288 -0.10 0.917 -.0221518 .0199045 H a tr h h i | -.0537163 .0484399 -1.11 0.267 -.1486568 .0412242 k_ le xp | -.6746324 .3004057 -2.25 0.025 -1.263417 -.085848 frm S y rte | 16.8509 2.779757 6.06 0.000 11.40268 22.29913 a t r | 16.20015 7.920954 2.05 0.041 .6753659 31.72494 H a t r | 724.565 43.23616 16.76 0.000 639.8237 809.3063 1 2 a tr | 14.54236 7.935858 1.83 0.067 -1.01164 30.09635 liatrdoccl | -698.7145 18.6938 -37.38 0.000 -735.3537 -662.0753 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 259 Hatrdocc2 | -648.5642 18.6938 -34.69 0.000 -685.2033 -611.925 llatrdocc3 | -624.315 18.6938 -33.40 0.000 -660.9542 -587.6759 Hatrdocc4 | -704.8069 18.6938 -37.70 0.000 -741.4461 -668.1677 HatrdoccS | -622.0505 18.6938 -33.28 0.000 -658.6897 -585.4113 l1atrdocc6 | -718.1906 18.6938 -38.42 0.000 -754.8298 -681.5514 H a trd 2 1 | -46.8385 102.5519 -0.46 0.648 -247.8365 154.1595 I1 a trd 2 2 | -111.4732 43.04823 -2.59 0.010 -195.8461 -27.10018 H a trd 2 3 | -109.3738 32.57332 -3.36 0.001 -173.2163 -45.53126 I1 a trd 2 4 | -162.5342 41.31633 -3.93 0.000 -243.5127 -81.55566 Hatrd25 | -194.451 47.08508 -4.13 0.000 -286.736 -102.1659 11a trd 2 6 | -123.6101 34.72874 -3.56 0.000 -191.6772 -55.54303 H a trd 2 7 | -52.98312 41.57757 -1.27 0.203 -134.4737 28.50741 I1 a trd 2 8 | -79.26493 25.53326 -3.10 0.002 -129.3092 -29.22066 I1 a trd 2 9 | -178.9202 34.82853 -5.14 0.000 -247.1829 -110.6576 11a trd 3 0 | -142.1474 35.64257 -3.99 0.000 -212.0056 -72.28926 11atrd32 | -146.4404 25.53988 -5.73 0.000 -196.4976 -96.38315 I1 a trd 3 3 | -115.3591 23.7983 -4.85 0.000 -162.003 -68.71531 I1 a trd 3 4 | -52.50455 29.67054 -1.77 0.077 -110.6577 5.648638 I1atrd35 | 33.94024 25.84104 1.31 0.189 -16.70726 84.58775 H a trd 3 6 | 20.73608 24.77743 0.84 0.403 -27.82679 69.29895 I1 a trd 3 7 | 45.38744 55.65675 0.82 0.415 -63.69779 154.4727 I1 a trd 3 8 | -149.4323 20.05005 -4.97 0.000 -208.3293 -90.53525 11a trd 3 9 | -206.4675 34.13901 -6.05 0.000 -273.3787 -139.5563 trend | -4.186384 4.698856 -0.89 0.373 -13.39597 5.023204 _cons | 58.29757 95.05921 0.61 0.540 -128.0151 244.6102 . local df=e(N_g) - 1 . lrtest, df('df') Xtgls: likelihood-ratio test chi2(6) = 4333.71 Prob > chi2 = 0.0000 . xtgls employmt realgdp avghhi H atrhhi k_lexp frmSyrte atr lia tr 12atr Hatrd21 llatrd22 I1atrd23 I1atrd24 I1atrd25 11atrd26 Hatrd27 I1atrd28 Hatrd29 Hatrd30 I1atrd32 I1atrd33 I1atrd34 I1atrd35 llatrd36 Hatrd37 I1atrd38 Hatrd39 trend, force Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: homoscedastic Correlation: no autocorrelation Estimated covariances = 1 Number o f obs = 2660 Estimated autocorrelations = 0 Number o f groups = 7 Estimated coefficients = 28 No. o f tim e periods= 380 Wald ch i2 (2 7 ) 210.82 Log lik e lih o o d = -18303.17 Prob > chi2 = 0.0000 employmt | Coef. Std. E rr. z P>|Z| [95% Conf. In te r v a l] re a lg d p | 1.589283 3.050126 0.52 0.602 -4.388854 7.56742 a vghhi | -.0011237 .0145781 -0.08 0.939 -.0296962 .0274489 H a tr h h i | -.0537163 .065819 -0.82 0.414 -.1827192 .0752866 k_ le xp | -.6746324 .4081844 -1.65 0.098 -1.474659 .1253942 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 260 frm S y rte | 16.8509 3.77707 4.46 0.000 9.447983 24.25383 a t r | 16.20015 10.76281 1 .51 0.132 -4.894569 37.29487 H a t r | 150.759 56.3458 2.68 0.007 40.32331 261.1948 1 2 a tr | 14.54236 10.78306 1 .35 0.177 -6.592055 35.67677 I1 a trd 2 1 | -46.8385 139.3451 -0 .3 4 0.737 -319.95 226.273 I1 a trd 2 2 | -111.4732 58.49294 -1 .91 0.057 -226.1172 3.1709 I1 a trd 2 3 | -109.3738 44.25988 -2 .4 7 0.013 -196.1216 -22.62603 I1 a trd 2 4 | -162.5342 56.13968 -2 .9 0 0.004 -272.5659 -52.50243 H a trd 2 5 | -194.451 63.97812 -3 .0 4 0.002 -319.8458 -69.05615 I1 a trd 2 6 | -123.6101 47.18861 -2 .62 0.009 -216.0981 -31.12213 H a trd 2 7 | -52.98312 56.49464 -0 .9 4 0.348 -163.7106 57.74434 H a trd 2 8 | -79.26493 34.69401 -2 .2 8 0.022 -147.2639 -11.26592 I1 a trd 2 9 | -178.9202 47.32421 -3 .7 8 0.000 -271.674 -86.16649 I1atrd30 | -142.1474 48.4303 -2 .9 4 0.003 -237.0691 -47.22577 I1atrd32 | -146.4404 34.703 -4 .2 2 0.000 -214.457 -78.42376 Hatrd33 | -115.3591 32.33659 -3 .5 7 0.000 - 178.7377 -51.98059 H a trd 3 4 | -52.50455 40.31564 -1 .30 0.193 -131.5218 26.51266 I1atrd35 | 33.94024 35.11221 0.9 7 0.334 -34.87842 102.7589 H a trd 3 6 ! 20.73608 33.667 0.6 2 0.538 -45.25003 66.72219 Hatrd37 | 45.38744 75.62511 0.6 0 0.548 -102.8351 193.6099 I1 a trd 3 8 | -149.4323 40.83132 -3 .6 6 0.000 -229.4602 -69.40435 H a trd 3 9 | -206.4675 46.38729 -4 .4 5 0.000 -297.3849 -115.5501 tre n d | -4.186384 6.384697 -0 .66 0.512 -16.70016 8.327392 cons | 58.29757 129.1643 0 .45 0.652 -194.8597 311.4549 . lrtest, saving(O) . xtgls employmt realgdp avghhi Hatrhhi k_lexp frmSyrte atr H atr 12atr Hatrd2l I1atrd22 I1atrd23 I1atrd24 I1atrd25 Hatrd26 l1atrd27 I1atrd28 Hatrd29 I1atrd30 Hatrd32 I1atrd33 I1atrd34 Hatrd35 I1atrd36 Hatrd37 I1atrd38 I1atrd39 trend, force Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: homoscedastic Correlation: no autocorrelation Estimated covariances = 1 Number o f obs = 2660 Estimated autocorrelations = 0 Number o f groups = 7 Estimated coefficients = 28 No. o f tim e periods= 380 Wald chi2(27) 210.82 Log likelihood = -18303.17 Prob > chi2 = 0.0000 employmt | Coef. S td. E rr. z P>|Z| [95% Conf. In te r v a l) re a lg d p | 1.589283 3.050126 0.52 0.602 -4.388854 7.56742 avghhi | -.0011237 .0145781 -0.08 0.939 -.0296962 .0274489 H a tr h h i | -.0537163 .065819 -0.82 0.414 -.1827192 .0752866 k_ le xp | -.6746324 .4081844 -1 .65 0.098 -1.474659 .1253942 frm S y rte | 16.8509 3.77707 4.46 0.000 9.447983 24.25383 a t r | 16.20015 10.76281 1 .51 0.132 -4.894569 37.29487 H a t r | 150.759 56.3458 2.68 0.007 40.32331 261.1948 1 2 a tr | 14.54236 10.78306 1.35 0.177 -6.592055 35.67677 H a tr d 2 l | -46.8385 139.3451 -0.34 0.737 -319.95 226.273 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Hatrd22 | -111.4732 58.49294 -1.91 0.057 -226.1172 3.1709 I1atrd23 | -109.3738 44.25988 -2 .4 7 0.013 -196.1216 -22.62603 I1 a trd 2 4 | -162.5342 56.13968 -2.90 0.004 -272.5659 -52.50243 I1 a trd 2 5 | -194.451 63.97812 -3.04 0.002 -319.8458 -69.05615 11atrd26 | -123.6101 47.18861 -2.62 0.009 -216.0981 -31.12213 I1 a trd 2 7 | -52.98312 56.49464 -0.94 0.348 -163.7106 57.74434 Hatrd28 | -79.26493 34.69401 -2.28 0.022 -147.2639 -11.26592 11atrd29 | -178.9202 47.32421 -3.78 0.000 -271.674 -86.16649 H a trd 3 0 | -142.1474 48.4303 -2 .9 4 0.003 -237.0691 -47.22577 H a trd 3 2 | -146.4404 34.703 -4.22 0.000 -214.457 -78.42376 Hatrd33 | -115.3591 32.33659 -3 .5 7 0.000 -178.7377 -51.98059 H a trd 3 4 | -52.50455 40.31564 -1 .30 0.193 -131.5218 26.51266 I1atrd35 | 33.94024 35.11221 0 .9 7 0.334 -34.87842 102.7589 Hatrd36 | 20.73608 33.667 0.62 0.538 -45.25003 86.72219 Hatrd37 | 45.38744 75.62511 0.60 0.548 -102.8351 193.6099 11atrd38 | -149.4323 40.83132 -3 .6 6 0.000 -229.4602 -69.40435 I1atrd39 | -206.4675 46.38729 -4.45 0.000 -297.3849 -115.5501 trend | -4.186384 6.384697 -0.66 0.512 -16.70016 8.327392 _cons | 58.29757 129.1643 0 .4 5 0.652 -194.8597 311.4549 . local df=e(N_g) - 1 . lrtest, df( df') Xtgls: likelihood-ratio test chi2(6) = 0.00 Prob > chi2 = 1.0000 . / ‘ homoskedastic*/ . /’testing for autocorrelation--comparing the no autocorrelation assumption to arl to see if the output changes, if so, there would be autocorrelation*/ . xtgls employmt realgdp avghhi k_lexp frm5yrte atr H atr 12atr trend, force corr(O) Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: homoscedastic Correlation: no autocorrelation Estimated covariances = 1 Number o f obs = 2660 Estimated autocorrelations = 0 Number o f groups = 7 Estimated coefficients = 9 No. o f tim e periods= 380 Wald ch i2 (8 ) 114.84 Log likelihood = -18348.39 Prob > chi2 = 0.0000 employmt | Coef. S td. E rr. z P>|Z| [95% Conf. Interval] realgdp | 4.214686 3.028808 1 .39 0.164 -1 .721669 10.15104 avghhi | .0164357 .0110254 1 .49 0.136 -.0051737 .038045 k_lexp | -1.343262 .3338354 -4 .0 2 0.000 -1.997568 -.6889571 frm S yrte | 15.28844 3.158289 4.84 0.000 9.098303 21.47857 a tr | 31.46115 10.37503 3.03 0.002 11.12647 51.79583 l i a t r | 31.70023 10.40908 3 .05 0.002 11.2988 52.10167 1 2 a tr | 32.79819 10.29526 3.19 0.001 12.61984 52.97654 tre n d | -10.63902 6.319941 -1 .68 0.092 -23.02588 1.747837 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 262 cons | -49.20382 128.5221 -0.38 0.702 -301.1025 202.6948 . xtgls employmt realgdp avghhi k_lexp frmSyrte atr H atr 12atr trend, force corr(arl) Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: homoscedastic Correlation: common AR(1) coefficient for all panels (0.9526) Estimated covariances 1 Number o f obs = 2660 Estimated autocorrelations = 1 Number of groups = 7 Estimated coefficients 9 No. o f time periods= 380 Wald c h i2 (8 ) = 195.71 Log lik e lih o o d = -13844.09 Prob >chi2 = 0.0000 employmt | Coef. Std. Err. z P>|z| [95% Conf. In t e r v a l| realgdp | 1.753502 .7406904 2.37 0.018 .3017755 3.205229 avghhi | -.0208425 .0073098 -2.85 0.004 -.0351695 -.0065155 k_lexp | -.9131083 .1606294 -5.68 0.000 -1.227936 -.5982805 frm S yrte | 11.72392 1.551585 7.56 0.000 8.682866 14.76497 a tr | 4.790512 1.756678 2.73 0.006 1.347488 8.233537 H a t r | 5.730523 1.89895 3.02 0.003 2.008649 9.452397 I2 a tr | 2.315773 1.743809 1.33 0.184 -1.10203 5.733576 trend | -4.35849 1.600794 -2.72 0.006 -7.495988 -1.220991 _cons | 94.78116 35.55719 2.67 0.008 25.09035 164.472 . xtgls employmt realgdp avghhi H atrhhi k_lexp frm5yrte atr H atr 12atr trend,force co rr(O ) Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: homoscedastic Correlation: no autocorrelation Estimated covariances 1 Number o f obs = 2660 Estimated autocorrelations 0 Number o f groups = 7 Estimated coefficients 10 No. of time periods= 380 Wald c h i2 (9 ) = 119.08 Log lik e lih o o d = -18346.36 Prob > chi2 = 0.0000 employmt | Coef. S td. E rr. z P >|z| [95% Conf. In t e r v a ll realgdp | 4.294134 3.026757 1 .42 0.156 -1.6382 10.22647 avghhi | -.0024523 .0144659 -0.17 0.865 -.0308049 .0259003 H a tr h h i | .0462711 .0229662 2.01 0.044 .0012582 .0912841 k_lexp | -1.332889 .3336207 -4 .0 0 0.000 -1.986774 -.6790043 frm S yrte | 13.56705 3.269492 4.15 0.000 7.158962 19.97514 a tr | 30.91067 10.37072 2.98 0.003 10.58443 51 .23691 H a t r | 1.555348 18.2222 0.09 0.932 -34.15951 37.27021 1 2 a tr | 32.77727 10.28742 3.19 0.001 12.61429 52.94024 tre n d | -10.67331 6.315148 -1.69 0.091 -23.05077 1.704157 _cons | -41.5686 128.48 -0.32 0.746 •293.3848 210.2476 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 263 . xtgls employmt realgdp avghhi Hatrhhi k_lexp frmSyrte atr lla tr 12atr trend.force c o r r ( a n ) Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: homoscedastic Correlation: common AR(1) coefficient for a ll panels (0.9490) Estimated covariances = 1 Number of obs = 2660 Estimated autocorrelations = 1 Number of groups = 7 Estimated coefficients = 10 No. of time periods= 380 Wald chi2(9) = 216.23 Log lik e lih o o d = -13843.96 Prob > ch i2 = 0.0000 employmt | Coef. Std. E rr. z P »|z| [95% Conf. In te rv a l) realgdp | 1.852391 .7420475 2.50 0.013 .3980041 3.306777 avghhi | -.0290426 .0075354 -3.85 0.000 - .0438117 -.0142735 H a tr h h i | .016296 .0035828 4.55 0.000 .0092739 .0233181 k_lexp | -.9058203 .1607202 -5.64 0.000 -1.220826 -.5908146 frm S yrte | 11.35163 1.554489 7.30 0.000 8.304892 14.39838 a t r | 4.716548 1.759368 2.68 0.007 1.268251 8.164845 l l a t r | -4.930413 3.024799 -1.63 0.103 -10.85891 .9980828 1 2 a tr | 2.345919 1.746357 1.34 0.179 -1.076878 5.768716 tre n d | -4.604915 1.603889 -2.87 0.004 -7.748479 -1.461351 _cons | 96.27528 35.05935 2.75 0.006 27.56021 164.9903 . xtgls employmt realgdp avghhi k_lexp frmSyrte atr lla tr 12atr 13atr 14atr 15atr 16atr 17atr, force corr(O) Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: