BOOK REVIEWS
Myth and Measurement The New Economics of the
Minimum Wage
David Card and Alan B. Krueger
Princeton, N.J.: PrInceton University Press, 1995, 422 pp.
This volume (beginning with Its title) exudes hubris. The myth, which
the authors promise to debunk, is that Increases In the minimum wage
have negative employment effects. This Is to be replaced by a “new”
economics of the minimum wage. Their book probes aspects of Intellectual
history, propounds some novel theoretical and methodological Ideas,
lectures its readers on technical aspects of econometrics, and prescribes
public policy, despite frequent disclaimers.
In the Introductory chapter. Card and Krueger resurrect the 1946-47
Stlgler-Machlup-Lester debate on the role of marginalism In the analysis
of wages and employment. Richard Lester,something of an apologist for
the trade union movement argued that marginalism was Irrelevant while
Fritz Machiup and George Stigler claimed It was crucial. Card and
Krueger unabashedly try to revive Lester’s arguments. They praise him,
dedicate their book to ~him, and complain that he has been neglected
and too Infrequently cited In the professional literature (while themselves
ignoring Machiup). Their attempt at rehabilitation begIns with a discussion
of theoretical and methodological Issues. The authors dismiss the
marginal productivity theory that underpins the concept of the Individual
firms downward slopIng demand curve for’ labor; they dispense with
aggregating employers’ labor demands to produce a “market demand for
labor which would also be negatively sloped. Instead they assume that
all employers behave independently and function as monopsonists In the
Joan Robinson mode. These few paragraphs are a substitute for a derivation
of a model and a rigorous application of the model to empirical situations.
On the methodological front, Card and Krueger advocate the use of
a technique that Invokes comparisons of the behavior of a sample group
with some control group In “before and after” scenarios. They call these
“natural” experiments and assert their superiority to the more conventional
modes of analysis,econometric studies of time-series and cross sectional
data.
Of course, both their natural experiments and the more
standard analytical methods are simply different ways of attempting to
satisfy the ceteris paribus conditions that are so critical In economic
analysis. Why their method should be Inherently superior Is not obvious.
As Card and Krueger describe them, natural experiments occur when
there are two situations that are essentially Identical except for a change
In a critical policy variable. In such circumstances, they maintain that the
ceteris paribus conditions are satisfied. The concept of natural experiments is useful,
although the question of who Is to decide whether they satisfy the ceteris paribus assumption
is as difficult as the creation of adequate statistical controls In econometric analysis of data.
Much more than a simple assertion Is requIred to establish the superiority of one technique over another.
Chapters 2 and 3 present case studies of the use of this “new” methodology’.
Comparisons are made between employment In fast-food establishments
before and after minimum wage Increases In New Jersey (April
1992), Texas (April 1991), and California (January1988). In each of these
cases, the authors chose what they regarded as an appropriate “control
group.” For New Jersey, they chose fast-food establishments In eastern
Pennsylvania~for Texas (In a—conducted by Lawrence Katz and
Krueger), they chose fast-food firms unaffected by the April 1, 1991
Increase In the federal minimum wage and for California, they chose
fast-food workers In Arizona, Florida, Georgia, New Mexico, and the
Dallas-Fort Worth area In Texas.
The Pennsylvania-New Jersey comparison has drawn the most attention.
In this case the authors chose to construct and use data obtained
from a 24-question telephone survey. Only one question In this survey
pertained to minimum-wage employment, and It Is somewhat Imprecise
with respect to what constitutes full-time and part-time employment.
The authors are to be commended for making their data sets available
(Ina sanitized fashion) through e-mail, but that is not enough. The quality
of their data may be critically flawed so as not to satisfy the ceteris paribus assumption.
The Pennsylvania-New Jersey data were collected In two waves of
telephone Interviews, one before the minimum-wage Increase (In February)
and one after (In November). There are serious questions concerning
the accuracy of that Information. Studies using the official payroll records
of fast-food firms In the relevant geographic areas had significantly different
results and reached opposite conclusions. The major challenge has
come from the Employment Policies Institute (EPI), which Issued a
report titled, “New Evidence on the Minimum Wage: The Crippling
Flaws In the New Jersey Fast Food Study,” In April 1996. EPI made its
data available to David Neumark and William Wascher, They found that
the employment effects of the New Jersey minimum-wage Increase were
negative and quite consistent with the prevailing wisdom.
In the case of the payroll data, employers and the tax-collecting agency
have strong financial incentives to ensure that the total dollar volume of
payrolls Is accurately stated, There Is no such Incentive for accuracy In
the telephone surveys. Thus, there Is a strong presumption of correctness
In favor of the payroll data We would urge the disputants In this issue
to attempt to reconcile the two databases, perhaps by making them
available In sufficient detail to enable some neutral third party, sworn to
confidentiality to exactly match the data, record the aggregate totals,
and then destroy the Individual firm Information, Until the questions
concerning the Card-Krueger data are resolved, their natural experiment
analysis must remain suspect.
In chapter 6, Card and Krueger evaluate the validity of the various
time-series analyses of the employment effects of minimum wages. They
painstakingly reproduce and update several of the numerous studies that
have explored the Issue In that way. The sample of studies examined is
limited and focuses on those with less robust statistical results. In choosing
which ones to Include In their examination, the authors claim to have
used a “rigorous selection process;” They Invoke the 1982 survey article
by Charles Brown, Curtis Gilroy, and Andrew Kohen, In which 26 articles
are examined. They select only the 18 studies that examine the effect on
employment rather than unemployment. Next, they limit their analysis
to studies that focus on “all teens,” ignoring any analysis dealing with
nonwhite, nonwhite male, or nonwhite female teens, groups where the
effect of minimum-wage Increases are much greater.
To those 18 studs, they add three more recent Investigations to bring
the total to 21. They then subtract 6 studies, those not using quarterly
data, to arrive at their final sample of 15 studies. In an effort to discredit
those studies they employ a variety of tests, such as regressing the square
root of the degrees of freedom against the t-ratio associated with the
minimum-wage variable. They conclude that, contrary to expectations,
the t-ratio fills, rather than rises, as the sample size Increases. However,
an Inspection of their regression results reveals that In no case is the
relationship they observe statistically significant. Further, the scatter diagram
they provide suggests quite strongly that without the Inclusion of
study number 15 (by Jacob Klerman (1992]) , the last—Inserted, not
even the negative regression coefficients would have been obtained. There
Is nothing about their results that is statistically robust, despite frequent
claims to the contrary.
Despite the above mentioned shortcomings, the Card-Krueger findings
have been seized upon, both In the United States and abroad, as providing
support for Increasing the minimum wage. Initially, there was a rush to
judgment and a rush to celebrate and acclaim the results. With the
emergence of major questions concerning the accuracy of the basic data
In their natural experiment, there has been some retreat from that
position. We think this retreat Is wise. Certainly, until some of the major
questions are resolved, it might be well to accept the statement made
by Krueger at a Milken Institute conference, where he stated, “I want
to emphasize that my comments should not be Interpreted as support
for the position that Increasing the minimum wage Is sound public policy”
(Krueger 1993:11).
Douglas K. Adie and
Lowell Galloway
Ohio University
object.cato.org |