homoscedastic Correlation: no autocorrelation Estimated covariances 1 Number o f obs = 1960 Estimated autocorrelations 0 Number o f groups = 7 Estimated coefficients 13 No. of time periods= 280 Wald chi2(l2) = 118.00 Log lik e lih o o d = -13473.77 Prob > chi2 = 0.0000 employmt | Coef. S td. E rr. z P> |z | [95% Conf. In te rv a l] realgdp | -1.210823 .6596942 -1.84 0.066 -2.5038 .0821533 avghhi | .0173831 .0128283 1.36 0.175 -.0077599 .042526 k_lexp | -1.562608 .3924632 -3.98 0.000 -2.331822 -.793394 frm 5 y rte | 16.38129 3.984645 4.11 0.000 8.571531 24.19105 a t r | 18.04643 12.1717 1 .48 0.138 -5.809673 41.90253 l l a t r | 14.83127 12.40943 1.20 0.232 -9.490763 39.15331 1 2 a tr | 14.16529 12.1967 1.16 0.245 -9.739801 38.07038 1 3 a tr | 17.2562 12.57326 1.37 0.170 -7.386935 41.89934 1 4 a tr | 13.25057 12.47303 1.06 0.288 -11.19613 37.69726 1 5 a tr | 16.1378 12.65433 1.28 0.202 -8.664238 40.93984 1 6 a tr | 25.16551 12.3398 2.04 0.041 .9799512 49.35106 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 264 17atr | 25.88928 12.04022 2.15 0.032 2.290884 49.48768 _cons | 168.8727 47.17401 3.58 0.000 76.41339 261.3321 . xtgls employmt realgdp avghhi k_lexp frm5yrte atr lla tr 12atr 13atr 14atr 15atr 16atr 17atr, force corr(arl) Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: homoscedastic Correlation: common AH(1) coefficient for a ll panels (0.9601) E stim ated covariances = 1 Number o f obs = 1960 Estimated autocorrelations = 1 Number of groups = 7 Estimated coefficients = 13 No. of time periods= 280 Wald c h i2 ( 12) 277.73 Log lik e lih o o d = -10220.49 Prob > chi2 = 0.0000 employmt | Coef. Std. Err. z P>|z| [95% Conf. In t e r v a ll realgdp | -.195348 .1570218 -1.24 0.213 -.503105 .1124091 avghhi | -.0090547 .0073748 -1.23 0.220 -.0235091 .0053996 k_lexp | -.9441045 .2179564 -4.33 0.000 -1.371291 - .5169178 frm S yrte | 15.58545 2.037179 7.65 0.000 11.59265 19.57824 a t r | 10.49525 2.060941 5.09 0.000 6.455883 14.53462 l l a t r | 10.10265 2.304783 4.38 0.000 5.585361 14.61995 1 2 a tr | 12.55968 2.423153 5.18 0.000 7.810391 17.30898 1 3 a tr | 10.98116 2.459271 4.47 0.000 6.16108 15.80125 l4 a t r | 7.721989 2.557433 3.02 0.003 2.709513 12.73447 1 5 a tr | 6.610441 2.626062 2.52 0.012 1.463453 11.75743 1 6 a tr | 6.728353 2.483284 2.71 0.007 1.861207 11.5955 1 7 a tr | 1.114671 2.150732 0.52 0.604 -3.100686 5.330027 _cons | 134.0705 25.39604 5.28 0.000 84.29514 183.8458 . xtgls employmt realgdp avghhi H atrhhi k_lexp frmSyrte atr lla tr 12atr Hatrdocd Hatrdocc2 !1atrdocc3 Hatrdocc4 Hatrdocc5 Hatrdocc6 trend, force corr(O) Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: homoscedastic Correlation: no autocorrelation Estimated covariances = 1 Number o f obs = 2660 Estimated autocorrelations = 0 Number o f groups = 7 Estimated coefficients = 16 No. of time periods= 380 Wald c h i2 ( 15) = 2335.83 Log lik e lih o o d = -17566.34 Prob > chi2 = 0.0000 employmt | Coef. Std. Err. z p * |z | [95% Conf. In te r v a l] realgdp | 4.294134 2.257481 1.90 0.057 -.130447 8.718715 avghhi | -.0024523 .0107893 -0.23 0.820 -.0235989 .0186942 H a tr h h i | .0462711 .0171291 2.70 0.007 .0126986 .0798436 k_lexp | •1.332889 .2488282 -5.36 0.000 -1.820583 -.8451946 frm 5 y rte | 13.56705 2.438523 5.56 0.000 8.787631 18.34647 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 265 atr | 30.91067 7.734914 4.00 0.000 15.75051 46.07082 llatr | 575.3613 18.53667 31 .04 0.000 539.0301 611.6925 12atr | 32.77727 7.672787 4.27 0.000 17.73888 47.81565 Hatrdocd | -698.7145 19.25507 -36.29 0.000 -736.4538 -660.9753 Hatrdocc2 | -648.5642 19.25507 -33.68 0.000 -686.3034 -610.8249 Hatrdocc3 | -624.315 19.25507 -32.42 0.000 -662.0543 -586.5758 Hatrdocc4 | -704.8069 19.25507 -36.60 0.000 -742.5461 -667.0676 HatrdoccS | -622.0505 19.25507 -32.31 0.000 -659.7898 -584.3113 l1atrdocc6 | -718.1906 19.25507 -37.30 0.000 -755.9299 -680.4514 trend | -10.67331 4.710099 -2.27 0.023 -19.90493 -1.441681 _cons | -41.5686 95.82574 -0.43 0.664 -229.3836 146.2464 . xtgls employmt realgdp avghhi Hatrhhi k_lexp frmSyrte atr lla tr 12atr Hatrdocd Hatrdocc2 l1atrdocc3 llatrdocc4 HatrdoccS Hatrdocc6 trend, force corr(arl) Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: homoscedastic Correlation: common AR(1) coefficient for a ll panels (0.6581) Estimated covariances = 1 Number of obs = 2660 Estimated autocorrelations = 1 Number of groups = 7 Estimated coefficients = 16 No. of time periods= 380 Wald c h i2 ( 15) 378.41 Log lik e lih o o d = -15618.56 Prob > c h i2 = 0.0000 employmt | Coef. Std. Err. z P>|z| [95% C o n f. In te r v a l] realgdp | 2.109637 1.509065 1 .40 0.162 -.8480753 5.067349 avghhi | - .0017183 .0102276 -0.17 0.867 -.021764 .0183275 H a tr h h i | .0192993 .0079051 2.44 0.015 .0038055 .0347931 k_lexp | -1.005254 .2550055 -3.94 0.000 -1.505056 -.5054526 frm S yrte | 14.3746 2.436684 5.90 0.000 9.598787 19.15041 a tr | 12.82239 3.663674 3.50 0.000 5.64172 20.00306 l l a t r | 127.6861 10.0597 12.69 0.000 107.9694 147.4027 1 2 a tr | 10.5592 3.628133 2.91 0.004 3.448193 17.67021 H a tr d o c d | -151.9726 11.83685 -12.84 0.000 -175.1724 -128.7728 Hatrdocc2 | -140.6018 11.83685 -11.88 0.000 -163.8016 -117.402 Hatrdocc3 | -135.26 11.83685 -11 .43 0.000 -158.4598 -112.0602 l1atrdocc4 | -152.3408 11.83685 -12.87 0.000 -175.5406 -129.141 HatrdoccS | -135.3087 11.83685 -11.43 0.000 -158.5085 -112.1089 Hatrdocc6 | -155.3004 11.83685 -13.12 0.000 -178.5002 -132.1006 tre n d | -5.419227 3.230616 -1 .68 0.093 -11 .75112 .9126651 _cons | 59.55417 63.43144 0.94 0.348 -64.76918 183.8775 . xtgls employmt realgdp avghhi Hatrhhi k_lexp frmSyrte atr lla tr 12atr Hatrdocd l1atrdocc2 11atrdocc3 l1atrdocc4 l1atrdocc5 Hatrdocc6 Hatrd21 I1atrd22 Hatrd23 liatrd24 Hatrd25 Hatrd26 Hatrd27 Hatrd28 Hatrd29 Hatrd30 Hatrd32 Hatrd33 Hatrd34 I1atrd35 I1atrd36 I1atrd37 I1atrd38 Hatrd39 trend, force corr(O) Cross-sectional time-series FGLS regression Coefficients: generalized least squares Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 266 Panels: homoscedastic Correlation: no autocorrelation Estimated covariances = 1 Number o f obs = 2660 Estimated autocorrelations = 0 Number o f groups = 7 Estimated coefficients = 34 No. o f tim e periods= 380 Wald c h i2 (3 3 ) 2640.33 Log lik e lih o o d = -17487.65 Prob > chi2 = 0.0000 employmt | Coef. S td . E rr. z P>|Z| [95% Conf. In te r v a l] realgdp | 1.589283 2.244758 0.71 0.479 -2.810363 5.988929 avghhi | -.0011237 .0107288 -0.10 0.917 -.0221518 .0199045 Hatrhhi | -.0537163 .0484399 -1.11 0.267 -.1486568 .0412242 k_lexp | -.6746324 .3004057 -2.25 0.025 -1.263417 -.085848 frmSyrte | 16.8509 2.779757 6.06 0.000 11.40268 22.29913 a t r | 16.20015 7.920954 2.05 0.041 .6753659 31.72494 l l a t r | 724.565 43.23616 16.76 0.000 639.8237 809.3063 12 a tr | 14.54236 7.935858 1.83 0.067 -1.01164 30.09635 Hatrdocd | -698.7145 18.6938 -37.38 0.000 -735.3537 -662.0753 l1atrdocc2 | -648.5642 18.6938 -34.69 0.000 -685.2033 -611.925 HatrdoccS | -624.315 18.6938 -33.40 0.000 -660.9542 -587.6759 llatrdocc4 | -704.8069 18.6938 -37.70 0.000 -741.4461 -668.1677 HatrdoccS | -622.0505 18.6938 -33.28 0.000 -658.6897 -585.4113 Hatrdocc6 | -718.1906 18.6938 -38.42 0.000 -754.8298 -681.5514 Hatrd21 | -46.8385 102.5519 -0.46 0.648 -247.8365 154.1595 Hatrd22 | -111.4732 43.04823 -2.59 0.010 -195.8461 -27.10018 I1 a trd 2 3 | -109.3738 32.57332 -3 .3 6 0.001 -173.2163 -45.53126 I1atrd24 | -162.5342 41.31633 -3.93 0.000 -243.5127 -81.55566 I1 a trd25 | -194.451 47.08508 -4.13 0.000 -286.736 -102.1659 I1atrd26 | -123.6101 34.72874 -3 .5 6 0.000 -191.6772 -55.54303 I1 a trd 2 7 | -52.98312 41.57757 -1 .27 0.203 -134.4737 28.50741 11atrd28 | -79.26493 25.53326 -3.10 0.002 -129.3092 -29.22066 Hatrd29 | -178.9202 34.82853 -5.14 0.000 -247.1829 -110.6576 l1 a trd 3 0 | -142.1474 35.64257 -3 .9 9 0.000 -212.0056 -72.28926 H a trd 3 2 | -146.4404 25.53988 -5 .7 3 0.000 -196.4976 -96.38315 I1atrd33 | -115.3591 23.7983 -4.85 0.000 -162.003 -68.71531 H a trd 3 4 | -52.50455 29.67054 -1 .77 0.077 -110.6577 5.648638 H a trd 3 5 | 33.94024 25.84104 1 .31 0.189 -16.70726 84.58775 I1atrd36 | 20.73608 24.77743 0.84 0.403 -27.82679 69.29895 I1 a trd37 | 45.38744 55.65675 0.82 0.415 -63.69779 154.4727 Hatrd38 | -149.4323 30.05005 -4.97 0.000 -208.3293 -90.53525 I1atrd39 | -206.4675 34.13901 -6 .05 0.000 -273.3787 -139.5563 trend | -4.186384 4.698856 -0 .8 9 0.373 -13.39597 5.023204 _cons | 58.29757 95.05921 0.61 0.540 -128.0151 244.6102 . xtgls employmt realgdp avghhi Hatrhhi k_lexp frmSyrte atr lla tr 12atr Hatrdocd Hatrdocc2 Hatrdocc3 l1atrdocc4 Hatrdocc5 l1atrdocc6 Hatrd21 I1atrd22 I1atrd23 Hatrd24 Hatrd25 Hatrd26 liatrd27 Hatrd28 Hatrd29 Hatrd30 Hatrd32 Hatrd33 Hatrd34 I1atrd35 Hatrd36 I1atrd37 I1atrd38 I1atrd39 trend, force corr(arl) Cross-sectional tim e-series FGLS regression Coefficients: generalized least squares Panels: homoscedastic Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 267 Correlation: common AR(1) coefficient for a ll panels (0.5672) Estim ated co variances = 1 Number o f obs = 2660 Estimated autocorrelations = 1 Number of groups = 7 Estim ated c o e ffic ie n ts = 34 No. o f tim e p e rio d s= 380 Wald chi2(33) = 621.70 Log likelihood = -16037.54 Prob > Chi2 = 0.0000 employmt | Coef. Std. Err. z P *|z | (95% C onf. In te r v a l] re a lg d p | 1.422054 1.769514 0.80 0.422 -2.04613 4.890238 avghhi | .0038811 .0106719 0.36 0.716 -.0170353 .0247976 llatrhhi | .0200838 .0268835 0.75 0.455 -.0326069 .0727744 k_lexp | -.8891295 .2753869 -3.23 0.001 -1.428878 -.3493811 frmSyrte | 15.78529 2.665433 5.92 0.000 10.56113 21.00944 atr | 12.86466 4.470918 2.88 0.004 4.101827 21.6275 l l a t r | 265.1549 28.37424 9.34 0.000 209.5424 320.7674 12atr | 10.32502 4.453832 2.32 0.020 1.595669 19.05437 H a tr d o c d | -229.4985 13.82946 -16.59 0.000 -256.6037 -202.3932 Hatrdocc2 | -212.6468 13.82946 -15.38 0.000 -239.7521 -185.5416 Hatrdocc3 | -204.6058 13.82946 -14.79 0.000 -231.7111 -177.5006 Hatrdocc4 | -230.6611 13.82946 -16.68 0.000 -257.7663 -203.5558 HatrdoccS | -204.2935 13.82946 -14.77 0.000 -231.3988 -177.1883 l1atrdocc6 | -235.0859 13.82946 -17.00 0.000 -262.1912 -207.9807 I1 a trd 2 1 | -123.1981 56.75347 -2 .1 7 0.030 -234.4329 -11.96338 l1 a trd 2 2 | -78.39662 26.18217 -2 .99 0.003 -129.7127 -27.08051 I1atrd23 | -86.76505 26.86983 -3 .2 3 0.001 -139.4289 -34.10115 I1 a trd 2 4 | -87.91621 28.1859 -3 .1 2 0.002 -143.1596 -32.67286 Hatrd25 | -84.62256 27.90227 -3.03 0.002 -139.31 -29.93511 l1 a trd 2 6 | -87.18259 27.39451 -3.18 0.001 -140.8749 -33.49034 Hatrd27 | -56.65229 26.7527 -2.12 0.034 -109.0866 -4.217954 l1 a trd 2 8 | -73.13887 22.45714 -3.26 0.001 -117.1541 -29.12368 I1atrd29 | -87.55798 24.93356 -3.51 0.000 -136.4269 -38.68911 l1 a trd 3 0 | -74.71034 24.73122 -3.02 0.003 -123.1826 -26.23804 I1 a trd 3 2 | -96.17447 21.26177 -4.52 0.000 -137.8468 -54.50215 I1 a trd 3 3 | -89.57479 22.95948 -3 .9 0 0.000 -134.5745 -44.57504 Hatrd34 | -49.70272 23.89808 -2.08 0.038 -96.54211 -2.863337 l1atrd35 | -4.881098 23.52136 -0.21 0.836 -50.98212 41.21992 Hatrd36 | -45.47169 21.65105 -2.10 0.036 -87.90698 -3.036404 I1 a trd 3 7 | -63.33949 33.29779 -1 .90 0.057 -128.602 1.922979 I1 a trd 3 8 | -93.15163 21.9963 -4 .2 3 0.000 -136.2636 -50.03967 Hatrd39 | -102.9666 25.91724 -3.97 0.000 -153.7634 -52.16973 tre n d | -3.98841 3.779436 -1 .06 0.291 -11.39597 3.419148 _cons | 76.71904 74.38683 1.03 0.302 -69.07646 222.5145 . xtgls employmt realgdp avghhi Hatrhhi k_lexp frmSyrte atr lla tr 12atr Hatrd2l Hatrd22 Hatrd23 Hatrd24 Hatrd25 Hatrd26 Hatrd27 llatrd28 Hatrd29 liatrd30 Hatrd32 Hatrd33 Hatrd34 liatrd35 llatrd36 Hatrd37 Hatrd38 Hatrd39 trend, force corr(O) Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: homoscedastic Correlation: no autocorrelation Estim ated co variances = 1 Number o f obs = 2660 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 268 Estimated autocorrelations = 0 Number o f groups = 7 Estimated coefficients = 28 No. of time periods= 380 Wald chi2(27) = 210.82 Log likelihood = -18303.17 Prob > chi2 = 0.0000 employmt | Coef. Std. Err. z P >|z| [95% Conf. In te r v a l] realgdp | 1.589283 3.050126 0.52 0.602 -4.388854 7.56742 avghhi | -.0011237 .0145781 -0.08 0.939 -.0296962 .0274489 Hatrhhi | -.0537163 .065819 -0.82 0.414 -.1827192 .0752866 klexp | -.6746324 .4081844 -1 .65 0.098 -1.474659 . 1253942 frmSyrte | 16.8509 3.77707 4.46 0.000 9.447983 24.25383 atr | 16.20015 10.76281 1 .51 0.132 -4.894569 37.29487 l l a t r | 150.759 56.3458 2.68 0.007 40.32331 261.1948 1 2 a tr | 14.54236 10.78306 1 .35 0.177 -6.592055 35.67677 I1atrd21 | -46.8385 139.3451 -0.34 0.737 -319.95 226.273 Hatrd22 | -111.4732 58.49294 -1 .91 0.057 -226.1172 3.1709 H a trd 2 3 | -109.3738 44.25988 -2.47 0.013 -196.1216 -22.62603 I1 a trd 2 4 | -162.5342 56.13968 -2.90 0.004 -272.5659 -52.50243 I1atrd25 | -194.451 63.97812 -3.04 0.002 -319.8458 -69.05615 Hatrd26 | -123.6101 47.18861 -2.62 0.009 -216.0981 -31.12213 I1 a trd 2 7 | -52.98312 56.49464 -0.94 0.348 -163.7106 57.74434 I1atrd28 | -79.26493 34.69401 -2.28 0.022 -147.2639 -11.26592 I1atrd29 | -178.9202 47.32421 -3.78 0.000 -271.674 -86.16649 H a trd 3 0 | -142.1474 48.4303 -2.94 0.003 -237.0691 -47.22577 I1 a trd 3 2 | -146.4404 34.703 -4.22 0.000 -214.457 -78.42376 Hatrd33 | -115.3591 32.33659 -3.57 0.000 -178.7377 -51.98059 H a trd 3 4 | -52.50455 40.31564 -1 .30 0.193 -131.5218 26.51266 H a trd 3 5 | 33.94024 35.11221 0.97 0.334 -34.87842 102.7589 Hatrd36 | 20.73608 33.667 0.62 0.538 -45.25003 86.72219 Hatrd37 | 45.38744 75.62511 0.60 0.548 -102.8351 193.6099 Hatrd38 | -149.4323 40.83132 -3.66 0.000 -229.4602 -69.40435 H a trd 3 9 | -206.4675 46.38729 -4.45 0.000 -297.3849 -115.5501 trend | -4.186384 6.384697 -0.66 0.512 -16.70016 8.327392 _cons | 58.29757 129.1643 0.45 0.652 -194.8597 311.4549 . xtgls employmt realgdp avghhi H atrhhi k_lexp frm5yrte atr lla tr 12atr Hatrd Hatrd22 I1atrd23 Hatrd24 I1atrd25 11atrd26 I1atrd27 I1atrd28 Hatrd29 Hatrd3 I1atrd33 I1atrd34 Hatrd35 Hatrd36 I1atrd37 I1atrd38 I1atrd39 trend, force cor Cross-sectional time-seriesi FGLS regression Coefficients: generalized least squares Panels: homoscedastic Correlation: common AR(1) coefficient for a ll panels (0.8862) Estimated covariances 1 Number o f obs = 2660 Estimated autocorrelations 1 Number o f groups = 7 Estimated coefficients 28 No. o f time periods= 380 Wald ch i2 (2 7 ) 227.83 Log lik e lih o o d = -14128.83 Prob > c h i2 = 0.0000 employmt | Coef. S td. E rr. z P » |z| [95% Conf. In te rv a l] re a lg d p | 1.547352 . 8659796 1.79 0.074 -.1499371 3.24464 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 269 avghhi | - .0193719 .0087139 -2.22 0.026 - .0364508 -.002293 H a tr h h i | .0426116 .0116765 3.65 0.000 .019726 .0654972 k_lexp | - .8017243 .1818986 -4.41 0.000 -1.158239 -.4452097 frm S yrte | 11.90183 1.778947 6.69 0.000 8.415156 15.3885 a tr | 5.290996 2.083349 2.54 0.011 1.207706 9.374286 l l a t r | 21.17094 15.03412 1 .41 0.159 -8.295404 50.63728 1 2 a tr | 2.691087 2.072528 1.30 0.194 -1.370994 6.753168 I1 a trd21 | -121.3173 25.58942 -4.74 0.000 -171.4717 -71.16299 I1 a trd 2 2 | -42.21047 14.25346 -2.96 0.003 -70.14674 -14.2742 I1atrd23 | -48.38441 15.35377 -3.15 0.002 -78.47724 -18.29158 I1 a trd 2 4 | -35.20073 15.45138 -2.28 0.023 -65.48487 -4.916581 I1 a trd 2 5 | -33.49804 14.91796 -2.25 0.025 -62.7367 -4.259377 lla tr d 2 6 | -45.70038 15.19296 -3.01 0.003 -75.47803 -15.92273 I1 a trd 2 7 | -25.17105 14.58037 -1.73 0.084 -53.74805 3.405937 I1 a trd 2 8 | -42.0504 13.68412 -3.07 0.002 -68.87077 -15.23002 H a trd 2 9 | -36.19467 14.13236 -2.56 0.010 -63.8936 -8.495751 H a trd 3 0 | -30.98653 13.98552 -2.22 0.027 -58.39764 -3.575413 H a trd 3 2 | -52.96362 12.73966 -4.16 0.000 -77.93289 -27.99435 H a trd 3 3 | -60.50117 14.1387 -4.28 0.000 -88.21252 -32.78983 H a trd 3 4 | -37.31376 14.11993 -2.64 0.008 -64.98832 -9.639206 H a trd 3 5 | -16.70527 14.37918 -1.16 0.245 -44.88794 11.4774 H a trd 3 6 | -47.00167 13.01043 -3.61 0.000 -72.50165 -21.5017 H a trd 3 7 | -65.0726 16.71493 -3.89 0.000 -97.83327 -32.31194 lla tr d 3 8 | -49.20207 12.88542 -3.82 0.000 -74.45702 -23.94711 I1 a trd 3 9 | -38.1672 14.64648 -2.61 0.009 -66.87376 -9.460632 tre n d | -4.134986 1.877058 -2.20 0.028 -7.813951 -.4560202 cons | 98.51253 37.19881 2.65 0.008 25.60421 171.4209 . log close log: u:\dissertation\emp_levels_tests.log log type: text closed on: 20 Jun 2002, 18:52:22 log: u:\dissertation\avgwg_levels_tests.log log type: text opened on: 21 Jun 2002, 16:44:11 . / ‘ THIS ANALYSIS EVALUATES THE EFFECT OF ATR INDICTMENTS ON EMPLOYMENT BY OCCUPATION USING CROSS-SECTION TIME SERIES TECHNIQUES. THE VAR SICYEAR WAS CREATED TO ACCOMMODATE FOR THE INABILITY OF STATA TO COMPUTE THE SIC CODE AND OCCUPATION AND YEAR VARIATION THE VAR EMPLOYMT REPRESENTS NUMBER OF EMPLOYEES BY OCCUPATION AND SIC CODE * / . use ■\\\sparkynt\userfiles\ypho\diss\sas_prograins\datasets\bigatr_final.dta“, clear . egen occsic=concat(occ sic) /'concatenates sic and occ variables*/ . egen sicyear=concat(sic year) /'concatenates sic and year variables*/ . egen occyear=concat(occ year) /'concatenates occ and year variables*/ . egen sicyearocc=concat(sic year occ) /'concatenates sic, year, and occ variables*/ . encode occsic, gen(OCCSIC) /'transforms character value to numeric*/ . encode sicyear, gen(SICYEAR) /'transforms character value to numeric*/ . encode occyear, gen(OCCYEAR) /'transform s character value to numeric*/ . encode sicye a ro cc, gen(SICYEAROCC) /'tra n s fo rm s c h a ra c te r v a lu e to n u m e ric '/ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 270 . iis occ /‘allows one to take the difference without subtracting the 1999*/ . tsset OCCSIC year /‘value from 1979 across 2 difference SIC codes*/ panel variable: OCCSIC, 1 to 140 tim e v a ria b le : ye a r, 1979 to 1999 . /‘converting current wage to wages in 1996 constant dollars*/ . gen wage=(wavgwg/(74/l56.9)) if year==l979 (2801 missing values generated) . replace wage=(wavgwg/(82.4/l56.9)) if year=='<9eo (139 real changes made) . replace wage=(wavgwg/(90.9/156.9)) if year==198l (139 r e a l changes made) . replace wage=(wavgwg/(96.5/156.9)) if year==1982 (139 real changes made) . replace wage=(wavgwg/(99.6/156.9)) if year==1983 (140 r e a l changes made) . replace wage=(wavgwg/(103.9/156.9)) if year==l984 (140 real changes made) . replace wage=(wavgwg/(107.6/156.9)) if year==1985 (140 real changes made) . replace waga=(wavgwg/(109.6/156.9)) if year==1986 (139 real changes made) . replace wage=(wavgwg/(H3.6/l56.9)) if year==l987 (140 real changes made) . replace wage=(wavgwg/(1l8.3/l56.9)) if year==i988 (139 r e a l changes made) . replace wage=(wavgwg/(124/156.9)) if year==1989 (140 r e a l changes made) . replace Wage=(wavgwg/(130.7/156.9)) if year==l990 (140 real changes made) . replace wage=(wavgwg/(136.2/156.9)) if year==l99l (139 real changes made) . replace wage=(wavgwg/(140.3/156.9)) if year==l992 (140 r e a l changes made) . replace wage=(wavgwg/(144.5/156.9)) if year==1993 (140 r e a l changes made) . replace Wage=(wavgwg/(148.2/156.9)) if year==1994 (140 real changes made) . replace wage=(wavgwg/(152.4/l56.9)) if year==l995 (140 real changes made) . replace wage=(wavgwg/(156.9/156.9)) if year==i996 (138 real changes made) . replace wage=(wavgwg/(160.5/156.9)) if year==l997 (139 r e a l changes made) . replace wage=(wavgwg/(l63/156.9)) if year==l998 (139 r e a l changes made) . replace wage=(wavgwg/(l66.6/156.9)) if year==i999 (139 r e a l changes made) . replace avghhi=avghhi*100 (2940 r e a l changes made) . gen tre n d = y e a r-1978 . gen lnwage=log(wage) (12 missing values generated) . gen dlnwage=d.lnwage Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 271 (159 missing values generated) . gen davghhi=d.avghhi (140 missing values generated) . gen dk_lexp=d.k_lexp (140 missing values generated) . replace k_lexp=k_lexp/10 (2940 real changes made) . replace realgdp=realgdp/l000 (2940 real changes made) . gen drealgdp=d.realgdp (140 missing values generated) . replace drealgdp=drealgdp*lO (2800 real changes made) . replace davghhi=davghhi*i0 (2800 real changes made) . replace atrhhi=atr*avghhi (1855 real changes made) . gen Hatrhhi=llatr*avghhi (140 missing values generated) . replace atrtypeid if typeln2==l /'the way atrtypel was defined previously was incorrect because atr_case.sas takes the firs t.s ic when sorted by year and sic code to prevent repeated sic and year values, this code uses the typeln2 var which was created in excel to identify sic codes and years with both types of atr case types to change the value to 1 if tne first.sic code kept a 0 instead of 1*/ (168 re a l changes made) . replace atrtype2=1 if tvpeln2==l (168 real changes made) . gen H atrtl=l.atrtypel /•lags antitrust type 1*/ (140 missing values generated) . gen Hatrt2=l.atrtype2 /•lags antitrust type 2*/ (140 missing values generated) . gen I2 a trtl= l.lla trtl /•twice lags antitrust type ;*/ (280 missing values generated) . gen 12atrt2=l.Hatrt2 /•twice lags antitrust type 1*/ (280 missing values generated) . gen I1atrt1hhi-I1atrt1*avghhi /•interaction between lagged atr type 1 and contemporaneous concentration*/ (140 missing values generated) . gen Hatrt2hhi=l1atrt2*avghlu /'interaction between atr type 2 and contemporaneous concentration*/ (140 missing values generated) . gen workcat=1 if occ==5 | occ==6 | occ==7 /'lew wage worker category*/ (1680 missing values generated) . re place workcat=2 i f occ==1 | occ==4 / ‘ medium wage w orker c a te g o ry */ (840 real changes made) . replace workcat=3 if occ==2 | occ==3 /’high wage worker category*/ (840 real changes made) . /'ORIGINALLY, ATRDOCC# VAR WAS EQUAL TO THE PRODUCT OF L1ATR AND DOCC#. NOW THAT WE HAVE TWICE LAGGED AND CONTEMPORANEOUS VARIABLES IN THE ANALYSIS, THE VAR NAME IS CHANGED REFLECT AS SUCH*/ . rename a trd o c c i H a trd o c d . rename a trd o cc2 lia trd o c c 2 . rename a trd o cc3 H a trd o cc3 . rename atrdocc4 !1atrdocc4 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 272 . rename atrdocc5 HatrdoccS . rename atrdocc6 Hatrdocc6 . rename atrdocc7 Hatrdocc7 . /'ORIGINALLY, ATRDSIC# VAR WAS EQUAL TO THE PRODUCT OF L1ATR ANO OSICW. NOW THAT WE HAVE TWICE LAGGED AND CONTEMPORANEOUS VARIABLES IN THE ANALYSIS, THE VAR NAME IS CHANGED REFLECT AS SUCH'/ . rename atrd20 H a trd 2 0 . rename atrd21 l1 a trd 2 1 . rename a trd 2 2 I1 a trd 2 2 . rename atrd 2 3 H a trd 2 3 . rename atrd 2 4 I1 a trd 2 4 . rename a trd 2 5 I1 a trd 2 5 . rename atrd 2 6 I1 a trd 2 6 . rename a trd 2 7 I1 a trd 2 7 . rename atrd28 I1 a trd 2 8 . rename a trd 2 9 I1 a trd 2 9 . rename atrd3C H a trd 3 0 . rename a trd 3 2 H a trd 3 2 . rename atrd 3 3 I1 a trd 3 3 . rename atrd 3 4 H a trd 3 4 . rename atrd 3 5 H a trd 3 5 . rename atrd36 H a trd 3 6 . rename atrd 3 7 I1 a trd 3 7 . rename atrd38 I1 a trd 3 8 . rename atrd 3 9 H a t rd39 . gen byte low= workcat==1 . gen byte med= workcat==2 . gen byte high= workcat==3 . gen Hatrlow=llatr*low (140 missing values generated) . gen Hatrmed=l1atr*med (140 missing values generated) . gen I1atrhi=l1atr'high (140 missing values generated) . la b e l d e fin e w orkfm t 1 "Low wage w orker* 2 ‘ Medium Wage Worker" 3 "High Wage Worker* . label define occfmt 1 ‘Technicians* 2 *°rof Specify Occ" 3 *Mgrs & Admin* 4 "Sales* 5 "Admin Supp, Cler* 6 "Service* 7 "Proon* . label define sicfmt 20 "Focd & Kindred" 21 ‘Tobacco M frs‘ 22 "Textil M ill Products* 23 •Apparel & Other Textile Products" 24 "Lumber & Wood Products" 2E “Furniture & Fixtures* 26 "Paper & Allied Products" 27 "Printing & Publisning* 23 "Chemicals & Allied Products" 29 "Petroleum & Coal Products* 30 "Rubber & Misc" 31 "Leather & leather Products* 32 ‘Stone,Clay.Glass & Concrete* 33 "Primary Metal* 34 "Fabricated Metal* 35 "Industrial Machinery & Equip* 36 "Electrical & Electronic* 37 "Transportation Equip* 38 "Instruments 8 Related* 39 "Misc Mfrg Inds" . label define atrtypes 1 "Monopoly, Premerger N otification Failure, Acquisitions, Joint Ventures* 2 "Price fixing, Restraint of Trade, Bid Rigging, Territorial Allocation, Restricting Output" . label values occ occfmt . label values workcat workfmt . label values sic sicfmt Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 273 11S occ ts s e t occ SICYEAR panel variable: occ, 1 to 7 time variable: SICYEAR, 1 to 420 set matsize 700 /•TESTING FOR HETEROSKEDASTICITY*/ xtgls lnwage realgdp avghhi k_lexp frmSyrte atr lla tr 12atr trend, force panel(hetero) Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: heteroskedastic Correlation: no autocorrelation E stim ated co va ria n ce s = 7 Number o f obs = 2650 Estimated autocorrelations = 0 Number of groups = 7 Estimated coefficients = 9 Obs per group: min = 375 avg = 378.5849 max = 380 Wald chi2(8) = 301.74 Log likelihood = -668.3235 Prob > chi2 = 0.0000 lnwage | Coef. S td. E rr. z P > lz| '95% Conf in t e r v a ll re a lg d p j .1004637 .0349535 2.87 C.004 .031956 .1689714 avghhi | .0001358 .0000128 10.64 C.000 •0001108 .0001608 k_ le xp | .0467456 .0038625 12. 10 0.000 .0391754 .0543159 frmSyrte | .0185979 .0036972 5.03 0.000 .0113516 .0258443 a t r | .0216605 .0119491 1 .81 0.070 - .0017594 .0450803 H a t r | .0135712 .0119883 1.13 0.258 -.0099254 .0370679 1 2 a tr | .0094769 .0118579 0.80 0.424 .0137643 .032718 trend | -.024019 .0072957 -3.29 0.001 -.0383183 -.0097197 _ccns | 2.131675 .1483106 14.37 0.000 1.840992 2.422358 . lrtest, saving(O) . xtgls lnwage realgdp avghhi k_lexp frmSyrte atr H atr 12atr trend, fo rc e Cross-secticnal time-series FGLS regression Coefficients: generalized least squares Panels: homoscedastic Correlation: no autocorrelation Estimated covariances 1 Number o f obs = 2650 Estimated autocorrelations 0 Number of groups = 7 Estimated coefficients 9 Obs per ■ group: min = 375 avg = 378.5849 max = 380 Wald chi2(8) = 241 .14 Log lik e lih o o d = -890.6759 Prob > ch i2 = 0.0000 lnwage | Coef. Std. E rr. z P >|z| [95% Conf. In te r v a l) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. realgdp | .0905923 .0428923 2.11 0.035 .006525 .1746596 avghhi | .0001443 .0000156 9.24 0.000 .0001137 .000175 k_lexp | .0529257 .0047367 11.17 0.000 .0436419 .0622095 frm S yrte | .0152734 .0044976 3.40 0.001 .0064584 .0240885 a tr | .0196408 .0146672 1 .34 0.181 -.0091064 .0483879 H a t r | .0114173 .0147157 0.78 0.438 -.017425 .0402595 1 2 a tr | .0115506 .0145564 0.79 0.427 -.0169794 .0400807 trend | -.0232148 .0089494 -2.59 0.009 -.0407554 -.0056742 cons | 2.164893 .1820366 11.89 0.000 1.808108 2.521678 . local df = e(N_g) - 1 . Irtest, d f('d f) Xtgls: likelihood-ratio test c h i2 (6 ) 444.70 Prob > chi2 = 0.0000 . xtgls lnwage realgdp avghhi Hatrhhi k_lexp frmSyrte atr lla tr 12atr trend, force panel(hetero) Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: heteroskedastic Correlation: no autocorrelation Estimated covariances 7 Number o f obs = 2650 Estimated autocorrelations 0 Number o f groups = 7 Estimated coefficients 10 Obs per group: min = 375 avg = 378.5849 max = 380 Wald c h i2 (9 ) = 307.65 Log lik e lih o o d = -666.0182 Prob > c h i2 = 0.0000 lnwage | Coef. Std. Err. z P >|z| (95% Conf. In te rv a l] realgdp | .0995294 .0349017 2.85 0.004 .0311234 .1679354 avghhi | .0001595 .0000168 9.49 0.000 .0001266 .0001925 H a tr h h i | -.0000573 .0000265 -2.16 0.031 -.0001093 -5.24e-06 k_lexp | .0466696 .0038567 12.10 0.000 .0391106 .0542285 frmbyrte | .0207953 .0038272 5.43 0.000 .0132941 .0282966 a t r | .0223496 .0119345 1 .87 0.061 - .0010415 .0457407 H a t r | .0508981 .0210252 2.42 0.015 .0096894 .0921068 12atr | .0094798 .0118393 0.80 0.423 -.0137248 .0326845 tre n d | -.0239785 .0072843 -3.29 0.001 -.0382555 -.0097016 _cons | 2.121423 .1481488 14.32 0.000 1.831057 2.411789 . Irtest, saving(O) . xtgls lnwage realgdp avghhi llatrhhi k_lexp frmSyrte atr H atr 12atr trend, force Cross-sectional tim e-series FGLS regression Coefficients: generalized least squares Panels: homoscedastic Correlation: no autocorrelation with permission of the copyright owner. Further reproduction prohibited without permission. 275 Estimated covariances = 1 Number o f obs = 2650 Estimated autocorrelations = 0 Number o f groups = 7 Estimated coefficients = 10 Obs per group: min = 375 avg = 378.5849 max = 380 Wald c h i2 (9 ) = 244.70 Log lik e lih o o d = -889.0486 Prob > chi2 = 0.0000 lnwage | Coef. S td. E rr. z P>|z| [95% Conf. In te r v a l] realgdp | .0895718 .0428697 2.09 0.037 .0055488 .1735948 avghhi | .0001684 .0000205 8.20 0.000 .0001281 .0002087 Hatrhhi | -.0000587 .0000325 -1 .80 0.071 -.0001224 5.05e-06 k_ le xp | .0527977 .0047343 11.15 0.000 .0435186 .0620768 frm5yrte | .0174613 .0046554 3.75 0.000 .0083368 .0265857 a t r | .020332 .0146632 1 .39 0.166 -.0084073 .0490713 Hatr | .0496514 .0257909 1 .93 0.054 -.0008976 .1002006 12atr | .011568 .0145475 0 .80 0.427 - .0169445 .0400806 trend | -.0231651 .008944 -2.59 0.010 -.040695 -.0056352 _cons | 2.155135 .1820052 11 .84 0.000 1.798411 2.511858 . local df = e(N_g) - 1 . lrtest; df('df') Xtgls: likelihood-ratio test c h i2 (6 ) - 446.06 Prob > chi2 = 0.0 000 . xtgls lnwage realgdp avghhi Hatrhhi k_lexp frm5yrte atr lla tr 12atr Hatrdocd Hatrdocc2 !1atrdocc3 !1atrdocc4 HatrdoccS Hatrdocc6 trend, force panel(hetero) Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: heteroskedastic Correlation: no autocorrelation Estimated covariances = 7 Number o f obs = 2650 Estimated autocorrelations = 0 Number o f groups = 7 Estimated coefficients = 16 Obs per group: min = 375 avg = 378.5849 max = 380 Wald chi2(15) = 1689.47 Log lik e lih o o d = -175.3693 Prob > chi2 = 0.0000 lnwage | Coef. S td. E rr. z P>|Z| [95% Conf. In te r v a l] realgdp | .0959166 .0304614 3.15 0.002 .0362133 .15562 avghhi | .0001656 .0000146 11.31 0.000 .0001369 .0001944 H a tr h h i | -.0000593 .0000231 -2.56 0.010 - .0001046 -.0000139 k_lexp | .0493918 .0033648 14.68 0.000 .0427968 .0559868 frm S y rte | .0197191 .0033299 5.92 0.000 .0131926 .0262456 atr | .022292 .0104186 2.14 0.032 .001872 .0427121 H a t r | -.210195 .0258605 -8.13 0.000 - .2608806 -.1595094 1 2 a tr | .0106932 .0103356 1.03 0.301 -.0095643 .0309507 Hatrdocd | .3102476 .0228085 13.60 0.000 .2655438 .3549514 Hatrdocc2 | .6030962 .028977 20.81 0.000 .5463022 .6598901 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 276 l1atrdocc3 | .646395 .0311282 20.77 0.000 .5853849 .707405 l1atrdocc4 | .4434839 .028443 15.59 0.000 .3877366 .4992312 HatrdoccS | -.002741 .0257306 -0.11 0.915 -.0531719 .04769 Hatrdocc6 | -.1513592 .0339791 -4.45 0.000 -.2179571 -.0847613 tre n d | -.0235417 .006357 -3.70 0.000 -.0360012 -.0110822 _cons | 2.120261 .1293055 16.40 0.000 1.866827 2.373695 . Irtest, saving(O) . xtgls lnwage realgdp avghhi H atrhhi k_lexp frmSyrte atr lla tr 12atr Hatrdocd l1atrdocc2 Hatrdocc3 l1atrdocc4 HatrdoccS l1atrdocc6 trend, force Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: homoscedastic Correlation: no autocorrelation Estimated covariances = 1 Number o f obs = 2650 Estimated autocorrelations = 0 Number o f groups = 7 Estimated coefficients = 16 Obs per group: min = 375 avg = 378.5849 max = 380 wald chi2(l5) 1769.95 >g likelihood = -328.2441 Prob > chi2 = 0.0000 lnwage | Coef. Std. Err. z P>|z| [95% Conf. In te r v a l| realgdp | .0895718 .0346931 2.58 0.010 .0215746 .157569 avghhi | .0001684 .0000166 10.13 0.000 .0001358 .000201 H a tr h h i | -.0000587 .0000263 -2.23 0.026 -.0001103 -7 .1 1e-06 k_lexp | .0527977 .0038313 13.78 0.000 .0452884 .060307 frmSyrte | .0174613 .0037675 4.63 0.000 .0100771 .0248454 a t r | .020332 .0118665 1.71 0.087 -.0029259 .0435898 H a t r | -.214509 .0284505 -7.54 0.000 -.2702709 -.1587471 1 2 a tr | .011568 .0117728 0.98 0.326 -.0115063 .0346423 H a tr d o c d | .3102476 .029533 10.51 0.000 .252364 .3681312 l1atrdocc2 | .6030962 .029533 20.42 0.000 .5452125 .6609798 l1atrdocc3 | .646395 .029533 21.89 0.000 .5885113 .7042786 l1atrdocc4 | .4434839 .029533 15.02 0.000 .3856003 .5013676 HatrdoccS | -.002741 .029533 -0.09 0.926 -.0606246 .0551427 l1atrdocc6 | -.1513592 .029533 -5.13 0.000 -.2092429 -.0934756 trend | -.0231651 .0072381 -3.20 0.001 -.0373515 -.0089787 _cons | 2.155135 .1472911 1463 0.000 1.86645 2.44382 . local df = e(N_g) - 1 . Irtest, df( df') xtgls: likelihood-ratio test Chi2(6) = 305.75 Prob > chi2 = 0.0000 . xtgls lnwage realgdp avghhi H atrhhi k_lexp frmSyrte atr H atr 12atr Hatrdocd Hatrdocc2 Hatrdocc3 Hatrdocc4 HatrdoccS !1atrdocc6 Hatrd21 Hatrd22 Hatrd23 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Ill I1atrd24 Hatrd25 Hatrd26 Hatrd27 I1atrd28 I1atrd29 I1atrd30 I1atrd32 Hatrd33 Hatrd34 Hatrd35 I1atrd36 Hatrd37 llatrd38 Hatrd39 trend, force panel(hetero) Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: heteroskedastic Correlation: no autocorrelation Estimated covariances = 7 Number o f obs = 2650 Estimated autocorrelations = 0 Number o f groups = 7 Estimated coefficients = 34 Obs per group: min = 375 avg = 378.5849 max = 380 Wald c h i2 (3 3 ) = 1834.15 Log lik e lih o o d -129.6348 Prob > chi2 = 0.0000 lnwage | Coef. Std. Err. z P>|Z| [95% Conf. In te r v a l] realgdp | .0970935 .0307577 3.16 0.002 .0368095 .1573774 avghhi | .0001594 .0000148 10.77 0.000 .0001304 .0001884 H a tr h h i | -.0001746 .0000663 -2.63 0.008 -.0003045 -.0000446 k_lexp | .0454153 .0041277 11 .00 0.000 .0373251 .0535055 frm S yrte | .0152329 .0038654 3.94 0.000 .0076569 .0228089 a tr | .0256198 .0108329 2.36 0.018 .0043877 .0468519 l l a t r | -.250752 .0594581 -4.22 0.000 -.3672878 -.1342162 1 2 a tr | .0104555 .0108552 0.96 0.335 -.0108204 .0317313 H a tr d o c d | .3102476 .0223243 13.90 0.000 .2664928 .3540024 l1atrdocc2 | .6030962 .028351 21.27 0.000 .5475293 .6586631 l1atrdocc3 | .646395 .0306271 21 .11 0.000 .586367 .7064229 Hatrdocc4 | .4434839 .0276121 16.06 0.000 .3893652 .4976027 HatrdoccS | -.002741 .0252313 -0.11 0.913 -.0521934 .0467115 Hatrdocc6 | -.1513592 .0334387 -4.53 0.000 -.216898 -.0858205 H a trd 2 1 | .2642273 .1402714 1 .88 0.060 -.0106997 .5391543 H a trd 2 2 | .0208471 .0588673 0.35 0.723 -.0945307 .1362249 I1 a trd 2 3 | .0200744 .0445437 0.45 0.652 -.0672297 .1073785 11atrd24 | .0012112 .0565109 0.02 0.983 -.1095481 .1119705 I1 a trd 2 5 | .0288376 .0644075 0.45 0.654 -.0973988 .155074 I1 a trd 2 6 | .1142955 .0475061 2.41 0.016 .0211854 .2074057 H a trd 2 7 | .0678516 .0568841 1.19 0.233 -.0436392 .1793424 H a trd 2 8 | .1833431 .0349312 5.25 0.000 .1148793 .2518069 I1 a trd 2 9 | .139911 .047676 2.93 0.003 .0464678 .2333543 I1 a trd 3 0 | .052272 .0487829 1.07 0.284 -.0433408 .1478847 I1 a trd 3 2 | .1063641 .0349245 3.05 0.002 .0379134 .1748149 l1 a trd 3 3 | .1581075 .0325422 4.86 0.000 .094326 .221889 I1 a trd 3 4 | .0983778 .0405812 2.42 0.015 .0188402 .1779154 H a trd 3 5 | .2151543 .0353416 6.09 0.000 .1458861 .2844226 H a trd 3 6 | .1582898 .033884 4.67 0.000 .0918785 .2247012 H a trd 3 7 | .2888575 .0761108 3.80 0.000 .1396829 .438032 I1 a trd 3 8 | .1833567 .0411627 4.45 0.000 .1026792 .2640341 H a trd 3 9 | .0465093 .0466992 1.00 0.319 -.0450194 .138038 tre n d | -.0235222 .0064402 -3.65 0.000 -.0361448 -.0108996 _cons | 2.123764 .1302387 16.31 0.000 1.868501 2.379027 Irtest, saving(O) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. . xtgls lnwage realgdp avghhi Hatrhhi k_lexp frmSyrte atr H atr 12atr Hatrdocd l1atrdocc2 l1atrdocc3 l1atrdocc4 HatrdoccS l1atrdocc6 Hatrd21 I1atrd22 Hatrd23 Hatrd24 Hatrd25 Hatrd26 I1atrd27 Hatrd28 I1atrd29 l1atrd30 Hatrd32 Hatrd33 I1atrd34 l1atrd35 I1atrd36 I1atrd37 I1atrd38 Hatrd39 trend, force Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: homoscedastic Correlation: no autocorrelation Estimated covariances = 1 Number of obs = 2650 Estimated autocorrelations - 0 Number of groups = 7 Estimated coefficients = 34 Obs per group: min = 375 avg = 378.5849 max = 380 Wald chi2(33) 1915.54 Log likelihood = -285.3045 Prob > chi2 = 0.0000 lnwage | Coef. Std. E rr. z P>|z| [95% Conf. Interval) realgdp | .0884659 .0349594 2.53 0.011 .0199467 .1569851 avghhi | .0001618 .0000168 9.66 0.000 .000129 .0001947 H a tr h h i | -.0001634 .0000753 -2 .17 0.030 -.0003111 -.0000158 k_lexp | .048097 .0046954 10.24 0.000 .0383942 .0572998 frmSyrte | .0123698 .004359 2.84 0.005 .0038263 .0209133 atr | .0228338 .0123151 . .85 0.064 -.0013033 .046971 H a t r | -.2638514 .0672352 -3 .9 2 0.000 -.3956299 -.1320729 12 a tr | .0105529 .0123411 0.86 0.392 -.0136353 .034741 H a trd o c d | .3102476 .0290583 10.68 0.000 .2532943 .3672009 l1atrdocc2 | .6030962 .0290583 20.75 0.000 .5461429 .6600494 Hatrdocc3 | .646395 .0290583 22.24 0 000 5894417 .7033482 l1atrdocc4 | .4434839 .0290583 15.26 0.000 .3865306 .5004372 HatrdoccS | -.002741 .0290583 -0 .0 9 0.925 -.0596942 .0542123 l1atrdocc6 | -.1513592 .0290583 -5.21 0.000 -.2083125 - .0944059 H a trd 2 l | .1759262 .1594319 1.10 0.270 -.1365546 488407 I1 a trd 2 2 | .0426025 .0669201 0.64 0.524 -.0885586 .1737635 I1atrd23 | -.0242596 .0506377 -0.48 0.632 -.1235077 .0749885 I1atrd24 | .0154133 .0642314 0.24 0.810 -.1104779 . 1413045 Hatrd25 | -.0068489 .0732034 -0 .0 9 0.925 -.150325 .1366272 I1 a trd 2 6 | .1217181 .0539962 2.25 0.024 .0158875 .2275488 I1atrd27 | .042745 .064645 0.6 6 0.508 -.3839568 . 1694469 H a trd 2 8 | .1946008 .0397021 4.9 0 0.000 .1167861 .2724154 H a trd 2 9 | .1406006 .0541995 2.59 0.009 .0343716 .2468296 H a trd 3 0 | .0707836 .0554268 1 .28 0.202 -.0378508 .1794181 H a trd 3 2 | .1030628 .0397013 2.60 0.009 .0252497 .1808759 H a trd 3 3 | .1687482 .0369941 4.56 0.000 .0962411 .2412554 H a trd 3 4 | .1139184 .0461264 2.47 0.014 .0235125 .2043244 H a trd 3 5 | .2163684 .0401736 5.39 0.000 .1376297 .2951071 H a trd 3 6 | .1778123 .0385176 4 .62 0.000 .1023192 .2533054 Hatrd37 | .2814405 .0865238 3.25 0.001 .111857 .4510241 H a trd 3 8 | .2023723 .0467506 4.3 3 0.000 .1107429 .2940017 Hatrd39 | .0390438 .0530763 0.74 0.462 -.0649838 .1430714 tre n d | -.022624 .0073178 -3.09 0.002 -.0369666 -.0082815 _cons | 2.169406 .1480707 14.65 0.000 1.879192 2.459619 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 279 local df = e(N_g) • 1 . Irtest, df('df') Xtgls: likelihood-ratio test ch i2 (6 ) = 311.34 Prob > chi2 ■= 0.0 000 . xtgls lnwage realgdp avghhi Hatrhhi k_lexp frmSyrte atr H at- 12atr I1atrd21 Hatrd22 Hatrd23 I1atrd24 Hatrd25 llatrd26 Hatrd27 llatrd28 l 1atrd29 liatrdoO Hatrd32 I1atrd33 I1atrd34 Hatrd35 I1atrd36 I1atrd37 Hatrd33 11atrd39 trend, force panel(hetero) Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: heteroskedastic Correlation: no autocorrelation Estimated covariances = 7 Number of obs = 2650 Estimated autocorrelations = 0 Number of groups = 7 Estimated coefficients = 28 Obs per group: min = 375 avg = 378.5849 max = 380 Wald cni2(27) 382.71 Log lik e lih o o d = -631.3878 Prob > chi2 = 0.0000 lnwage | Coef. Std. Err. z P> | z | [95% Conf. In te r v a l| realgdp | .101983 .0353787 2.88 0.004 .032642 .171324 avghhi | .0001533 .000017 8.99 0.000 .0001199 .0001867 Hatrhhi | - .0001757 .0000762 -2.30 0.021 -.0003251 -. 0000263 k_lexp | .0426258 .0047489 8.98 0.000 .033318 .0519335 frm S yrte | .0163909 .0044638 3.67 0.000 .007642 .0251397 a tr | .0263313 .0124567 2.11 0.035 .0019166 .050746 H a t r | .0081113 .0652847 0.12 0.901 -.1198444 .1360671 1 2 a tr | .0097983 .0124825 0.78 0.432 -.014667 .0342636 I1 a trd 2 l | .2965616 .1613084 1.84 0.066 -.019597 .6127202 Hatrd22 | .0180672 .0676905 0.27 0.790 -.1146036 .1507381 H a trd 2 3 | .0455044 .0512201 0.89 0.374 -.0548851 .1458939 H a trd 2 4 | -.0008389 .0649853 -0.01 0.990 -.1282077 .12653 I1atrd25 | .0478424 .074068 0.65 0.518 -.0973282 .193013 I1atrd26 | .1190201 .054631 2.18 0.029 .0119452 .2260949 H a trd 2 7 | .0791791 .0654197 1 .21 0.226 -.0490412 . 2073994 Hatrd28 | .1862494 .0401711 4.64 0.000 .1075155 .2649833 11atrd29 | .1488569 .0548296 2.71 0.007 .0413929 .2563209 H a trd 3 0 | .0504659 .0561089 0.90 0.368 -.0595055 .1604373 H a trd 3 2 | .1140689 .0401592 2.84 0.005 .0353584 .1927795 H a trd 3 3 | .1581028 .0374194 4.23 0.000 .0847621 .2314435 I1atrd34 | .0969367 .0466663 2.08 0.038 .0054724 .188401 I1 a trd 3 5 | .2196879 .0406401 5.41 0.000 .1400347 .2993412 I1 a trd 3 6 | .1582615 .0389632 4.06 0.000 .0818949 .234628 H a trd 3 7 | .2969168 .0875185 3.39 0.001 .1253836 .46845 H a trd 3 8 | .1817856 .0473544 3.84 0.000 .0889726 .2745986 I1 a trd 3 9 | .0567036 .0537038 1 .06 0.291 -.0485539 .1619612 tre n d | -.0242398 .0074087 -3.27 0.001 -.0387605 -.0097192 _cons | 2.119784 .149794 14.15 0.000 1.826193 2.413375 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 280 . Irtest, saving(O) . xtgls lnwage realgdp avghhi Hatrhhi k_lexp frmSyrte atr lla tr 12atr I1atrd21 Hatrd22 I1atrd23 llatrd24 Hatrd25 I1atrd26 Hatrd27 I1atrd28 Hatrd29 I1atrd30 11atrd32 I1atrd33 I1atrd34 Hatrd35 I1atrd36 I1atrd37 I1atrd38 llatrd39 trend, force Cross-sectional tim e-series FGLS regression Coefficients: generalized least squares Panels: homoscedastic Correlation: no autocorrelation Estimated covariances = 1 Number o f obs = 2650 Estimated autocorrelations = 0 Number o f groups = 7 Estimated coefficients = 28 Obs per group: min = 375 avg = 378.5849 max = 380 Wald chi2(27) = 306.44 Log lik e lih o o d = -861.0845 Prob > chi2 = 0.0000 lnwage | Coef. S td. E rr. z P>|Z| [95% Conf. In te rv a l] realgdp | .0884659 .0434436 2.04 0.042 .003318 .1736138 avghhi | .0001618 .0000208 7.77 0.000 .000121 .0002026 Hatrhhi | -.0001634 .0000936 -1 .75 0.081 -.0003469 .0000201 k_lexp | .048097 .0058349 8.24 0.000 .0366608 .0595332 frm S yrte | .0123698 .0054169 2.28 0.022 .0017529 .0229867 a tr | .0228338 .0153038 1.49 0.136 -.0071611 .0528288 H a t r | .0003089 .0801382 0.00 0.997 -.1567591 .1573769 1 2 a tr | .0105529 .0153361 0.69 0.491 -.0195054 .0406112 Hatrd2l | .1759262 .1981239 0.89 0.375 -.2123894 .5642419 I1 a trd 2 2 | .0426025 .0831607 0.51 0.608 -.1203896 .2055945 H a trd 2 3 | -.0242596 .0629268 -0.39 0.700 -.1475938 .0990746 I1 a trd 2 4 | .0154133 .0798195 0.19 0.847 -.14103 .1718566 I1 a trd 2 5 | -.0068489 .0909689 -0.08 0.940 -.1851447 .1714469 H a trd 2 6 | .1217181 .0671003 1 .81 0.070 -.0097961 .2532324 H a trd 2 7 | .042745 .0803335 0.53 0.595 -.1147057 .2001957 H a trd 2 8 | .1946008 .0493372 3.94 0.000 .0979015 .2913 I1 a trd 2 9 | .1406006 .0673529 2.09 0.037 .0085913 .2726099 I1 a trd 3 0 | .0707836 .0698781 1.03 0.304 -.0642149 .2057822 Hatrd32 | .1030628 .0493362 2.09 0.037 .0063656 .1997601 I1atrd33 | .1687482 .0459721 3.67 0.000 .0786446 .2588519 I1atrd34 | .1139184 .0573206 1.99 0.047 .0015722 .2262647 I1atrd35 | .2163684 .0499231 4.33 0.000 .1185209 .314216 I1 a trd 3 6 | .1778123 .0478653 3.71 0.000 .083998 .2716266 I1 a trd 3 7 | .2814405 .107522 2.62 0.009 .0707013 .4921797 I1 a trd 3 8 | .2023723 .0580963 3.48 0.000 .0885057 .3162389 I1atrd39 | .0390438 .0659572 0.59 0.554 -.0902298 .1683175 tre n d | -.022624 .0090937 -2.49 0.013 -.0404473 -.0048007 _cons | 2.169406 .1840054 11.79 0.000 1.808762 2.53005 local df = e(N_g) - 1 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 281 . Irtest, d f('d f) Xtgls: likelihood-ratio test c h i2 (6 ) 459.39 Prob > chi2 0.0000 . / ‘ TESTING FOR AUTOCORRELATION*/ . xtgls lnwage realgdp avghhi k_lexp frmSyrte atr H atr 12atr trend, force panel(hetero) corr(O ) Cross-sectional tim e-series FGLS regression Coefficients: generalized least squares Panels: heteroskedastic Correlation: no autocorrelation Estimated covariances = 7 Number o f obs = 2650 Estimated autocorrelations = 0 Number o f groups = 7 Estimated coefficients = 9 Obs per g ro u p : min = 375 avg = 378.5849 max = 380 Wald c h i2 (8 ) = 301.74 Log lik e lih o o d -668.3235 Prob > chi2 = 0.0000 lnwage | Coef. Std. E rr. 7 P>|z| [95% Conf. In te r v a l] realgdp | .1004637 .0349535 2.87 0.004 .031956 .1689714 avghhi | .0001358 .0000128 10.64 0.000 .0001108 .0001608 k_lexp | .0467456 .0038625 12.10 0.000 .0391754 .0543159 frmSyrte | .0185979 .0036972 5.03 0.000 .0113516 .0258443 a tr | .0216605 .0119491 1.81 0.070 -.0017594 .0450803 H a t r | .0135712 .0119883 1.13 0.258 -.0099254 .0370679 1 2 a tr | .0094769 .0118579 0.80 0.424 -.0137643 .032718 trend | -.024019 .0072957 -3 .2 9 0.001 -.0383183 -.0097197 _cons | 2.131675 .1483106 14.37 0.000 1.840992 2.422358 . xtgls lnwage realgdp avghhi k_lexp frmSyrte atr H atr 12atr trend, force panel(hetero) c o rr (a r1 ) Cross-sectional tim e-series FGLS regression Coefficients: generalized least squares Panels: heteroskedastic Correlation: common AR(1) coefficient for a ll panels (0.8193) Estimated covariances = 7 Number of obs = 2650 Estimated autocorrelations = 1 Number o f groups = 7 Estimated coefficients = 9 Obs per group: min = 375 avg = 378.5849 max = 380 Wald c h i2 (8 ) = 139.21 Log likelihood = 1445.796 Prob > chi2 = 0.0000 lnwage | Coef. S td. E rr. z P>|z| [95% Conf. In te r v a l] realgdp | .0225884 .022248 1.02 0.310 -.0210169 .0661936 avghhi | .0001453 .0000188 7.73 0.000 .0001084 .0001821 k_lexp | .0359429 .0044608 8.06 0.000 .0271998 .0446859 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 282 frmSyrte | .0151206 .0042825 3.53 0.000 .006727 .0235142 a tr | -.0030295 .0053024 -0.57 0.568 -.0134219 .0073629 llatr | -.0033635 .0056669 -0.59 0.553 -.0144704 .0077434 1 2 a tr | .0031669 .0052639 0.61 0.545 -.0071302 .0135039 tre n d | -.0052476 .004796 -1.09 0.274 -.0146477 .0041524 _cons | 2.384679 .0941204 25.34 0.000 2.200206 2.569151 . xtgls lnwage realgdp avghhi H atrhhi k_lexp frmSyrte atr H atr 12atr trend, panel(hetero) corr(O ) Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: heteroskedastic Correlation: no a u to c o rre la tio n Estimated covariances 7 Number o f obs = 2650 Estimated autocorrelations 0 Numberof groups = 7 Estimated coefficients 10 Obs per group: min = 375 avg = 378.5849 max = 380 Wald c h i2 (9 ) = 307.65 Log lik e lih o o d = -666.0182 Prob > chi2 = 0.0000 lnwage | Coef. S td . E rr. z P>iz| [95% Conf. In te r v a l] realgdp | .0995294 .0349017 2.85 0.004 .0311234 .1679354 avghhi | .0001595 .0000168 9.49 0.000 .0001266 .0001925 H a tr h h i | -.0000573 .0000265 -2.16 0.031 -.0001093 -5.24e-06 k_lexp | .0466696 .0038567 12.10 0.000 .0391106 . 0542285 frm S yrte | .0207953 .0038272 5.43 0.000 .0132941 .0282966 a tr | .0223496 .0119345 1 .87 0.061 -.0010415 .0457407 H a tr | .0508981 .0210252 2.42 0.015 .0096894 .0921068 12atr | .0094798 .0118393 0.80 0.423 -.0137248 .0326845 trend | -.0239785 .0072843 -3.29 0.001 -.0382555 -.0097016 _cons | 2.121423 .1481488 14.32 0.000 1.831057 2.411789 . xtgls lnwage realgdp avghhi H atrhhi k_lexp frm5yrte atr H a tr 12atr trend, panel(hetero) corr(arl) Cross-sectional time-series FGLS re g re ssio n Coefficients: generalized least squares Panels: heteroskedastic Correlation: common AR(1) coefficient for a ll panels (0.8169) Estimated covariances = 7 Number o f obs = 2650 Estimated autocorrelations = 1 Number o f groups = 7 Estimated coefficients = 10 Obs per group: min = 375 avg = 378.5849 max = 380 Wald c h i2 (9 ) = 139.44 Log lik e lih o o d = 1440.8 Prob > chi2 = 0.0000 lnwage | Coef. Std. Err. z P>|z| [95% Conf. Interval] Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 283 realgdp | .0227887 .0223407 1.02 0.308 -.0209982 .0665756 avghhi | .0001452 .0000195 7.43 0.000 .0001069 .0001835 H a tr h h i | 3.90e-07 .000011 0.04 0.972 -.0000212 .000022 k_lexp | .0360441 .0044684 8.07 0.000 .0272862 .044802 frm S yrte | .015152 .0042984 3.53 0.000 .0067272 .0235767 a t r | -.0029793 .0053237 -0.56 0.576 -.0134135 .007455 H a t r | -.0035621 .0092056 -0.39 0.699 -.0216048 .0144806 1 2 a tr | .0032412 .0052846 0.61 0.540 -.0071164 .0135988 tre n d | -.0052919 .0048162 -1.10 0.272 -.0147315 .0041476 _cons | 2.384176 .0944583 25.24 0.000 2.199041 2.569311 . xtgls lnwage realgdp avghhi Hatrhhi k_lexp frmSyrte atr It atr 12atr Hatrdocd !1atrdocc2 Hatrdocc3 Hatrdocc4 HatrdoccS Hatrdocc6 trend, force panel(hetero) corr(O) Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: heteroskedastic Correlation: no autocorrelation Estimated covariances = 7 Number o f obs = 2650 Estimated autocorrelations = 0 Number o f groups = 7 Estimated coefficients = 16 Obs per group: min = 375 avg = 378.5849 max = 380 Wald chi2(15) = 1689.47 Log lik e lih o o d = -175.3693 Prob > chi2 = 0.0000 lnwage | C oef. S td. E rr. z P »|z| [95% C onf. In te rv a l] realgdp | .0959166 .0304614 3.15 0.002 .0362133 .15562 avghhi | .0001656 .0000146 11 .31 0.000 .0001369 .0001944 H a tr h h i | - .0000593 .0000231 -2.56 0.010 - .0001046 -.0000139 k_lexp | .0493918 .0033648 14.68 0.000 .0427968 .0559868 frm 5 y rte | .0197191 .0033299 5.92 0.000 .0131926 .0262456 a t r | .022292 .0104186 2.14 0.032 .001872 .0427121 H a t r | -.210195 .0258605 CO to 0.000 -.2608806 -.1595094 1 2 a tr | .0106932 .0103356 1.03 0.301 -.0095643 .0309507 H a tr d o c d | .3102476 .0228085 13.60 0.000 .2655438 .3549514 Hatrdocc2 | .6030962 .028977 20.81 0.000 .5463022 .6598901 Hatrdocc3 | .646395 .0311282 20.77 0.000 .5853849 .707405 Hatrdocc4 | .4434839 .028443 15.59 0.000 .3877366 .4992312 HatrdoccS | -.002741 .0257306 -0.11 0.915 -.0531719 .04769 Hatrdocc6 | -.1513592 .0339791 -4.45 0.000 -.2179571 -.0847613 tre n d | -.0235417 .006357 -3.70 0.000 -.0360012 -.0110822 cons | 2.120261 .1293055 16.40 0.000 1.866827 2.373695 . xtgls lnwage realgdp avghhi Hatrhhi k_lexp frmSyrte atr H atr 12atr Hatrdocd Hatrdocc2 Hatrdocc3 Hatrdocc4 HatrdoccS HatrdoccS trend, force panel(hetero) c o r r ( a r l) Cross-sectional time-series FGLS regression Coefficients: generalized least squares Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 284 Panels: heteroskedastic Correlation: common AR(1) coefficient for a ll panels (0.5973) Estimated covariances = 7 Number o f obs = 2650 Estimated autocorrelations = 1 Number o f groups = 7 Estimated coefficients = 16 Obs per group: min = 375 avg = 378.5849 max = 380 Wald c h i2 ( 15) = 342.78 Log lik e lih o o d = 928.7902 Prob > chi2 = 0.0000 lnwage | Coef. Std. Err. z P>|Z| [95% Conf. Interval| re a lg d p | .0466611 .0285887 1 .63 0.103 -.0093717 .102694 avghhi | .0001546 .0000179 8.65 0.000 .0001196 .0001897 H a tr h h i | -7.33e-06 .0000155 -0.47 0.636 -.0000376 .000023 k_lexp | .044937 .0045302 9.92 0.000 .036058 .053816 frm 5 y rte | .0180201 .0043436 4.15 0.000 .0095069 .0265333 a t r | .0033901 .0070494 0.48 0.631 -.0104265 .0172067 H a t r | -.049321 .0168223 -2.93 0.003 -.0822921 -.0163498 1 2 a tr | .0075812 .0069726 1 .09 0.277 -.0060849 .0212472 H a tr d o c d | .0911884 .0196774 4.63 0.000 .0526215 .1297554 Hatrdocc2 | .1511501 .0227968 6.63 0.000 .1064692 .195831 H a trd o c c 3 | .1734414 .021437 8.09 0.000 .1314256 .2154572 !1atrdocc4 | .1029319 .0239691 4.29 0.000 .0559534 .1499105 HatrdoccS | - .01088 .0182184 -0.60 0.550 -.0465874 .0248274 Hatrdocc6 | -.0342119 .0259856 -1 .32 0.188 -.0851428 .0167191 tre n d | -.0109047 .0061039 -1 .79 0.074 -.0228682 .0010588 _cons | 2.290183 .12018 19.06 0.000 2.054634 2.525731 . xtgls lnwage realgdp avghhi H atrhhi k_lexp frmSyrte atr H atr 12atr Hatrdocd l1atrdocc2 l1atrdocc3 Hatrdocc4 HatrdoccS Hatrdocc6 l1atrd2l Hatrd22 Hatrd23 Hatrd24 Hatrd25 Hatrd26 I1atrd27 Hatrd28 Hatrd29 I1atrd30 Hatrd32 Hatrd33 I1atrd34 I1atrd35 I1atrd36 I1atrd37 Hatrd38 I1atrd39 trend, force panel(hetero) corr(O) Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: heteroskedastic Correlation: no autocorrelation Estimated covariances = 7 Number o f obs = 2650 Estimated autocorrelations = 0 Number o f groups = 7 Estimated coefficients = 34 Obs per group: min = 375 avg = 378.5849 max = 380 Wald C hi2(33) = 1834.15 Log lik e lih o o d -129.6348 Prob > chi2 = 0.0000 lnwage | Coef. Std. E rr. z P>|Z| [95% Conf. In te r v a l] re a lg d p | .0970935 .0307577 3.16 0.002 .0368095 .1573774 avghhi | .0001594 .0000148 10.77 0.000 .0001304 .0001884 H a tr h h i | -.0001746 .0000663 -2.63 0.008 -.0003045 -.0000446 k_ lexp | .0454153 .0041277 11.00 0.000 .0373251 .0535055 frm S y rte | .0152329 .0038654 3.94 0.000 .0076569 .0228089 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. a tr | .0256198 .0108329 2.36 0.018 .0043877 .0468519 H a t r | - .250752 .0594581 -4 .2 2 0.000 -.3672878 -.1342162 1 2 a tr | .0104555 .0108552 0.96 0.335 -.0108204 .0317313 H a tr d o c d | .3102476 .0223243 13.90 0.000 .2664928 .3540024 Hatrdocc2 | .6030962 .028351 21.27 0.000 . 5475293 .6586631 Hatrdocc3 | .646395 .0306271 21.11 0.000 .586367 .7064229 Hatrdocc4 | .4434839 .0276121 16.06 0.000 .3893652 .4976027 HatrdoccS | - .002741 .0252313 -0.11 0.913 - .0521934 .0467115 l1atrdocc6 | - .1513592 .0334387 -4.53 0.000 -.216898 -.0858205 lla t r d 2 l | 2642273 .1402714 1.88 0.060 -.0106997 .5391543 I1atrd22 | .0208471 .0588673 0.35 0.723 -.0945307 .1362249 I1atrd23 | .0200744 .0445437 0.45 0.652 -.0672297 .1073785 i1 a trd 2 4 | .0012112 .0565109 0.02 0.983 - .1095481 .1119705 !1atrd25 | .0288376 .0644075 0.45 0 654 -.0973988 .155074 I1 a trd 2 6 | .1142955 .0475061 2.41 0.016 .0211854 .2074057 H a trd 2 7 | .0678516 .0568841 1 . 19 0.233 - .0436392 .1793424 I1 a trd 2 8 | .1833431 .0349312 5.25 0.000 .1148793 .2518069 I1 a trd 2 9 | .139911 .047676 2.93 0.003 .0464678 .2333543 H a trd 3 0 | .052272 .0487829 1.07 0.284 - .0433408 .1478847 Hatrd32 | .1063641 .0349245 3.05 0.002 .0379134 .1748149 I1 a trd 3 3 | .1581075 .0325422 4.86 0.000 .094326 .221889 H a tra 3 4 | .0983778 .0405812 2.42 0.015 .0188402 .1779154 I1 a trd 3 5 | .2151543 .0353416 6.09 0.000 .1458861 .2844226 Hatrd36 | .1582898 .033884 4.57 0.000 .0918785 .2247012 H a trd 3 7 | .2888575 .0761108 3.80 0.000 .1396829 .438032 Hatrd38 | .1833567 .0411627 4.45 0.000 .1026792 .2640341 I1 a trd 3 9 | .0465093 .0466992 1.00 0.319 - .0450194 .138038 tre n d | -.0235222 .0064402 -3.65 0.000 - .0361448 - 0108996 cons | 2.123764 .1302387 16.31 0.000 1.868501 2.379027 . xtgls lnwage realgdp avghhi Hatrhhi k_lexp frmSyrte atr H atr 12atr Hatrdocd Hatrdocc2 Hatrdocc3 Hatrdocc4 HatrdoccS Hatrdocc6 Hatrd21 Hatrd22 Hatrd23 Hatrd24 I1atrd25 Hatrd26 I1atrd27 Hatrd28 I1atrd29 Hatrd30 llatrd32 Hatrd33 Hatrd34 Hatrd35 I1atrd36 Hatrd37 Hatrd38 Hatrd39 trend, force panel(hetero) corr(arl) Cross-sectional tim e-series FGLS regression Coefficients: generalized least squares Panels: heteroskedastic Correlation: common AR(1) coefficient for a ll panels (0.5716) Estim ated covariances = 7 Number o f obs =■ 2650 E stim ated a u to c o rre la tio n s = * Number o f groups = 7 Estimated coefficients = 34 Obs per group: min = 375 avg = 378.5849 max = 380 Wald ch i2 (3 3 ) = 404.72 Log likelihood = 878.9047 Prob > chi2 = 0.0000 lnwage | C oef. S td. E rr. z P>|Z| [95% C onf. In te r v a l] re a lg d p | .0548368 .029596 1.85 0.064 -.0031703 .1128439 avghhi | .0001523 .000018 8.46 0.000 .0001171 .0001876 H a tr h h i | -.0000523 .0000449 -1.16 0.244 - .0001402 .0000357 k_lexp | .0440081 .004633 9.50 0.000 .0349277 .0530886 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 286 frmSyrte | .0163396 .004504 3.63 0.000 .0075119 .0251674 a t r | .0069951 .0074712 0.94 0.349 -.0076482 .0216385 H a t r | -.082584 .0465467 -1.77 0.076 -.1738139 .0086459 12atr | .010335 .0074448 1.39 0.165 -.0042566 .0249266 H a tr d o c d | .1012935 .0199395 5.08 0.000 .0622127 .1403742 Hatrdocc2 | . 1720204 .0232429 7.40 0.000 .1264652 .2175756 Hatrdocc3 | .1952805 .0222143 8.79 0.000 .1517412 .2388198 Hatrdocc4 | .1186374 .0241828 4.91 0.000 .07124 .1660348 HatrdoccS | - .0105144 .0187818 -0.56 0.576 -.0473261 .0262972 l1atrdocc6 | - .0396214 .0266301 -1.49 0.137 -.0918154 .0125726 I1atrd21 | .1920854 .0947652 2.03 0.043 .006349 .3778218 11atrd22 | .0513494 .0438029 1.17 0.241 -.0345027 .1372014 I1atrd23 | - .0167102 .0450244 -0.37 0.711 -.1049565 .071536 H a trd 2 4 | .0406009 .0471736 . 0.86 0.389 -.0518575 .1330594 H a trd 2 5 | .0187687 .0466696 0.40 0.688 -.072702 .1102394 I1atrd26 | .0433813 .0458795 0.95 0.344 -.0465409 .1333035 I1 a trd 2 7 | .0313086 .0447679 0.70 0.484 -.0564348 .1190521 H a trd 2 8 | .08887 .0376608 ’ 2.36 0.018 .015056 .1626839 I1atrd29 | .0595921 .0417578 1 .43 0.154 -.0222517 .1414358 Hatrd30 | .0387865 .0414148 0.94 0.349 -.0423849 .119958 H a trd 3 2 | .0411941 .0356443 1.16 0.248 -.0286674 . 11 10556 H a trd 3 3 | .0702757 .0385298 1.82 0.068 -.0052413 .1457927 Hatrd34 | .0545893 .0400585 1 .36 0.173 -.0239238 .1331025 liatrd35 | .1010909 .0394563 2.56 0.010 .023758 .1784239 I1 a trd 3 6 | .0634606 .0363038 1.75 0.080 -.0076935 .1346146 I1 a trd 3 7 | .0954063 .0556634 1.71 0.087 -.013692 .2045045 H a trd 3 8 | .0946628 .0368594 2.57 0.010 .0224197 .1669059 Hatrd39 | .0310986 .0434069 0.72 0.474 -.0539774 .1161746 tre n d | -.012449 .0063257 -1 .97 0.049 -.0248472 -.0000509 cons | 2.258128 .1243592 18.16 0.000 2.014388 2.501868 . xtgls lnwage realgdp avghhi Hatrhhi k_lexp frmSyrte atr H atr 12atr Hatrd2l I1atrd22 I1atrd23 Hatrd24 I1atrd25 llatrd26 Hatrd27 Hatrd28 I1atrd29 Hatrd30 I1atrd32 I1atrd33 Hatrd34 Hatrd35 I1atrd36 Hatrd37 Hatrd38 llatrd39 trend, force panel(hetero) corr(O) Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: heteroskedastic Correlation: no autocorrelation Estimated covariances = 7 Number o f obs - 2650 Estimated autocorrelations = 0 Number o f groups = 7 Estimated coefficients = 28 Obs per group: min = 375 avg = 378.5849 max = 380 Wald c h i2 (2 7 ) = 382.71 Log lik e lih o o d = -631.3879 Prob > chi2 = 0.0000 lnwage | Coef. S td. E rr. z P>|Z| [95%. Conf. In te r v a l] realgdp | .101983 .0353787 2.88 0.004 .032642 .171324 avghhi | .0001533 .000017 8.99 0.000 .0001199 .0001867 Hatrhhi | -.0001757 .0000762 -2.30 0.021 -.0003251 -.0000263 k_lexp | .0426258 .0047489 8.98 0.000 .033318 .0519335 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 287 frm S yrte | .0163909 .0044638 3.67 0.000 .007642 .0251397 a t r | .0263313 .0124567 2.11 0.035 .0019166 .050746 H a tr | .0081113 .0652847 0.12 0.901 -.1198444 .1360671 1 2 a tr | .0097983 .0124825 0.78 0.432 -.014667 .0342636 H a trd 2 1 | .2965616 .1613084 1 .84 C.C66 -.019597 .6127202 Hatrd22 | .0180672 .0676905 0.27 0.790 -.1146036 .1507381 Hatrd23 | .0455044 .0512201 0.89 0.374 -.0548851 .1458939 H a trd 2 4 | -.0008389 .0649853 -0.01 0.990 -.1282077 .12653 H a trd 2 5 | .0478424 .074068 0.65 0.518 -.0973282 .193013 Hatrd26 | .1190201 .054631 2.18 0.029 .0119452 .2260949 H a trd 2 7 | .0791791 .0654197 1 .21 0.226 - .0490412 .2073994 Hatrd28 | .1862494 .0401711 4.64 0.000 .1075155 .2649833 Hatrd29 i .1488569 .0548296 2.71 C.007 .0413929 .2563209 Hatrd30 | .0504659 .0561089 0.90 0.368 -.0595055 .1604373 Hatrd32 | .1140689 .0401592 2.84 0.005 .0353584 .1927795 Hatrd33 | .1581028 .0374194 4.23 0.000 . 0847621 .2314435 lia tr d 3 4 | .0969367 .0466663 2.08 0.038 .0054724 .188401 H a trd 3 5 | .2196879 .0406401 5.41 0.000 .1400347 .2993412 H a trd 3 6 | .1582615 .0389632 4.06 0.000 .0818949 .234628 H a trd 3 7 | .2969168 .0875185 3.39 0.001 .1253836 .46845 H a trd 3 8 I .1817856 .0473544 3.84 0.000 .0889726 .2745986 H a trd 3 9 | .0567036 .0537038 1.06 0.291 -.0485539 . 1619612 tre n d | -.0242398 .0074087 -3.27 0.001 -.0387605 - .0097192 cons I 2.119784 .149794 14.15 0.000 1.826193 2.413375 . xtgls lnwage realgdp avghhi Hatrhhi k_lexp frmSyrte atr H atr 12atr Hatrd2l Hatrd22 Hatrd23 Hatrd24 Hatrd25 Hatrd26 I1atrd27 Hatrd28 I1atrd29 I1atrd30 l1atrd32 Hatrd33 Hatrd34 Hatrd35 Hatrd36 I1atrd37 I1atrd38 Hatrd39 trend, force panel(hetero) c o r r ( a r l) Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: heteroskedastic Correlation: common AR(1) coefficient for a ll panels (0.8022) Estimated covariances 7 Number of obs = 2650 Estimated autocorrelations 1 Number o f groups = 7 Estimated coefficients 28 Obs per group: min = 375 avg = 378.5849 max - 380 Wald ch i2 (2 7 ) = 158.14 Log lik e lih o o d = 1417.795 Prob >chi2 = 0.0000 lnwage | Coef. Std. Err. z P>|z| (95% Conf In te r v a l] realgdp | .0261237 .0233432 1.12 0.263 -.0196282 .0718756 avghhi | .0001455 .0000204 7.12 0.000 .0001055 .0001856 H a tr h h i | -.0000336 .000032 -1.05 0.294 -.0000964 .0000292 k_lexp | .0358559 .004575 7.84 0.00C .026889 .0448229 frm S yrte | .0139885 .004474 3.13 0.002 .0052197 .0227574 a t r | -.0026292 .0056341 -0.47 0.641 -.0136718 .0084134 Hatr | .0208719 .0384647 0.54 0.587 -.0545175 .0962613 1 2 a tr | .0039592 .0056138 0.71 0.481 -.0070436 .014962 H a trd 2 1 | .1066551 .0695092 1.53 0.125 -.0295805 .2428906 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 288 I1 a trd 2 2 | .0085433 .0367006 0.23 0.816 -.0633885 .0804751 I1 a trd 2 3 | -.0426935 .0394542 -1 .08 0.279 -.1200223 .0346353 lla tr d 2 4 | .0261922 .0398102 0.66 0.511 -.0518344 .1042187 H a trd 2 5 | -.0260554 .0385327 -0 .6 8 0.499 -.1015782 .0494674 I1atrd26 | -.0167123 .0392458 -0.43 0.670 -.0936327 .0602082 H a trd 2 7 | -.0250392 .0375682 -0.67 0.505 -.0986715 .048593 I1atrd28 | .016608 .0345739 0.48 0.631 -.0511557 .0843716 Hatrd29 | -.0156473 .0361739 -0.43 0.665 -.0865469 .0552523 H a trd 3 0 | -.0214523 .0357377 -0 .6 0 0.548 -.0914969 .0485923 Hatrd32 | -.013798 .0323207 -0 .43 0.669 -.0771454 .0495495 I1atrd33 | -.0123521 .0359138 -0.34 0.731 -.0827418 .0580376 I1 a trd 3 4 | -.017558 .0359479 -0.49 0.625 -.0880145 .0528985 H a trd 3 5 | .0160804 .0364209 0.44 0.659 -.0553033 .087464 11atrd36 | -.0048368 .0330446 -0.15 0.884 -.069603 .0599294 I1 a trd 3 7 | .0239149 .0440444 0.54 0.587 -.0624105 .1102402 lla tr d 3 8 | .0255571 .0327282 0.78 0.435 -.038589 .0897031 Hatrd39 | -.002409 .0375551 -0.06 0.949 -.0760156 .0711976 tre n d | -.0057124 .0050473 -1 .1 3 0.258 -.0156049 .0041801 _cons | 2.369515 .0984292 24.07 0.000 2.176597 2.562432 . log close Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 289 APPENDIX G: EMPLOYMENT LEVELS AND NATURAL LOG OF WAGES ANALYSES BY WAGE CATEGORY Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 290 . / ‘ ANALYSIS ON TWICE LAGGED ATR VARIABLE WITH AVG WAGE CATEGORIES INSTEAD OF OCC CATEGS AND LAGGED ATRHHI*/ . iis workcat . ts s e t workcat SICYEAROCC panel variable: workcat, 1 to 3 time variable: SICYEAROCC, 1 to 2940, but with gaps . xtgls employmt realgdp avghhi k_lexp frm5yrte atr H atr 12atr trend,force corr(arl) panel(hetero) Cross-sectional time-senes FGLS regression Coefficients: generalized least squares Panels: heteroskedastic Correlation: common AR(1) coefficient for a ll panels (0.4250) Estimated covariances = 3 Number of obs = 2660 Estimated autocorrelations = 1 Number of groups = 3 Estimated coefficients = 9 Obs per group: min = 760 avg = 922.8571 max = 1140 Wald c h i2 (8 ) = 96.25 Log likelihood = -16585.1 Prob > chi2 = 0.0000 employmt | Coef. S td. E rr. z P>|Z| [95% C ont. In te r v a l| realgdp | -.3499034 1.563474 -0.22 0.823 -3.414256 2.714449 avghhi | .0229176 .0065769 3.48 0.000 .010027 .0358081 k_lexp | -.2994828 .19427 -1.54 0.123 - .6802449 .0812794 frmSyrte | 12.0957 1.838151 6.58 0.000 8.492989 15.69841 atr | 13.26464 4.387938 3.02 0.003 4.664441 21.86484 H a tr | 13.18775 4.253448 3.10 0.002 4.851144 21.52435 12atr | 9.022256 4.328634 2.08 0.037 .5382885 17.50622 trend | .3020248 3.291147 0.09 0.927 -6.148504 6.752554 _cons | 53.90702 66.08014 0.82 0.415 -75.60766 183.4217 . xtgls employmt realgdp avghhi Hatrhhi k_lexp frmSyrte atr H atr 12atr trend,force corr(arl) panel(hetero) Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: heteroskedastic Correlation: common AR(1) coefficient for a ll panels (0.4230) Estimated covariances = 3 Number of obs = 2660 Estimated autocorrelations = 1 Number of groups = 3 Estimated coefficients = 10 Obs per group: min = 760 avg = 922.8571 max = 1140 Wald c h i2 (9 ) = 98.79 Log likelihood = -16587.26 Prob > chi2 = 0.0000 employmt | Coef. Std. Err. z P>|z| [95%Conf. Interval) + - ...... realgdp | -.309325 1.564622 -0.20 0.843 -3.375928 2.757278 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 291 avghhi | .0172728 .0077899 2.22 0.027 .002005 .0325407 Hatrhhi | .0138159 .010099 1.37 0.171 -.0059778 .0336096 k_lexp | -.2971659 .1941918 -1 .53 0.126 -.6777748 .083443 frmSyrte | 11.62666 1 .872823 6.21 0.000 7.955998 15.29733 a tr | 13.19664 4.396865 3.00 0.003 4.578948 21.81434 H a t r | 4.217191 7.853995 0.54 0.591 -11.17636 19.61074 12atr | 9.130263 4.336719 2.11 0.035 .6304493 17.63008 tre n d | .2349955 3.293167 0.07 0.943 -6.219494 6.689485 cons | 55.77672 66.14978 0.84 0.399 -73.87447 185.4279 . xtgls employmt realgdp avghhi H atrhhi k_lexp frmSyrte atr H atr 12atr liatrmed lla trh i trend, force corr(arl) panel(hetero) Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: heteroskedastic Correlation: common AR(1) coefficient for a ll panels (0.2780) Estimated covariances = 3 Number o f obs = 2660 Estimated autocorrelations = 1 Number o f groups = 3 Estimated coefficients = 12 Obs per group: min = 760 avg = 922.8571 max = 1140 Wald c h i2 ( 11) 316.92 Log lik e lih o o d = -16469.38 Prob > chi2 = 0.0000 employmt | Coef. Std. Err. z P>|z| [95% Conf. Interval) realgdp | .4981644 1.313379 0.38 0.704 -2.07601 3.072339 avghhi | .0148182 .006344 2.34 0.020 .0023841 .0272523 H a tr h h i | .0149863 .0091332 1.64 0.101 -.0029145 .032887 K_lexp | - .2854455 .1532273 -1.86 0.062 -.5857656 .0148745 frm 5 yrte | 10.74019 1.487437 7.22 0.000 7.824868 13.65552 a tr | 13.91229 3.995518 3.48 0.000 6.081221 21.74336 H a t r | 197.1761 22.25381 8.86 0.000 153.5595 240.7928 1 2 a tr | 10.43289 3.946928 2.64 0.008 2.697049 18.16872 lia trm e d | -225.1852 21.70466 -10.37 0.000 -267.7255 -182.6448 H a t r h i | -170.6473 21.67393 -7.87 0.000 -213.1274 -128.1672 tre n d | -1.455868 2.752311 -0.53 0.597 -6.850298 3.938561 _cons | 19.10753 55.63958 0.34 0.731 -89.94406 128.1591 . xtgls employmt realgdp avghhi H atrhhi k_lexp frmSyrte atr H atr 12atr liatrmed H atrhi I1atrd21 I1atrd22 l1atrd23 I1atrd24 I1atrd25 Hatrd26 11atrd27 I1atrd28 Hatrd29 Hatrd30 I1atrd32 I1atrd33 11atrd34 I1atrd35 I1atrd36 I1atrd37 Hatrd38 Hatrd39 trend, force c o r r ( a r l) Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: homoscedastic Correlation: common AH(1) coefficient for a ll panels (0.2204) Estim ated covariances = 1 Number o f obs = 2660 Estimated autocorrelations = 1 Number of groups = 3 Estimated coefficients = 30 Obs per group: min = 760 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 292 avg = 922.8571 max = 1140 Wald chi2(29) = 266.06 Log lik e lih o o d = -18353.97 Prob > chi2 = 0 . 0 0 0 0 employmt | Coef. S td. E rr. z P »|z| [95% Conf. In te rv a l) re algdp | 1.323996 3.794937 0.35 0.727 -6.113944 8.761937 avghhi | -.0001525 .0182343 -0.01 0.993 -.035891 .035586 H a tr h h i | -.0405619 .0766961 -0.53 0.597 -.1908835 .1097596 k_lexp | -.7498007 .5071822 -1 .48 0.139 -1.743859 .2442581 frmSyrte | 16.80217 4.732008 3.55 0 . 0 0 0 7.52761 26.07674 atr | 14.19406 12.35859 1.15 0.251 -10.02832 38.41645 H a t r | 260.5682 67.17487 3.88 0 . 0 0 0 128.9079 392.2285 1 2 a tr | 11.66186 12.34359 0.94 0.345 -12.53112 35.85485 lia trm e d | -230.8447 20.27887 -11.38 0 . 0 0 0 -270.5906 -191.0989 H a t r h i | -173.272 20.27887 -8 .5 4 0 . 0 0 0 •213.0178 •133.5261 H a tr d 2 l | -60.87497 161.2078 -0.38 0.706 -376.8364 255.0865 I1 a trd 2 2 | -108.1506 67.98732 -1 .59 0.112 -241.4033 25.10212 l1 a trd 2 3 | -108.6363 54.87754 -1 .98 0.048 -216.1943 -1.078254 llatrd24 | -152.6427 67.20985 -2 .27 0.023 -284.3716 -20.91384 11 itrd25 I -171.7494 74.1012 -2.32 0.020 -316.9851 -26.51377 11atrd26 | -119.2788 58.37931 -2.04 0.041 -233.7002 -4.857506 O CD o I1 a trd 2 7 | -60.08373 66.73384 0.368 -190.8796 70.71219 H a trd 2 8 i -79.96322 43.48439 -1 .84 0.066 -165.1911 5.264605 11atrd29 | -163.1079 57.01861 -2.86 0.004 -274.8623 -51.35344 I1atrd30 | -129.8642 58.03211 -2 .24 0.025 -243.6051 -16.12336 I1 a trd 3 2 | -138.1308 43.06395 -3.21 0.001 -222.5345 -53.72697 I1 a trd 3 3 i -112.9756 41.07587 -2 .7 5 0.006 -193.4828 -32.46835 I1 a trd 3 4 | -51.27033 49.42343 -1 .04 0.300 -148.1385 45.59781 I1atrd35 | 29.0303 44.13634 0.66 0.511 -57.47535 115.5359 Hatrd36 | 7.651247 42.19039 0.18 0.856 •75.04039 90.34289 I1 a trd 3 7 | 21.83448 88.43629 0.25 0.805 -151.4975 195.1664 liatrd38 | -140.9959 49.24392 -2.86 0.004 -237.5122 -44.47959 Hatrd39 | -192.0297 56.73746 -3.38 0.001 -303.2331 -80.82632 trend | -3.686537 7.961484 -0 .4 6 0.643 -19.29076 11.91768 _cons | 71.21065 160.5611 0.44 0.657 -243.4334 385.3047 . /'ANALYSIS ON TWICE LAGGED ATR VARIABLE WITH AVG WAGE CATEGORIES INSTEAD OF OCC CATEGS AND LAGGED ATRHHI*/ . iis workcat . ts s e t w orkcat SICYEAROCC panel variable: workcat, 1 to 3 time variable: SICYEAROCC, 1 to 2940, but with gaps . xtgls olnwage drealgdp davghhi dk_lexp frmSyrte atr H atr 12atr Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: homoscedastic Correlation: no autocorrelation E stim ated covariances = i Number o f obs = 2643 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 293 Estimated autocorrelations 0 Number o f groups = 3 Estimated coefficients 8 Obs per group: min = 750 avg = 917.0598 max = 1133 Wald c h i2 (7 ) 15.06 Log likelihood = -10903.37 Prob > chi2 = 0.0352 dlnwage | Coef. Std. Err. z P>|z| [95% Conf. Interval] drealgdp | .399531 .2436502 1 .64 0.101 -.0780147 .8770767 davghhi | .000977 .0008034 1 .22 0.224 -.0005977 .0025517 dk_lexp | -2.296497 .8321691 -2.76 0.006 -3.927518 -.6654756 frm S yrte | .0728794 .1919606 0.38 0.704 -.3033565 .4491154 a tr | -.8693626 .6461013 -1.35 0.178 -2.135698 .3969728 l l a t r ] -.5457432 .6472736 -0.84 0.399 -1.814376 .7228898 1 2 a tr | .498571 .6389365 0.78 0.435 -.7537216 1.750864 _cons | -.420862 .6989572 -0.60 0.547 -1.790793 .9490689 . xtgis dlnwage drealgdp davghhi H atrhhi dk_Jex? frmSyrte atr lla tr 12atr Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: homoscedastic Correlation: no autocorrelation Estimated covariances - 1 Number o f obs = 2643 Estimated autocorrelations = 0 Number o f groups = 3 Estimated coefficients = 9 Obs per group: min = 750 avg = 917.0598 max = 1133 Wald c iu 2 (8 ) = 15.07 Log likelinood = -10903.37 Prob > chi2 = 0.0578 dlnwage | Coef. Std. E rr. z P>|Z| [95% Conf. In te r v a l] d re algdp | .4007645 •2440C72 1 .64 0.101 - .0774808 .8790098 davghhi | .0009543 .0008394 1.14 0.256 -.0006909 .0025996 H a tr h h i | .000107 .0011448 0.09 0.926 -.0021368 .0023507 dk_lexp | -2.294848 .8323549 -2.76 C.006 -3.926233 -.6634621 frm 5 y rte | .0723072 .192058 0.38 0.707 -.3041194 .4487339 a t r | -.8707915 .6462812 -1.35 0.178 -2.137479 .3958964 H a t r | -.6156929 .9896226 -0.62 0.534 -2.555318 1.323932 1 2 a tr | .4964524 .6393376 0.78 0.437 -.7566262 1.749531 _cons | -.4215146 .6989909 -0.60 0.546 -1.791512 .9484824 . xtgls dlnwage drealgdp davghhi H atrhhi dk_lexp frmSyrte atr H atr 12atr liatrmed H a t r h i Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: homoscedastic Correlation: no autocorrelation Estimated covariances = 1 Number o f obs = 2643 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 294 Estimated autocorrelations = 0 Number o f groups = 3 Estimated coefficients = 11 Obs per group: min = 750 avg = 917.0598 max = 1133 Wald c h i2 ( 10) = 15.40 Log lik e lih o o d = -10903.2 Prob > c h i2 = 0.1182 dlnwage | Coef. S td. E rr. z P >|zi [95% Conf. In te r v a l! drealgdp | .4007645 .2439921 1 .64 C. 100 -.0774512 .8789802 davghhi | .0009543 .0008394 1 .14 0.256 -.0006908 .0025994 Hatrhhi | .000107 .0011447 0.09 0.926 -.0021366 .0023506 dk_lexp ! -2.294848 .8323035 -2.76 0.006 -3.926132 -.6635628 frm S yrte | .0723072 .1920461 0 3 8 0.707 -.3040962 .4487107 a tr | .8707915 .6462413 -1 .35 0.173 -2.137401 .3958182 Hatr | -.8103714 1.107957 0.73 0.465 2.981927 1.361184 12 a tr | .4964524 .6392981 0.78 0.437 -.7565488 1.749454 lia trm e d [ .1026506 1.042339 0.10 0.922 -1.940296 2.145597 H a t r h i | .5787243 1.042339 0.56 0.579 -1.464222 2.621671 _cons | -.4215146 .6989478 -0.60 0.546 -1.791427 . 9483978 . xtgls dlnwage drealgdp davghhi H atrhhi dk_lexp frmSyrte atr Hatr- 12atr liatrmed H atrhi Hatrd21 Hatrd22 I1atrd23 llatrd24 Hatrd25 Hatrd26 Hatrd27 lUtrd28 Hatrd29 I1atrd30 Hatrd32 I1atrd33 11atrd34 I1atrd35 I1atrd36 Hatrd37 Hatrd38 Hatrd39 Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: homoscedastic Correlation: no autocorrelation Estimated covariances = 1 Number o f obs = 2643 Estimated autocorrelations = 0 Number o f groups = 3 Estimated coefficients = 29 Obs per group: min = 750 avg = 917.0598 max = 1133 Wald c h i2 (2 8 ) 23.95 Log lik e lih o o d = -10898.96 Prob > chi2 = 0.6844 dlnwage | Coef. S td. E rr. z P>|z| [95% Conf In te rv a l) drealgdp j .4458835 .2472077 1 .80 0.071 -.0386346 .9304017 davghhi | .0012441 .0008641 1 .44 0.150 -.0004496 .0029378 H a tr h h i | -.0008856 .0040588 -0.22 0.827 -.0088408 .0070695 dk_lexp | -2.096365 .8467492 -2.48 0.013 -3.755963 -.4367674 frm S yrte | .0574468 .222838 0.26 0.797 -.3793078 .4942013 a tr | -.883736 .6765093 -1.31 0.191 -2.20967 .4421979 H a t r | .3174986 3.508978 0.09 0.928 -6.559971 7.194969 1 2 a tr | .5285678 .6721172 0.79 0.432 -.7887577 1.845893 lia trm e d | .1026506 1.040667 0.10 0.921 -1.93702 2.142321 H a t r h i | .5787243 1.040667 0.56 0.576 -1.460946 2.618395 I1atrd21 | 8.060634 8.826807 0.91 0.361 -9.23959 25.36086 I1 a trd 2 2 | .8172924 3.684495 0.22 0.824 -6.404184 8.038769 I1 a trd 2 3 | -.4114367 2.771025 -0.15 0.882 -5.842547 5.019673 I1 a trd 2 4 | .6958498 3.53194 0.20 0.844 -6.226625 7.618325 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 295 l1 a trd 2 5 | -5.728247 4.018549 -1.43 d ID -13.60446 2.147965 Hatrd26 | -2.284511 2.928798 0 7P 0.43= -8.024849 3.455828 11atrd27 | -1.031324 3.528789 -0.29 0.770 -7.947624 5.884976 X latrd28 | -.1469463 2.170486 -0.07 0.946 -4.40102 4.107127 H a trd 2 9 | -2.743606 2.725334 -1.01 0.314 -8.C85162 2.59795 I1 a trd 3 0 | -.0316161 3.02528 -0.01 0.992 -5.961055 5.897823 I1 a trd 3 2 | -1.084338 2.203482 -0.49 0.623 -5.403084 3.234408 I1 a trd 3 3 | -.7344953 2.051652 -0.36 0.720 -4.755659 3.286668 Hatrd34 | -.5303141 2.528002 -0.21 0.834 -5.485107 4.424479 H a trd 3 5 | -.9594432 2.212883 -0.43 0.665 -5.296615 3.377728 Hatrd36 | .6154846 2.137641 0.29 0.773 -3.574216 4.805185 Hatrd37 | -.4679754 4.779421 -0.10 0.922 -9.835468 3.899517 I1atrd38 | 1.547217 2.563314 0.60 0.546 -3.476786 6.57122 H a trd 3 9 | -.7663889 2.894118 -0.26 0.791 -6.438756 4.905978 _cons | -.5293822 .7181862 -0.74 0.461 -1.937001 .8782369 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 296 WORKS CITED Armentano, Domonick T. Antitrust and Monopoly: Anatomy of a Policy’ Failure. New York: John Wiley & Sons, 1982. Asch, Peter. “The Determinants and Effects of Antitrust Activity.” Journal of Law and Economics, 17 (1975): 575-81. ______. Economic Theory and the Antitrust Dilemma. New York: John Wiley & Sons, 1970. Askildsen, Jan Erik. “Regulating LVtonopolies with Worker Participation." Economic Analysis 1, no. 3 (1998): 203-20. Bartel, Ann and Nachum Sichemian “Technological Change and Wages: An Interindustry Analysis.” Hie Journal o f Political Economy 107, no. 2 (April 1999): 285-325. Baumol, William J., and Janusz Ordover. “Use of .Antitrust to Subvert Competition." Journal of Law and Economics, 28 (1985): 247-65. Bittlingmayer, George. “The Real and Monetary Consequences of Antitrust.” Kansas: University of Kansas, 1990. Photocopied. ______. “Government and The Stock Market: The Effects of Antritrust.” Center for the Study of the Economy and the State at The University of Chicago. Working Paper No. 89 (June 1993). ______. “The Output and Stock Price Effects of Loose Antritrust: Experience Under the NRA." Center for the Study of the Economy and the State at The University of Chicago. Working Paper No. 87 (May 1993). ______. “The Stock Market and Early Antitrust Enforcement.” Journal o f Law & Economics 36, no. 1 (April 1993): 1-32. Bizjak, John M and Jeffrey Coles. “The Effect of Private Antitrust Litigation on the Stock-Market Valuation of the Fiim.” American Economic Review 85, no.3 (June 1995): 436-61. Block, Michael and Jonathan Feinstein. “The Spillover Effect of Antitrust Enforcement.” The Review of Economics and Statistics 68, no. 1 (February' 1986): 122-131. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 297 Block, Michael, Frederick Nold and Joseph Sidak. “The Deterrent Effect of Antitrust Enforcement.” Journal o f Political Economy, 89 (1981): 429-45. Bradley, Robert Jr. “On the Origins of the Sherman Antitrust Act.” Cato Journal. 9: 737- 42. Brazen, Yale. “The Antitrust Tradition: Entrepreneurial Restraint.” Harvard Journal of Law and Public Policy, 9 (1986): 337-56. Choi, Dosoung and George Philippatos. “Financial Consequences of Antitnist Enforcement.” The Review o f Economics and Statistics 65, no. 3 (August 1983): 501-506. DeBow, Michael. “The Social Costs of Populist Antitrust: A Public Choice Perspective.” Harvard Journal of Law and Public Policy, 14 (1991): 205-25. DiLorenzo, Thomas. “The Origins of Antitrust: Rhetoric vs. Reality.” Regulation, 13 (1990): 26-34. ______. “Antitrust and Competition, Historically Considered.” Economic Inquin. 26 (1988): 423-35. Feinberg, Robert. “Antitrust Enforcement and Subsequent Price Behavior.” The Review o f Economics and Statistics 62, no. 4 (November 1980): 609-612. ______. “Strategic and Deterrent Pricing Responses to Antitrust Investigations.” International Journal o f Industrial Organization, 26 (1984): 75-84. Frey, Bruno S. “The Employment Consequences of the Sherman and Clayton Acts, Comment.” Journal o f Institutional and Theoretical Economics, 147 (1991): 53- 57. Fuhr. Joseph P, Jr. “Stee-rike Four! What’s Wrong with the Business of Baseball?” Atlantic Economic Journal 27, no. 2 (June 1999): 221-31. Galavan Supplements Ltd. v. Archer-Daniels-Midland Company, 1997 U.S. Dist LEXIS 18585 [N.D. Cal. Nov. 18, 1997], Gallo, Joseph, Joseph Craycraft and Steven Bush. “Guess Who Came to Dinner: An Empirical Study of Federal Antitrust Enforcement for the Period 1963-1984.” Review of Industrial Organization, 2 (1985): 106-30. Ginsburg, Douglas H. “Goals of Antitrust Revisited - Comment.” Journal of Institutional & Theoretical Economics 147, no. I (March 1991): 24-30. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 298 Gollop, Frank and Mark Robeits. “Firm Interdependence in Oligopolistic Markets.” Journal of Econometrics, 10(1979): 313-331. Grandy, Christopher. “Original Intent and the Sherman Antitrust Act: A Re-examination of the Consumer-Welfare Hypothesis.” Journal o f Economic History 53, no. 2 (June 1993): 359-376. Greene, William H. Econometric Analysis, Third Edition. New Jersey: Prentice Hall, 1997. Hazlett, Thomas W. “Is Antitrust Anticompetitive?” Harvard Journal of Law and Public Policy, 9 (1986): 277-87. Jorde, Thomas M., and David J. Teece. “Innovation, Cooperation and Antitrust: Balancing Competition and Cooperation.” High Technology Law Journal, 4 (1989): 1-113. Keezer, Dexter M. “The Effectiveness of the Federal Antitrust Laws: A Symposium.” The American Economic Review 39, no. 3 (June 1949): 689-724. Kennedy, Peter. A Guide to Econometrics, Fourth Edition. Massachusetts: The MIT Press, 1998. Krueger, Alan and Lawrence Summers. “Efficiency Wages and the Inter-Industry Wage Structure.” Econometrica 56, no. 2 (March 1988): 259-93. Lande, Robert. “The Risk and (Coming) Fall of Efficiency as the Ruler of Antitrust.” Antitrust Bulletin, 33 (1988): 429-65. Lewis, H.G. “Union Relative Wage Effects.” Handbook o f Labor Economics, eds. O. Ashenfelter and R. Layard. New York: Elsevier Science Publishers (1986): 1139-81. Long, William, Richard Schramm and Robert Tollison. “The Economic Determinants of Antitrust Activity.” Journal o f Law and Economics, 16(1973): 351 -64. Machin, Stephen and Sushil Wadhwani. “The Effects of Unions on Organisational Change and Employment.” The Economic Journal 101, no. 407 (July 1991): 835- 54. McCutcheon, Barbara. “Do Meetings in Smoke-Filled Rooms Facilitate Collusion?” Journal of Political Economy 105, no. 2 (April 1997): 330-50. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 299 McChesney, Fred and William F. Shughart II, eds. The Causes and Consequences of Antitrust: The Public Choice Perspective. Illinois: The University of Chicago Press, 1995. McWilliams, Abagail, Thomas Turk and Asghar Zardkooni. “Antitrust Policy and Mergers: The Wealth Effect of Supreme Court Decisions.” Economic Inquiry, 31 (1993): 517-33. Moschel, Wemhard. “Goals of Antitrust Revisited.” Journal of Institutional & Theoretical Economics 147, no. 1 (March 1991): 7-23 Nevvmark, Craig. “Is Antitrust Enforcement Effective?" Journal o f Political Economy, 96 (1988): 1315-28. Peterson, Shorey. “Antitrust and the Classical Model.” The American Economic Review 47, no. 1 (March 1957): 60-78. Posner, Richard. “A Statistical Study of Antitrust Enforcement.” Journal o f Law and Economics, 13 (1970): 365-419. Rcmer, David. Advanced Macroeconomics. New York: McGraw-Hill Companies, Inc., 1996. Ross. Stephen and Robert Lucke. “Why Highly Paid Athletes Deserve More Antitrust Protection than Ordinary Unionized Workers.” The Antitrust Bulletin (Fall 1997): 641-79. Scherer, F.M. and David Ross. Industrial Market Structure and Economic Performance. New Jersey: Houghton Mifflin Company, 1990. Schlieper, Ulrich. “The Employment Consequences of the Sherman and Clayton Acts, Comment.” Journal o f Institutional and Theoretical Economics, 147 (1991): 58- 59. Schmidtchen, Dieter. “Goals of Antitrust Revisited - Comment.” Journal o f Institutional & Theoretical Economics 147, no. 1 (March 1991): 31-7. Shughart, William II and Robert Tollison. “The Employment Consequences of the Sherman and Clayton Acts.” Journal of Institutional and Theoretical Economics. 147 (1991): 38-52. Siegfried, John. “The Determinants of .Antitrust Activity.” Journal o f Law and Economics, 17 (1975): 559-74. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 300 Stigler, George. “The Economic Effects of the Antitrust Laws.” Journal of Law and Economics, 9 (1966): 225-58. Williamson, Oliver E. “Allocative Efficiency and the Limits of Antitrust.” American Economic Review 59, no. 2 (May 1969): 105-118. ______. “Economies as an Antitrust Defense: The Welfare Tradeoffs.” American Economic Review, 58 (1968): 18-36. U.S. Department of Commerce. Bureau of the Census. Bridge Between NAICS and SIC. 1997 Economic Census. Core Business Statistics Series. Washington, D.C.: GPO. 2000. United Slates v. Archer-Daniels-Midland Company and Nabisco Brands Inc., 584 F.Supp. 1134 remanded F.Supp. 1400, 1991 U.S. Dist. LEXIS 19118, 1991-2 Trade Cas. [CCH] P69647 [S.D. Iowa 1991], United States v. Archer-Daniels-Midland Company. 1996 Trade Cas. [CCH] P 45096 [Criminal No. 96CR640], United States v. Merck £ Co. Inc.. 1980-81 Trade Cas. [CCH] P63682 [S.D. Cal. 1980], United States v. The Stanley Works et. al., 1994-1 Trade Cas. [CCH] P70574 [Criminal No. 90-92]. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.